Datasets:

pszemraj

/

scientific_lay_summarisation-plos-norm

like 6

Owing to their phylogenetic position, cartilaginous fishes (sharks, rays, skates, and chimaeras) provide a critical reference for our understanding of vertebrate genome evolution. The relatively small genome of the elephant shark, Callorhinchus milii, a chimaera, makes it an attractive model cartilaginous fish genome for whole-genome sequencing and comparative analysis. Here, the authors describe survey sequencing (1. 4× coverage) and comparative analysis of the elephant shark genome, one of the first cartilaginous fish genomes to be sequenced to this depth. Repetitive sequences, represented mainly by a novel family of short interspersed element–like and long interspersed element–like sequences, account for about 28% of the elephant shark genome. Fragments of approximately 15,000 elephant shark genes reveal specific examples of genes that have been lost differentially during the evolution of tetrapod and teleost fish lineages. Interestingly, the degree of conserved synteny and conserved sequences between the human and elephant shark genomes are higher than that between human and teleost fish genomes. Elephant shark contains putative four Hox clusters indicating that, unlike teleost fish genomes, the elephant shark genome has not experienced an additional whole-genome duplication. These findings underscore the importance of the elephant shark as a critical reference vertebrate genome for comparative analysis of the human and other vertebrate genomes. This study also demonstrates that a survey-sequencing approach can be applied productively for comparative analysis of distantly related vertebrate genomes. Our understanding of the human genome has benefited greatly from comparative studies with other vertebrate genomes. Comparison with closely related genomes can identify divergent sequences that may underlie unique phenotypes of human (e. g. , [1,2]), while comparison with distantly related genomes can highlight conserved elements that likely play fundamental roles in vertebrate development and physiology. Among the vertebrate taxa that are most distant from human, teleost fishes that shared a common ancestor with tetrapods about 416 million years (My) ago [3,4] have been valuable for discovering novel genes and conserved gene regulatory regions. Several hundred novel human genes were discovered by comparing the human genome with compact genomes of the pufferfishes, fugu and Tetraodon [5,6]. Genome-wide comparisons of human–fugu and human–zebrafish have been effective in identifying a large number of evolutionarily conserved putative regulatory elements in the human genome [7,8]. However, comparisons of the human and teleost fish genomes are complicated by the presence of many “fish-specific” duplicate gene loci in teleosts. These duplicate loci have been attributed to a “fish-specific” whole-genome duplication event that occurred in the ray-finned fish lineage approximately 350 My ago [9,10]. The extent and copies of “fish-specific” duplicated genes retained following the fish-specific genome duplication vary in different teleost lineages. For example, genome-wide comparison between zebrafish and Tetraodon has shown that different duplicated genes have been retained in these teleosts [11]. Analysis of Hox clusters show that compared to four Hox clusters (HoxA, HoxB, HoxC, and HoxD) with 39 Hox genes in mammals, fugu and zebrafish contain seven Hox clusters with 45 and 49 Hox genes, respectively [12–14]. Fugu has completely lost a copy of the duplicated HoxC cluster, whereas zebrafish has retained both HoxC clusters, and lost a copy of the duplicated HoxD cluster. Adding further complexity, the rates at which specific duplicated genes have mutated vary significantly among different teleost fish lineages [15,16]. Consequently, it is not always straightforward to define orthologous relationships between the genes of teleost fishes and human. The living jawed vertebrates (Gnathostomes) are represented by two lineages: the bony fishes (Osteichthyes) and cartilaginous fishes (Chondrichthyes). The bony fishes are divided into two groups, the lobe-finned fishes represented by lungfishes, coelacanths, and tetrapods, and the ray-finned fishes (e. g. , teleosts; see Figure 1). The cartilaginous fishes possess a body plan and complex physiological systems such as an adaptive immune system, pressurized circulatory system, and central nervous system that are similar to bony fishes, but distinct from the jawless vertebrates (Agnatha). The oldest fossil record of scales from cartilaginous fishes is dated to be about 450 My old [17]. The living cartilaginous fishes are a monophyletic group comprising two lineages: the elasmobranchs represented by sharks, rays, and skates; and the holocephalians, represented by chimaeras [18]. The two lineages of cartilaginous fishes diverged about 374 My ago [19]. By virtue of their phylogenetic position, cartilaginous fishes are an important group for our understanding of the origins of complex developmental and physiological systems of jawed vertebrates. They also serve as a critical outgroup in comparisons of tetrapods and teleost fishes, and help in identifying specialized genomic features (polarizing character states) that have contributed to the divergent evolution of tetrapod and teleost fish genomes. A major impediment to the characterization of genomes from cartilaginous fish is their large size. The dogfish shark (Squalus acanthias), nurse shark (Ginglystoma cirratum), horn shark (Heterodontus francisi), and little skate (Raja erinacea), which are all popular subjects for biological research, have genome sizes that range from 3,500 Mb to 7,000 Mb [20]. In order to identify a model cartilaginous fish genome that could be sequenced economically, we recently surveyed the genome sizes of many cartilaginous fishes, and showed that the genome of the elephant shark, Callorhinchus milii (also known as the elephant fish or ghost shark) is small relative to other cartilaginous fishes [21]. The elephant shark is a chimaerid holocephalian (Order Chimaeriformes; Family Callorhynchidae) [18]. Their natural habitat lies within the continental shelves of southern Australia and New Zealand at depths of 200 to 500 m. Elephant sharks grow to a maximum length of 120 cm. Mature adults migrate into large estuaries and inshore bays for spawning during spring and summer [22]. To further explore the elephant shark genome, and to evaluate its utility as a model for better understanding the human and other vertebrate genomes, we have conducted survey sequencing and analysis of the elephant shark genome. Previously, a survey sequencing approach was used to estimate several global parameters of the dog genome [23]. Here, we demonstrate that the survey-sequencing approach can also be applied productively for comparative analysis of much more distantly related vertebrate genomes. Whole-genome shotgun sequences for the elephant shark were derived mainly from paired end-reads of 0. 85 million fosmid clones. The reads were assembled with the Celera Assembler, yielding 0. 33 million contigs and 0. 24 million singletons. Contigs that were linked by at least two mated end-reads were ordered within larger scaffolds. The combined length of the assembly, including singletons, is 793. 4 Mb. Previously, we estimated the length of euchromatic DNA in the dog genome after survey-sequence coverage (2. 43 Gb after 1. 5× coverage [23]), and this value is very close to that estimated after more complete sequencing (2. 44 Gb after 7. 5× coverage [24]). A similar approach (see Materials and Methods) was used to estimate the length of euchromatic DNA in the elephant shark genome (0. 91 Gb). This value is similar to the length of the chicken genome (0. 96–1. 05 Gb [25]), and is consistent with FACScan data that showed elephant shark and chicken genomes are of similar length [21]. Assuming a haploid genome size of 0. 91 Gb, the sequence data represents 1. 4× coverage, and the assembly output (0. 329 million contigs of mean length 1. 72 kb) is comparable to a simple model assembly [26] with 40 base overlaps (0. 327 million contigs of mean length 1. 87 kb). Assuming 1. 4× coverage, the maximal possible genome coverage is ∼75%. RepeatMasker (version 3. 0. 8; http: //www. repeatmasker. org) uses a library that includes 310 known repeats from Chondrichthyes and Actinopterygii (ray-finned fishes). However, the elephant shark genome contains few homologs of these characterized repeats, and only 6. 0% of the elephant shark sequence was classified by RepeatMasker as repetitive (including 3. 0% that is merely simple or low-complexity sequence). In order to estimate the content of novel repetitive elements, a sample of 100,000 sequence reads was searched against itself using BLASTN. Reads that matched more than 500 other reads were aligned to build consensus sequences for novel repetitive elements. This yielded ten unique consensus sequences, consisting of two short interspersed element (SINE) –like repeats, three long interspersed element (LINE) –like repeats, four satellite-like sequences, and one sequence of unknown identity. When these ten sequences were added to the 310 known fish repeats, RepeatMasker classified 27. 8% of the elephant shark assembly as repetitive. Among the genomes of vertebrates, the content of retrotransposons in elephant shark appears to be much higher than for other nonmammalian species (Figure 2). However, these values are dependent on the level of curation that has been applied to the repeats of each genome, which may not be uniform. The most abundant SINE and LINE-like species each have homology with 7%–8% of the elephant shark genome. The SINE appears to be tRNA-derived, while the LINE encodes a reverse transcriptase with greatest similarity to CR1-like retrotransposons from fish [27]. Like several other vertebrate species [28], the major SINE and LINE species of elephant shark share significant sequence homology at their 3′ ends (41 of 46 identical bases). The content of protein-coding genes was assessed by comparing the translated assembly with known and predicted protein sequences. Nonrepetitive sequences were searched against annotated proteins from the genomes of human, chicken, fugu, zebrafish, Ciona intestinalis, fruit fly, and nematode, and all known proteins from cartilaginous fishes. A total of 60,705 “genic regions” were identified, with a majority representing partial gene sequences. Of the 608,147 sequences in the assembly, 55,298 contain a single genic region each, and 2,663 contain two or more genic regions. The combined length of coding sequence in these genic regions is 20. 6 Mb, representing 2. 6% of the assembled sequence data. This value is likely an underestimate because the homology-based approach used would fail to identify genes that are evolving faster than their homologs in other genomes. For example, when a homology-based approach was used to annotate the fugu genome, it failed to identify homologs for nearly 25% of human genes, particularly the cytokine genes, in the fugu genome [5]. However, many of these genes were subsequently identified in another pufferfish (Tetraodon) based on sequencing of cDNAs [6]. We therefore expect the fraction of coding sequences in the elephant shark genome to be greater than 2. 6%. We assigned putative orthology to genic regions based on their best matching protein sequences in other genomes. However, different fragments of the same gene can display best matches to proteins from different genomes. To avoid this redundancy, we first searched the conceptual protein sequences against the nonredundant human proteome. Of the 60,705 genic regions, 48,400 (80%) had significant similarity (cutoff at 1 × 10−10) to 11,805 human proteins. For the remaining genic regions, the assignment of putative orthology was based on significant matches to known proteins in cartilaginous fishes, chicken, fugu, zebrafish, and C. intestinalis. In total, the genic regions of the elephant shark assembly contain partial or complete sequences for 14,828 genes. This collection defines a minimal set of elephant shark genes that share strong sequence similarity with known vertebrate genes. A description of these genes can be found at http: //esharkgenome. imcb. a-star. edu. sg. Annotation of InterPro domains within the putative protein sequences identified 3,085 unique domains (http: //esharkgenome. imcb. a-star. edu. sg). Most of these domains are also found in annotated proteins of human, mouse, dog, fugu, Tetraodon, and zebrafish. However, 26 domains are absent only from teleost fishes (Table S1), five domains are absent only from mammals (Table S2), and ten domains are absent from both teleost fishes and mammals (Table S3). The elephant shark protein domains absent from teleost fishes or mammals are likely to be encoded by genes that have been lost, or have diverged extensively, in these lineages. Cartilaginous fishes are a useful outgroup for comparison of tetrapod and teleost fish genomes (Figure 1). Comparisons of the gene complements for elephant shark, mammals, and teleost fishes should help to identify ancient genes shared by the three groups of jawed vertebrates and genes that have undergone differential loss or expansion in mammalian and teleost fish lineages. Our analysis (see Materials and Methods) identified 154 human genes that have orthologs in mouse, dog, and the elephant shark, but not in the teleost fish genomes (Table S4). Out of the 154 genes, 85 (highlighted in Table S4) have no homologs in C. intestinalis, fruit fly, or the nematode worm. These are likely to be vertebrate-specific genes that have been lost (or are highly divergent) in the teleost lineage. Among these genes are notable examples, such as ribonuclease L (RNaseL) and 2′-5′oligoadenylate synthetase 1 (2′-5′OAS). The enzymes encoded by these genes are thought to play an important role in the innate immune response to viral infection. 2′-5′OAS is induced by interferon, and activated by double-stranded RNA [29]. Its activity catalyzes the synthesis of oligoadenylates that activate the latent endoribonuclease, RNaseL. The activated RNase degrades both viral and cellular RNA, and is thought to mediate apoptosis. Previously, the genes encoding 2′-5′OAS and RNaseL had been identified only in mammals and chicken. Orthologs of the two enzymes were not identified in the genomes of the three sequenced teleost fishes, or the amphibian, Xenopus tropicalis (http: //www. ensembl. org). This suggests that the relevant genes have been lost independently from at least two vertebrate lineages. This set of genes also includes three members of the amiloride-sensitive epithelial Na+ channel (ENaC) family. This family includes four members, ENaC α, β, γ, and δ subunits, and all members have been cloned from mammals, birds, and amphibians. However, none has been identified in teleost fishes. In contrast to the voltage-gated sodium channels that generate electrical signals in excitable cells, ENaC channels mediate electrogenic transport of Na+ across the apical membranes of polarized epithelial cells. The active transepithelial transport of Na+ is important for maintaining Na+ and K+ levels in the kidney and colon [30]. The mechanism of Na+ uptake in teleost fish cells is currently a subject of controversy. Two models have been proposed. The original model involves amiloride-sensitive electroneutral Na+/H+ exchanger (NHE), with the driving force derived from Na+-K+ ATPase and carbonic anhydrase [31]. A recent model involves ENaC, electrochemically coupled to H+-ATPase [32]. This is not supported by our observation of the loss of ancestral ENaC subunit genes from teleost fish genomes. On the other hand, since NHE has been cloned from a teleost fish, and is shown to express at high levels on the apical membrane of chloride cells [33], the original model seems to be a likely mechanism for Na+ uptake in teleost fishes. A significant number of human genes that have orthologs in the elephant shark but not in teleost fishes are associated with male germ cells and fertilization (Table 1). These include genes that encode zona pellucida (ZP) –binding protein (Sp38) and ZP–sperm-binding protein (ZP-1). These are respectively expressed in the acrosome of sperm [34] and the ZP of oocytes [35] where they mediate the binding of sperm to ZP. In mammals, several sperm initially bind to ZP but only one of them triggers the “acrosomal reaction” that leads to successful fertilization and prevention of other sperm from entering the oocyte. In contrast, sperm of teleost fishes enter the egg through a unique structure called the micropyle, which allows only one sperm to enter and fertilize the oocyte [36]. Micropyle does not exist in the oocytes of mammals and cartilaginous fishes. The conservation of genes essential for the binding of sperm to ZP in mammals and the elephant shark indicates that cartilaginous fishes use the ZP-mediated mode of fertilization similar to mammals. These genes seem to have been either lost or become divergent in teleost fishes following the invention of the micropyle. Our analysis identified 107 teleost fish genes that have orthologs in the elephant shark assembly, but not in the human, mouse, and dog genomes (Table S5). Twenty of these genes have no homologs in invertebrate genomes (C. intestinalis, fruit fly, and nematode worm) and are likely to be vertebrate-specific. The remaining 87 genes (Table S5) are ancient metazoan genes that have been conserved in the elephant shark and teleost fishes, but were lost or are highly divergent in the mammalian lineage. The loss of the ancient vertebrate-specific genes in mammals is likely to be related to some of the divergent phenotypes of mammals compared with cartilaginous fishes and teleost fishes. The vertebrate-specific genes absent from mammals include globinX (GbX), the recently identified fifth member of the vertebrate globin family that includes hemoglobin, myoglobin, neuroglobin, and cytoglobin. GbX has been cloned from teleost fishes and amphibians but has been reported to be absent in amniotes [37]. Although GbX shows expression in several nonneuronal tissues, its function is unknown. The existence of GbX in the elephant shark has confirmed that this is an ancient vertebrate gene that has been lost from the amniote lineage. The genes that are absent from mammals include a large number (80 of 107) that are either hypothetical or predicted novel genes with no known function (Table S5). It is possible that some of these genes may be necessary for aquatic life and should be targeted for functional analysis. After 1–2× sequence coverage of vertebrate genomes using conventional plasmid clones, the assembled sequence data has little long-range continuity that can be used to identify conserved synteny between species. For example, 1. 5× coverage of the dog genome yielded scaffolds with a mean span of only 8. 6 kb [23]. For our survey of the elephant shark genome, >95% of the sequence data was derived from fosmid clones, with inserts of 35–40 kb. Consequently, it was possible to derive much more information on the relative ordering of sequenced genes. For 10,708 fosmid clones, the paired end-reads are located in contigs that have significant homology to unique pairs of human genes. For most pairs (10,655), both genes have defined chromosomal locations. These include 3,059 unique pairs of genes (29%) that are separated by less than 1 Mb on the human genome (median separation, 48 kb). These 3,059 gene pairs could be collapsed further into 1,713 clusters, containing a total of 4,629 genes, in clusters of two to 23 genes per cluster (http: //esharkgenome. imcb. a-star. edu. sg). For comparison, conserved synteny between the elephant shark and zebrafish genomes was analyzed. There was a similar number of fosmid clones (13,773) with end-reads in contigs that have significant homology to unique pairs of zebrafish genes. For 7,916 pairs, both genes have defined chromosomal coordinates. Interestingly, only 848 of these gene pairs (11%) are separated by <1 Mb in the zebrafish genome (median separation, 22 kb), and these are consolidated into 657 clusters, containing 1,489 genes in clusters of two to six genes per cluster (http: //esharkgenome. imcb. a-star. edu. sg). When normalized to the number of unique gene pairs with defined chromosomal coordinates, the level of detectable conserved synteny for human is more than double that seen for zebrafish. These data suggest that elephant shark genome has experienced a lower level of rearrangements compared to teleost fish genomes. This is consistent with the observation that the major histocompatibility complex (MHC) class I and class II genes that are closely linked in mammals and cartilaginous fish such as nurse shark and banded houndshark (Triakis scyllium) are located on different chromosomes in zebrafish, carp, trout, and salmon [38]. Loss of some syntenic blocks in teleost fish could be explained by the differential loss of duplicate genes that arose due to a “fish-specific” whole-genome duplication event in the ray-finned fish lineage [9,10]. For instance, conserved synteny of genes X-Y between the elephant shark and human genomes could be lost in teleost fishes if alternative copies of duplicate genes on paralogous chromosome segments containing duplicate Xa-Ya and Xb-Yb genes are lost resulting in Xa-† and †-Yb genes († represents the lost gene). The higher level of synteny conservation between the elephant shark and human suggests that the elephant shark genome has not undergone whole-genome duplication, and that the identification of orthologous genes in the genomes of elephant shark and nonteleost vertebrates will benefit from the analysis of conserved synteny. Bejerano et al. [39] have identified 481 ultraconserved elements (UCEs) that are longer than 200 bp and perfectly conserved among the human, mouse, and rat genomes. These UCEs overlap transcribed and nontranscribed regions of the genome. To assess the extent of UCEs conserved in the cartilaginous fish genomes, we searched for UCEs in the elephant shark sequences, and fugu and zebrafish genomes (see Material and Methods). Of the 481 UCEs, 57% are found in the elephant shark sequences (83% coverage with an average identity of 86%), whereas 55% and 62% are found (81% coverage, average identity 84%) in the fugu and zebrafish, respectively. Of the 141 UCEs missing from both fugu and zebrafish, 46 (33%) are found in the elephant shark sequences. We predict that the whole genome of the elephant shark will contain ∼75% of the UCEs. Our analysis of the noncoding sequences in the elephant shark has shown that the elephant shark and human genomes contain twice as many conserved noncoding elements as that between human and zebrafish or fugu [40]. Taken together, these results suggest that a higher proportion of human sequences might be conserved in the elephant shark genome than in the teleost fish genomes. Cartilaginous fishes are the phylogenetically oldest group of living organisms known to possess an adaptive immune system based on rearranging antigen receptors. They possess all the four types of T-cell receptors identified in mammals (TcRα, β, γ, and δ); at least three types of Ig isotypes: IgM, IgW (also called IgX-long or IgNARC in some species) and new antigen receptor (IgNAR); the recombination-activating genes (RAG1 and RAG2); and polymorphic MHC genes. The IgNAR isotype, found only in cartilaginous fishes, is unique in that it does not form a heterotetramer (of two light chains and two heavy chains) but instead forms a homodimer of two heavy chains and binds to antigen as a single V domain [41]. A major difference between cartilaginous fishes and other jawed vertebrates is in the organization of Ig genes. In other jawed vertebrates each Ig locus is organized as a single “translocon” containing all the V genes in the 5′ region, followed by all the D, J, and then C region genes in the 3′ end. In contrast, the Ig genes in cartilaginous fishes are present in multiple “clusters, ” with each cluster typically consisting of one V, two D, one J, and one set of C exons [42]. In addition to the above distinct types of Ig and TcR antigen receptor chains, a unique antigen receptor chain comprising two V domains called new antigen receptor–T-cell receptor V domain (NAR-TcRV) and TcRδ V domain (TcRδV) has been recently identified in the nurse shark [43]. The two V domains in the NAR-TcR chain contain a combination of characteristics of both IgNAR and TcR and are generated by separate VDJ gene rearrangements. Such a combination between the Ig and TcR antigen receptor chains were previously thought to be incompatible. BLAST searches of the elephant shark assembly showed that the elephant shark contains homologs for all known cartilaginous fish adaptive immune system genes except IgNAR (see descriptions of genes at http: //esharkgenome. imcb. a-star. edu. sg). Since the elephant shark genome sequence is incomplete, it is unclear whether IgNAR genes are absent in the elephant shark. The discovery of the NAR-TcR genes in the elephant shark assembly is particularly significant since previous attempts to identify this gene in the spotted ratfish (Hydrolagus colliei), a chimaera, by Southern blot analysis using probes from the nurse shark had suggested that this family may be absent in chimaeras [43]. Alignments of peptide sequences of representative elephant shark NAR-TcRVs and associated TcRδVs, together with their homologs from the nurse shark, are shown in Figure 3. Similar to the nurse shark NAR-TcRV, the peptides encoded by the elephant shark NAR-TcR gene contain a typical leader peptide and a cysteine residue in the a-b loop, and lack the canonical tryptophan of the “WYRK” motif. The associated elephant shark TcRδVs lack the leader peptide and share a conserved cysteine residue in the CDR1 similar to their nurse shark homologs (Figure 3). The identification of homologs of NAR-TcR in the elephant shark confirms that this unique doubly rearranging antigen receptor evolved in a common ancestor of elasmobranchs and chimaeras. Hox genes are transcription factors that play a crucial role in the control of pattern formation along the anterior–posterior axis of metazoans. In vertebrates and most nonvertebrates, Hox genes are arranged in clusters and thus are central to the characterization of genome duplications during vertebrate evolution. The amphioxus, a cephalochordate, contains a single cluster of 14 Hox genes [44], whereas coelacanth (a lobe-finned fish) and mammals contain four Hox clusters (HoxA, HoxB, HoxC, and HoxD) that have arisen through two rounds of duplication during the evolution of vertebrates [45,46]. Teleost fishes such as zebrafish and pufferfish contain almost twice the number of Hox clusters found in mammals [5,6, 47], due to the additional “fish-specific” whole-genome duplication in the ray-finned fish lineage [9,10]. Jawless vertebrates (e. g. , the sea lamprey) contain at least three Hox clusters [48,49], one of which seems to be the result of a lineage-specific duplication event [50]. Among the cartilaginous fishes, a complete HoxA cluster and a partial HoxD cluster (HoxD5 to HoxD14) have been sequenced from the horn shark [51,52]. The total number of Hox clusters and Hox genes in cartilaginous fishes is currently unknown. Hox genes typically consist of two exons, and their orthology can be identified reliably based even only on the second exon, which codes for the Hox domain. We identified Hox genes in the elephant shark assembly using a combination of manual annotation, reciprocal BLAST searches, and phylogenetic analysis. A total of 37 partial or complete sequences of Hox genes that were located on different contigs could be identified. These genes belong to putative four Hox clusters (HoxA, HoxB, HoxC, and HoxD), and include a maximum of four members for each of the 14 paralogy groups (Hox1 to Hox14; Figure 4). Thus, elephant shark is likely to contain only four Hox clusters similar to coelacanth and mammals. The presence of four Hox clusters in the elephant shark suggests that, unlike teleost fishes, the elephant shark lineage has not experienced additional whole-genome duplication. Although Hox genes identified in the elephant shark assembly may not include all the Hox genes in the genome, they provide the first glimpse of Hox genes belonging to the four clusters in a cartilaginous fish. The HoxA cluster genes identified in the elephant shark include orthologs of all the HoxA genes identified in the horn shark, while the elephant shark HoxD cluster genes include two genes (HoxD3 and HoxD4) whose orthologs are yet to be identified in the horn shark (Figure 4). The elephant shark HoxB and HoxC cluster genes are the first members of these clusters to be identified in a cartilaginous fish. Comparisons of the elephant shark Hox genes with genes from the completely sequenced Hox clusters from mammals and ray-finned fishes have identified several Hox genes that have been differentially lost in mammals and ray-finned fishes (Figure 5). For example, HoxD5 and HoxD14 genes present in the elephant shark have been lost in both mammalian and teleost lineages, whereas HoxA6, HoxA7, and HoxD8 have been lost only in the teleost lineage. Interestingly, the single HoxA cluster in a nonteleost ray-finned fish, bichir, contains a functional HoxA6 gene and a HoxA7 pseudogene, indicating that HoxA6 was lost in the ray-finned fish lineage after the divergence of the bichir lineage [53]. An ortholog for the elephant shark HoxC1 gene is absent in both mammals and fugu, and is on the way to becoming a pseudogene in zebrafish [54]. However, the presence of this gene in the coelacanth indicates that it has been lost independently in the mammalian lineage after the divergence of the coelacanth and in the lineage leading to teleosts. The presence of HoxB10 in the elephant shark and zebrafish and its absence in mammals and fugu suggest that this gene was lost independently in the teleost lineage leading to fugu after the divergence of the zebrafish lineage and in the mammalian lineage. These comparisons show that duplication of Hox clusters and differential loss of Hox genes is a continuous process in the evolution of vertebrates. The ancestral jawed vertebrate Hox genes that have been differentially lost in different lineages are potential targets for studies aimed at understanding the molecular basis of morphological phenotypic differences between different vertebrate lineages. The extant jawed vertebrates are represented by three major lineages, the cartilaginous fishes, the lobe-finned fishes, and the ray-finned fishes, with the cartilaginous fishes constituting an outgroup to the other two groups. Cartilaginous fishes thus constitute a critical reference for understanding the evolution of jawed vertebrates. The survey sequencing of the elephant shark, the first cartilaginous fish genome to be characterized to this depth, has provided useful information regarding the length, gene complement, and organization of the genome, and highlighted specific examples of vertebrate genes and gene families that have been lost differentially in the mammalian and teleost fish lineages. The 1. 4× coverage elephant shark sequence generated in this study contains partial or complete sequences for about 15,000 unique genes. These sequences can serve as probes for isolating genomic clones and for obtaining complete sequences of gene loci of interest on a priority basis. At 0. 91 Gb, the length of elephant shark genome is similar to that of the chicken (1. 05 Gb), half that of the zebrafish (∼1. 7 Gb), and one-third the length of the human genome (2. 9 Gb). It is about twice the length of the fugu and Tetraodon genomes (∼0. 4 Gb), which are the smallest among vertebrates. The elephant shark genome is the smallest among known cartilaginous fish genomes, and thus is an ideal cartilaginous fish genome for economically sequencing the whole genome and for comparative analysis. A major drawback in comparisons between human and teleost fish genomes is the presence of many duplicate gene loci in teleost fishes due to the additional fish-specific whole-genome duplication event in the ray-finned fish lineage. Analysis of Hox genes in the elephant shark assembly has indicated that the elephant shark genome has not undergone a lineage-specific whole-genome duplication. Interestingly, the human and elephant shark genomes exhibit a higher level of conserved synteny compared with human and zebrafish genomes, even though humans are more closely related to zebrafish than they are to the elephant shark. The disruption of syntenic blocks in the teleosts may be partly related to differential loss of duplicate copies of genes following the fish-specific genome duplication event. The elephant shark also exhibits a higher level of sequence similarity with humans. A higher number of mammalian UCEs, which include both coding and noncoding sequences, were identified in the elephant shark genome compared with the zebrafish and fugu genomes. In a related study, we have shown that twice as many noncoding elements are conserved between human and elephant shark genomes compared with that between human and zebrafish or fugu genomes [40]. The higher level of sequence similarity between the elephant shark and humans could be due to a decelerated evolutionary rate of the elephant shark DNA compared with human and teleost DNA or an accelerated evolutionary rate of teleost sequences compared with the elephant shark and human genomes. Analysis of mitochondrial DNA sequences from 12 lineages of sharks belonging to the elasmobranch lineage has shown that the nucleotide substitution rate in sharks is 7- to 8-fold slower than in mammals [55]. The evolutionary rate of mitochondrial proteins ND2 and Cytb was also found to be slower (about one-fourth) in these sharks compared with mammals [56]. These studies suggest that the evolutionary rate of DNA in cartilaginous fishes is slower than that in mammals. Comparisons of evolutionary rates of protein-coding genes in Tetraodon, fugu, zebrafish, and other teleosts have shown that the fish coding sequences have been evolving at a faster rate than their mammalian orthologs, and that the duplicated pairs of fish genes are evolving at an asymmetric rate [6,15,16,57,58]. Duplicated fish genes also tend to accumulate complementary degenerate mutations in the coding and noncoding sequences, resulting in partitioning of regulatory elements and exons between the two copies [59–62]. Such partitioning could result in a reduced level of sequence conservation between each of the duplicate copies and its ortholog in humans. Thus, the higher level of sequence similarity between the elephant shark and humans compared with that between teleost fish and humans could be the result of both a decelerated evolutionary rate of elephant shark DNA and an accelerated evolutionary rate of teleost fish sequences. The higher degree of conservation of synteny and conserved sequences between the human and elephant shark genomes compared with human and teleost fish genomes, and the absence of evidence for a lineage-specific whole-genome duplication event in the elephant shark lineage, underscore the importance of the elephant shark genome as a model jawed vertebrate genome for comparative analysis of human and other jawed vertebrate genomes. Cartilaginous fishes are the oldest phylogenetic group of jawed vertebrates that possess an adaptive immune system. Analysis of the elephant shark genome sequences has identified all components of the adaptive immune system genes (e. g. , T-cell receptors, immunoglobulins, and RAG and MHC genes) known in tetrapods and teleosts, as well as a unique family of doubly rearranging antigen receptor (NAR-TcR) genes previously reported only in elasmobranch cartilaginous fishes [43]. The presence of this unique family of genes in the elephant shark, a holocephalian, indicates that NAR-TcR existed in a common ancestor of all cartilaginous fishes. Thus, cartilaginous fishes appear to have evolved a distinct type of adaptive immune system after they diverged from their common ancestor with bony fishes. The physiological significance of such a unique adaptive immune system remains to be understood. The number of Hox gene clusters in vertebrates illuminate the history of genome duplications during vertebrate evolution (Figure 5). It has been proposed that the evolution of phenotypic complexity in vertebrates was accomplished through two rounds of whole-genome duplication (the “2R” hypothesis) during the evolution of vertebrates from invertebrates [63]. Although the presence of four mammalian paralogs for many single genes in invertebrates [64] and four Hox clusters in mammals compared with a single Hox cluster in amphioxus is consistent with this hypothesis, the exact timings of the two rounds of genome duplication are unclear. The identification of four putative clusters of Hox genes in the elephant shark in the present study indicates that the two rounds of genome duplication occurred before the divergence of the cartilaginous fish and bony fish lineages (Figure 5). Since the analyses of Hox genes in jawless vertebrates such as the lamprey show that at least one round of genome duplication (“1R”) occurred before the divergence of the jawless and jawed vertebrate lineages, it can be inferred that the second round of duplication (“2R”) occurred after the divergence of the jawless and jawed vertebrate lineages but before the split of cartilaginous fish and bony fish lineages (Figure 5). The presence of almost twice the number of Hox clusters in teleost fishes as in mammals and the elephant shark supports an additional whole-genome duplication event in the ray-finned fish lineage. This more recent fish-specific genome duplication event, referred to as “3R, ” has been hypothesized to be responsible for the rapid speciation and diversity of teleosts [61]. Thus, genome duplication has continued to play an important role in the evolution of vertebrates even after the emergence of bony vertebrates. In this project, we have taken a survey sequencing approach to characterize the elephant shark genome. Previously, a survey sequencing approach was used to estimate several global parameters of the dog genome, such as its length, repeat content, and neutral mutation rate [23]. The coverage (1. 5×) included partial sequence data for dog orthologs of ∼75% of annotated human genes, and revealed that >4% of intergenic sequence is conserved between the dog, human and mouse. More complete sequencing of the dog genome has confirmed the accuracy of these estimates [24]. The survey sequencing approach has now been recognized as an effective and economical way of rapidly characterizing the large genomes of closely related vertebrates for which there is little or no genomic sequences or genetic/physical maps. Here, we have shown that a survey sequencing approach can also be productively used for characterizing most distantly related vertebrate genomes. In contrast to sequencing of paired-ends of short-insert plasmid libraries in conventional whole-genome shotgun sequencing strategy, survey sequencing of the elephant shark genome was based on sequencing of paired-end sequences of fosmid clones. This approach allows accurate assembly of dispersed repeats that are larger than 2–3 kb and provides long-range linkage information that can be used to determine conserved synteny between species. Fosmid clones are also valuable templates for filling gaps in the assembly and for obtaining complete sequences of gene loci of interest. We propose survey sequencing to a depth of 1. 5–2× based on paired-end sequencing of large-insert libraries as an effective and economical approach for characterizing distantly related vertebrate genomes. Genomic DNA was extracted from the testis of an adult elephant shark collected in Hobart, Tasmania. Fosmid libraries (containing 35- to 40-kb inserts) and a plasmid library (3- to 4-kb inserts) were prepared from sheared genomic DNA. End sequencing of clones from each library was conducted using standard procedures, and yielded 1. 54 million reads (93. 7% paired) from the fosmid clones, and 0. 20 million reads (93. 1% paired) from the plasmid clones. The finished sequence data consisted of 1. 73 million reads, with a mean read length of 763 bases. The reads were assembled with Celera Assembler (http: //wgs-assembler. sourceforge. net) [23,65,66]. The assembly output consisted of 0. 327 million contigs (mean length, 1,720 bases; mean content, 4. 3 reads per contig), 0. 245 million singletons, and 0. 037 million mini-scaffolds (paired end-reads that were otherwise unassembled). A small number of contigs (2,113) that were linked by at least two mated end-reads were ordered within scaffolds that spanned a total of 33. 6 Mb. Previously, we estimated the length of euchromatic DNA in the dog genome after survey sequence coverage (2. 43 Gb after 1. 5× coverage [23]), and this value is very close to that estimated after more complete sequencing (2. 44 Gb after 7. 5× coverage [24]). A similar approach was used to estimate the length of euchromatic DNA in the elephant shark genome. The numbers and positions of overlaps that began five or more bases downstream from the 5′ end of each of 200,000 reads were computed. In order to eliminate reads from repetitive regions, only “qualifying” reads with fewer than k = 5 overlaps beginning in this region were considered. For the first 100 bases of the region, the number of overlaps beginning in that window is tabulated for each of the N qualifying reads. Letting ni equal the number of qualifying reads with i overlaps, the mean number of overlaps per read is calculated. For the current dataset, λ5 = 0. 18 ± 0. 01. Although for k = 5 the effect is small, λk is an underestimate due to the truncation of the sum at i = k − 1. To correct for this truncation, λk = λk′/P (x < k|λk′) may be solved for a final estimate, λk′. Here, λk′ = 0. 19. Equating λk′ to np, the mean of the binomial distribution, with n = 1,730,917 reads, and probability of a read beginning in a window of length 100 being p = 100/Gk, where Gk is the estimated genome length, yields Gk = 100n/λk′ (i. e. , G5 = 9. 1 × 108). Estimates based on other values of k, ranging from 3 to 6, result in very similar estimates. The assembly output (0. 329 million contigs of mean length 1. 72 kb) is comparable to a simple model assembly [26] with 40 base overlaps (0. 327 million contigs of mean length 1. 87 kb). We first delineated “genic regions” in the elephant shark sequences by mapping the extreme start and end positions of individual protein matches from BLASTX alignments. Overlapping genic regions were then clustered to identify the longest non-overlapping genic regions. All the BLASTX high-scoring segment pairs (HSPs) that lay within a genic region were grouped together, and the best matching non-overlapping HSPs were retained to represent the coding regions in that particular genic region. The conceptual protein sequences of HSPs that fall within each genic region were joined to obtain the protein sequences encoded by the genic regions. These genic regions may include some pseudogenes that have retained significant homology to their parent genes. Protein domains in the elephant shark proteins were predicted using the FPrintScan, ScanRegExp, and HMMPfam applications of the InterProScan (version 4. 0; http: //www. ebi. ac. uk/InterProScan) package. The InterPro domains predicted in human, mouse, dog, fugu, Tetraodon, and zebrafish were extracted from Ensembl version 35 (http: //www. ensembl. org) and compared with the elephant shark InterPro domains. To identify genes that are orthologous in the elephant shark and mammals, but absent from teleost fishes, we started with 3,708 human genes that have annotated orthologs in the genomes of dog and mouse, but not fugu, Tetraodon, or zebrafish (Ensembl, version 35). These genes were used for reciprocal BLAST searches, consisting of a TBLASTN search of the human proteins against the elephant shark assembly (1 × 10−7 cutoff), followed by a BLASTX search of the aligned elephant shark sequences against the human proteome (1 × 10−7 cutoff). Putative orthologs for 423 of the human genes were found in the elephant shark assembly. In order to discount genes that have partial homologs in fugu, Tetraodon, or zebrafish, the 423 human protein sequences were again searched against the three fish genomes using TBLASTN at a less stringent cutoff of 1 × 10−3. These assemblies of fugu, Tetraodon, and zebrafish genomes are predicted to contain 22,008,28,005, and 22,877 protein-coding genes, respectively. Of the 423 proteins, 85 had no significant similarity to any of the genomes. The remaining 338 human proteins had similarity to sequences in at least one of the fish genomes. A reciprocal BLASTX search of these fish sequences indicated that 69 of them showed significant similarity to a different sequence in the human proteome. These fish sequences contain domains that are shared by multiple proteins in addition to their true orthologs. To identify genes that are conserved in the elephant shark and teleost fishes, but divergent or lost from mammals, we first identified 2,967 zebrafish genes that have annotated orthologs in the genomes of fugu and Tetraodon, but not human, mouse, and dog (Ensembl, version 35). Reciprocal BLAST searches were conducted using the approach described for orthologs that are absent from teleost fishes. All elephant shark contigs and singletons (571,269) and miniscaffold reads (73,756 from 36,878 miniscaffolds) were searched against Ensembl-predicted peptides (version 37) from the human genome (National Center for Biotechnology Information version 35; 33,869 peptides from 22,218 genes) and the zebrafish genome (Zv5; 32,143 peptides from 22,877 genes) using BLASTX [67]. Zebrafish was chosen as a representative teleost for this analysis since more genes in the zebrafish assembly have been assigned chromosome coordinates (18,009 of 22,877 predicted) compared to Tetraodon (16,275 of 28,005 predicted) and fugu (no chromosome coordinates) assemblies. For the search against human peptides, 122,804 elephant shark sequences produced good alignments with e < 1 × 10−6 and a HSP of >50 bits. Of the clones that contributed to these sequences, there were 10,708 where both end reads were linked to unique pairs of human proteins. For the search against zebrafish peptides, 92,291 elephant shark sequences produced good alignments with e < 1 × 10−6 and a HSP of >50 bits. Of the clones that contributed to these sequences, there were 13,773 where both end reads were linked to unique pairs of zebrafish proteins [68]. UCEs identified in the mammalian genomes [39] were searched against the elephant shark, fugu, and zebrafish genomes using BLASTN to identify elements that showed a minimum 100 bp alignment with UCEs. This Whole-Genome Shotgun project has been deposited at DNA Databank of Japan/EMBL/GenBank under the project accession AAVX00000000. The version described in this paper is the first version, AAVX01000000. The whole-genome shotgun sequences can also be BLAST-searched on our webpage at http: //esharkgenome. imcb. a-star. edu. sg. The repetitive sequences identified have been deposited in GenBank (http: //www. ncbi. nlm. nih. gov/Genbank) under the accession numbers DQ524329 to DQ524339.

Cartilaginous fishes (sharks, rays, skates, and chimaeras) are the phylogenetically oldest group of living jawed vertebrates. They are also an important outgroup for understanding the evolution of bony vertebrates such as human and teleost fishes. We performed survey sequencing (1. 4× coverage) of a chimaera, the elephant shark (Callorhinchus milii). The elephant shark genome, estimated to be about 910 Mb long, comprises about 28% repetitive elements. Comparative analysis of approximately 15,000 elephant shark gene fragments revealed examples of several ancient genes that have been lost differentially during the evolution of human and teleost fish lineages. Interestingly, the human and elephant shark genomes exhibit a higher degree of synteny and sequence conservation than human and teleost fish (zebrafish and fugu) genomes, even though humans are more closely related to teleost fishes than to the elephant shark. Unlike teleost fish genomes, the elephant shark genome does not seem to have experienced an additional round of whole-genome duplication. These findings underscore the importance of the elephant shark as a useful “model” cartilaginous fish genome for understanding vertebrate genome evolution.

Abstract Introduction Results Discussion Materials and Methods Supporting Information

teleost fishes eukaryotes vertebrates evolutionary biology genetics and genomics

2007

Survey Sequencing and Comparative Analysis of the Elephant Shark (Callorhinchus milii) Genome

DNA damage resulting from intrinsic or extrinsic sources activates DNA damage responses (DDRs) centered on protein kinase signaling cascades. The usual consequences of inducing DDRs include the activation of cell cycle checkpoints together with repair of the damaged DNA or induction of apoptosis. Many DNA viruses elicit host DDRs during infection and some viruses require the DDR for efficient replication. However, the mechanism by which DDRs are activated by viral infection is poorly understood. Human cytomegalovirus (HCMV) infection induces a DDR centered on the activation of ataxia telangiectasia mutated (ATM) protein kinase. Here we show that HCMV replication is compromised in cells with inactivated or depleted ATM and that ATM is essential for the host DDR early during infection. Likewise, a downstream target of ATM phosphorylation, H2AX, also contributes to viral replication. The ATM-dependent DDR is detected as discrete, nuclear γH2AX foci early in infection and can be activated by IE proteins. By 24 hpi, γH2AX is observed primarily in HCMV DNA replication compartments. We identified a role for the E2F1 transcription factor in mediating this DDR and viral replication. E2F1, but not E2F2 or E2F3, promotes the accumulation of γH2AX during HCMV infection or IE protein expression. Moreover, E2F1 expression, but not the expression of E2F2 or E2F3, is required for efficient HCMV replication. These results reveal a novel role for E2F1 in mediating an ATM-dependent DDR that contributes to viral replication. Given that E2F activity is often deregulated by infection with DNA viruses, these observations raise the possibility that an E2F1-mediated mechanism of DDR activation may be conserved among DNA viruses. Cellular DNA is constantly bombarded by insults from both intrinsic sources, such as reactive oxygen species, and extrinsic sources, like genotoxic chemicals. DNA damage resulting from these challenge produces a complex protein kinase signaling cascade that promotes repair of the damaged DNA and activates cell cycle checkpoints or apoptosis [1]. A central mediator of certain DNA damage response (DDR) pathways is the ataxia telangiectasia mutated (ATM) protein kinase [2]. ATM activation leads to the phosphorylation of numerous proteins that ultimately signal cell cycle arrest and DNA repair and/or apoptosis. Recent data have shown that several viruses, including herpes simplex virus type 1 (HSV-1), polyomavirus, human papillomavirus (HPV), and human immunodeficiency virus type 1 (HIV-1) require the activation of ATM and/or downstream proteins for a fully permissive infection [3], [4], [5], [6], [7]. Presumably, these viruses also encode proteins that interfere with downstream DDR signaling that antagonize virus replication through the activation of cell cycle checkpoints or the induction of apoptosis. Human cytomegalovirus (HCMV) infection activates multiple DDR proteins, including ATM and the downstream effector protein, p53 [8], [9], [10]. The p53 transcription factor plays an important role in responding to certain cellular stresses as well as in regulating cell cycle progression. It has been proposed that the activation of p53 helps to elicit the cell cycle arrest in HCMV infected fibroblasts by modulating p21 levels [11] or by facilitating viral gene expression [12]. However, the functional relevance of ATM in HCMV replication has been unclear. Although others have concluded that ATM does not contribute to HCMV replication [9], it seems reasonable to reconsider the role of ATM in this process given that downstream factors of ATM activation are required for efficient replication of HCMV and that ATM contributes to the replication of other DNA viruses. It has been noted that the cellular environment of HCMV infected cells is “G1/S-arrest”, yet these cells exhibit some biochemical properties of S and G2 phase, such as cyclin E and cyclin B kinase activation and pRb hyperphosphorylation [13], [14], [15], [16]. One consequence of these events is the induction of E2F activator complexes following HCMV infection [17]. The RB-regulated activator class of proteins within the E2F family of transcription factors includes E2F1, E2F2, and E2F3a [18], [19]. These proteins regulate the transcription of many genes, such as those required for S-phase progression and DNA repair [19]. In addition, it has been shown that RB inactivation and deregulation of E2F1, but not E2F2 or E2F3, leads to DNA double strand break (DSB) accumulation and cell cycle checkpoint signaling [20], [21], [22], [23]. Although it is well established that one of the initial effects of HCMV infection is to inactivate the RB family of proteins, whether the consequential deregulation of the E2F proteins affects HCMV replication is unknown. In this study, we asked whether there is a functional role for the host DDR in HCMV replication. We find that efficient HCMV replication requires a host DDR that centers on the presence of ATM and E2F1 protein. E2F2 and E2F3 do not influence the infection-associated DDR or viral replication. We show that expression of the HCMV IE proteins is sufficient to activate the host DDR. Our data suggests a model wherein HCMV infection stimulates an E2F1-mediated DDR to activate downstream pathways that facilitate the replication or maturation of nascent virus. Many viruses require ATM activation for a fully permissive infection, and it has been reported that ATM is activated by IE1 expression or HCMV infection [8], [9], [10]. We asked whether HCMV replication is affected by functional changes in ATM. Initially, we examined the effects of caffeine, an inhibitor of PI3-like kinases including ATM, on HCMV replication. Following virus absorption, infected HEL fibroblasts were treated with 10 mM caffeine and virus yield was examined by plaque assay. As shown in Figure 1A, little or no virus replication was observed in the caffeine treated cells at either a low or moderate MOI, whereas HCMV replicated to expected levels in sham treated HEL fibroblasts. These results suggest that PI3-like kinase activity is necessary for HCMV replication. Next we determined the contribution of ATM to HCMV replication by assessing viral replication in dermal fibroblasts from a normal donor compared to fibroblasts from a patient with ataxia telangiectasia (AT) that do not express detectable levels of the ATM protein. As shown in Figure 1B, much lower yields of infectious virus were generated in the AT fibroblasts compared to the control fibroblasts. The difference in HCMV replication was dose dependent with higher infectious doses (MOIs of 1. 0 or 3. 0) resulting in 2 to 3 log reductions in infectious virus production. Low MOI infection (MOI of 0. 3) of AT fibroblasts resulted in little detectable viral progeny. At all time points and MOIs tested, infection of the AT fibroblasts resulted in reduced levels of IE2, pp65 and gB55, representing IE, E, and L viral gene products, respectively (Figure 1C). However, IE1 levels were not dramatically affected by the absence of ATM. Given that UL123, which encodes IE1, is the first viral gene expressed in infected cells, the lack of sustained changes in IE1 accumulation raises the possibility that later gene expression events are compromised. These results suggest that functional ATM is necessary for efficient HCMV replication. Our observations suggesting a role for ATM in HCMV replication is contrary to another study [9]. Moreover, there is a concern with using AT fibroblasts as a model because the prolonged absence of functional ATM in cells from AT patients may have resulted in secondary genetic and/or biochemical changes that alter cellular environments and thereby influence HCMV replication. We addressed these issues by using siRNAs to transiently deplete ATM protein levels (siATM) in HEL fibroblasts. Cells were transfected with siATM 24 h prior to HCMV infection and viral replication (Figure 1D) and gene expression (Figure 1E) were monitored during a 5-day time course. Of the siRNAs designed to deplete ATM levels, only siATM-c was effective. This ATM-specific siRNA inhibited progeny virus production ∼10-fold throughout the replication time course (Figure 1D). Another siRNA, siATM-a, which did not consistently affect ATM levels, produced replication results comparable to a nonspecific siRNA (siNS). Similar to what we observed in dermal fibroblasts (Figure 1C), we found reduced levels of IE2, pp65 and gB55, but little change in IE1 levels when ATM levels were depleted by siATM-c (Figure 1E). We conclude that ATM is required for efficient HCMV replication. We next determined whether cells deficient in ATM are compromised in the formation of replication compartments (RCs), which are sites of viral DNA replication and maturation. HEL fibroblasts were treated with siATM-c or control siRNA (siNS) and infected with HCMV and immunostained with anti-pUL44 antibody to detect HCMV replication compartments and scored (Figure 2). pUL44 is a virally encoded PCNA-like processivity factor of the viral DNA polymerase [24], [25]. In addition, dermal fibroblasts from normal and AT individuals were infected with virus, RC structures identified and scored. Under conditions of ATM depletion, the percentage of merged, “mature” RCs was reduced relative to control cells (Figure 2B). This difference was more apparent in AT fibroblasts where very few mature RCs were observed. The change in the percentages of mature RCs between siATM treated HEL fibroblasts and AT fibroblasts may explain why the replication phenotype observed in AT fibroblasts is dramatically different (compare viral replication curves in Figures 1B and 1D) whereas viral protein expression is less divergent (compare Figures 1C and 1E). Because it was reported that HCMV infection or expression of IE1 or IE2 can activate ATM as measured by autophosphorylation on Ser1981 [8], [10], we asked whether HCMV could induce the formation of DNA damage sensing foci containing γH2AX, an event downstream of ATM activation and other DNA damage-activated kinases [26], [27], [28]. γH2AX is the phosphorylated form of H2AX that is mediated by PI3-like kinases, including ATM. Infected cell cultures were co-immunostained for IE expression (both IE1 and IE2) to mark infected cells and γH2AX. As shown in Figure 3A, γH2AX staining was visible in the nuclei of cells expressing IE antigens. The levels of γH2AX protein increased over time and accumulated with infectious dose as measured by immunoblotting (Figure 3D). The pattern of γH2AX immunostaining in Figure 3 is different from the punctate foci observed when cells are treated with DNA damaging agents that cause dsDNA breaks [28], [29]. In HCMV-infected cells, γH2AX appears to accumulate in larger “domains” of the nucleus and by 72 hpi, much of the nucleus appears to be reactive to the γH2AX antibody. Although this pattern of γH2AX immunostaining is unusual, it is reminiscent of viral RCs. To determine whether the γH2AX localization observed in infected cells is coincident with viral RCs, we co-immunostained infected cells for both γH2AX and pUL44. Although the IE proteins were not restricted to the RCs, γH2AX accumulated predominantly in these nuclear compartments based on co-immunostaining for pUL44 (Figures 3B–C). Thus, γH2AX accumulates in HCMV RCs. One would anticipate that activated, autophosphorylated ATM would colocalize with γH2AX in RCs if ATM were responsible for γH2AX phosphorylation. Co-immunostaining for phosphoserine 1918-ATM and γH2AX in infected HEL fibroblasts showed that phosphoserine 1918-ATM and γH2AX colocalized at 24 hpi, but this pattern was diminished at 48–72 hpi (Figure S1). Next, we determined whether ATM is responsible for γH2AX accumulation following infection. As observed in HEL fibroblasts, infections of normal dermal fibroblasts showed partially overlapping co-immunostaining for phospho-ATM and γH2AX (Figure 3E). However, γH2AX still accumulated in AT dermal fibroblasts at 48 hpi in a pattern suggestive of co-localization in RCs. We also noticed that γH2AX accumulated in punctate foci early during infection (5 hpi) in both HEL and normal dermal fibroblasts (Figure S1 and Figure 3E). In contrast to the results observed at later times pi, no γH2AX was detected in AT dermal fibroblasts at 5 hpi (Figure 3E). Similar results were obtained in HEL fibroblasts when ATM was depleted with an siRNA (Figure S2). Therefore, ATM is responsible for the infection-associated DDR prior to the formation of mature RCs. However, this conclusion does not preclude the possibility that ATM may functionally contribute viral replication during other stages of the replication cycle. Given the contribution of ATM to viral replication, we determined whether downstream targets in the ATM-mediated DDR also influences replication. Here we focused on H2AX because it is responsive to ATM signaling (Figure 3E, and S2). Depletion of H2AX in HEL fibroblasts with either of two siRNAs reduced HCMV replication approximately 10 fold (Figure 4A). H2AX depletion also decreased the levels of IE2, and to a lesser extent pp65 and gB55 protein levels (Figure 4B). Another downstream target of ATM is p53. We had previously shown that p53 is phosphorylated by ATM during HCMV infection [8] and others have shown that p53 contributes to HCMV replication [12]. Phosphorylated H2AX may stabilize DNA damage recognition structures including MRE11-NBS1-RAD50 (“MRN”). We determined whether depletion of H2AX would impact the localization of NBS1 and DNA PKCS, a PI3-like kinase, during infection. HCMV infection appeared to increase the levels of both DNA PKCS and NBS1 and the nuclear distribution of DNA PKCS (Figure S3). The localization of NBS1 (Figure S3A) did not appear to be grossly impacted by reduced levels of γH2AX in infected cells. In contrast, the level and distribution of DNA PKcs (Figure S3B) appear to be similar to mock infected cells following treatment with siH2AX and HCMV infection (Figure S3). However, it is not apparent whether these patterns of protein localization are relevant to HCMV infection. While, H2AX, a cellular target of ATM-mediated signaling contributes to HCMV replication, the mechanism (s) by which this factor modulates replication is unclear. Given the rapid formation of γH2AX foci and protein accumulation after HCMV infection (5 hpi; Figure 5A–B) and given that it has been previously reported that ectopic IE1 expression results in ATM autophosphorylation [8], we further examined the DDR associated with expression of IE gene products by monitoring the accumulation of γH2AX and p-ATM. Transduction of cells with Ad-IE1 or Ad-IE2 resulted in a time dependent nuclear accumulation of p-ATM and γH2AX (Figure 5C–H). Initially the p-ATM immunostaining pattern was punctate in the presence of IE1 or IE2 expression (Figure 5D). At 48 hpi, cells transduced with Ad-IE1 produced a broad, punctate pattern of p-ATM immunostaining, whereas p-ATM appeared to co-localize with IE2 in Ad-IE2 transduced cells (Figure 5D). Thus, ATM and H2AX phosphorylation occur early during HCMV infection and both IE1 and IE2 have the capacity to promote these events. We previously reported that altering RB function or increasing E2F1 levels leads to an ATM-dependent DDR [20], [21], [22], [30]. Given that HCMV infection or ectopic expression of IE1 or IE2 leads to increased E2F activity [17], [31], [32], [33], we asked whether E2F1 or other RB-associated, activator E2Fs were responsible for the DDR following HCMV infection or IE cDNA transduction. To address this question, we individually blocked the expression of E2F1, E2F2 or E2F3 with one of two different siRNAs prior to infection with HCMV or transduction with recombinant adenoviruses, and then scored cells for a host DDR by γH2AX immunostaining. A low basal percentage (<10%) of HEL fibroblasts stained positive for γH2AX in mock-infected HEL cells (Figure 6A). This level of γH2AX immunostaining most likely represents DNA damage signaling that normally occurs in human fibroblasts replicating their own DNA [30]. Depletion of individual E2Fs did not affect this background staining (Figure 6A). Infection with HCMV resulted in increased γH2AX immunostaining, with ∼35% of the cells positive for γH2AX (Figure 6A). This percentage dropped to ∼16% when either of two siRNAs targeting E2F1 expression was transfected into cells prior to infection. Depletion of either E2F2 or E2F3 with specific siRNAs did not significantly alter the percentage of cells staining positive for γH2AX. Transduction with a control recombinant adenovirus encoding β-gal did not affect the background of levels of γH2AX staining, but transduction with Ad-IE1 resulted in the majority of HEL fibroblasts immunostaining positive for γH2AX (Figure 6B). Only depletion of E2F1 reduced the percentage of γH2AX-positive cells, depleting E2F2 or E2F3 with siRNAs had no effect on the host DDR (Figure 6B). Likewise, only E2F1 depletion reduced the percentage of γH2AX-positive HEL fibroblasts when transduced with AD-IE2 (Figure 6C). Multiple time points are shown for this experiment because of the lower percentages of γH2AX-positive cells at earlier times post transduction. Ad-E7, which encodes HPV type 16 E7, was included as a positive control for E2F1-mediated DDR [20], [21]. Therefore, HCMV infection and IE1 or IE2 expression activate an E2F1-mediated DNA damage response. Given the observations that ATM is required for HCMV replication and that E2F1 contributes to the DDR in infected and IE1 or IE2 transduced cells, we next determined whether E2F1 specifically contributes to HCMV replication. HEL fibroblasts were transfected with siRNAs specific for E2F1 or with a control siRNA 24 h prior to infection with HCMV. Although virus infection increases the levels of E2F1, depletion of E2F1 in infected cells reduced its levels approaching that observed in the mock-infected sample (Figure 7B). Transfection of either E2F1-specific siRNA also reduced viral IE, E and L gene expression as measured by immunoblotting for IE2, pp65 and gB55, respectively. However, IE1 levels were not consistently affected by E2F1 depletion. Depletion of E2F1 also altered HCMV replication with a ∼5 to ∼50-fold reduction in progeny virus production that was dependent on the siRNA used to deplete E2F1 levels (Figure 7A). These results are consistent with the patterns observed for ATM depletion (Figures 1D–E). E2Fs are generally thought to function as transcription factors with E2F1 having additional, less well-characterized roles in DNA damage accumulation and apoptosis [34]. To begin to differentiate whether the effects of E2F1 depletion on virus replication were due to reduced levels of an “activator E2F” (i. e. , E2F1, E2F2, and E2F3a) or due to unique functions of E2F1, we determined whether depletion of E2F2 or E2F3 would affect HCMV protein expression and replication. Depletion of E2F2, E2F3a, E2F3b (an E2F3 isoform that does not contribute to proliferation [35]), or a combination of E2F3a and E2F3b reduced the levels of the targeted protein to approximately that observed in mock-infected samples (Figures S4A–B). Targeting of E2F2 or E2F3a or E2F3b, or the combination of E2F3a and E2F3b had no discernable effect on the accumulation of viral proteins (Figure S4A–B) or the production of progeny virus (Figures 7C–D). These results suggest that the specific deregulation of E2F1 levels is required for efficient replication of HCMV. In this study, we find that HCMV infection stimulates an E2F1-mediated DDR that centers on activation of the ATM kinase early in infection and subsequently coordinates with nuclear viral replication compartments. Moreover, we show that ATM and downstream signaling are required for replication following infection at a low MOI and contributes to HCMV replication at higher doses (Figure 8). Our results are consistent with ATM contributing to the replication of other viruses (for review, see [36]). This conclusion contrasts with what has been previously reported for HCMV infection [9], where it was determined that ATM is not required for the progression of HCMV infection. It is unclear why there is a discrepancy between these studies, but we have confirmed our results using multiple approaches (Figure 1). It appears that ATM activation represents a general response to infection by DNA viruses or viruses that have a DNA stage in the replication strategy, such as retroviruses. The question remains as to why viruses activate ATM and other DDR proteins for replication. Indeed, activation of the host DDR is an obstacle for the replication of at least one DNA virus, adenovirus, which blocks the host DDR during infection [37]. One reason for infection-associated ATM activation may be to utilize the consequential stimulation of cellular DNA repair and recombination enzymes [38] to benefit viral replication [3], [39], [40]. Perhaps, in the case of HCMV, repair and recombination enzymes may aid in circularizing the viral DNA after it has entered the cell and/or facilitate the maturation of nascent viral genomes. A DNA repair complex of DNA ligase IV and XRCC4 circularizes herpes simplex virus genomes early in infection [41]. If correct, one would predict that γH2AX, as well as other DDR factors would be bound to virion-delivered HCMV DNA once uncoated in the nucleus. However, others have concluded that parental viral DNA and γH2AX do not co-localize [9]. It remains an open question as to whether there is a contribution of the host DDR to very early events in HCMV replication. HCMV gene expression patterns in infected cells lacking or depleted for ATM may offer clues to the stage (s) in infection that depend on ATM function. IE1 expression is largely unaffected by ATM status. One interpretation of this observation is that ATM does not influence IE events that affect viral replication. Our observation that mature RCs and marker E and L gene products are reduced during infection of AT fibroblasts is consistent with a model wherein ATM influences events associated with DNA replication, presumably by stimulating host (or viral) factors to aid in the repair or recombination of nascent viral DNA. A role for ATM in DNA repair or recombination post replication is also a possibility, although the pattern of viral gene expression argues against this idea. It also cannot be excluded that ATM may have a novel function in phosphorylating/activating an essential host or viral factor not associated with DNA replication, repair, or recombination. A number of mechanisms have been proposed for how viral infections lead to ATM activation. Upon HIV infection, ATM activation requires the viral integrase and it is proposed that ATM functions in post integration DNA repair [41]. For polyomaviruses like SV40, it is thought that the onset of viral DNA replication activates ATM, which then phosphorylates an essential serine residue on large T antigen [42]. HPV genome replication appears to switch from theta to rolling circle replication [43], which may activate ATM. Alternately, infection by DNA viruses may cause damage to host chromosomes, which would stimulate a host DDR. In this situation, targets of ATM phosphorylation should also contribute to viral replication. Both H2AX (Figure 3) and p53 [8] are substrates of the ATM kinase during HCMV infection and both H2AX (Figure 4) and p53 [12] contribute to HCMV replication. While the mechanism by which H2AX contributes to HCMV replication is unknown, p53 is found in RCs, binds viral DNA and evidence suggests that p53 influences the expression of viral genes [44]. However, the roles of ATM-mediated phosphorylation of H2AX or p53 to productive replication are not known at this point. Somewhat surprisingly, we find that the initial accumulation of γH2AX following HCMV infection is dependent on ATM whereas ATM is dispensable for γH2AX accumulation once mature viral DNA replication compartments are formed. The kinase (s) responsible for phosphorylating H2AX in the absence of ATM at these later times is unknown but it is possible that another PI3-like kinase, perhaps ATR [45], is responsible for H2AX phosphorylation. DNA PKcs, another PI3-like kinase, is known to phosphorylate H2AX in response to DNA damage signaling, but it has been shown that DNA PKcs does not localize to HCMV RCs [9]. However, even though H2AX can be phosphorylated by other kinases later during infection, activated ATM is mostly located in HCMV RCs at these times pi, leaving open the possibility that ATM is influencing activities in these nuclear compartments. ATM is required for efficient H2AX phosphorylation in MHV68-infected primary marcrophages and ATM is relocalized to sites of viral genome deposition, although a viral kinase also contributes to H2AX phosphorylation [46]. ATM is also rapidly relocalized to replication compartments during HSV infection [3]. It will be interesting to determine whether ATM is only transiently responsible for host DDR signaling (and viral replication) during infections with herpesviruses. Deregulation of E2F activity is a hallmark of infections with many DNA viruses that replicate in the nucleus. HCMV infection and expression of its major IE proteins, particular IE1 and IE2, have been shown to inactivate RB family members and induce the expression of E2F regulated genes [17], [33], [47], [48] possibly by providing host factors that contribute to virus replication. Our data reveal another consequence of inactivating RB family members and the specific deregulation of E2F1, the activation of a host DDR that facilitates the replication of HCMV. The mechanism by which E2F1 stimulates host DDR is not known. Inactivation of RB and the subsequent deregulation of E2F1—but not the related family members, E2F2 or E2F3, which also interact with RB—leads to an accumulation of DNA double-strand breaks in human fibroblasts [21]. Although it is not clear if HCMV infection causes extensive host DNA damage, infection can result in a DNA double strand break on chromosome 1 [49]. Whether this single DNA break is sufficient to initiate the observed host DDR is unclear. Alternately, it has been shown that activation of a DDR does not necessarily require DNA lesions. Rather, the physical interaction of DNA repair factors with chromatin can be sufficient to activate the DDR signaling cascade [50]. Therefore, host chromosomal changes mediated by disruption of RB/E2F1 complexes or other mechanisms of E2F1 deregulation should also be considered as possible ATM activators during infection. Most productive infections by DNA viruses result in deregulation of E2F activity through inactivation of RB and RB family members [51]. These viruses also activate an ATM-centric DDR, although some viruses, including MHV68, KSHV, and adenovirus, encode factors that can block signals from reaching ATM or its targets [37], [52], [53], [54]. The herpesviral proteins responsible for this inhibition are often expressed during latency, which raises the possibility that the host DDR interferes with aspects of latency such as cell survival, proliferation or, perhaps, the maintenance of viral episomes. Most of these viruses have in common infection-associated E2F deregulation, DDR activation, and a contribution of ATM to productive infections. These shared features raise the possibility that E2F1 contributes to the replication of many viruses through its activation of the ATM-associated DDR. It will be interesting to determine how common the E2F1-mediated DDR is to productive viral infections. AT dermal fibroblasts from an ataxia-telangiectasia patient (GM05823C; termed “AT”), age-matched primary human dermal fibroblasts (GM00316B; termed “CONB”) and human embryonic lung fibroblasts (HEL fibroblasts) were obtained from the Coriell Institute for Medical Research (Camden, N. J). Dermal fibroblasts were maintained in Minimum Essential Media (MEM) supplemented with 15% fetal bovine serum (FBS) and 1% penicillin-streptomycin. HEL fibroblasts were cultured in Dulbecco modified Eagle medium (DMEM) supplemented with 10% FBS and 1% penicillin-streptomycin. All media, FBS, and antibiotics were from GIBCO. HCMV strain AD169 was obtained from the American Type Culture Collection (ATCC, Manassas, VA). Fibroblasts were infected with HCMV AD169 at various multiplicities of infection (MOI). Viral infections were performed in growth media with 2% FBS for 2 hours. The viral inoculum was removed and replaced with normal grow medium. Cells pellets were collected at different time post infection and lysates were generated as described as below. Cells were treated with caffeine (Sigma) at a dose of 10 mM following virus absorption. The drug was replenished every 24 h. Recombinant adenoviruses encoding HCMV IE1-72 (Ad-IE1), HCMV IE2-86 (Ad-IE2), β-galactosidase (Ad β-gal), and HPV16 E7 (Ad-E7) have been described [11], [20], [55], [56]. Recombinant adenovirus stocks were generated, purified and titered as described [20], [57]. All recombinant adenovirus infections were done at a MOI of 250 unless otherwise noted. siRNA were transfected at 50–100 nM using Lipofectamine 2000 (Invitrogen) or by electroporation in the presence of siPort transfection buffer (Ambion). The nonspecific siRNAs (NS) were composed of a nonsense sequence and had no effect on parameters tested relative to mock transfection. Transfection conditions for individual siRNAs were optimized. The sequences of the siRNAs used in this study are as follows: siNS (5′-CTTCCTCTCTTTCTCTCCCTTGTGA-3′) [used as a control for siATM], siATM-a (5′-GGAGTTATTGATGACGTTACATGAG-3′), siATM-c (5′-CGCATGTGATTAAAGCAACATTTGC-3′), siE2F1A (5′-GGCCCGATCGATGTTTTCC-3′), siE2F1C (5′-GTCACGCTATGAGACCTCA-3′), siE2F2A (5′-GTGCATCAGAGTGGATGGC-3′), siE2F2B (5′-CAAGAGGCTGGCCTATGTG-3′), siE2F3a (5′-GCGTACATCCAGATCCTCA-3′), siE2F3b (5′-GGAAATGCCCTTACAGCAG-3′), siE2F3 (a+b) (5′-GACCAAACTGTTATAGTTG-3′), siH2AXa (5′-CAACAAGAAGACGCGAATC-3′) siH2AXb (5′-CGACGAGGAGCTCAACAAG-3′) NS (5′-TTTTTTTCCCCAAAGGGGG-3′) [used as a control for siE2F and siH2AX treatments]. Cells were seeded and infected at the listed MOI for each experiment. Triplicate infections were performed for each time point. At the indicated times pi, a small aliquot (200 ul) of supernatant was harvested from each dish and stored at −80°C. Viral titers were then determined on HEL fibroblasts using standard techniques. Plaques were counted 7 dpi using Giemsa stain (Sigma) to enhance the visualization of plaques. Plotted values represent the average of triplicate infections. Infected cells were harvested at the indicated time point and pellet cells were stored at −80°C. Thawed cell pellets were resuspended in radioimmunoprecipitation assay buffer (RIPA: PBS, 0. 1%NP-40,1% sodium dodecyl sulfate, 0. 5% sodium deoxycholate, sodium vanadate, phenylmethylsulfonyl fluoride, and aprotinin) and incubated on ice for 1 h. Samples were sonicated for 15 sec, and soluble proteins were collected by centrifugation for 10 min at 13,000 rpm in a microcentrifuge. Proteins were resolved by SDS-PAGE, and the proteins were transferred to a polyvinylidene difluoride membrane (Perkin-Elmer) by electroblotting. Detection of E2F1, E2F2, E2F3, IE, pp65, gB55, ATM, γH2AX and actin proteins was performed with antibodies specific for E2F1 (C-20, Santa Cruz Biotechnology), E2F2 (C-20, Santa Cruz Biotechnology), E2F3 (C-18, Santa Cruz Biotechnology), IE1-72 and IE2-86 (MAB8130, Chemicon International), pp65 (CA003-100, Virusys), gB55 (Shan Lu, UMass Medical School), ATM (D2E2, Cell Signaling), histone γH2AX (Upstate Biotechnology), actin (A5316, Sigma) and HRP-conjugated secondary antibodies. Protein bands were visualized by chemiluminescence with ECL reagent (Amersham). Cells were plated on glass coverslips that were pretreated with 40% HCl for 2 min followed by a 5 min wash in 70% ethanol. Cells were infected with recombinant adenoviruses or HCMV at the indicated MOIs. Cells were washed three times with PBS and fixed with 2% paraformaldehyde. Fixed cells were blocked in 10% FBS for 1 h at room temperature and incubated with antibodies against IE (MAB8130, Chemicon International), pUL44 (Virusys), γH2AX (Upstate Biotechnology), phospho-Ser1981 ATM (Rockland Immunochemicals), NBS1 (GeneTex), and DNA PKcs (Thermo Scientific). FITC conjugated goat anti-rabbit, Texas Red-conjugated goat anti-mouse IgG1 or IgG2a secondary antibodies (Southern Biotechnology Associates, Inc) were used to detect bound primary antibody by immunofluorescence. Images were captured on a Nikon microscope and analyzed using Improvision software. Over 200 cells were counted per sample when quantifying cell staining. Statistical analyses were performed using unpaired t-tests. Values are expressed as mean ± SD of three independent experiments. A P value of ≤0. 05 was considered statistically significant.

As intracellular parasites, viruses often redirect cellular pathways to facilitate their own replication. Infection by DNA viruses often lead to the activation of host DNA damage response pathways, which normally function to repair damage to host chromosomes. Some DNA viruses depend on this infection-induced DNA damage response to efficiently replicate. How infection activates the DNA damage response is poorly understood. To address this limitation, we first determined whether the DNA damage response affects the replication of human cytomegalovirus (HCMV) and then addressed how infection induces this response in cells. We find that HCMV infection results in a host DNA damage response centered on the Ataxia Telangiectasia Mutated (ATM) protein kinase. We also show that HCMV requires ATM for efficient replication. Unexpectedly, we find that the mechanism responsible for ATM activation is the expression of E2F1, a cellular transcription factor. Moreover, expression of E2F1, like ATM, is required for HCMV replication. These observations may be of fundamental importance because infection by most DNA viruses result in both E2F1 expression and an ATM-mediated DNA damage response.

Abstract Introduction Results Discussion Materials and Methods

virology/viral replication and gene regulation virology

2011

An E2F1-Mediated DNA Damage Response Contributes to the Replication of Human Cytomegalovirus

Behavioural anomalies suggesting an inner ear disorder were observed in a colony of transgenic mice. Affected animals were profoundly deaf. Severe hair bundle defects were identified in all outer and inner hair cells (OHC, IHC) in the cochlea and in hair cells of vestibular macular organs, but hair cells in cristae were essentially unaffected. Evidence suggested the disorder was likely due to gene disruption by a randomly inserted transgene construct. Whole-genome sequencing identified interruption of the SorCS2 (Sortilin-related VPS-10 domain containing protein) locus. Real-time-qPCR demonstrated disrupted expression of SorCS2 RNA in cochlear tissue from affected mice and this was confirmed by SorCS2 immuno-labelling. In all affected hair cells, stereocilia were shorter than normal, but abnormalities of bundle morphology and organisation differed between hair cell types. Bundles on OHC were grossly misshapen with significantly fewer stereocilia than normal. However, stereocilia were organised in rows of increasing height. Bundles on IHC contained significantly more stereocilia than normal with some longer stereocilia towards the centre, or with minimal height differentials. In early postnatal mice, kinocilia (primary cilia) of IHC and of OHC were initially located towards the lateral edge of the hair cell surface but often became surrounded by stereocilia as bundle shape and apical surface contour changed. In macular organs the kinocilium was positioned in the centre of the cell surface throughout maturation. There was disruption of the signalling pathway controlling intrinsic hair cell apical asymmetry. LGN and Gαi3 were largely absent, and atypical Protein Kinase C (aPKC) lost its asymmetric distribution. The results suggest that SorCS2 plays a role upstream of the intrinsic polarity pathway and that there are differences between hair cell types in the deployment of the machinery that generates a precisely organised hair bundle. The sensory “hair” cells of the hearing and balance (vestibular) organs in the inner ears of vertebrates convert movements, initiated by sound waves in the cochlea or by translational or rotational motion in the vestibular system, into electrical signals. Fundamental to their function is the organised bundle of stiff, erect projections from their apical (luminal) surface [1] from which hair cells derive their name. This “hair bundle” is formed of stereocilia, microvillus-like protrusions formed of closely packed actin filaments [2,3], organised in rows that increase in height in one particular direction across the apical surface of the cell. Eccentrically positioned behind the row of longest stereocilia is a single specialised true cilium known as the kinocilium, except uniquely in the mature auditory epithelium in mammals, the organ of Corti, where a kinocilium is present as hair cells first differentiate but subsequently retracts to leave only the basal body in the apical cytoplasm [4]. The asymmetry deriving from the increasing height of stereocilia in one particular direction towards the position of the kinocilium/basal body defines a cellular “polarity” which is of fundamental functional significance [1,5]. Deflections of the stereocilia along the line of polarity, towards and away from the position of the kinocilium/basal body, opens and closes “mechano-transduction” channels [6], initiating hair cell responses. The polarity of hair-bundles on different hair cells with respect to each other is not random. Hair-bundles show a distinct “orientation”, as defined by the position of the kinocilium/basal body, that is precisely related to that of their immediate neighbours. In the organ of Corti the hair bundles on all hair cells are oriented with the row of longest stereocilia on the side of the bundle towards the outside of the spiralling sensory epithelial strip (the lateral side). In the maculae, flat sheets of sensory epithelium in the utricle and saccule of the vestibular system, all the hair bundles located on one side of a region running along the length of the epithelial sheet are oriented the same way with respect to the periphery and at 180° to those on the other side of that region. In the saddle-shaped cristae of the semi-circular canals, all the hair bundles are oriented in the same direction across the epithelium. When hair cells first begin to differentiate during embryonic life, around embryonic day (E) 12. 5–13. 5 in mice, the kinocilium emerges in the centre of the hair cell surface surrounded by nascent stereocilia [7,8]. Subsequently the kinocilium assumes an eccentric position and stereocilia grow differentially to produce the morphologically polarised and oriented bundle [7,9, 10]. In the organ of Corti (and the basilar papilla, the auditory epithelium of birds) the length of the longest stereocilia and the number of stereocilia on each hair cell are precisely defined [10–12]. In auditory hair cells, the number of stereocilia that initially emerge exceeds the number ultimately present on the fully differentiated cell; the supernumerary presumptive stereocilia on the inner side (the side of the shorter stereocilia) are retracted into the cell as the remaining ones lengthen [9,10]. In addition, while the early immature bundles are essentially rounded, mature hair bundles have a characteristic shape, particularly those in the organ of Corti. As the extra stereocilia are withdrawn, hair bundles on cochlear inner hair cells (IHC) become like a wide U-shape or linear and reduced to 2–3 rows (in mice), while bundles on the three rows of outer hair cells (OHC) each consist generally of three rows of stereocilia and assume a characteristic M-shape. The establishment and maintenance of hair bundle polarity and orientation are thought to be regulated through a number of interacting pathways. Vertebrate homologues of the planar cell polarity (PCP) proteins that are involved in the eccentric positioning of hairs on cells of the wing in Drosophila, such as Vangl2, flamingo, frizzled and prickle [13,14] localise in distinct asymmetric patterns that match hair bundle orientation at the level of the tight-adherens junctional specialisations between a hair cell and the non-sensory supporting cells that surround it [15,16]. Disruption of the genes that encode these proteins result in mild to moderate disturbance of bundle orientation [15,16] in a proportion of (mainly) OHC but does not affect cellular polarity—the ranking of stereocilia on individual hair cells. These core PCP proteins are thought to be involved primarily in co-ordinating orientation across the tissue. Bundle polarity at the cellular level may be regulated through the kinocilium/basal body. Mutations in the genes encoding ciliopathy-associated proteins, ift88 (intraflagellar transport 88) and Kif3a (a component of the kinesin II motor) all of which are involved in maintenance of cilia and which localise in hair cells to the kinocilium or its basal body, result in varying degrees of mis-orientation and/or disturbance of bundle shape in some OHC [17–22]. These observations have led to investigation of proteins that regulate orientation of the mitotic spindle and centrosome positioning during asymmetric cell division [23–25]. Mammalian inscuteable (mInsc), LGN (Leu-Gly-Asn repeat-enriched protein, also known as mammalian Partner of inscuteable [mPins] or G-protein signalling modulator 2 [Gpsm2]), Gαi3 (GTP-binding protein alpha-I subunit 3), aPKC (atypical protein kinase C) and Par3/Par6 (partitioning-defect-3 and 6), which have roles in coupling the centrosome to the cortical cytoskeleton, are expressed in cochlear hair cells. They are asymmetrically distributed across the apical surface of both OHC and IHC during the formation and maturation of the hair bundles in complementary compartments that outline the hair bundle [23,25]. Ablation of the genes encoding mInsc and LGN and loss or inactivation of Gαi3 all lead to disruptions of varying severity to the shape and orientation of the bundles in a proportion of OHC and some IHC. These “intrinsic polarity pathway proteins” [25] are thought to interact with the cortical cytoskeleton underlying the apical surface plasma membrane to position the kinocilium and shape the bundles of these cochlear hair cells, thereby controlling hair cell apical asymmetry. A similar asymmetric distribution of the intrinsic polarity proteins is also evident in macular hair cells [23] but the effects of their inactivation upon hair bundle morphology and organisation in vestibular sensory epithelia has not been reported. In this paper we demonstrate disturbances of hair bundle morphology more severe than those described for mutations in other genes involved in hair bundle formation, which affect every cochlear IHC, cochlear OHC and vestibular macular hair cell, with different effects on each of these hair cell types, but which does not affect the hair bundles of hair cells in the cristae. This phenotype is associated with the mis-expression or mis-localisation of intrinsic polarity pathway proteins. We also show that the phenotype is associated with a disruption in the gene encoding SorCS2 (Sortilin-related VPS-10 domain containing protein). SorCS2 has not previously been implicated in hair bundle differentiation. As an initial step to identify the genetic defect in the mouse colony in which the hair bundle anomalies arose, the inner ear phenotypes of all possible genotypes with respect to the presence or absence of the transgenic inserts and in the various combinations were assessed. This revealed that the common feature of all animals that showed bundle anomalies was homozygosity for the Tie2-cre insert. The Tie-2-cre mice in the colony had been derived from animals in which two copies of the transgene had been inserted randomly into the genome [29]. It was reasoned therefore that the insertion had caused a gene disruption. Consequently, next generation (Illumina HiSeq) whole genome sequencing was applied to identify the site of the disruption. Total genomic DNA extracted from deaf mice that showed hair bundle anomalies was analysed by Next Gen DNA sequencing. All of the paired end sequence reads that contained any part of the Tie2-cre vector were computationally identified. Sequences that contained the internal transgene at both ends were discarded. Of the remaining paired end reads, 21 had reads that at one end contained the transgene vector and at the other end mapped within an Ig Kappa gene. The Ig Kappa family has ~180 closely related members which makes unequivocal mapping difficult. There are no reports of a role for IgK in actin assemblies or similar functions, and immunolabelling for IgK in frozen sections of early postnatal inner ear was confined to capillaries and none was detectable in cells of the sensory epithelium. An additional 13 vector-host paired end sequences mapped to the SorCS2 gene encoding the Sortilin-Related VPS10 Domain Containing Receptor 2. Specifically, these insertions occurred~15kb into the first intron of SorCS2 at positions 36382105 to 36382309 on mouse chromosome 5 (NCBI37/MM9) (Fig 1D). Both ends of the insertion were detected in multiple sequence reads. The intervening 204bp of intronic sequences had been deleted and subsequent BLAST analysis of the whole genome DNA sequence (of the homozygous transgenic mouse) confirmed this. SorCS2 is a transmembrane glycoprotein receptor that is a member of the mammalian vacuolar protein sorting 10 protein (VPS 10P) family that have a role in intracellular trafficking of proteins [30], but is also a pro-neurotrophin receptor involved in regulation of actin dynamics associated with neural growth cone collapse [31]. The association of the protein with regulation of actin assemblies therefore suggested disruption of SorCS2 as a likely cause of the hair bundle anomalies. Evidence for the expression of SorCS2 in the inner ear of affected and unaffected littermates was therefore sought. In order to assess whether the transgenic insertion within intron 1 of the SorCS2 gene affects expression of the major SorCS2 transcript (Ensembl ID: Sorcs2-001, ENSMUST00000037370. 13) RT-qPCR was performed on cDNA made from whole inner ear tissue removed from P0-P1 animals that were the offspring of crosses between homozygous affected animals and heterozygotes (as identified from the ratios of the phenotypes in their litters). The samples included the cochlea, the vestibular maculae and the cristae. RT-qPCR was performed on tissue from one ear of each animal; the opposite ear was prepared for subsequent phenotypic assessment. RT-qPCR revealed that SorCS2 was expressed in the inner ear tissues of unaffected animals at this age (Fig 1E). Expression of SorCS2 varied in affected animals. No expression was detected in 4 out of 9 affected animals. In the remaining 5, expression was much reduced compared to unaffected ones (Fig 1E) with the mean relative level of expression 19 fold lower. Very similar results were obtained using two different expression assays, one spanning exons 2 and 3 of the SorCS2 gene (i. e. the first two exons immediately following the intronic insertion) and one spanning exons 23 and 24 at the 3’ end of the gene. There was no difference between unaffected and affected animals in the relative expression of the gene encoding the hair cell “marker” protein myosin 7a (Fig 1E). Using previously characterised antibodies that target distinct regions of the SorCS2 molecule [32,33] immuno-labelling of inner ear whole-mount preparations from late embryonic and early postnatal wild-type mice confirmed expression of SorCS2 protein in the organ of Corti, utricular maculae and cristae (Fig 2 and S2 Fig). The expression of SorCS2 was evident in hair cells in all three sensory regions in the embryonic inner ear around the time at which hair bundles are forming (E15-18; Fig 2A–2F). In early postnatal organ of Corti (P0-P1) SorCS2 became increasingly evident in organ of Corti supporting cells (Fig 2G; S2C and S2D Fig). This expression pattern persisted in the mature organ of Corti (P30, S2G Fig). The organ of Corti of unaffected P1 mice expressed relatively higher amounts of SorCS2 in comparison to that of affected littermates (Fig 2G and 2H; S2C–S2F Fig). These results demonstrate (i) that SorCS2 is expressed in all sensory epithelia of the inner ear, and (ii) that disruption of the SorCS2 gene locus results in reduced SorCS2 protein expression. In the stereociliary bundles of hair cells of the utricular maculae in normal animals (Fig 7A) the longest stereocilia were very long—more than 5μm in those across the striola (the region where the line of reversal of hair bundle orientation occurs), those in extrastriolar regions reaching ca 10 μm—and the height gradients of the rows down to the shortest stereocilia were quite distinct. The kinocilium was located behind the longest stereocilia. The number of stereocilia on the utricular hair cells in the normal animals was ca. 50 (50. 5±11. 5sd (n = 23;) 8 animals), the large variation (37–76) a likely reflection of different populations of bundles associated with the two different hair cell types [34,35] (but this was not analysed). In inner ears with anomalous hair bundles in the cochlea, hair bundles in maculae were substantially different from those of normal littermates. As on affected cochlear hair cells, stereocilia were significantly shorter than normal (Fig 7B and 7C): at P35 no stereocilium exceeded 3. 5μm in height. Affected bundles were rounded in shape (Fig 7B, 7C and 7D) and the stereocilia entirely encircled the kinocilium which, unlike the kinocilia of affected cochlear hair cells, was located almost centrally (Fig 7B, 7C and 7E). There was thus no bundle orientation. In some cells the kinocilium was very long (Fig 7B) but in many others it was shorter than the longest stereocilia, so that it did not emerge above the stereociliary bundle (Fig 7B and 7C). Some bundles showed a small gradient in the height of stereocilia from the periphery towards the centre, i. e. the longest stereocilia were those closest to the kinocilium (Fig 7B, 7C and 7F), and tip links were arranged along this gradient (Fig 7F). However, unlike the affected bundles on cochlear hair cells, the number of stereocilia comprising each affected macular hair bundle, was the same as normal (52. 9±8. 9; range 38–79, n = 25/6 animals). In vestibular maculae, hair cells are generated into the early postnatal period such that at P1-P2 all stages in the maturation of hair bundles are exhibited. In the utricular maculae of normal unaffected animals at this age (Fig 7G, S4E Fig) in many of the most immature bundles, the kinocilium was located eccentrically and “behind” the bundle of short stereocilia which even at this stage showed distinct height gradients. A change in the shape of the hair cell surface and of the shape of the hair bundles from rounded to the more elongate contour of the mature bundle (Fig 7A) was apparent co-incident with relocation of the kinocilium. In the utricular maculae of affected animals of the same age (Fig 7H; S4C, S4D and S4F Fig) always the bundle occupied almost the entire apical surface of the cell with the kinocilium located centrally at the cell’s surface encircled by a cluster of stereocilia of almost equal height. The kinocilium never became re-located to an eccentric position behind a clearly polarised bundle as occurs during normal maturation of macular hair cells [7]. The difference between the normal and affected bundles was particularly clearly seen by stereoimaging (S4E and S4F Fig). The hair bundles of hair cells in the cristae appeared essentially unaffected in inner ears where bundles of cochlear and macular hair cells were abnormal. As early as P1 stereocilia were arranged in rows of ascending height and the direction in which the stereocilia increased in height, bundle orientation, was essentially the same for all hair cells (Fig 8A). The kinocilium appeared behind the longest streocilia in most bundles though sometimes could be seen to emerge from in front of the longest stereocilia (Fig 8B) suggesting a more central position and some variability in the relationship between the siting of the longest stereocilia and the position of the kinocilium. Nevertheless, the hair bundles of the cristae in affected animals showed both polarisation and orientation. By P3 (Fig 8C) the stereocilia (and the kinocilium) had grown rapidly to the same great lengths as in normal animals while retaining pronounced height gradients, so that at all ages examined up to 1 year, the hair bundles in the cristae of animals where cochlear and macular hair bundles were abnormal were indistinguishable from those in the cristae of normal littermates (Fig 8D and 8E). Brief exposure of the organ of Corti of affected animals to the styryl dye FM1-43, which is taken up into hair cells via the transduction channel at the tips of the stereocilia [36], resulted in fluorescent labelling of both IHC and OHC (Fig 9A) in an identical manner to normal hair cells [36], indicating that despite the bundle anomalies the stereocilia bore active transduction channels. Immunolabelling for the actin crosslinking protein espin, which when defective or absent causes anomalies of hair bundle morphology and stereociliary length [37], was present in the stereocilia of morphologically disrupted hair bundles (Fig 9B). Thin sections showed the filamentous actin in stereocilia was tightly packed as in normal stereocilia (Fig 9C and 9D) and “top connectors” [38] between the shafts of stereocilia at their apical ends were present (Fig 9D inset). These observations suggest that the defect did not affect the structure of individual stereocilia. Thin sections of the organ of Corti also suggested that morphological anomalies both of IHC and of OHC were confined to hair bundle organisation. The cuticular plates, upon which the stereocilia are mounted, were of the same size and apparently of the same density of actin cross-linking as in normal cochlear and macular hair cells (Fig 9C and 9D). Stereociliary rootlets running through the proximal end of each stereocilium into the cuticular plate were prominent (Fig 9C and 9D). In the maculae of older animals with hair bundle anomalies, the length of stereocilia was greater than that in young mature animals; by P200 some stereocilia reached 4. 5μm. Also, from ca. 3 months of age, an increasing number of hair bundles contained very long, thick structures indicating fusion and elongation of stereocilia and fusion of stereocilia with the kinocilium (Fig 10A–10E). Similar fused, elongated stereocilia were present in the maculae of older normal animals (Fig 10F) and have been described in other studies of age-related effects upon macular hair cells [39]. Likewise, stereocilia in the anomalous bundles of cochlear IHC, particularly those in the apical coils, appeared to increase in length with age; by 6 months the height of longer stereocilia in the bundles was ca. 5 μm (mean 4. 8μm; n = 12 bundles) (Fig 10G). Elongated, fused stereocilia, similar to those in the maculae, also appeared in the anomalous hair bundles of cochlear IHC in older affected animals (Fig 10H and 10I) as well as in their normal litter mates where stereociliary fusion amongst IHC was evident from as early as 3 months of age (Fig 10J and 10K). However, neither increase in length nor fusion of stereocilia was obvious in the cochlear OHC in the older animals. With ageing, in animals with hair bundle anomalies there was an increasing number of OHC in which shortened and stump-like stereocilia were evident at the outer edges of the bundle (Fig 10L). This suggested a continuing retraction of stereocilia into the cell and a decrease in the number of stereocilia in the bundle. By one year of age (Fig 10M), the mean number of stereocilia in the bundles of apical coil OHC in affected animals was about half of that at P15-P30,27. 4±5. 4 (n = 30 from 3 animals). Since polarity and bundle orientation were disrupted in the organs of Corti and vestibular maculae of affected animals, the presence and distribution of core PCP [16,40] and intrinsic bundle polarity pathway proteins [23,25] were investigated by immuno-labelling these tissues from early postnatal animals. In the organ of Corti of the normal animals, the core PCP protein Vangl2 was localised in an asymmetrical pattern at the junction between hair cells and supporting cells (S5 Fig), as described elsewhere [40]. Vangl2 immunofluorescence was also present in the affected animals (S5 Fig), where it localised in a comparable pattern to that in normal animals, suggesting core PCP signalling was largely unaffected. There were differences between control and affected animals, however, in the expression of proteins regulating the intrinsic hair bundle polarity pathway [23,25] (Figs 11–13). In the normal organ of Corti LGN was localised asymmetrically, in a distinct lateral compartment at the apical surface of IHC and OHC (Fig 11A and 11A’), and it was also detectable at the tips of the longest stereocilia of OHC (Fig 11A’). In affected animals, immuno-labelling for LGN was almost entirely absent from the organ of Corti, save for a peri-centrosomal density in a few IHC (Fig 11B and 11B’). Immuno-labelling for Gαi3 localised to the same compartment as LGN in the normal organ of Corti (Fig 11C and 11C’), but was absent from the organ of Corti of affected animals (Fig 11D and 11D’). In normal animals, aPKC expression was restricted to a medial compartment at the apical surface of IHC and OHC (Fig 11E and 11E’). aPKC was present in the cochlear hair cells of affected animals, but it was evenly distributed around the entire circumference and across their apical surface (Fig 11F and 11F’). LGN was detected in a lateral compartment on the apical surface of utricular hair cells of normal animals (Fig 12A and 12A’), but this asymmetric labelling was absent from hair cells of affected animals (Fig 12B and 12B’). Gαi3 was localised to a comparable lateral compartment in normal animals (Fig 12C and 12C’), but also was absent in the utricular hair cells of affected animals (Fig 12D and 12D’). aPKC expression was localised to one hemisphere of utricular hair cells of normal animals (Fig 12E and 12E’), and although aPKC was present in the utricular hair cells of affected animals it was evenly distributed around the entire circumference and across their apical surface (Fig 12F and 12F’). In cristae ampullares (Fig 13), the distribution of these proteins in normal animals, and their disruption in affected animals were comparable to those observed in utricular hair cells. Thus, disruption of SorCS2 is associated with loss or mis-expression of intrinsic polarity pathway proteins in the organ of Corti, vestibular maculae and cristae. SorCS2 is widely expressed in the CNS [30,42]. One role is as a co-receptor with p75NTR for pro-neurotrophins (NTs) mediating extracellular signals regulating actin dynamics to induce growth cone collapse and terminate neurite outgrowth [31,33]. Genome-wide association studies have linked SORCS2 as a risk factor for ADHD [43] which has been related to the role of the protein as a proNT receptor [33]. Likewise its proNT binding function may underlie an association of SorCS2 with epileptic seizures in mice [44]. However, members of the Vps10 family, to which SorCS2 belongs, also function in intracellular signalling and protein trafficking, sorting cargo from endosomes to the trans-Golgi-network or to the cell surface [45,46]. Genetic variation in SORCS2 has been associated with Alzheimer’s disease where the protein may affect the intracellular processing of amyloid precursor protein [46]. Nucleotide polymorphisms in SORCS2 have also been associated with circulating levels of insulin-like growth factor binding protein 3 (IGFBP3) which regulates the level of IGF-1 in plasma [47]. Overexpression of IGFBP3 in mice leads to growth retardation [48]. The possibility that disruption of SorCS2 affects the availability of IGF-1 thereby retarding growth could explain the present observation that the animals that showed hair bundle anomalies were invariably the lightest ones in litters. In humans, SORCS2 is located on the distal short arm of chromosome 4 within a gene-rich region subject to large chromosomal deletions that are associated with Wolf-Hirschhorn syndrome (WHS), a rare multisyndromic disorder [49]. The mouse homologues of the affected genes, including SorCS2, map to a syntenic region on mouse chromosome 5 [50,51]. The nature and severity of the symptoms of WHS vary with the size of the deletion but they include seizures, skeletal defects that affect craniofacial development, growth retardation and hearing loss. While there is conductive hearing loss due to cranio-facial anomalies affecting the outer and middle ear, sensorineural hearing loss has also been recorded in ca. 15% of cases [52]. Recently this has been attributed to mutations in the gene encoding Wolf Hirschhorn Syndrome Candidate 1 (WHSC1) a histone methyl transferase that regulates gene expression [53]. In a mouse model, that mutation is associated with mis-orientation of hair bundles, but also, unlike the disruptions described here, in disturbance of the arrangement of OHC in the basal and middle cochlear turns. We suggest that in cases where its locus is included in the deletion, SORCS2 is a second candidate gene that may underlie the sensorineural hearing loss in WHS through additional effects on hair bundle organisation. A consequence of disruption of SorCS2 is loss of expression or mis-localisation of all the intrinsic polarity pathway proteins in hair cells, indicating SorCS2 likely plays a role in events upstream of their deployment. Whether it is acting as a receptor for some unidentified ligand, or whether it is involved in intracellular processing and/or trafficking of the proteins in hair cells is not known but our observations of cytoplasmic localisation of the protein at the time when hair bundles are forming may indicate a role in trafficking. Since the core PCP protein Vangl2 was asymmetrically localised in the normal pattern in both the cochlea and the vestibular maculae, it is likely that the defects in the intrinsic polarity pathway were the major factor contributing to the hair bundle anomalies. LGN and Gαi3, together with mInsc, are thought normally to form a complex [25]; loss of any one of them reduces, but does not ablate, expression of the others. Loss of expression of any one of them also produces effects that resemble those observed in the present study, but not as extensively since many OHC were only mildly or not affected and severe effects on IHC were not described [25]. The severity and extent of the hair bundle anomalies observed in the present study may, therefore, be because the expression or localisation of all the intrinsic polarity pathway proteins is affected simultaneously. LGN, Gαi3 and mInsc are normally distributed together across the apical surface of the hair cell in the compartment at the “back” (kinocilial) side of the bundle, confining aPKC to the opposite hemisphere “in front” of the bundle [23,25]. These complementary asymmetric patterns of distributions are thought to define the position of the bundle and to regulate its shape, as well as that of the apical surface of the hair cell, in particular that of OHC, through interactions with the acto-myosin cortical cytoskeleton beneath the apical surface plasma membrane [23,25]. Polarised distributions of myosins and f-actin have been reported to guide co-ordinated reshaping of the contour of the apical surface and generation of the precise shape and orientation of the hair bundle that occur during the later stages of maturation of OHC [54]. The present work clearly shows a close relationship between the shape of the outline of the bundle and the contour of the apical surface in all affected hair cell types, most notably in OHC where the contour of the apical surface reflected the particular bundle shape even when it was quite tortuous or even split into separate clusters of stereocilia. Not only does this emphasise the close linkage between bundle shape and hair cell apical surface contour during the re-modelling. It also suggests that the loss of the intrinsic polarity proteins does not impair activity of the mechanisms underlying reshaping the apical surface of hair cells as they mature but leads to mis-regulation of the directions in which the forces underlying it are applied. This has implications for the development of the heads of the supporting cells that surround each hair cell. Normally, as the contour of the apical surface of the immature hair cell changes from round to the mature shape, the heads of the supporting cells undergo coincident remodelling to accommodate the shape changes in the hair cell. With loss or mis-expression of the intrinsic polarity proteins, the remodelling of the heads of the supporting cells accommodated the mis-shaping of the hair cell apex and maintained separation between adjacent hair cells, but occasionally it failed to do so resulting in apparent direct contact between neighbouring OHC. However, the normal maturation of supporting cells that results in systematic increases in the size of the heads of the pillar cells and of Deiters’ cells (the supporting cells in the region of the OHC) from base to apex of the cochlea [55] was unaffected. Clearly, however, the hair bundles of cochlear IHC, cochlear OHC and macular hair cells are each affected differently by the same genetic defect and misexpression of intrinsic polarity proteins. Together with the observation that hair bundles in cristae were largely unaffected despite loss of these proteins. It would seem that there are differences between hair cell types in the way that the molecular machinery which generates a precisely organised hair bundle is deployed. In both outer and inner cochlear hair cells in affected animals at the earliest postnatal ages kinocilia were located eccentrically on the medial side of stereociliary bundles. Very early in the differentiation of cochlear hair cells, the kinocilium that is initially located towards the centre of the cell becomes positioned very close to one edge, before extensive expression of LGN/MInsc/Gαi3 [25] so is likely directed by the core PCP proteins [16] not those of the intrinsic polarity pathway. As differentiation proceeded and the shape of the cell surface changed, in IHC the kinocilium was initially close to the bundle and the edge of the cell, with only a small “bare area” free of microvilli around and behind it. Additionally, while there was some initial reduction in the number of presumptive stereocilia, no bare area was created on the opposite side of the bundle by retraction of stereocilia. Retraction appeared to be terminated early in IHC such that there was a significantly greater number of stereocilia in the bundle of adult animals than normal. The failure to generate the opposing smooth membrane areas either side of the bundle would be consistent with disruption of the projected role of the asymmetric distribution of intrinsic polarity pathway proteins in regulating delineation of bundle shape and position. In OHC, on the other hand, areas of smooth surface membrane were prominent in the region around the kinocilium, defining an “outer” side of the bundle, and on the opposite side through retraction of the supernumerary “minivilli” proceeding in a manner similar to normal as cell surface contour was re-shaped. In consequence, the hair bundle of the OHC retained a number of normal features: it became confined across the central region of the surface and to an almost normal number of rows of stereocilia; and it exhibited a “polarity” as defined by height differentials in one direction across the bundle (albeit streociliary lengths being significantly shorter than normal). Furthermore, in contrast to the situation in IHC, retraction of stereocilia not only proceeded but appeared to be excessive such that the total number of stereocilia on an OHC in adult animals was significantly less than normal, and retraction appeared to continue slowly throughout life, suggesting that the mechanism regulating retraction and defining the final number of stereocilia was not properly terminated. These observations suggest that in OHC there are elements additional to the intrinsic polarity pathway proteins that regulate hair bundle positioning and the number of rows of stereocilia. In contrast to the situation in cochlear hair cells, in macula hair cells the kinocilium remained centrally at the surface surrounded by stereocilia. During normal maturation of stereociliary bundles in macular hair cells, the re-positioning of the kinocilium from the centre to one side of the bundle occurs as the apical surface area of the cell expands and the hair bundle re-shapes[7]. With the translocation of the stereocilia, the kinocilium becomes positioned behind the longest row. It is possible therefore, that normally rather than the kinocilium moving, it is fixed in position while, as the bundle reshapes and the apical surface expands, directed movements of the apical surface membrane around it leads to its eccentric position in relation to the bundle of stereocilia. A similar mechanism could account for the variable location of the kinocilium at the apical surface of IHC which occurred during apical surface re-modelling in affected animals. If the asymmetric distribution of intrinsic polarity proteins normally instructs directed movements of the apical surface membrane then it is possible that the retention of the kinocilium at the centre of the hair cell surface in macular hair cells in the affected animals is because of disruption of the apical surface membrane re-modelling as a consequence of mis-expression these proteins. Although the shapes of bundles and the number of stereocilia of which they were comprised were affected, height differentials amongst stereocilia, though less marked than normal, were evident in individual anomalous bundles. In the bundles of macular hair cells, the longest stereocilia were those in the innermost ring encircling the central kinocilium. Such an organisation would be consistent with a hypothesis that the position of the kinocilium defines the location of the longest stereocilia in a bundle. In OHC differential heights were marked and the longest were in the row on that side of the bundle closer to the position of the kinocilium/basal body. In IHC, height differentials were initially less marked, and sometimes not obvious. With greater maturity, clearly longer stereocilia grew on many, but not all, bundles, with the longest stereocilia most commonly within the middle rather than the lateral (OHC-facing) side, as is normal. The re-location of the longest stereocilia to the middle of the bundle occurred as the shape of the apical surface changed from rounded to a more elongated, oval-like shape. Thus, the movements associated with re-modelling of the apical surface appeared to have led to re-distribution of stereocilia within a bundle, suggesting the position of stereocilia is not initially fixed. Since some of those longer, more centrally positioned stereocilia in a bundle increased markedly in length while others elongated little or not at all, it may imply that the longest stereocilia in the bundle are defined early and regardless of their position maintain the machinery to elongate differentially from their neighbours. Another feature common to the abnormal hair bundles of inner and outer cochlear hair cells and hair cells of the maculae was a significant reduction in the length of stereocilia, without effects on width. Since the normal differential in width of stereocilia between IHC and OHC (and between vestibular hair cell types) was maintained in affected animals this implies that the number of actin filaments that form a stereocilium, and determine it width, is regulated through those genetic networks that define hair cell type separately from those that regulate bundle morphology. Stereocilia grow by the addition of monomers to the plus ends of actin filaments, which are located at the distal tips [3] and is thought to be regulated by a complex of proteins that includes myosin XVa [56]; whirlin, a scaffolding protein [57,58]; and Esp8, which caps elongating actin filaments [59]. Severely shortened stereocilia result from the absence of any one of these proteins. Shortening of stereocilia may also occur with absence of actin cross-linking proteins fascin [60,61] and most notably espin [37,62]. However, labelling for espin was intense in stereocilia of the defective hair bundles of the organ of Corti and the packing of actin filaments within the stereocilia was unaffected suggesting that actin filaments are cross-linked normally. It has been reported (but not illustrated) that hair bundles in cristae, as well as those in maculae and the organ of Corti, show shortened stereocilia in the absence of myoXVa [56] but in the affected animals described here the stereocilia in the cristae grew to their normal great lengths. Thus, it may be that the molecular assembly that includes myoXVa is not directly affected by the mutation. In normal hair bundles, LGN and Gαi3 localise to the tips of the longest stereocilia [25,63] and they may be additional players in the macromolecular assembly that regulates the length of stereocilia. Thus, the loss of expression of LGN and Gαi3 from inner ear sensory epithelia in the affected animals could be a contributory factor to the shortened stereocilia in the organ of Corti and maculae. However, if this is so then these proteins may not play a role in regulating stereociliary lengthening in cristae. Previous studies of hair bundle anomalies caused by defects in core PCP proteins, intrinsic polarity pathway proteins or in proteins involved in the formation of individual stereocilia, have not drawn particular attention to possible differences in the effects on the different sensory epithelia of the inner ear. Strikingly, although cristae normally appear to express SorCS2 and intrinsic polarity pathway proteins as in other sensory patches, the loss of these proteins did not have a significant effect on the morphology of their hair bundles when those in vestibular maculae and the organ of Corti were severely defective. Cristae are evolutionarily ancient in the vertebrate line—jawless fish possess only two cristae as well as a single common macula [64,65]—and it could be argued they are functionally the simplest of the sensory epithelia in the inner ear since their role is solely vestibular whereas in bony fish and amphibia the saccular macula, and in some fish species the utricular macula as well, have a role in audition in addition to detecting position in space [66]. Furthermore, hair bundles in an individual crista are all oriented in the same direction and appear to be of essentially uniform morphology (except, in mammals at least, for differences in stereociliary width, associated with the two hair cell types, and length related to their location across the epithelium). In contrast bundles of macular hair cells show more than one orientation and often more than one distinct morphological type [67]. Thus, the machinery generating a bundle in the crista may be in its simplest form whereas the requirements for more precise bundle organisation with the acquisition of greater functional specialisation in maculae and auditory epithelia may have led to increasing/additional levels of regulation during hair bundle formation. It may be that cristae provide a better model than other sensory patches to explore the basic molecular machinery by which hair bundle polarity and orientation are established. The work involving animals was performed in accordance with guidelines laid out by the British Home Office and with commitment to the application of the principle of replacement, reduction and refinement in the use of animals in research. In accordance with this commitment the number of animals used was the minimum needed to achieve the results sought; and the procedures used were refined as much as possible to minimise suffering of the animals used. The work with animals formed part of a project that had been approved by the UCL Animal Welfare and Ethical Review Body (AWERB) prior to the award of a Project Licence from the British Home Office, No. PPL70/8144, issued in accordance with United Kingdom Animal (Scientific Procedures) Act of 1986 that gave approval of the work undertaken. The initial colony of mice consisted of Tie2-cre transgenic animals that were crossed with animals containing floxed CNP (c-type natriuretic peptide), both strains on a C57Bl/6 background. The colony included all possible homozygous and heterozygous genotypes of these animals as well as wild types. They were obtained from Prof Adrian Hobbs (Pharmacology Dept UCL, now at Queen Mary University of London) and were bred and maintained in the UCL Biological Services Unit. All procedures involving the use of animals were approved by the UCL Animal Ethics Committee and were performed under the terms of a project licence granted by the British Home Office. Screening of animals was performed by the “ear twitch response” to an 18. 5kHz tone burst delivered from a “click box” [68] which initial ABR analyses showed was a reliable indicator of affected animals. Animals that showed no response were selected as “affected” animals and subsequent morphological analyses of their inner ears, or those of their offspring, were used to confirm phenotypes. In all further work, phenotypic assessment of one ear from each animal was used to identify affected animals. Early postnatal animals were from the litters of breeding pairs the genotypes of which had been determined from assessment of responses to click-box testing and the Mendelian ratios of the phenotypes in previous litters. ABRs were recorded blind to phenotype. Animals aged P17- P35 were anaesthetized and placed in a sound isolated chamber. Subdermal needle electrodes (Rochester Electro-Medical) were inserted at the vertex (active), mastoid (reference) and with the ground needle electrode in the hind leg. Recordings were obtained using TDT system3 equipment and software (Tucker-Davis Tech. , Alachua FL). Responses to click stimuli and to tone pips at 8,12,24,32 and 40 kHz were recorded and threshold determined by the lowest level at which the ABR waveform could be recognised. The auditory bullae of mice of various ages from P0 to 1 year were isolated and the cochleae exposed. To provide access for fixative to the inner ear tissues, an opening was made at the apex of the cochlea, the vestibule was widely opened at the round and oval windows and small cuts were made through the lateral wall of the cochlea. Fixative was slowly injected into the inner ear via the openings at base and apex of the cochlea and then the opened bulla was immersed in fixative. Inner ears were also obtained from animals of embryonic ages E15-E18. Their heads were bisected longitudinally and placed in fixative. The fixative was either 2–4% paraformaldehyde in phosphate buffered saline (PBS) prior to immunohistochemical labelling, or 2. 5% glutaraldehyde in 0. 1M cacodylate buffer, pH7. 3 with 3mM CaCl2 in preparation for electron microscopy. Fixation continued for a maximum of 2 hours at room temperature. The bullae from animals older than 10 days were decalcified in 4% EDTA in either PBS or cacodylate buffer, pH7. 3, as appropriate, for no more than 48 hours, before further processing. For scanning electron microscopy (SEM) organs of Corti, utricular and saccular maculae and cristae were dissected from the bullae and postfixed in 1% OsO4 in cacodylate buffer. Samples were then processed through the thiocarbohydrazide-OsO4 repeated procedure [69] before dehydration in an ethanol series, critical point drying and mounting on SEM sample stubs with conductive silver paint. A thin (2-3nm) layer of Pt was applied by sputter coating before examination. For transmission EM (TEM) of thin sections in some cases tannic acid at 1% was added to the glutaraldehyde fixative. Bullae were trimmed to remove excess material from around the cochlea and vestibular system, which were partially opened, and the bullae were then processed intact without isolation of the inner ear tissues. The samples were post-fixed in 1% OsO4, partially dehydrated in an ethanol series to 70% ethanol, incubated overnight at 4°C in a saturated solution of uranyl acetate in 70% ethanol before completing dehydration and embedding in plastic. Thin sections of the entire cochlea were cut. From SEM images, the number of stereocilia in individual IHC and OHC and the surface areas of these cells were obtained from organ of Corti samples tilted and rotated in the microscope to provide views from as close as possible 90° to the apical surface of the hair cells (i. e. from directly above) for surface area or additionally for counting stereocilia towards the inner aspect of the bundles, when all stereocilia in a bundle are visible. Images were collected at a nominal consistent magnification of 4000x, which provided for all three rows of OHC to be included in a single image. The height and width of IHC and OHC stereocilia were measured from images of samples tilted and rotated to view the stereocilia as closely as possible perpendicular to their long axes to reduce the effects of parallax on measurements. Images were obtained at a nominal consistent magnification of 10000x. Microscope magnification was calibrated with a cross-grating replica. To enable comparisons between animals in the location along the organ of Corti from which the quantitative data was obtained, the measured width of the inner pillar cell head between the IHC and first row of OHC, and that of the OHC region from the medial border of the first row of OHC to the lateral border of the 3rd row OHC were used as surrogates for position; the width of the pillar cell heads and the width across the OHC region increase systematically from base to apex along the cochlea [55] and thus provide an indicator of relative position along the organ of Corti. Generally three images for each assessment at each location for each animal were obtained from 5 different litters. Counting and measurements were made from the micrographs with the aid of AnalySIS image analysis software. Statistical comparison was by t-test. Results are presented as mean ± standard deviation, and significance level was set at p<0. 05. Organs of Corti, utricular maculae and cristae were dissected from the bullae, after decalcification when used, and prepared for whole mount examination. Entire opened bullae were prepared for vibratome sectioning and sections 150μm thick were cut. Whole mount samples or sections were permeabilised and blocked with 0. 1% Triton X-100 and 10% normal goat serum in PBS for 30 minutes at room temperature, before overnight incubation at 4°C with primary antibodies diluted in 100 mM L-lysine in PBS solution. Following extensive washing in PBS, samples were incubated with appropriate Alexa-conjugated secondary antibodies (Invitrogen, Paisley, UK) diluted at 1: 400 in 100 mM L-lysine in PBS solution for 1 hour at room temperature. Fluorescently-tagged phalloidin (Sigma) was added to the secondary antibody solution at 1 μg/ml. The samples were mounted using Vectashield (Vector Laboratories) containing DAPI to label nuclei. Samples were imaged using a laser scanning confocal microscope (LSM 510; Zeiss, Jena, Germany) or by wide-field fluorescence microscopy. The primary antibodies used were: rabbit polyclonal anti-espin (1: 100; kind gift of James Bartles, Northwestern University, Chicago, USA); rabbit polyclonal anti-Alms1 (1: 100; gift from David Wilson, University of Southampton, UK); rabbit polyclonal anti-LGN (1: 400; gift from Fumio Matsuzaki, RIKEN Center for Developmental Biology, Kobe, Japan); rabbit polyclonal anti-Gαi3 (1: 400; Sigma Aldrich, Poole, UK; G4040); rabbit polyclonal anti-atypical Protein Kinase C (1: 400; Santa Cruz Biotechnology Inc, Heidelberg, Germany; sc-216, C-20); rabbit polyclonal anti-SorCS2-CT (cytoplasmic tail) (1: 1000; gift from Simon Glerup, Aarhus University, Denmark); sheep polyclonal anti-SorCS2-ED (extracellular domain) (1: 50; R&D Systems, Abingdon, UK; AF4237); rabbit polyclonal anti-Vangl2 (1: 500; gift from Mireille Montcouquiol, INSERM, Bordeaux, France). Segments of organ of Corti were isolated from the cochleae of animals aged P10 under Hepes buffered Hanks Balanced salt solution (HBHBSS). They were immersed in 3μM FM1-43 (ThermoFisher Scientific) for 20 secs, immediately transferred to HBHBSS and washed three times before examination by confocal microscopy. Tail snips from animals that showed the abnormal phenotype were assessed. Genomic DNA was extracted. This was fragmented to ~300bp and was processed for Illumina Next Gen DNA sequencing by standard library preparation protocols. The resulting library was DNA sequenced in one lane of an Illumina 2500 sequencer and yielded ~100 million paired end reads, each of 100 bases. The Fastq files from this were converted to a Fasta database. These were searched by BLAST for sequences homologous to the transgene (the Tie2 promoter, Cre cDNA, MT-1 polyA signal sequence, and Tie2 intron 1 enhancer). All of the resulting detected paired end sequence reads were extracted and further analyzed. Auditory bullae were obtained from mice aged P0-P4. One bulla from each mouse was removed to RNA-later (Qiagen) and stored at -80°C until analysis. The opposite ear from each of these animals was fixed in paraformaldehyde and processed to determine the inner ear phenotype. Nine heterozygote and 9 homozygote animals were included in gene expression analysis. The qPCR analysis was performed blind to the phenotype. RNA was extracted using an RNeasy kit (Qiagen), treated with RQ1 RNase-free DNase (Promega), and cDNA made using Omniscript reverse transcriptase (Qiagen) and using random primers (Promega). Relative levels of SorCS2 gene expression were determined by RT-qPCR using Taqman® gene expression assays from Thermo Fisher Scientific [assay ID: Mm00473051_m1 and Mm00473072_m1] to detect the 5’ and 3’ regions of the major predicted transcript respectively, and Myo7a (assay ID: Rn00596450_m1). Reactions were multiplexed with a primer-limited eukaryotic 18S rRNA endogenous control, performed in triplicate for each animal and amplified on a SDS7500 real-time PCR System (Thermo Fisher Scientific). Relative quantification studies were performed using SDS1. 2. 1 software using the 2-ΔΔCT method (Thermo Fisher Scientific).

Sensory “hair” cells in the inner ear derive their name from an organised bundle of mechano-sensory “stereocilia” on their apical surface. The finger-like stereocilia are composed of actin filaments and increase in height staircase-like in one direction across the cell surface, a system which provides a defined polarity across the sensory epithelium. The establishment of hair bundle polarity and orientation is a highly regulated process involving several signalling pathways. There are subtle differences in bundle morphology between hair cell subtypes, suggesting local variations exist within these regulatory pathways. In a transgenic mouse colony we found profoundly deaf animals displaying severe hair bundle abnormalities on all auditory hair cells. Hair cells in the gravity sensing tissues of the balance system also had abnormal bundles, but there was no effect on bundles of hair cells sensing rotation. Hair bundle abnormalities coincided with mis- expression of proteins of an intrinsic apical asymmetry pathway. These defects were associated with disruption of the SorCS2 gene, and consequent mis-expression of the coded transmembrane receptor in the inner ear. The results suggest that SorCS2 regulates hair cell development, and that there are previously unrecognised differences in the way subtypes of hair cells form their stereociliary bundles.

Abstract Introduction Results Discussion Materials and methods

cell physiology medicine and health sciences integumentary system ears neuroscience cell polarity outer hair cells organ of corti inner ear immunologic techniques research and analysis methods hair animal cells head inner hair cells immunohistochemistry techniques cellular neuroscience anatomy cell biology cochlea neurons biology and life sciences cellular types afferent neurons histochemistry and cytochemistry techniques

2017

Disruption of SorCS2 reveals differences in the regulation of stereociliary bundle formation between hair cell types in the inner ear

DNA methylation is an epigenetic mark that is associated with transcriptional repression of transposable elements and protein-coding genes. Conversely, transcriptionally active regulatory regions are strongly correlated with histone 3 lysine 4 di- and trimethylation (H3K4m2/m3). We previously showed that Arabidopsis thaliana plants with mutations in the H3K4m2/m3 demethylase JUMONJI 14 (JMJ14) exhibit a mild reduction in RNA-directed DNA methylation (RdDM) that is associated with an increase in H3K4m2/m3 levels. To determine whether this incomplete RdDM reduction was the result of redundancy with other demethylases, we examined the genetic interaction of JMJ14 with another class of H3K4 demethylases: LYSINE-SPECIFIC DEMETHYLASE 1-LIKE 1 and LYSINE-SPECIFIC DEMETHYLASE 1-LIKE 2 (LDL1 and LDL2). Genome-wide DNA methylation analyses reveal that both families cooperate to maintain RdDM patterns. ChIP-seq experiments show that regions that exhibit an observable DNA methylation decrease are co-incidental with increases in H3K4m2/m3. Interestingly, the impact on DNA methylation was stronger at DNA-methylated regions adjacent to H3K4m2/m3-marked protein-coding genes, suggesting that the activity of H3K4 demethylases may be particularly crucial to prevent spreading of active epigenetic marks. Finally, RNA sequencing analyses indicate that at RdDM targets, the increase of H3K4m2/m3 is not generally associated with transcriptional de-repression. This suggests that the histone mark itself—not transcription—impacts the extent of RdDM. Cytosine DNA methylation is an epigenetic mark that is conserved across all kingdoms of eukaryotes. Depending on its location in the genome, DNA methylation can be broadly classified as either genic or non-genic. Genic—or gene-body—methylation has been observed in several species of plants and animals, and generally correlates with transcriptionally active loci [1]–[3]. Conversely, non-genic methylation is associated with transcriptional repression at repetitive elements such as transposons [4]. Both plants and animals also have examples of non-genic DNA methylation repressing protein-coding gene transcription when the mark is present in the gene' s regulatory regions [5], [6]. In the model plant Arabidopsis thaliana, gene-body methylation is exclusively found in the CG dinucleotide context and is maintained by METHYLTRANSFERASE 1 (MET1), the plant ortholog of mammalian DNA Methyltransferase 1 (DNMT1) [7]. In contrast, non-genic methylation is maintained by at least four methyltransferases: MET1, CHROMOMETHYLASE 2 (CMT2), CHROMOMETHYLASE 3 (CMT3), and DOMAINS REARRANGED METHYLTRANSFERASE 2 (DRM2). The four methyltransferases have distinct cytosine contexts that they preferentially act upon: CG (MET1), CHG (CMT3) and CHH (CMT2 and DRM2), where H is any base that is not a G [8], [9]. In addition to these well-characterized context preferences there is also a degree of redundancy for maintenance of non-CG methylation between CMT3 and DRM2 [8], [10], as well as CMT2 at some loci [9]. Furthermore, while all of the methyltransferases act in DNA methylation maintenance, only DRM2 is required for establishment of DNA methylation in all three sequence contexts [11]. In Arabidopsis, DNA methylation is correlated with specific histone marks that vary depending on the context and genomic location of the DNA methylation. Gene-body DNA methylation, for example, is largely co-incidental with histone 3 lysine 4 monomethylation (H3K4m1) [12]. Conversely, non-genic DNA methylation is strongly enriched in regions of histone 3 lysine 9 dimethylation (H3K9m2) [13]. Non-genic methylation is also inversely correlated with H3K4m2/m3—a mark that is associated with the 5′ ends of RNA Polymerase II (Pol II) genes [12], [14]. The link between H3K9m2 and CHG DNA methylation has been well established with regards to the CMT3 pathway: CMT3 binds to H3K9m2 through its eponymous chromodomain, as well as its bromo-adjacent homology (BAH) domain [15], [16]. Null mutant lines for the H3K9m2 histone methyltransferases recapitulate the cmt3 DNA methylation phenotype, which illuminates the tight correlation between the two marks [7]. Links between histone modifications and the DRM2 pathway are also emerging. DRM2-dependent methylation depends on two plant specific RNA polymerases: RNA Polymerase IV and V (Pol IV and V). Pol IV generates a transcript that is processed into 24-nucleotide small interfering RNAs (siRNAs), and Pol V produces a transcript that serves as a scaffold for ARGONAUTE 4 (AGO4) loaded siRNAs that are generated by Pol IV [17], [18]. This dual-RNA polymerase system targets DRM2 to methylate DNA, although the specific mechanism for the targeting is not yet clear. Recent evidence suggests that Pol IV occupancy requires a factor, SAWADEE HOMEODOMAIN HOMOLOG 1 (SHH1), which is a dual histone modification sensor, preferentially binding to histones containing H3K9 methylation as well as lacking in H3K4 di- or trimethylation [19], [20]. We and others previously showed that mutation of the H3K4m2/m3 demethylase JMJ14 causes a partial reduction of DRM2-dependent RdDM, but does not affect the MET1 or CMT3 pathways [21], [22]. Since the observed decrease in DNA methylation correlated with a partial gain of H3K4 methylation, we concluded that H3K4m2/m3 might negatively impact RdDM. In this report, we tested whether the modest DNA methylation reduction phenotype of the jmj14 mutant might be due to redundant activity of other histone demethylases. Arabidopsis contains a family of H3K4 demethylases distinct from JUMONJI proteins known as LYSINE-SPECIFIC DEMETHYLASE 1-LIKE (LDL). We show that mutation of two partially redundant members of the LDL family, LDL1 and LDL2, causes a DNA methylation phenotype that is similar to jmj14, and that the jmj14 ldl1 ldl2 triple mutant shows an enhanced methylation-loss phenotype. Interestingly, like the jmj14 single mutant [21], the jmj14 ldl1 ldl2 triple mutant reduced the maintenance of RdDM, but did not affect the establishment of DRM2-mediated methylation. Genomic analysis showed that the histone demethylase mutations only affect methylation at a subset of RdDM targets and that these targets are close to protein-coding genes. These results suggest that the JMJ14 and LDL histone demethylases reinforce RNA-directed DNA methylation near genes by counteracting nearby activating H3K4 epigenetic marks. We previously screened T-DNA insertional mutant lines in genes containing JmjC histone demethylase domains to determine whether perturbations in histone modifications might influence the establishment or maintenance of DNA methylation [21]. These results showed that mutation of the JMJ14 gene reduced DRM2-mediated DNA methylation, but did not affect the MET1 or CMT3 pathways, and the effects were correlated with increased H3K4 di- and trimethylation. Interestingly, the DNA methylation reduction was not as strong as that observed in drm2 mutants, suggesting the possibility that JMJ14 might be acting redundantly with other histone demethylases [21]. Lysine Specific Demethylase 1 (LSD1) is a well-characterized H3K4 demethylase in mammals [23], [24], and Arabidopsis contains four LSD1 homologs termed LDL1, LDL2, LDL3, and FLOWERING LOCUS D (FLD). Biochemical analysis suggests that LDL1 is exclusively an H3K4 demethylase with preference for mono- and dimethylation [25], and a previous report described LDL1 and LDL2 as partially redundant H3K4 demethylases that reduced DNA methylation at the FLOWERING WAGENINGEN (FWA) gene [26]. It should be noted however that, even though mammalian LSD1 only demethylates H3K4 in vitro, it has both H3K4 and H3K9 demethylase activity in vivo [23], [27]. Thus we cannot rule out the possibility that LDL1/LDL2 may have more diverse biological functions in planta. We observed that the jmj14-1 mutant shows reduced DNA methylation and increased H3K4 methylation at FWA to about the same degree as that reported for ldl1-2 ldl2 double mutants [21], [26]. To study possible genetic interactions between the two families of demethylases, we generated a jmj14-1 ldl1-2 ldl2 triple mutant line. Consistent with the data reported by Jiang et al. , we did observe an increase in H3K4 dimethylation and trimethylation (m2/m3) in the ldl1-2 ldl2 double mutant (Figure S1A) [26]. However, while Jiang et al. reported a CG methylation defect at the FWA repeats, we did not observe any such effect by bisulfite sequencing analysis (Figure 1A). Rather, we only observed a reduction in non-CG methylation that is much more similar to that observed in jmj14-1 (Figure 1A). Consistent with a defect in RdDM, this same study did show a reduction in non-CG methylation at the FWA transgene in ldl1 ldl2 double mutants [26]. In analysis of other RdDM targets, we observed a similar phenomenon. At the MEDEA-INTERGENIC SUBTELOMERIC REPEATS (MEA-ISR), there was no reduction in MET1-dependent CG methylation, and a decrease in non-CG methylation, once again, similar to that observed in jmj14-1 (Figure 1B). We also analyzed the AtSN1 transposon using a quantitative PCR (qPCR) based assay in which we digested genomic DNA with the HaeIII endonuclease that is sensitive to CHH methylation at three restriction sites within the amplified region (Figure 1C). We observed increased digestion in the ldl1-2 ldl2 double mutant, the jmj14-1 single mutant, and the jmj14-1 ldl1-2 ldl2 triple mutant. We also analyzed the Ta3 locus by bisulfite sequencing (Figure 1D). Ta3 is methylated by MET1 and CMT3, but not DRM2 [8]. Similar to jmj14-1, the ldl1-2 ldl2 double mutant showed no impact on Ta3 methylation. Prior to our initial study describing JMJ14, all mutations that caused a reduction in the maintenance of RNA-directed DNA methylation were also found to be required for the establishment of DNA methylation [28]. In order to examine the requirements of DNA methylation establishment, we take advantage of a transgenic version of the FWA gene. FWA is a homeodomain transcription factor with tandem repeats in its 5′ UTR. In unmethylated epialleles, the FWA gene is ectopically expressed, causing a delay in flowering time [29]. Unmethylated transgenes introduced into wild-type plants are recognized by the RdDM machinery, and methylated and silenced [11], [30]. However, in mutants such as drm2 that are unable to establish DNA methylation, transgenic FWA expression leads to late flowering. Previous FWA transformation assays on the jmj14-1 mutant showed the surprising result that flowering time and DNA methylation establishment were not affected [21], even though jmj14-1 reduces the maintenance of RdDM at the FWA locus. To test whether the other histone demethylase gene mutations might affect de novo methylation of FWA, we transformed ldl1-2 ldl2 and jmj14-1 ldl1-2 ldl2 with FWA and scored for flowering time (Figure S2). Despite previously published evidence that LDL1 and LDL2 were required for DNA methylation establishment, we observed that each untransformed mutant line exhibited a flowering-time phenotype that deviated only slightly from wild type [26], [31]–[33]. More importantly, the flowering time after FWA transformation was comparable to the slight delay also observed in wild-type Col-0 plants, showing that none of the mutations caused a block in de novo silencing of FWA. However, one cannot completely exclude the possibility that de-repression of FLOWERING LOCUS T (FT) in the jmj14 mutant background [31]–[33] may partially offset ectopic FWA expression. While it is not clear why the histone demethylase mutations cause decreases in the maintenance of RdDM but do not affect methylation establishment, it is possible that the nature of chromatin at the time of DNA methylation establishment (early zygotic development [34]) is such that histone demethylases are not required at this particular stage. To ascertain the global impact on DNA methylation of the various histone demethylase mutants, we performed whole genome shotgun bisulfite sequencing (BS-Seq). We analyzed the data in parallel with those generated from drm2-2 mutants and nrpe1-11 mutants (the largest subunit of Pol V) in order to draw a direct comparison with canonical RdDM factors [10]. Since the RdDM pathway primarily impacts CHH methylation, we defined differentially methylated regions (DMRs) for CHH context methylation in each mutant (Table S1; Figure 2). Although JMJ14 and LDL1/LDL2 appear to have some preferential targets, there was a large degree of overlap observed by comparing their respective DMRs. Moreover, the greatest number of DMRs appear in the triple mutant, suggesting that JMJ14 and LDL1/LDL2 act in a mostly redundant fashion to maintain DNA methylation (Figure 2A). To determine whether the histone demethylase mutants have a generally weak DNA methylation defect at all RdDM sites or if they might be acting more strongly at only a subset of RdDM targets, we assayed CHH methylation levels at DMRs defined in the demethylase mutants and compared these with DMRs in a strong RdDM mutant, nrpe1-11 (Figure 2A and B). We observed a strong loss of CHH methylation in the demethylase mutants which was enhanced in the triple mutant. However, we observed only a slight loss of CHH methylation in the demethylase mutants at the bulk of canonical RdDM sites (Figure 2A and B), suggesting that demethylases do not serve as general effectors of RdDM, but act in a locus specific manner. Furthermore, we found that the DMRs in the jmj14-1 ldl1-2 ldl2 triple mutant overwhelmingly overlapped with the nrpe1-11 DMRs (Figure 2A), strongly suggesting that DNA methylation defects in the histone demethylase mutants are mainly limited to RdDM targets. We further analyzed the CHH methylation defects in the histone demethylase mutants by heat-map analysis of the DMRs. As expected, drm2-2 mutants exhibited virtually a complete loss of CHH methylation at all NRPE1 sites (Figure 2C). Consistent with our other analyses, the histone demethylase mutants uniformly displayed a much more minor methylation loss at most sites. Conversely, at DMRs defined in jmj14-1 ldl1-2 ldl2, there was a dramatic loss in CHH methylation in drm2-2 and nrpe1-11 (Figure 2C). In addition, at the histone demethylase DMRs, there were many examples of synergistic effects between the jmj14-1 single mutant and the ldl1-2 ldl2 double mutant, with a loss of CHH methylation only apparent in the jmj14-1 ldl1-2 ldl2 triple mutant (Figure 2C). Taken together, these data strongly suggest that histone demethylases act in a partially redundant fashion to control RNA-directed DNA methylation at a subset of loci. To determine the nature of the histone demethylase mutant DMRs and to see if they had unique characteristics compared to DMRs in canonical RdDM mutants, we further analyzed them relative to annotated genes, based on our initial observations at individual loci that suggested that they tended to be closer to protein-coding genes. When we calculated the distance of the DMRs to protein-coding genes, we found that the average distance is significantly smaller for jmj14-1 ldl1-2 ldl2 DMRs than for drm2 or nrpe1 DMRs (Figure 2D, Figure S3). It is known that JMJ14 and LDL1/LDL2 regulate non-DNA methylated protein-coding genes through their H3K4 demethylase activity [21], [26], [31]–[33]. Thus a likely explanation for these results is that transposons or other silent elements in proximity to actively H3K4-demethylated genes might be more sensitive to mutations in H3K4 demethylases than other sites. Given that two families of histone demethylases appear to functionally overlap at a large number of RdDM targets, we sought to use our sets of genomic data to better understand their genetic interaction. Our DMR identification analysis suggests that JMJ14 and LDL1/LDL2 have some sites of preferential action (Figure 2A). However, further analyses of these data indicate that at a majority of identified demethylase DMRs, both the jmj14-1 single mutant and the ldl1-2 ldl2 double mutants reduce CHH methylation to varying degrees that are enhanced in the jmj14 ldl1-2 ldl2 triple mutant (Figure S4). Thus the number of overlapping regions of DNA hypomethylation for the histone demethylases reported in Figure 2A is likely an underestimate, as some regions were omitted for not meeting our thresholds for calling DMRs in a given genotype, yet still show a subtle methylation effect in that genotype. In addition, these data indicate a high degree of overlap in the sites affected in jmj14 and ldl mutants and, given the general enhancement seen in the triple mutant, suggest that these demethylases act in a largely additive fashion with regards to RdDM sites in the genome. The exceptions to this interpretation are those small number of DMRs unique to either the jmj14-1 or ldl1-2 ldl2 genotypes which show CHH methylation defects that are not strongly enhanced in the demethylase triple mutant (Figure S4). This suggests a low level of locus-specific preference for either class of demethylase that is as yet not understood. In mammals, it is known that LSD1 can exist in both the CoREST and NuRD repressive complexes in which they perform different activities [35], [36]. Similarly, LDL1/LDL2 and JMJ14 could exist in different complexes depending on the locus of action, and thus have differential effects on DNA methylation in a locus-specific manner. This possibility will deserve further investigation in future studies. In order to further understand the relationship between H3K4 methylation and DNA methylation at RdDM targets, we performed chromatin immunoprecipitation (ChIP) against H3K4 methylation marks in the suite of histone methylation mutants. At three individual loci analyzed, FWA, MEA-ISR, and AtSN1, we observed only moderate gains in H3K4m2/m3 in the demethylase mutants (Figure S1). A comparison of these data with the DNA methylation data at these three loci (Figure 1) suggests that even slight increases in H3K4m2 or H3K4m3 are associated with reduced RNA-directed DNA methylation at these sites. To gain a more global perspective on the relationship between DNA methylation and H3K4 methylation, we performed ChIP against H3K4m2 and H3K4m3 marks coupled with high-throughput sequencing (ChIP-seq). Across DMRs defined in nrpe1-11 and jmj14-1 ldl1-2 ldl2, H3K4 di- and trimethylation were depleted (Figures 3A). This result was expected given a previous study demonstrating that H3K4 di- and trimethylation and DNA methylation are anti-correlated [12]. In addition, in the jmj14-1 ldl1-2 ldl2 triple mutant we observed an increase of H3K4 di- and trimethylation specifically at DMRs defined from the jmj14-1 ldl1-2 ldl2 triple mutant BS-seq data (Figures 3A and C, Figure S5). These results show that genomic regions experiencing the largest alterations of DNA methylation levels in the histone demethylase triple mutant mutants generally showed the largest increase in H3K4m2/m3. Although we did see some increases in H3K4m2 and H3K4m3 in the jmj14-1 and/or in the ldl1-2 ldl2 mutants, the affects were more variable and were not as strong as those seen in the jmj14-1 ldl1-2 ldl2 triple mutant (Figure 3B and D). We also wanted to rule out the possibility that the increase in H3K4 methylation might be an indirect result of DNA hypomethylation. Therefore we also performed H3K4m2 and H3K4m3 ChIP-seq in drm2-2, nrpe1-11, as well as an upstream RdDM mutant nrpd1-4 (the largest subunit of Pol IV), and analyzed the H3K4 methylation profile at both nrpe1-11 and jmj14-1 ldl1-2 ldl2 DMRs (Figure S6). While the loss of CHH methylation at these DMRs is dramatic in nrpe1-11 and in drm2-2 mutants (Figure 2B), there was little increase of H3K4 methylation in these mutants, but significant gain of methylation in the triple histone demethylase mutant (Figure S6). Therefore, we conclude that the increase in H3K4 methylation antagonizes RdDM, and not vice versa. Finally, these ChIP data provide further insight into the nature of the small number of specific DMRs found in the histone demethylase mutants compared to canonical RdDM factors (Figure 2A). As we showed above, the DMRs in the triple mutant are on average closer to genes (Figure 2D). Consistent with this, we observed a higher level of H3K4 methylation flanking the midpoint of DMRs defined in jmj14-1 ldl1-2 ldl2 mutants than in those defined in nrpe1-11 (Figure 3A and C). This was true in wild-type plants and demethylase mutants, showing that the jmj14-1 ldl1-2 ldl2 DMRs are in regions that are closer to highly H3K4 methylated areas, which are primarily represented by protein-coding genes. These data suggest that JMJ14 and LDL1/LDL2 act at a specific subset of RdDM targets to prevent accumulation of H3K4 methylation in silent regions, which would otherwise antagonize the DNA methylation machinery. In more gene-poor regions, histone demethylases would not be required, thus the mutants do not display a DNA methylation phenotype. Even though a majority of RdDM targets are not over genic regions but are instead upstream of genes in promoter regions, we reasoned that one possible explanation for the gain of H3K4 at demethylase DMRs would be aberrant transcription at these sites in the demethylase mutants. Indeed, many RdDM targets are transposable elements that may be competent for transcription in a demethylase mutant background. Such aberrant transcription by Pol II could potentially displace Pol IV or Pol V activity, resulting in reduced DNA methylation targeting. To test the hypothesis that demethylase mutant effects on RdDM were caused indirectly by a transcription-based mechanism, we performed a comparative transcriptome analysis by RNA sequencing (RNA-seq) between wild type and mutants (Figure 4). Strikingly, we observed no increase in transcription at DMRs defined in either nrpe1-11 or in jmj14-1 ldl1-2 ldl2 (Figure 4A). In addition, we did not observe increased transcription of protein-coding genes nearest the defined DMRs (Figure 4B). Thus the increase of H3K4 methylation in the proximity of DNA methylated regions does not affect transcriptional activity of these regions. Together, these data indicate that it is not transcription per se that affects RdDM, but more likely the nature of the chromatin itself. RdDM is an siRNA-driven process and previous studies have shown that siRNA profiles in different DNA methylation mutants can help position these mutations in the RdDM pathway. We tested whether mutations in H3K4 demethylases impacted siRNA accumulation by generating small RNA (sRNA) libraries and performing high-throughput sequencing (sRNA-seq). At DMRs defined in nrpe1-11, there was only a very small decrease in sRNA levels in the three histone demethylase mutant lines analyzed (Figure 5A). However, there was a marked reduction of sRNA levels at jmj14-1 ldl1-2 ldl2 DMRs (Figure 5A). Heat map analysis similarly showed that at the great majority of DMRs from the jmj14-1 ldl1-2 ldl2 triple mutant, there was depletion of sRNA reads, and that at many DMRs, there was nearly a complete loss of sRNAs (Figure 5B). Mutations in certain RdDM components have been shown to exhibit locus specific reductions of siRNAs, including mutation in SHH1, as well as mutation of downstream RdDM factors such as DRM2 and NRPE1 [20]. SHH1 encodes a factor that facilitates recruitment of Pol IV to a subset of loci, and shh1 mutants show losses of siRNAs at largely the same subset of loci as do drm2 and nrpe1 mutants [20]. To analyze whether there is an overlap between the loci affected in these other RdDM components and the histone demethylase mutations, we included in our analysis sRNA-seq data from shh1-1, drm2-2, and nrpe1-11, as well as nrpd1-4, which is a mutation in the largest subunit of Pol IV (Figure 5A). We observed a nearly total loss of siRNAs from the jmj14-1 ldl1-2 ldl2 triple mutant DMRs in all the RdDM mutants tested (Figure 5A), indicating these sites require SHH1 and downstream RdDM factors for normal RdDM pathway function and suggesting that a common mechanism may regulate siRNA levels at these sites. In this study we have described the relationship between H3K4 methylation and DRM2-mediated DNA methylation through the study of histone demethylase mutants. Genome-wide ChIP and bisulfite sequencing analyses show that H3K4m2 and H3K4m3 marks antagonize the RdDM pathway at a large number of sites in the genome, and these sites are enriched near the 5′ end of K4-methylated protein-coding genes, which suggests that the activity of histone demethylases is more crucial for DNA methylation integrity nearby H3K4m2/m3 rich promoters. Our transcriptome data also indicate that H3K4m2 and H3K4m3 marks per se, and not associated transcriptional changes, act to prevent complete RNA-directed DNA methylation. Together, these data raise interesting questions about the mechanisms by which active chromatin marks like H3K4 methylation may affect the stable maintenance of repressive DNA methylation. In mammals, Dnmt3L—the binding partner of the de novo methyltransferase Dnmt3A—specifically binds to unmodified H3K4 (H3K4m0) [37], [38]. Although it cannot be ruled out that a convergent mechanism evolved in plants, there is no evidence for such a relationship between H3K4m0 and DRM2, and DRM2 lacks the Plant Homeodomain (PHD) found in Dnmt3L that recognizes H3K4m0. Because regions that lose methylation in jmj14 ldl1 ldl2 triple mutants also showed a reduction in siRNA levels, we propose that SHH1 recruitment of Pol IV may be a link that can explain the effect of histone demethylases on RNA-directed DNA methylation. SHH1 interacts with Pol IV, and is required for 24-nt siRNA biogenesis at a large subset of RdDM targets [19], [20], including at those sites that lose methylation in jmj14-1 ldl1-2 ldl2 triple mutants (Figure 5A). Structural and biochemical data indicate that SHH1 binds to H3K9m2 through its tandem-tudor-like SAWADEE domain. Interestingly, SHH1 is also inhibited from binding to histone tails in vitro when H3K4m2/m3 is also present. Therefore, we propose that the DNA methylation defect observed in H3K4 demethylase mutants may be due to impaired SHH1 binding, reduced Pol IV recruitment, and reduced siRNA biogenesis. In this way, JMJ14 and LDL1/2 can serve to reinforce methylation of silent transposons and other repeated sequences that are nearest to protein-coding genes. All plants utilized in this study are in the Col-0 ecotype, and grown under long day conditions. The following mutant lines were used: jmj14-1 (SALK_135712), ldl1-2 (SALK_034869), ldl2 (SALK_135831), drm2-2 (SALK_150863), nrpd1-4 (SALK_08305), nrpe1-11 (SALK_02991), and shh1-1 (SALK_074540C). All whole-genome sequencing datasets were submitted to the Gene Expression Omnibus (GEO) database and are accessible as part of the GSE49090 accession. For sodium bisulfite sequencing, DNA was treated using the EZ DNA Methylation Gold kit (Zymo Research) by following the manufacturer' s instructions. Amplified PCR fragments from each analyzed locus were cloned into pCR2. 1-TOPO (Invitrogen) and sequenced. We analyzed 15 to 22 clone sequences per sample using Lasergene SeqMan software. In order to distinguish the FWA transgene from the endogene, we destroyed a BglII restriction site in the transgenic copy in the region of PCR amplification. We then bisulfite treated genomic DNA of transgenic plants following a BglII digestion (37°C, overnight), which prevented amplification of the endogenous gene. Additionally, the transgenic copy of FWA was derived from the Landsberg ecotype, thus we could distinguish between the transgene and endogene based on the existence of three single nucleotide polymorphisms within the amplicon in case BglII digestion was not complete. Primers used for amplification are listed in Table S2. Analysis of asymmetric methylation at the AtSN1 locus was performed exactly as described in [21]. Primers used for amplification are listed in Table S2. Analysis of non-CG methylation at AT5G35935 was performed by extracting DNA from young flowers using a standard Cetyl trimethyl ammonium bromide protocol. A total of 200 ng of genomic DNA was digested overnight at 37°C with MspI side-by-side with samples containing buffer and no enzyme (undigested). Quantitative real-time PCR validation of uncut DNA after MspI digestion was performed using the Bio-Rad Synergy Brands Green SuperMix on an MX3000 Stratagene cycler. The PCR parameters are as follows: one cycle of 10 min at 95°C, 40 cycles of 30 s at 95°C, 1 min at 55°C, and 1 min at 72°C. PCR primers sequences are listed in Table S2. Transgenic plants were generated as described in [39]. We measured flowering time of plants as the total number of leaves (rosette and cauline leaves) developed by a plant at the time of flowering. Plants transformed with the FWA transgene were selected for by spraying with a 1∶1000 dilution of Basta soon after germination. ChIP assays were performed as described in [40] with modifications. For immunoprecipitation, the following antibodies were used: H3K4m2, Abcam AB32356; H3K4m3, Diagenode pAb-003-050. Primers used for amplification of ChIP targets are listed in Table S2. Total RNA was prepared using a standard Trizol extraction from 0. 5 grams of 3-week-old plant aerial tissue. 4 µg of total RNA was then used to prepare libraries for Illumina sequencing, following the Illumina TruSeq RNA Sample Prep guidelines. Multiplexed samples were sequenced at 50-nt length on an Illumina HiSeq 2000 instrument. ChIP of the Col, jmj14, ldl1 ldl2, and triple mutants shown in Figure 3 was performed as described above. Libraries were generated as described in [41]. For the ChIP-seq analysis of RdDM mutants done in parallel with Col and the demethylase triple mutant (Figure S6), ChIP was carried out as described in [20] using 10-day-old seedlings and the following antibodies: H3K4m2, Millipore 07-030; H3K4m3, Millipore 04-745. Subsequent ChIP-seq libraries were generated as described in [20]. Genomic DNA was extracted from one gram of 3-week-old plant aerial tissue using a DNeasy Plant Maxi Kit (Qiagen). Libraries for bisulfite sequencing were generated and sequenced as described in [42], with the change that sequencing was carried out on an Illumina HiSeq 2000 instrument. Small RNAs were purified from Trizol-purified total RNA by fractionation with one volume of 25% PEG 8000, followed by gel purification of ∼15–30-nt RNA species from a 20% polyacrylamide gel (7M urea). The initial total RNA was isolated from 100 mg of immature floral buds, and the resulting siRNA fraction was resuspended in 10 uL TE buffer, all of which was used for library generation. sRNA-seq libraries were generated using the small RNA TruSeq kit (Illumina) following the manufacturer instructions with the exception that 15 cycles were used during the amplification step. Sequenced reads were base-called using the standard Illumina pipeline. For ChIP-seq, mRNA-seq and BS-seq libraries, only full 50-nt reads were retained. For sRNA-seq, reads had adapter sequence removed and were retained if between 18 and 28-nt in length. For ChIP-seq and mRNA-seq, only uniquely mapping reads, allowing for 1 mismatch, were mapped to the Arabidopsis genome (TAIR10 – www. arabidopsis. org) with Bowtie [43] and retained for further analysis. For sRNA-seq, both unique and non-unique reads were mapped to the genome, allowing for one mismatch, and for downstream read density calculations, only the unique reads were considered with the total reads (unique+non-unique) being used for normalization purposes between libraries. For BS-seq libraries, reads were mapped using the BSseeker wrapper for Bowtie [44]. For ChIP-seq, mRNA-seq, and BS-seq, identical reads were collapsed into one read to avoid optical or PCR based duplicates. For the calculation of read density in sRNA-seq libraries where duplicate reads may be of biological significance, up to 100 identical reads were retained as distinct reads in any given region, with any reads exceeding that flattened to the 100 read cap. For methylation analysis, percent methylation was calculated as described [10]. DMRs were defined using the bsseq package of the R-based BSmooth pipeline [45]. For the purposes of defining CHH hypomethylated DMRs, only cytosines with 2× coverage in a majority of the wild-type libraries as well as the mutant library in question were utilized after data smoothing. Variances were estimated from the wild-type library group, and initial identified DMRs were filtered for t-statistics <−2 or >2. These filtered DMRs were then further filtered by keeping only those covering > = 20 assayed CHH context cytosines and a mean difference > = 0. 1. Finally, the DMRs were filtered once more and only those with an area statistic = <−100 or > = 100 were retained for the final set of DMRs. For each mutant genotype the corresponding BS-seq library was compared to four wild-type (ecotype Col-0) libraries. One of these wild-type libraries is submitted as part of the current study GEO record (GSE49090), while the other 3 wild-type replicates were previously published and are thus part of other GEO records (GSE39901 – listed as “WT replicate 2”; GSE38286; GSE36129). For Figure 2, the wild type represented is the Col-0 replicate from the GSE36129 study as these plants were grown side-by-side with the demethylase mutants. For all libraries and analyses, the list of mRNA used, along with genomic coordinates, were obtained from TAIR (TAIR10). For all analyses, overlap was considered to > = 1 bp overlap of defined regions. All statistical analysis was conducted within the R environment.

A number of factors contribute to the organization of eukaryotic genomes and the expression state of the underlying genes. For example, cytosine bases can be modified with the addition of a methyl-group. In the model plant Arabidopsis thaliana, methylated cytosines are typically associated with transcriptionally repressed regions—so called “heterochromatin. ” Additionally, genomic DNA is wrapped around nucleosomes; each nucleosome consists of a complex of eight histone proteins. In turn, amino acid residues on histone proteins can be modified by a number of means, one of which is methylation. A methyl modification on lysine four of histone three (H3K4) is associated with transcriptional activation. Genome-wide studies in Arabidopsis have previously shown that DNA methylation and H3K4 methylation are highly anti-correlated. In this paper we examine a set of Arabidopsis mutants in which H3K4 methylation is abnormally high at a number of loci in the genome. At several of these loci, DNA methylation levels are decreased in the same mutants. These data suggest that H3K4 methylation antagonizes DNA methylation, which may contribute to mechanisms that distinguish active from silent regions of the genome.

Abstract Introduction Results Discussion Materials and Methods

2013

Interplay between Active Chromatin Marks and RNA-Directed DNA Methylation in Arabidopsis thaliana

Antigenic variation is employed by many pathogens to evade the host immune response, and Trypanosoma brucei has evolved a complex system to achieve this phenotype, involving sequential use of variant surface glycoprotein (VSG) genes encoded from a large repertoire of ~2,000 genes. T. brucei express multiple, sometimes closely related, VSGs in a population at any one time, and the ability to resolve and analyse this diversity has been limited. We applied long read sequencing (PacBio) to VSG amplicons generated from blood extracted from batches of mice sacrificed at time points (days 3,6, 10 and 12) post-infection with T. brucei TREU927. The data showed that long read sequencing is reliable for resolving variant differences between VSGs, and demonstrated that there is significant expressed diversity (449 VSGs detected across 20 mice) and across the timeframe of study there was a clear semi-reproducible pattern of expressed diversity (median of 27 VSGs per sample at day 3 post infection (p. i.), 82 VSGs at day 6 p. i. , 187 VSGs at day 10 p. i. and 132 VSGs by day 12 p. i.). There was also consistent detection of one VSG dominating expression across replicates at days 3 and 6, and emergence of a second dominant VSG across replicates by day 12. The innovative application of ecological diversity analysis to VSG reads enabled characterisation of hierarchical VSG expression in the dataset, and resulted in a novel method for analysing such patterns of variation. Additionally, the long read approach allowed detection of mosaic VSG expression from very few reads–the earliest in infection that such events have been detected. Therefore, our results indicate that long read analysis is a reliable tool for resolving diverse gene expression profiles, and provides novel insights into the complexity and nature of VSG expression in trypanosomes, revealing significantly higher diversity than previously shown and the ability to identify mosaic gene formation early during the infection process. Antigenic variation is used by many pathogens as a means of staying one step ahead of the host’s adaptive immune response. Trypanosoma brucei has long been a paradigm for the study of antigenic variation, and the protein responsible, the variable surface glycoprotein (VSG) has been the focus of much research [1–3]. Each trypanosome in a population expresses a single species of protein, and an inherent, parasite-driven switching process causes a proportion of the population to replace their active VSG gene with a different VSG gene, resulting in the expression of a protein in those cells with different epitopes exposed to the host immune system (at a rate of up to 10−2 switches per cell/generation [4]). The post-genomic era has revealed T. brucei’s antigenic variation system to be unrivalled in its elaboration, particularly in terms of the scale of the numbers of genes that comprise the VSG family. Sequencing the genome of T. brucei has uncovered a gene family much greater in numbers and complexity than was previously thought. Characterisation to date suggests that at least 2,000 VSG genes are in the genome of each trypanosome, providing a spectacularly large repertoire of potential antigens [5,6], particularly when compared to other pathogens that undergo antigenic variation, such as Plasmodium Falciparum (60 genes in PfEMP1 family [7]), Anaplasma marginale (~10 members in the msp2 & msp3 gene families [8]), and Borrelia burgdorferi (15 members in the vls gene family[9]). The scale of the gene family size is also reflected in the complexity of switching mechanisms employed to change the identity of the surface antigen. The VSGs are expressed from one of approximately 20 bloodstream expression sites (BES) [10], the active expression occurring in a dedicated sub-nuclear organelle, the expression site body (ESB) [11], with the remainder of BESs being transcriptionally silent. A minor mechanism of VSG switching, accounting for only approximately 10% of events in wild type trypanosomes [12], is to turn off the transcription of the active BES and activate one of the silent BESs (‘transcriptional switching’). However, the majority of switching is through replacing the gene sequence in the active BES via gene duplication, which involves the copying of variable amounts of sequence, ranging from within the gene to the whole telomere [13,14]. Insights into mechanisms involved in switching suggest that replacing expressed VSG sequence is driven by DNA recombination, and DNA repair/homologous recombination pathways and proteins (e. g. RAD51) have been identified to be involved in the gene duplication process of VSG switching [15] (reviewed in [16]). A further layer of complexity is the construction of novel VSG sequences in the BES from multiple donor VSG sequences, a form of segmental gene conversion termed ‘mosaic’ gene formation [17,18]. Mosaic gene formation was previously considered to be a rare and minor mechanistic component of overall VSG switching in an infection (e. g. [14]). However, the revelation upon the sequencing of the T. brucei genome that a significant proportion of the VSG repertoire (80–90%) consisted of pseudogenes [19] that cannot be expressed as functional proteins began to alter that perception [5,20]. It has become clear from subsequent experimental work that mosaic gene formation is an integral component of VSG switching, particularly after the early stages of infection (i. e. beyond the first peak of parasitaemia in mouse infections) [5,21]. One of the challenges of analysing VSG expression in vivo, and in particular gaining an accurate measurement of the level of expressed diversity given the extent of the VSG repertoire (i. e. to what extent is the repertoire actually used during infection), has been the relatively limited resolution of available techniques–in particular the manual cloning and sequencing of individual VSG cDNAs that has been undertaken in recent studies [5,21]). While this approach clearly provides accurate data at the level of individual VSG transcripts, the limitations have undoubtedly resulted in a low estimate of the diversity and complexity of VSG expression at the population level, and particularly with respect to minor variant populations. Additionally, although transcriptomics potentially provides the ability to overcome the resolution limitations of manually cloning and sequencing transcripts, the application of RNAseq to VSG expression from in vivo samples has long been deemed challenging, due to the requirement for assembling multiple closely related gene variants from a mixed population using short reads of 100–200 base pairs (e. g. Illumina) –this has similarly been an issue when attempting to resolve, for example, the diversity of the mammalian immunoglobulin gene repertoire underpinning the antibody response (e. g. [22]). However, a recent study subjected in vivo samples to Illumina sequencing (100bp, single-end reads) and demonstrated the utility of transcriptomics in terms of increased resolution [23], and were able to detect minor variants (0. 1% of population) and up to 79 variants at a time point, although they were not able to identify significant mosaic gene expression. Long read sequencing potentially provides the ability to further increase our resolution, particularly as the length of reads commonly reached with such technologies (average read length in Pacbio, for example, is quoted as 10,000–20,000 bp; http: //www. pacb. com/smrt-science/smrt-sequencing/read-lengths/) far exceeds the length of the VSG transcript (approximately 1600 bp), meaning that the issue of assembly of closely related VSGs from multiple reads should be bypassed. Here, we present analysis of VSG expression from replicate in vivo T. brucei TREU927 infections in mice at 4 time points over 12 days using almost 500,000 Pacbio Sequencing reads. We demonstrate that long read technologies provide significant advantages for analysing the diversity of VSG expression. Our data suggest that the VSG population comprises significantly more variants even at an early stage of infection (up to 190 variants at day 10 post-infection), that the pattern of VSG expression is surprisingly reproducible (using the novel application of ecological diversity indices), and that mosaic gene expression can be detected much earlier in infection than has been possible previously. Our data also provide insights into the nature of mutations introduced by Pacbio long-read sequencing technology, as the dataset includes significant coverage of one sequence (>140,000 reads). Using PacBio long read RNA sequencing of 20 blood samples enriched for VSG transcripts from replicate in vivo T. brucei TREU927 infections in mice at 3,6, 10 and 12 days post infection, we obtained 486,343 reads with an average read length of insert of 1,569 bp (Table 1, Fig 1B). Reads were filtered by length (1400-2000bp) based upon both literature on VSG genes [21,24] and the read distribution in our dataset (Fig 1B) to remove reads resulting from sequencing artefacts and shorter fragments (i. e. partial reads), and on the basis of similarity to known VSGs (blastn ≥60% alignment against TriTrypDB-v26 [25]–note that the reads include both N-Terminal and C-Terminal domain sequences) (Fig 1C), resulting in a dataset of 296,937 ‘VSG’ reads. Of the reads that were of the appropriate length (1400-2000bp) but did not have ≥60% match to VSGs in the reference database (n = 102,940), 90,810 (88. 2%) mapped partially to VSGs, 3,513 (3. 4%) mapped to non-VSGs, and 8,617 (8. 3%) did not produce any match to the TREU927 reference genome. Within the dataset of 296,937 VSG reads, each read on average represented the consensus sequence from 6. 50 passes of the full length fragment by the DNA polymerase (‘full passes per read’; summarised in Table 1; full data in S1 Table), and for each of these reads there was robust identification of a donor gene for the N-Terminal domain (NTD); therefore, for these 296,937 reads we have high confidence that they contain all of the features necessary to be consistent with being full length VSG transcripts. The 296,937 reads represent a total of 449 VSGs (74. 77% of VSG a-type and 25. 22% VSG b-type [24]) across 20 samples, with the number of reads per VSG following a power-law distribution (Fig 1D), and provide a unique insight into the in vivo VSG transcriptome across time and animal replicate. ORFs were identified in the 296,937 reads with a conservative minimum nucleotide size of 1200 nucleotides (reported size ranges of VSG NTDs and C-Terminal domains [CTDs] are approximately 300–350 and 100 amino acids, respectively [5,21,24]). Surprisingly, only 33,234 reads (11%) resulted in predicted ORFs. Although the percentage of reads with predicted ORF increased with increasing number of full passes, it remained well below 50% even for reads having 10 full passes or more (Fig 2A). Since the distribution of the number of reads with a detectable ORF over all VSGs was similar to total expression level distribution (Table 2), we hypothesize that the lack of identified ORFs was due to random sequencing errors rather than any systematic biases in the data, despite PacBio claiming an accuracy of more than 99% for reads with 15-fold coverage [26]. To investigate this hypothesis in more detail, we focused on the most abundant VSG (Tb08. 27P2. 380,1551bp, 141,822 high-confidence reads) and annotated each discrepant base pair of each aligned read as either an insertion, deletion or mismatch with respect to the Tb08. 27P2. 380 reference genome sequence. All reads had an alignment score greater than 90% over the first 1266bp (the N-Terminal domain) (Fig 2B). The distribution of sequence errors showed a clear bimodal pattern across the N-Terminal domain, with 145 nucleotide positions having a consistent mismatch (131), deletion (10) or insertion (4) across more than 80% of the reads, and 1,112 nucleotide positions having errors in at least one but fewer than 2% of reads (Fig 2B). This suggests that the former represent genuine mutations already present in our inoculum (with respect to the reference genome sequence), whereas the latter represent either random sequencing errors introduced by Pacbio or low level genuine mutations that we cannot currently distinguish from Pacbio error. Previous studies have indicated accumulation of mutations over time in expressed VSGs, and we examined this in our data for reads aligning to Tb08. 27P2. 380 (for the N-terminal domain) by assessing the error rate for mismatches, insertions and deletions (S1 Fig). While these data indicated statistical support for differences in the data distribution across time points for all 3 mutation classes, due to the skewed nature of the data distribution (most bases have an error rate close to zero) this conclusion must be treated with a degree of caution. The assertion that the errors present in >80% of reads were ‘genuine’ mutations was further supported by these 145 mutations being consistently present in PCR amplicons sequenced by Sanger sequencing. These PCR amplicons had been generated from cDNA extracted from multiple samples (n = 7 for Tb08. 27P2. 380; representing sequences independently cloned and sequenced from 4 mice on days 3 and 10, S2 Fig). Insertions were the most common Pacbio-introduced error (average per-base error rate of 0. 79% across the N-terminal domain sequence), followed by deletions (0. 73%) and mismatches (0. 33%) (Fig 2C), in agreement with what has been reported before [27]. Consistent with the ORF prediction pattern (Fig 2A), the overall error percentage was lower for reads with higher number of passes, but introduced sequencing errors (i. e. interpreted as mutations not present in the genome of the inoculated trypanosomes) remained present at more than 1000 nucleotide positions even for reads with 10 passes (Fig 2D). The nature of our data, comprising >141,000 reads of the same sequence, therefore provides an unusually robust insight into the nature of Pacbio errors and the caveats that must be placed upon interpretation of such data, as most studies involve much less coverage per single base pair. Our data demonstrate that we can detect multiple VSGs in each sample, and that we can identify changes in VSG expression and diversity over time. We identified a median of 27 unique VSGs per sample at day 3 post infection (p. i.), which progressed to 82 VSGs at day 6 p. i. , peaking at 187 VSGs at day 10 p. i. and reducing to 132 VSGs by day 12 p. i. (Fig 3A). When identified VSGs that mapped to single reads from single samples were removed, this resulted in an identification of 334 VSGs (median of 27,81,170 and 126 VSGs per sample at 3,6, 10, and 12 days p. i. , respectively). Not only were the number of distinct VSGs consistent across samples for the same time point, but the expression pattern (proportion of reads per sample mapping to particular VSGs) was also highly reproducible between samples and over time (Fig 3B), albeit bearing in mind that these analyses are of batches of mice at four different time points rather than longitudinal samples of the same mice. The VSG that is dominant at day 3 (Tb08. 27P2. 380), presumably introduced as the dominant VSG in the inoculum, remains dominant in all mice at day 6, but is the single VSG with the most reads aligned in only two of five mice at day 10. Interestingly, by day 12, the VSG with the most reads per sample is the same in all five mice (Tb09. v4. 0077) and this VSG was also most common at similar timepoints in previous analyses [21]. Additionally, the other eight VSGs that reads map to in mice at days 10 and 12 (Tb927. 4. 5730, Tb927. 10. 10, Tb11. v5. 0932, Tb927. 9. 300, Tb09. v4. 0088, Tb05. 5K5. 330, Tb927. 9. 16490 and Tb927. 3. 480; Fig 3B) are present in all ten mice suggesting a degree of conservation in the sequential expression of VSGs in independent infections, consistent with previous observations [21,28,29]. However, in all mice there were reads that mapped to VSGs distinct to these most favoured 10 VSGs (‘others’ in Fig 3B, which account for 10. 36% of all VSG-mapped reads), and in some mice this proportion was particularly high (e. g. mice 3. 5,6. 1 and 10. 4; Fig 3B). This is particularly evident at day 6, where although the dominant VSG (Tb08. 27P2. 380) makes up most reads, the majority of reads that do not map to Tb08. 27P2. 380 map to VSGs other than the other top 9 VSGs in all mice. Additionally, at Day 10 we observe both the greatest number of VSGs and the least domination by any single VSG, but the proportion of ‘others’ either reduces or remains stable. These analyses combine to indicate that while there is a broad predictability in expression, with dominant VSGs at the beginning and end of infections, in between these timepoints there is a degree of stochasticity in the system–although eight VSGs comprise the majority of reads that do not map to either of the two dominant VSGs, the relative proportion of these ‘minority’ VSGs is not consistent, and there are furthermore many VSGs that are expressed at very low levels in all mice. The analysis described thus far (Fig 3) has not taken into account any sequence similarity between VSGs, but relied on mapping reads to identified VSGs in the reference database. To analyse the population diversity of VSGs within and across samples using a method that is independent of mapping to existing databases (which are likely to be incomplete), we applied information theoretic measures more commonly used to quantify the biodiversity of ecosystems [30]. This approach initially applied a clustering algorithm to a proportion of reads (n = 33,205; comprising reads with predicted ORF) in order to enable identification of the reads that clustered on the basis of sequence similarity, as putatively distinct VSGs (Fig 4A). These data showed significant congruity with those described for the VSG mapping approach described above (Table 2). The top 10 clusters comprised 89. 34% of all reads, compared to 89. 68% for the VSG mapping approach, and the relative proportion of reads that either map to VSGs or cluster by sequence similarity is very comparable for the 10 most abundant VSGs (Table 2). These data indicate that the clustering algorithm applied was robust in terms of identifying individual VSGs, and therefore indicated a very similar pattern of a dominant early VSG, followed by an intermediate period of significant greater VSG diversity, ending up with a second dominant VSG by day 12. The sequence similarity data also allowed the analysis of variability between mice using a new measure of population differentiation called normalised beta diversity [30] (Fig 4B). When looking at a single day, beta diversity is the effective number of distinct VSG profiles present on that day, giving information on the differentiation between the animals. This analysis indicates (similar to the VSG mapping data) the greatest beta diversity across individuals is at day 10 (Fig 4B solid line). Further exploring each time point and variation between mice (Fig 4B dots), we can see that although the mice at day 3 show some distinct VSG profiles (albeit with overexpressed VSGs in individual mice common to all mice, SI Fig 3), at day 6 most mice (except for mouse 6. 5) are broadly consistent with respect to which VSGs are present and how common they are. The effective number of VSG profiles increases further on day 10 with maximal divergence between mice at any time point, (Fig 4B, solid line). This value then decreases on day 12 (though mouse 12. 2 is distinct), as the mice begin to express similar profiles again. These analyses again indicate that there is stochasticity in the process of VSG expression considered as a progression over 12 days, and there is semi-predictability rather than strict hierarchical progression through VSG expression, as has been described previously [21,29,31,32] Mosaic genes were considered identified where BLAST hits for a particular read demonstrated non-overlapping homology to more than one distinct VSG in the reference database. This was commonly seen in the C-Terminal domain, where the same N-Terminal domain was in many instances observed with different C-Terminal domains (“3’ donation” in [21]). Using pairwise alignments of all reads that mapped to Tb08. 27P2. 380, based on the alignment coverage over the gene, donors were filtered based on the region representing the C-Terminal domain (the 3’ region approximating to 30% of the gene shown in Fig 5A). Donors were selected based on at least 80% alignment coverage to the CTD. These data show that the reads aligning to Tb08. 27P2. 380 consists of three subgroups based on their CTD donors, which are derived from either the reference gene Tb08. 27P2. 380 (43% of all reads), but also from Tb10. v4. 0158 (29%) or Tb927. 6. 5210 (28%). The proportion of the three donor CTDs varies across time points, with the proportion of reads deriving from the donor Tb08. 27P2. 380 gene decreasing by days 10 and 12 (reducing from 46. 55% at day 3 to day 26. 19% at day 12, although the number of reads in total aligning to Tb08. 27P2. 380 is low by days 10 and 12). The frequent nature of this recombination has been observed previously [21]. We detected N-Terminal domain mosaics (within the constraints of our stringent selection criteria) at a much lower frequency (n = 45 over all 20 mice; three sequences at day 3, five at day 6,13 at day 10 and 23 and day 12 –S2 Table), and in most cases these are single read examples, and so must be treated with some caution (albeit 12 of the putative mosaic reads have coverage of at least 7 full passes, a coverage level at which our analysis–Fig 2D–suggests should effectively remove sequencing-derived error). However, we have two examples where we have more than one read indicating N-Terminal domain mosaicism, with the additional support for one of these sequences that it is only detected in one mouse–given the complex nature of previously identified mosaic N-Terminal domains [5,21], it is unlikely that identical mosaics would emerge in separate individual infections. Nevertheless, we do also have one putative mosaic sequence that occurs in two separate mice (balbc_6_0/100673/ccs5 and balbc_12_1/30571/ccs9 in mice 6. 1 and 12. 1, respectively; S2 Table) –this may either represent a gene currently not annotated in the TREU927 genome or be a true mosaic gene that was present in the initial inoculum and has remained at low levels throughout infection. The N-Terminal domain mosaic examples we have detected are mostly relatively simple mosaic genes (e. g. Fig 5B). Although we cannot formally rule out that at least a proportion of these mosaic genes were present in the original inoculum, the increased frequency over time is consistent with expectations that this process is rarest early in infection but becomes more prevalent as infections progress. The results illustrate the power of long read sequencing when applied to expressed gene diversity–we identified 449 VSGs across 20 individual samples, covering four time points post-infection (3,6, 10 and 12 days). The identification of the VSGs was achieved by two approaches; mapping reads to a reference VSG database, and secondly clustering read sequences to identify distinct variants–importantly without reliance upon a reference genome or sequence database. These independent approaches were highly congruent in the number of VSGs and the proportion of reads that were attributed to individual VSGs (Table 2), meaning that the clustering approach may be particularly valuable for analysis of long read data generated from infections with trypanosome strains (or species) where a genome is either not available or is incomplete. When compared with previous approaches, such as manual cloning (801 VSG sequences that comprised 93 distinct VSGs or ‘sets’ across 11 mice across 19 days of sampling each [21]) or short read Illumina sequencing (289 VSGs for 4 mice– 3 mice sampled 9 times over 30 days and one mouse sampled 13 times over 105 days [23]), the Pacbio approach gives significantly higher resolution per sample. It must be acknowledged that in the present study the starting volume of infected blood for each sample was higher (200 μl versus 50–100 μl in [23] and approximately 15 μl in [21]), and additionally the inoculum in the current study was significantly greater and not clonal, meaning the study design may predispose to more expressed variants being detectable. The TREU927 clone used was also highly virulent, giving rise to a high parasitaemia early that was maintained for the 12 days of infection. This is not representative of the classical fluctuating profile of less virulent strains (or clones of this strain, e. g. [33]); however, for the purposes of assessing the utility of Pacbio this was advantageous. A proportion of the identified VSGs (115/449; 25. 6%) derive from single reads in single samples, and therefore a degree of caution must be employed with these variants. However, when the singleton VSGs are removed, we can confidently conclude that we have identified 334 VSGs across our datasets–this ranges from a median of 27 VSGs in day 3 samples to 170 in day 10 samples. Therefore, despite these caveats, we can still conclude that the resolution in terms of diversity is significant for the long read approach, and likely to be of great utility for studies incorporating VSG diversity going forward. Despite the limitations of the study design, where we have analysed batches of mice at four time points rather than longitudinal surveys of individual mice, our data across 20 mice and four time points are very consistent with a highly reproducible pattern of VSG expression over time (Fig 3 & Table 2). There was a remarkable degree of consistency in identity of dominant VSGs across independent infections–particularly as the inoculum used was not a single cell or a cloned inoculum (this is very distinct from, for example, Borrelia, where Pacbio analysis has indicated very little overlap in expressed antigen diversity across replicates from the same starting inoculum [34]). The data demonstrated a consistent emergence of the two sequentially dominant variants at the beginning and end of the infection period (Tb08. 27P2. 380 and Tb09. v4. 0077), although during the period in between the dominant VSGs there was significant diversity in expressed VSGs that was consistent with an inherent degree of stochasticity in the system. This was reinforced by the application of biodiversity analysis (Fig 4), which illustrated the semi-predictable nature of the variant progression across the mice and timepoints. This chimes with previous work that described the semi-predictable expression of VSGs in T. brucei [21,28,29], and modelling approaches that have also reflected semi-predictable use of the VSG repertoire [31,32,35]. When analysing our data set and comparing with that of Hall et al, 2013, who used the same TREU927 strain, we have significant overlap in detected expressed variants. 90% of our reads correspond with a VSG detected in Hall et al. The dominant early VSG is different (corresponding to ‘Set_23’ in the Hall data, Table 2), although the Tb09. v4. 0077 which becomes dominant by day 12 was similarly dominant by ~day 20 in Hall et al; differences are presumably due to the use of either a stabilate with a distinct passage history, or the use of a larger inoculum rather than single trypanosomes (i. e. inoculation of a population from a previous infection presumably expressing the dominant VSG at that particular stage). The dominant VSG in our dataset (Tb08. 27P2. 380) was annotated as a pseudogene in the reference genome (predicted to be truncated due to insertion of a stop codon). The annotation as a pseudogene is not consistent with our data as a dominant early VSG, as it would suggest mosaic gene formation providing a dominant early gene–indeed, recent reannotation has classified this gene as intact, which would be more in keeping with early expression favouring intact over pseudogene or mosaic VSGs [5,21]. However, given the 1 × 105 inoculum used in this study, it is also feasible that the transfer of Tb08. 27P2. 380 as the dominant expressed VSG from the donor mouse infection may have given rise to it being the dominant expressed VSG in the infections analysed. We have identified mosaic genes (classified as reads demonstrated non-overlapping homology to more than one distinct VSG N-Terminal domain in the reference database) earlier in infection than has previously been identified, although we cannot formally rule out that at least some of these were introduced in the original inoculum. The rate of mosaic gene detection was very low in our study, mostly either single or very few reads, which probably reflects our timeframe being only 12 days post-infection; this also pertains if the trypanosomes expressing mosaic VSGs derived from the inoculum, which was also generated over a short duration (5–7 days) in the donor mouse. However, these data do indicate that the nature of the long read sequencing is highly beneficial in terms of mosaic gene identification; even low frequency expressed genes (within the limitation of the four orders of magnitude of coverage that the read number per sample provides) can be identified with some confidence due to the acquisition of the whole gene sequence–in order to achieve this with short read approaches a reasonable degree of read coverage would be required to identify and confirm putative mosaic genes. This has potential implications for the application of long read sequencing to significantly further our understanding of infection dynamics and the role of mosaic genes as infections progress. This is likely to be important in terms of ability to gain insights into the mechanisms of mosaic gene formation because of consequent increased ability to resolve defects in switching rate (e. g. analysis of DNA recombination gene mutants such as RAD51 that have been implicated in DNA recombination-based VSG switching [15]) –at present it is not known if mosaic gene formation involves a mechanistic switch in terms of pathways; the ability to detect low frequency mosaic gene expression should provide the ability to study this. Additionally, detection of low frequency VSGs would enhance the ability overall to more fully analyse the temporal kinetics of VSG switching–providing an avenue for improved quality of inputs into modelling dynamics of VSG expression. The clustering approach developed in this study that does not rely upon a reference database would also make analysing expressed VSG diversity in the animal trypanosomes, T. congolense & T. vivax, feasible—the reference genome (and therefore genomic VSG repertoire) is less well annotated in these species than in T. brucei. One challenge for taking a similar approach in these species is the lack of conserved 3’ UTR sequence in expressed VSGs to enrich transcripts. However, such analyses may be particularly enlightening given the different structure and content of VSG repertoires recently described between the three genomes [36,37], as well as the strikingly different arrangement of VSG expression sites in T. congolense compared to T. brucei [38]. We detected indels consistently when comparing Pacbio transcripts to the reference gene (Fig 2). While these differences may indeed be real, with our protocol we have somewhat limited resolution for conclusively differentiating indels introduced by the trypanosome from those potentially introduced by PCR. However, PCR is unlikely to be the sole cause of the observed mutations, because in the dominant VSG in our dataset (Tb08. 27P2. 380), which represents 141,822 reads across all 20 samples–therefore, 20 independent PCR reactions—we observe a consistent set of variations from the reference genome sequence (145 nucleotide positions across the NTD having a consistent mismatch (131), deletion (10) or insertion (4) across more than 80% of reads with respect to the reference sequence) across all reads–these are consistently present across all reads for this variant, including those reads with high fold coverage (i. e. greater than 10 full passes per read) (Fig 2). These data, across technical and biological replicates, lead us to conclude that these differences were present in most likely the genome copy, but also potentially a distinct BES-resident copy of this VSG that has accumulated mutations distinct to the genome basic copy of the gene, and mutations were not introduced by PCR. One possible explanation for this is that there is very likely a significant (and unknown) divergence in passage history between the sequenced reference genome TREU927 trypanosomes and those used in this experiment. This would be consistent with data from many pathogens of the increased mutability of telomeric/subtelomeric gene families [39]. Previous data indicated accumulation of point mutations in expressed VSGs over time within infections [5,21], and in our data we saw some support for this process, but the skewed nature of the data distribution limits our ability to conclude increased mutations over time as an important aspect of VSG expression (it should be noted that a timeframe of 12 days is relatively short and will have limited our resolution). However, our data indicate that application of long read analysis over longer infection timeframes is likely to be a useful means of characterising the nature and role of this mechanism. However, the multiple mutations that were present across multiple VSG sequences in our data, did enable detailed analysis of the nature of mutations detected in Pacbio sequencing (Fig 2). Ideally, to enable clear differentiation of PCR bias and artefact, errors introduced by Pacbio, and mutations introduced by the trypanosome, unique molecular identifiers (UMIs) would be added prior to PCR amplification (e. g. [22,40]). While we did not incorporate this step, we can draw some conclusions from analysis of our data. When data for Tb08. 27P2. 380, which represents 141,822 reads, is analysed across the range of fold coverage per read, it is clear that most of these mutations are removed as the coverage increases (Fig 2) –although notably even at a high number of passes some introduced mutations remain. This strongly suggests that most of these are errors that are introduced by the Pacbio process, and the proportion we observed across the dataset (insertion 0. 79%, deletion 0. 73%, mismatches 0. 33% per base pair) is consistent with that reported in other studies (e. g. [41]). The mutations also directly influenced the ability to predict open reading frames in our data—ORFs only being detected in 11. 22% of VSG reads (33,234 of 296,937). Clearly, with these reads being generated from cDNA one would have expected most if not all to have identifiable ORFs. Therefore, these data indicate some of the limitations when using Pacbio, even with data that comprises multiple passes–the introduction of mutations does provide a layer of complexity to the analysis that must be addressed with care. This is particularly pertinent when trying to analyse multiple closely related genes, as in the case of VSGs. We were able to draw conclusions on the basis of sufficient coverage of a highly expressed dominant gene, combined with the inclusion of multiple biological replicates; without these elements interpretation would have been very difficult without parallel short read sequencing to correct errors introduced by the technology. A further issue for consideration for the application of long read technologies to the analysis of expressed gene diversity is the number of reads per sample. Our data provided coverage over four orders of magnitude—although significantly greater in resolution than previous manual and laborious methods, this contrasts relatively poorly with the numbers of reads that short read applications deliver (millions). However, it should be noted that with the short read approach many reads will be required to robustly identify full length single variants (in particular to enable differentiation of closely related transcripts, either similar genome-encoded variants or related lineages of mosaic genes [5,21]), whereas in theory at least a single pacbio read should provide the ability to robustly identify a particular VSG transcript. While the coverage is being improved with the newer platforms (e. g. the Pacbio Sequel potentially delivers a further tenfold increase in data per run), this may limit resolution in terms of detecting minor variants, for example. We did detect significant expressed diversity, and this is partly explained by our use of a relatively large inoculum, which was not cloned, of a virulent isolate that resulted in high and sustained parasitaemia. Therefore, we started with what was probably a relatively diverse population (albeit dominated by expression of Tb08. 27P2. 380), reflected in the diversity of VSGs detected at day 3 post-infection, which would be significantly lower in the event of a clonal or smaller initial inoculum. While our data indicate that long read sequencing provides increased resolution in terms of identifying VSG diversity, clearly questions still remain. For example, why the VSG repertoire is so evolved and large? Our data suggest an increased proportion of repertoire is involved, even at early stages, compared to previous studies, which indicates a bigger proportion of the repertoire may be utilised during the lifetime of an infection (which in cattle can be many hundreds of days) than previous data suggests. This is consistent with the data of Mugnier et al [23], where multiple minor variants were observed using an Illumina sequencing approach. However, that study and ours both have limitations, one with relatively few biological replicates (albeit one mouse was followed for ~120 days) and one that only ventured to 12 days post-infection. Therefore, assessing antigen dynamics in the chronic phase of infections with tools that give significant resolution of expressed antigen diversity will be critical to furthering our understanding of the mechanisms of trypanosome antigenic variation. Key to studying this will be analysing the picture in the truly chronic stages of infection (as was done by Mugnier et al in the context of mouse infections), but particularly doing so in relevant hosts (e. g. cattle [42]) where the total population of trypanosomes in the animal will be potentially 1,000 times greater at peak parasitaemia and where infections may last for 100s of days–this will have a profound influence on the usage of the repertoire (our data, for example, was representative of a total population of approximately 1 × 108 parasites per mouse). Additionally, recent studies indicates T. brucei populations inhabit different niches in the mammalian host (e. g. skin and adipose [43,44]), to the extent that some show evidence of local adaptation with respect to metabolism ([44]) –how this population compartmentalisation interacts with antigenic variation and immunity is likely to be important for parasite maintenance and transmission. Therefore, understanding the dynamics in both the chronic stages of infection and in clinically relevant hosts will potentially provide ideas on the selective pressures that maintain such an elaborate system. Additionally, given the significant advantages described above in terms of identifying low frequency variants (including mosaic VSGs), it may be that a combined long and short read approach is likely to be the optimal way of holistically and accurately identifying expressed VSG diversity; the increased read number of short read technologies in combination with the better resolution of long read technologies would provide significant power to examine the complexity of VSG expression in trypanosomes. Animal experiments were carried out at the University of Glasgow under the auspices of Home Office Project License number 60/3760. Care and maintenance of animals complied with University regulations and the Animals (Scientific Procedures) Act (1986; revised 2013). All mice were infected with Trypanosoma brucei brucei TREU927, the genome reference strain [19,45]. A cryostabilate from liquid nitrogen was thawed and inoculated into BALB/c mice in order to amplify a viable in vivo population. Donor mice were euthanased at first peak parasitaemia (approximately 1 × 107 trypanosomes/ml), and blood extracted. Trypanosomes were counted in triplicate under an improved Neubauer haemocytometer, diluted to inocula of 1 × 105 trypanosomes in 200 μl Carter’s Balanced Salt Solution, which were then inoculated via the intraperitoneal route into 20 recipient BALB/c mice. Mice were maintained for 12 days post-infection, and groups of 5 mice were euthanased at 3,6, 10 and 12 days post-infection. Parasitaemia was monitored daily by venesection of the lateral tail veins using the rapid matching technique [46], and was counted in triplicate under an improved Neubauer haemocytometer on the sampling days. At each sampling day, RNA was extracted from 200 μl infected blood using the Qiagen RNeasy kit (Qiagen), according to the manufacturer’s instructions. Approximately 1 μg RNA was treated with DNase Turbo (Ambion), according to manufacturer’s instructions, and cDNA was generated as in Hall et al, 2013 [21], including a column purification step on generated cDNA using the PCR Purification kit, according to the manufacturer’s instructions (Qiagen). VSG transcripts were enriched by carrying out PCR with proof reading Herculase II Fusion polymerase (Agilent) on the cDNA template with oligonucleotide primers specific to the T. brucei spliced leader sequence (TbSL) and a reverse primers complementary to a 13 base pair conserved region in VSG 3’ untranslated regions (3UTR); primer sequences and PCR conditions were as previously described [21,23]. A subset of PCR transcripts was subjected to cloning and sequencing; PCR products were ligated into pGEMT-Easy vectors, transfected into One Shot TOP10 cells, bacteria were grown up and cloned under suitable antibiotic selection (all using the TOPO cloning kit, Invitrogen), and plasmid DNA extracted using a Miniprep kit (Qiagen); these procedures were all carried out according to manufacturer’s instructions. Extracted plasmid DNA of appropriate concentration was sent for sequencing (Eurofins MWG). 1 μg of PCR amplicon template as measured by Nanodrop (ThermoScientific) and Bioanalyser (Agilent) was submitted to the Centre for Genomic Research, University of Liverpool for sequencing using the Pacbio RSII platform (Pacific Biosciences). DNA was purified with 1x cleaned Ampure beads (Agencourt) and the quantity and quality was assessed using Nanodrop and Qubit assay. Fragment Analyser (using a high sensitivity genomic kit) was used to determine the average size of the DNA and the extent of degradation. DNA was treated with Exonuclease V11 at 37°C for 15 minutes. The ends of the DNA were repaired as described by the Pacific Biosciences protocol. Samples were incubated for 20 minutes at 37°C with damage repair mix supplied in the SMRTbell library kit (Pacific Biosciences). This was followed by a 5 minute incubation at 25°C with end repair mix. DNA was cleaned using 0. 5x Ampure beads and 70% ethanol washes. DNA was ligated to adapter overnight at 25°C. Ligation was terminated by incubation at 65°C for 10 minutes followed by exonuclease treatment for 1 hour at 37°C. The SMRTbell libraries were purified with 0. 5x Ampure beads. The quantity of library and therefore the recovery was determined by Qubit assay and the average fragment size determined by Fragment Analyser. SMRTbell libraries were then annealed to the sequencing primer at values predetermined by the Binding Calculator (Pacific Biosciences) and a complex made with the DNA Polymerase (P4/C2chemistry). The complex was bound to Magbeads and this was used to set up 3 SMRT cells for sequencing. Sequencing was done using 180 minute movie times. Data (raw sequencing files) is available through Gene Expression Omnibus (https: //www. ncbi. nlm. nih. gov/geo/ - accession number GSE114843).

Antigenic variation is a system whereby pathogens switch identity of a protein that is exposed to the host adaptive immune response as a way of remaining one step ahead and avoiding being detected. African trypanosomes have evolved a spectacularly elaborate system of antigenic variation, with variants being used from a library of ~2,000 genes. Our ability to understand how this rich repository is used has been hampered by the resolution of available technologies to discriminate between what can be closely related gene variants. We have applied a long read sequencing technology, which generates sequence information for the whole length of the antigen gene variants, thereby avoiding having to try and piece together antigen sequences from lots of small fragments, the pitfall of standard sequencing. Applying this technology to material taken at specific time points from batches of mice infected with trypanosomes reveals that the diversity of variants is much higher than previously suspected, and that there is a clear semi-predictable pattern in the gene expression. Additionally, using this technology we have been able to detect the presence of ‘mosaic’ genes, which are created by stitching together fragments from several donor genes in the library, much earlier in infection than has been shown previously. Therefore, we shed new light on the complexity of antigenic variation and show that long read sequencing will be a very useful tool in analysing and understanding the expression patterns of closely related genes, and how pathogens use them to cause persistent infections and disease.

Abstract Introduction Results Discussion Materials and methods

medicine and health sciences immune physiology immunology antigenic variation parasitic protozoans genetic mapping protozoans genome analysis molecular biology techniques research and analysis methods sequence analysis immune system proteins genomics sequence alignment bioinformatics proteins artificial gene amplification and extension antigens biological databases mutant genotypes molecular biology biochemistry trypanosoma eukaryota sequence databases polymerase chain reaction heredity database and informatics methods physiology genetics biology and life sciences computational biology genomic databases organisms

2019

Application of long read sequencing to determine expressed antigen diversity in Trypanosoma brucei infections

Signal crosstalk within biological communication networks is common, and such crosstalk can have unexpected consequences for decision making in heterogeneous communities of cells. Here we examined crosstalk within a bacterial community composed of five strains of Bacillus subtilis, with each strain producing a variant of the quorum sensing peptide ComX. In isolation, each strain produced one variant of the ComX signal to induce expression of genes associated with bacterial competence. When strains were combined, a mixture of ComX variants was produced resulting in variable levels of gene expression. To examine gene regulation in mixed communities, we implemented a neural network model. Experimental quantification of asymmetric crosstalk between pairs of strains parametrized the model, enabling the accurate prediction of activity within the full five-strain network. Unlike the single strain system in which quorum sensing activated upon exceeding a threshold concentration of the signal, crosstalk within the five-strain community resulted in multiple community-level quorum sensing states, each with a unique combination of quorum sensing activation among the five strains. Quorum sensing activity of the strains within the community was influenced by the combination and ratio of strains as well as community dynamics. The community-level signaling state was altered through an external signal perturbation, and the output state depended on the timing of the perturbation. Given the ubiquity of signal crosstalk in diverse microbial communities, the application of such neural network models will increase accuracy of predicting activity within microbial consortia and enable new strategies for control and design of bacterial signaling networks. In microbiology, quorum sensing (QS) is a process in which bacteria produce and secrete small chemical molecules known as autoinducers. Many bacteria regulate gene expression in response to the external concentration of autoinducer, including regulation of processes related to biofilm formation, virulence, and horizontal gene transfer [1–4]. Although QS is historically viewed as a process of a single species regulating its own gene expression, numerous reports have shown signal exchange between species contributed to regulation of QS phenotypes [5–10]. Such crosstalk between cells is usually the result of two bacterial strains producing chemical variants of a QS signal. QS signals have many naturally occurring chemical variants, including 56 distinct variations of acyl homoserine lactones and 231 variants of auto-inducing peptides (AIP) [11,12]. Chemically similar variants of a signal interact with QS receptors, leading to excitation or inhibition of QS activation to a variable degree [5,8, 13,14]. Multiple signal inputs to a given receptor protein lead to variable levels of gene expression, making it difficult to predict community-level behaviors in the presence of two or more signaling molecules. Signal crosstalk was first recognized when Vibrio cholerae and Vibrio parahaemolyticus produced a QS response in Vibrio harveyi [15]. Riedel et al. [16] observed QS crosstalk between Burkholderia cepacia and Pseudomonas aeruginosa where P. aeruginosa activated QS in B. cepacia. Mclean et al. [8] showed that a Chromobacterium violeceum biosensor produced different levels of QS activation when introduced with a variety of distinct Acyl-Homoserine Lactones (AHLs) separately. Geisinger et al. [17] detected QS activity in pairwise combinations of the Agr-I-IV QS AIP system to uncover the contribution of divergent QS alleles to variant expression of virulence determinants within Staphylococcus aureus. Although these studies identified the potential for crosstalk in the presence of pairwise combinations of cells, in a given environment QS crosstalk can be more complex when several QS bacteria coexist in the same community. For instance, in the human microbial gut 300–500 bacterial species are present [18] and among these populations at least ten QS species have been identified and more than eight QS signal variants have been recognized [19]. Thompson et al. [20] showed that the ratio between Firmicutes and Bacteroidetes can be influenced by introducing an external source of QS signals, demonstrating that changes in QS signals can have community-level influences on activity. Furthermore, in rhizosphere soil out of the 350–550 bacterial species, at least 30 species were capable of producing multiple QS signals [21], and another study found 8% of genomes in the soil contained the genes needed to activate an AHL reporter strain [22]. Interest in engineering microbial communities to utilize multiple QS signals led to a broader characterization of gene regulation in contexts with multiple signals. Several early examples of cellular networks with multiple signals utilized the LasRI and RhlRI networks from Pseudomonas aeruginosa, which produce nearly orthogonal signals C4-HSL and 3-oxo-C12-HSL, with essentially zero crosstalk. Later work combined signaling networks with measurable crosstalk [23–26]. Scott et al. [27] conducted an extensive study on the pairwise effects of the AHLs 3-oxo-C6 HSL, 3-oxo-C8-HSL and 3-oxo-C12-HSL on the LuxR, LasR, RpaR and TraR QS systems to understand how to construct higher-level genetic circuitry for the use in microbial consortia. Wu et al. [7] also characterized the QS pairwise interactions of the AHLs 3-oxo-C6-HSL and 3-oxo-C12-HSL on the LuxI/R and LasI/R QS systems to better understand new directions in engineering gene networks. Our previous study investigated the robustness of signaling networks to interference by quantifying crosstalk between the LuxI/R QS system and the AHLs 3-oxo-C6-HSL and C4-HSL [5]. Although these studies identified the pairwise interactions of QS in the presence of one or two auto-inducers, there is limited knowledge on how species composition, signal diversity and external perturbations would affect QS activation when more than two QS species are present, see Fig 1A. In such heterogeneous environments, with several QS signals, activation of QS in each species is interdependent, making it a challenge to predict community-wide QS activity. Here we introduce a neural network model to examine the consequences of QS crosstalk within bacterial communities producing mixtures of QS signals, see Fig 1B. Neural network models have been commonly used to understand the network-level consequences of interactions in many complex systems, including both biological and non-biological contexts. Neural networks have been implemented in advanced analytical techniques such as deep learning, pattern recognition and image compression [28–31]. In finance and economics, neural networks are trained with historical market data to discover trends and to make successful current market predictions [32]. In the pharmaceutical industry, based on the bio activity of a large set of chemicals, neural networks are used to identify new types of drugs that can be used to treat diseases [33]. In a neural network, components have a variable state, in simplest cases active or inactive. Interactions between network components influence state dynamics and are represented as weights, with the magnitude of the weight indicating the strength of the interaction and the sign of the weight indicating whether the interaction promotes or inhibits activation. We have previously implemented a neural network to theoretically analyze the information capacity within a QS networks composed to multiple Staphylococcus aureus strains [34]. Here we extend these ideas, combining both experimental and theoretical results, to test whether neural network models can be used to predict and control the activation of QS in communities of bacteria producing multiple signals. To test our ability to predict QS activation within a community producing multiple variants of a signal, we used five strains of Bacillus subtilis previously reported by Ansaldi et al. [9]. Each of the five strains produces a unique variant of the ComX QS signal [9,35], see Fig 1C. Each strain also had variation in the sequence of the ComP receptor protein [35,36]. In previous work [9], crosstalk between pairs of these five strains was reported, revealing a mix of both excitatory and inhibitory crosstalk between strains of variable strength. These measurements were not sufficient to construct a neural network representation of QS activity within the community, as the ratio of signals was not varied, so pairwise crosstalk between each signal was measured. We measured QS activity within mixtures of two ComX signals by combining ratios of supernatant from stationary phase culture of two “producer” strains and measuring QS activation using a “tester” strain containing a lacZ QS reporter (as used in the Tortosa et al. [35]), see Fig 2A and 2B. Producer strains produce the ComX signals and self-activate QS while the tester strains do not produce the ComX signal, but can activate QS if there is an external supply of ComX. QS activity was measured with a fluorogenic LacZ assay in a 96 well plate reader, see Fig A in S1 Text and methods for tester responses with cognate signal. The fold change in LacZ expression is the ratio of LacZ expression in the presence and absence of ComX signal, see methods for further details. In individual strains, QS activity increased and approached saturation, Fig 2C and Fig B in S1 Text. The yellow circles shown in Fig 2C are a representation used throughout the manuscript to symbolize the addition of either a producer cells (P) or a supernatant from producer cells (S) to a tester (T). The tester is indicated in the middle of the circle while the producer or supernatant strains are indicated on the circumference. Each producer supernatants was extracted only once and stored at -20 oC. For all experiments the same batch of supernatant was used such that signal concentrations were consistent for all measurements, see methods for further details. The activity of the supernatant remained consistent throughout the study, see Fig C in S1 Text. Next we examined QS activity upon exposure to mixtures of signal. As shown in Fig 2C, in the presence of second signal D, the QS activity in strain C is inhibited. To quantify the interaction weight between strains, we constructed a mathematical model. The model is based on a set of differential equations and accounts for signal crosstalk by introducing a crosstalk weight for each pair of receptor and signal, similar to models used previously [5,34]. Specifically, the expression of the QS-regulated gene lacZ follows: ∂L∂t=ρLni (fi (Ceff, imCeff, im+θim) +1) −γLL, (1) where the effective concentration of signal as the result of crosstalk is, Ceff, i=∑jwi, jcj. (2) In a population of ni cells, lacZ expression occurs at a basal rate of ρL and upon QS activation the production rate is increased by a fold change fi. A Hill’s function is used to represent the scaling of QS-activity with signal concentration, with a Hill coefficient of m and the concentration of half maximum of θi. LacZ will degrade at a rate of γL. An effective concentration, Ceff, i is used to account for the excitatory or inhibitory influence of each signal on QS activation in the ith strain. The interaction weight wi, j accounts for the magnitude and sign of the interaction between a ComX signal from B. subtilis strain j on QS activation in B. subtilis strain i. As in a neural network mode, the sum of these weighted interactions predicts the activity of each node (strain) for mixtures of inputs (signals). The self-weight (wi, i) is one for all strains, and if Ceff<0, we assume that Ceff = 0. Similar to previous studies [5,37–39], the ith tester grows at a rate μ in a volume v through a logistic growth equation, ∂ni∂t=μini (1−ntotalsv). (3) Here ni is the amount of the ith tester cell, μi is the growth rate of the cells (given in Fig D in S1 Text), ntotal is the total number of cells in the system, and s is the maximum density of cells reached by the culture. In simulations, tester strains in the well of a plate grew from a cell density of 108 to 109 cell per mL. Over time, QS-regulated lacZ was produced following Eqs 1,2 and 3. LacZ concentrations in the culture were simulated for tester cells exposed to no signal as well as for a specified mixture of signal to calculate the fold change in lacZ expression, see methods section and Table A in S1 Text for model parameters [37,40,41]. We first simulated the response of the testers when mixed only with the cognate signal supernatant and these simulations were used to fit experimental data to obtain fi and θi for each strain, further details given in the methods section, see Figs E and F in S1 Text and Table 1. The best fit fi and θi values were used to generate the simulation curve, and the fold change in LacZ of this simulated curve at θi was defined as the threshold value of LacZ fold change needed for QS activation, see Fig 2C and Fig F in S1 Text. In the methods section we describe how to convert from supernatant volume to a relative signal concentration. This is an approximation of the signal concentration in the supernatant and does not take into account variability of signal production for each strain. The five strains had five distinct threshold values needed for QS activation, see Fig F in S1 Text. Simulations were done to obtain the response of the testers when the cognate signal was mixed with an interacting signal, see Fig 3A. In Fig 3B, we have simulated a simplified representation with QS on (yellow boxes) and off (blue boxes), for such a case, and we observe that this pattern of activity (or QS activation landscape) changes depending on the weight of the interacting signal. QS on and off were determined by testing whether the QS activity for a given combination of cognate and interacting signal would be higher or lower than the threshold calculated previously. We obtained similar activation landscapes for experiments considering the same thresholds for each strain, Fig 3C and Figs G-K in S1 Text. Fig 3C shows the activation landscape of TC. For each strain, between 0 and 25 μL of supernatant from each producer strain was mixed with 0 to 25 μL of supernatant from a second strain. 25 μL was chosen as the maximum volume of supernatant as individual strains required 10 μL or 15 μL to activate QS, giving sufficient dynamic range to measure even strong inhibition. We compared the experimental activation landscapes with the simulated activation landscapes to extract the weights of each B. subtilis tester and the corresponding interacting strain. For example, simulations for the response of strain C to mixtures of signal from strains A and C indicate wC, A between 0. 363 and 0. 527 reproduce the experimental measurements of quorum sensing activation. Therefore, wC, A is reported as the mean of these values with error bars indicating the range of possible values. Using this method, the interaction weight was calculated for each pair of strains, see Fig 3D. The weights calculated for each strain revealed a rich network of signaling interactions within the 5 strain community, Fig 3D and Fig L in S1 Text. For example QS activation in strain C was activated by signal from strains A or C, strongly inhibited by signal from strains B or D, and only weakly responded to signal from strain E. Strain E on the other hand only responded to its own signal and was not influenced by the signal from any other strain tested. Note that these simulations predict QS activation to a static input of each signal. Later we discuss how quorum sensing activity depends on the dynamics of signal production and cell growth. The extraction of the pairwise interaction weights in the previous section enabled us to apply a neural network model to predict QS activation patterns in groups of 3 or more strains. This model is a fully connected, single-layer network without any hidden layers. In the model the nodes represent each strain in the community with the weighted connections representing signal exchange between each strain. In Fig 4A, we have predicted the response of TC in the presence of SC, SD and SE. The interaction weights are not retrained with each new set of experimental data, instead the interaction weights calculated in Fig 3 are inherited for all subsequent model predictions. Hence in Fig 4A, we use the weights wC, C, wC, D, wC, E to simulate the response of Tc. As shown in Fig 4B and 4C, model predictions were verified in experimental measurements. In experiments, supernatants from three different strains were mixed at a specific ratio with a tester strain, and the expression of QS genes was measured using the fluorogenic LacZ indicator. Each measurement used only a single tester strain, and separate experiments were carried out in parallel to determine the response of the full community. Experimental measurements reproduced the predicted pattern of QS activation as ratios of supernatant were varied. This shows that a simple one layered network is capable of predicting the QS output in the presence of three signals. Analysis of QS activation in the presence of three signals revealed the concept of a community-level signaling state. In a microbial community producing multiple signal variants, crosstalk between these signaling systems potentially leads to the activation of QS in subsets of the community. The exact combination and ratio of signals, as well as the structure of the crosstalk network will determine which strains activate QS. The neural network model, with the pairwise weights, predicted the community-level signaling state and its sensitivity to changes in signal concentrations, as shown in Fig 4D and Fig M in S1 Text. The signaling state of the 5-strain community is shown as the amount of signal from strain E was varied. The community-level QS state can be represented as a binary string, with a 1 or 0 in each position of the string indicating whether QS will activate or not activate, respectively, for each strain. For example, at 0 μL of supernatant from strain E, the predicted string was (1,0, 1,0, 0), which indicated that only strains A and C would activate QS under these conditions. As the amount of strain E supernatant was increased, the community-level state changed twice, first to (1,0, 0,0, 1) and then to (1,1, 0,0, 1). These predictions from the model were borne out in experimental measurements, as the two transitions of the community-level signaling state were observed as the volume of supernatant from strain E increased. In the previous sections, we determined that a neural network model could predict the consequences of QS crosstalk within a 5-strain community of B. subtilis. These predictions were tested in experiments in which specified ratios of supernatant from stationary phase cultures were combined with tester strains to measure QS activity. Next we wanted to verify that the model could also predict QS activity for mixtures of strains growing from low density culture, see Fig 5A. These experiments revealed how inoculation ratios influenced QS activation in a multi-strain community. As cells grow, release signals, and potentially activate QS, the signal concentrations will change over time until reaching a steady state [5,37,39]. To simulate signal production, cell growth, and the expression of QS-regulated lacZ, Eqs 1–4 were used. Eq 4 describes the change in signal concentration (ci) over time for the ith strain as the result of signal production and degradation. The signal production of the ith strain will be influenced by all the other strains that are present, as captured by ceff, i defined in Eq 2. In a population of ni cells of the ith strain, signal production occurs at a basal rate of ρci and at QS activation, this rate is increased by a fold change fi. A Hill’s function is used to represent the scaling of signal production with effective signal concentration ceff, i, with a Hill coefficient of m and the concentration of half maximum of θi. ceff, i incorporates the QS crosstalk as mentioned previously. The signals will degrade at a rate of γci. For this case, in Eq 3, ntotal=∑j=1knj, where k is the total amount of producer strains mixed together. Parameter values are listed in Table A in S1 Text. In simulations, using Eqs 3 and 4, PA and PB were mixed at different ratios and the concentration profile of the ComX signals for both were plotted with respect to time, see Fig N in S1 Text and methods for further details. Fig 5B shows how the inoculation ratio of PA to PB influenced expression of the QS-regulated lacZ gene. The simulated fold change in LacZ was normalized by the threshold fold change in LacZ needed for QS activation. Here the fold change in LacZ is determined by applying the supernatant of the mixture PA and PB after 10 hours growth to the testers TA and TB separately. We observed that above a 6: 1 inoculation ratio of PA to PB, strain B does not produce a sufficient concentration for QS activation, whereas strain A activated QS at all strain ratios. In experiments, we mixed PA and PB at ratios of 1: 1,2: 1,10: 1,100: 1,1000: 1, and grew them for 10 hours, see Fig 5C. Supernatants of these mixtures activated QS in TA at all ratios, whereas TB was only activated at 1: 1 and 2: 1 ratios. PA and PB as both are based on the B. subtilis 168 strain such that any other competition is minimized. As seen in Fig 5A, mixing these strains would impact the ComX production of each other. These results demonstrate that the inoculation ratio of strains dictated the final QS state of the community. We have shown that the QS activation of species depends on species composition, ratio of the number of species, signal composition and signal concentration. Next we tested if the QS activity of a multi-strain community could be influenced by a perturbation of added signal, and whether the timing of signal addition impacted the response to the perturbation. As depicted in Fig 6A, in both simulation and experiments we mixed together producers A and B at a ratio of 1: 1, and measure the QS activity of tester strains A and B. In the control experiment, the coculture was not perturbed, and we expected results as in Fig 5, strains A and B both activate QS. Duplicate cocultures were perturbed by the addition of 200 μl of supernatant from strain C at various points of time after inoculation. As shown in Fig 6B, both simulations and experiments confirm that the addition of signal C had the potential to alter the activation of QS in the coculture, see Fig O in S1 Text. Signal C inhibits QS activation in strain B and promotes QS activation in strain A. However, whether the community-level signaling state changed from (1,0) to (1,1) depended on the time of the perturbation. The simulations showed either QS on or QS off for TB, since after 10 hours the signal concentration reached a steady-state that would either be above or below the threshold for QS activation, see Fig O in S1 Text. Addition of supernatant from strain C prior to 4 hours resulted in strain B not activating QS, whereas a perturbation at 4 hours or later did not alter QS activation of either strain. As above, the QS activity for these cases was tested with supernatants extracted from cultures that were grown 2 hours past exponential phase. In this study, we analyzed QS activation within a community of five different B. subtilis strains that produce distinct ComX signals. The pairwise interactions measured within the 5-strain community were consistent with crosstalk patterns reported in Tortosa et al. and Ansaldi et al. [9,35], Fig P in S1 Text. Although crosstalk between pairs of QS systems have been observed previously [5,6, 9,35,37,42], here for the first time we predict the consequences of signal crosstalk on QS activity in a community utilizing multiple signal variants. In natural environments, QS crosstalk is prominent and often regulates genes associated with virulence and biofilm formation [14,43–47]. Many natural communities produce diverse sets of signaling molecules [48,49], resulting in entangled and interdependent gene expression within the community. For instance, in a recent report [50], it was suggested that QS in B. subtilis can regulate ComX degradative enzymes, which in turn will inhibit QS activation. Therefore, QS crosstalk could potentially be entangled with such ComX degradation. Neural networks, inspired by the decision making within densely interconnected neurons, enable predictions of activity within complex systems with many interacting components. Here we demonstrate the consequences of signal crosstalk on community-level gene expression states can be accurately predicted using a neural network model with pairwise interactions. Many previous studies on QS crosstalk have implied that QS activation state is solely dependent on the type, or in our case sign, of the crosstalk [7,45]. In the presence of all five ComX signals, we observed that the QS activation state of one strain will be different depending on the exact mixture of ComX concentrations. Therefore simply knowing the community composition will not give an accurate picture of the potential for QS activation. Communities with identical membership can be driven towards multiple community-level signaling states; ratios of strains and the structure of the interaction network determine which strains or species activate QS. The history of the community also dictates activation patterns, as the temporal dynamics of signal accumulation, as influenced by strain inoculation ratios and perturbations, dictate QS activation states. The neural network model may enable the design of perturbations to redirect QS activation within bacterial communities. As QS activation is known to influence the real-world problems such as infections and biofouling, this predictive model of community-level activity should be relevant for industrial and medical applications [45,51,52]. Although the utility of a neural network model of QS was demonstrated for only one class of QS signals, cyclic autoinducing peptides used by several species of Gram-positive bacteria, the model should be easily extended to communities using the acyl-homoserine lactone signals common in Gram-negative QS. Crosstalk as the result of two cells producing chemically related signaling molecules is not unique to cyclic autoinducing peptides [7,53]. Crosstalk within AHL networks or other signaling networks may have a different distribution of weights, which would impact the number and sensitivity of community-level signaling states. Here we assumed each signal accumulates to the same concentration in the supernatant, however each strain might accumulate different concentrations of signal. Variability in signal concentration in the supernatant therefore modulated each interaction weight by an unknown multiplication factor, a factor which could be determined through measurements of signal concentrations. In addition, although the 5-strain B. subtilis community studied here was densely connected, i. e. each strain was connected to every other strain in the network, and the modelling framework could be applied to networks in which only subsets of strains participate in quorum sensing crosstalk. The model might also be relevant to other types of microbial interaction [54]. Some bacterial species detect more than one signal, such as strains which utilize AI-2 and AHLs or for example V. cholerae which combines information from four chemically distinct signals [55]. Mapping these situations onto the neural network model should give deeper insights into QS regulation in diverse microbial communities and reveal how the structure of the network determines the community’s ability to exchange information and coordinate group activity. All strains used in this study were obtained from the study Ansaldi et al. [9]. The producer cells produced the ComX pro-protein which is modified and processed by ComQ, in turn, releasing the ComX pheromone to the extracellular environment. Released ComX pheromones bind to the ComP membrane protein to phosphorylate ComA and activate QS. The testers cannot produce the ComX signal due to the disruption of the comQ and comX genes but can activate QS when the signal is exogenously added to produce LacZ under the QS-regulated srfA promoter. All the testers and producers were isogenic apart from producing distinct pheromones and receptors. To construct the strains labeled from B-E, the comQXP genes of the Bacillus subtilis 168 were replaced with the foreign comQXP genes from four distinct Bacillus species [9,35]. The Bacillus subtilis strains were grown in the Bacillus competence media as described previously [56]. This media contained (w/v) 1. 00% sodium Lactate, 0. 25% yeast extract, 0. 20% ammonium sulfate, 1. 40% dipotassium phosphate, 0. 60%; mono potassium phosphate, 0. 10% sodium citrate-2H20,0. 02% magnesium sulfate*7H20 and 0. 40% of glucose. All cultures were grown at 37°C and 200 RPM in competence media. Tester strains grown overnight, diluted 1/1000 in fresh media, and grown for an additional 10 hours before measurements. To extract supernatant, producers were grown for 10 hours after 1/1000 dilution of overnight culture. Cultures were centrifuged for 4 min at 4000 RPM to collect supernatant. The supernatant was passed through a 0. 2 μm VWR syringe filter. Filtered supernatants were stored at -20°C. A single batch of supernatant from each strain was used for all experiments. When mixing the producers strains together producers were grown in separate cultures for 10 hours, diluted 1/1000 in fresh competence media and grown for an additional 3 hours. Growth for 3 hours at low density was to ensure QS was not active prior to mixing strains together. After 3 hours of growth, cultures reached a final OD of 0. 2–0. 3. Cultures of producer strains were mixed together at different ratios. For example, in the 1: 1000 case, we mixed 1 μl of PA with 999 μl of PB together into 3 mL of competence media. Mixed cultures were grown for 10 hours. Since all producers and testers were based on the B. subtilis 168 and they were near isogenic, no growth interactions occurred between strains when mixed together. To measure activity, supernatant was extracted from producer cocultures as described above. In perturbation experiments, supernatant from strain C was added to the culture from 0 and 7 hours at every 1 hour interval, after mixing producer strains together. For the indicator, we used fluorescein di-β-D-galactopyranoside (FDG) [35,57], and the fluorescence was detected by using a Tecan plate reader with a 96 well plate. In the 96 well plate, we mixed 25 μl of the testers, 25 μl of the FDG at 0. 04 mg ml-1 (FDG in the competence media), and 75 μl of fresh competence media. The remainder of the 200 μl volume was a variable combination of filtered supernatants and spent media. For consistency, all wells had a total of 75 μl of spent media, a combination of supernatant from producer strains and supernatant from the tester strain (which did not contain any signal). Testers were loaded after 10 hours growth, as described above. Fluorescence was detected with an excitation wavelength of 480 nm and an emission wavelength of 514 nm. For the mixed producer experiments, we added 150 μl of supernatant with the testers and FDG. The absorbance was detected at 600 nm to enable calculation of fluorescence per cell, as described previously [37]. Similar to previous studies [9,35], the LacZ expression level was calculated by obtaining the rate of fluorescence increase, which is the gradient of the fluorescence per cell vs time in the linear region between 0–5 hrs, see Fig A in S1 Text. The fold change in LacZ was calculated by taking the ratio of a given LacZ expression level with the LacZ expression level when no signal is present. To obtain the growth rates, the cells were grown in the Bacillus competence media, as mentioned above. We then diluted the cells 1000 fold in 3 mL of competence media and grew them at 37°C and 200 RPM. Dilutions of the culture were plated on competence media agar plates at every 1 hour interval during growth for 7 hrs, and these plates were incubated for 16 hrs at 37 0C. For each time point 2 sets of replicates were considered, and after 10 hours growth, the number of colonies were counted to get the colony forming units. The growth rates were obtained from a fit to the linear region in the growth curve. To perform the simulations described in the manuscript we used MATLAB 2016b. We used the finite difference method to simulate the change in LacZ expression with respect to time, using a time step of 1 min. We calculated the LacZ production rate by simulating the increase in LacZ using Eq 1. For each strain i, as shown in Table 1, fi and θi were distinct. We obtained fi and θi by minimizing the root-mean squared error and determining the best fit curve between the experimental data points and simulation curve, see Figs E and F in S1 Text and Table 1. The thresholds to determine QS activation for each strain were different based on the unique values for fi and θi, see Fig F in S1 Text. To plot the QS activation patterns of each tester for the pair-wise case, using unique fi and θi, values we simulated the fold change in LacZ for each combination of signals and tested if the fold change in LacZ was above or below the threshold of the tester. If it was above, we assigned the yellow color and if it was below, we assigned a blue color. The activation pattern of the simulations, as seen in Fig 3A, was compared with the experimental patterns to extract the weights. We observe the same activation pattern for a narrow range of weights, see Figs G-L in S1 Text. To convert the loading volume of the supernatant to the signal concentration we considered the following method. Previous studies have reported the final concentration of ComX in media after 2 hours growth from late exponential phase is approximately 30 nM [58,59]. Since a volume of x μl of the supernatant at 30 nM gets diluted in a total of 200 μl in the 96 well plates, see methods for further details, we can calculate the signal concentration = 30 nM (x/200). For the simulations with the PA and PB grown together, we considered an initial density of 106 cells/ml. This was based on the dilutions used in the experiments. We mixed PA and PB at different ratios and simulated the change in ComX concentrations over time. In mixed cultures of PA and PB, the production of ComX concentration depends on the crosstalk between the two producer strains, as signal production is regulated by quorum sensing. The crosstalk is taken into account by calculating Ceff, i using Eq 2. Using Eqs 1–4 and a finite difference method with a time step of 1 min, we simulated the change in cell number and signal concentration over time to determine if the mixture of signals that accumulated after 10 hrs would activate quorum sensing in strains A or B. A similar procedure was followed to simulate the effect of a signal perturbation in Fig 6. Perturbations were introduced between 0 to 7 hrs, and the response of the tester strains to the signal mixture that accumulated after 10 hrs was predicted.

Bacteria can communicate with each other using chemical signals to activate genetic expression in a process known as quorum sensing. Quorum sensing in bacteria is known to regulate a number collective behaviors in bacteria such as biofilm formation, antibiotic production and production of virulence factors which leads to bacterial infections. In a community, different species of bacteria can crosstalk using these signals, such that they regulate each other’s quorum sensing activation. Crosstalk can be either excitatory or inhibitory towards quorum sensing activation. Generally, in a bacterial community, it is not straightforward to understand how cells utilize mixtures of quorum sensing signals to regulate quorum sensing activation. To address this issue, we used a neural network approach in which we were able to predict patterns of quorum sensing activation in a diverse community of Bacillus subtilis cells producing five different signals and we observed that quorum sensing activation depended on signal concentration, species ratio and time sensitive external perturbations. Our findings can be useful in systematically controlling quorum sensing and potentially devising better strategies to fight bacterial infections.

Abstract Introduction Results Discussion Materials and methods

medicine and health sciences pathology and laboratory medicine neural networks engineering and technology pathogens signal processing bacillus signaling networks microbiology neuroscience prokaryotic models signal inhibition network analysis experimental organism systems bacteria bacterial pathogens research and analysis methods computer and information sciences microbial physiology animal studies medical microbiology gene expression microbial pathogens cellular crosstalk signal transduction cell biology bacillus subtilis genetics biology and life sciences cell signaling quorum sensing organisms

2019

A neural network model predicts community-level signaling states in a diverse microbial community

In eukaryotic cells, environmental and developmental signals alter chromatin structure and modulate gene expression. Heterochromatin constitutes the transcriptionally inactive state of the genome and in plants and mammals is generally characterized by DNA methylation and histone modifications such as histone H3 lysine 9 (H3K9) methylation. In Arabidopsis thaliana, DNA methylation and H3K9 methylation are usually colocated and set up a mutually self-reinforcing and stable state. Here, in contrast, we found that SUVR5, a plant Su (var) 3–9 homolog with a SET histone methyltransferase domain, mediates H3K9me2 deposition and regulates gene expression in a DNA methylation–independent manner. SUVR5 binds DNA through its zinc fingers and represses the expression of a subset of stimulus response genes. This represents a novel mechanism for plants to regulate their chromatin and transcriptional state, which may allow for the adaptability and modulation necessary to rapidly respond to extracellular cues. In eukaryotes, chromatin structure regulates the access of the transcriptional machinery to genetic elements, playing an important role in the regulation of gene expression. The transition between transcriptionally active (loosely packed) chromatin and repressed (tightly packed) chromatin states is controlled by covalent modifications of the histone tails, DNA cytosine methylation, and the differential use of histone variants [1]. In mammals and plants, transcriptionally inactive chromatin—or heterochromatin—is typically associated with DNA methylation and histone H3 lysine 9 methylation (H3K9me). These epigenetic silencing marks are generally thought to be coordinately regulated by cooperation between DNA methyltransferases and histone methyltransferases, contributing to their stability and self perpetuating nature. However, in order to readily adapt to environmental stimuli or developmental cues, some of these marks also need to be reversible, although how this is achieved is currently unclear. Most histone methyltransferases (HMTases) contain a catalytic SET domain (named after three Drosophila proteins: Suppressor of position effect variegation 3–9, SU (VAR) 3–9; Enhancer of zeste, and Trithorax) [2]. The enzymatic activity of the SET domain was first discovered in a mammalian homolog of SU (VAR) 3–9, SUV39H1, which was shown to methylate histone H3 at lysine 9 [3]. In plants, there is a relatively large family of SET domain-containing proteins that are closely related to Drosophila SU (VAR) 3–9 and its human and S. Pombe homologs (SUV39H and CLR4, respectively) [4]. In Arabidopsis thaliana, of the 14 SET domain-containing proteins most related to SU (VAR) 3–9, nine are classified as SU (VAR) 3–9 HOMOLOGS (SUVH1–SUVH9), and five as SU (VAR) 3–9-RELATED proteins (SUVR1–SUVR5). Arabidopsis SUVH proteins link the epigenetic silencing marks H3K9me2 and DNA methylation through the activity of their SRA domains (for SET and RING finger Associated), which bind different contexts and states of methylated DNA. Contrary to SUVHs, most of the SUVR proteins are of completely unknown function. In addition, because they lack the SRA domain, how they are recruited to chromatin is unknown. In Arabidopsis, DNA methylation occurs in three different sequence contexts: CG, CHG and CHH (where H is any base other than G). In all cases, de novo DNA methylation is established by DOMAINS REARRANGED METHYLTRANSFERASE 2 (DRM2), a homolog of the mammalian DNA METHYLTRANSFERASE 3 (DNMT3) family [5]. Subsequent to establishment, DNA methylation is maintained through the cell cycle by at least three different pathways depending on the sequence context [6]. The maintenance of CHH methylation is mostly carried out by DRM2 through persistent de novo methylation [6], [7]. The maintenance of CG methylation depends on METHYLTRANSFERASE 1 (MET1), the Arabidopsis homolog of mammalian DNA METHYLTRANSFERASE 1 (DNMT1), in collaboration with the VARIANT IN METHYLATION/ORTHRUS (VIM/ORTH) family [8], [9], [10], the Arabidopsis homologs of the mammalian UHRF1. These proteins contain SRA domains that bind to hemimethylated CG sites [11], [12], [13]. The maintenance of CHG methylation relies on CHROMOMETHYLASE 3 (CMT3), a plant specific DNA methyltransferase that acts together with some of the above mentioned SUVH proteins, KRYPTONITE (KYP) /SUVH4, SUVH5, and SUVH6 [14], [15], [16], [17], which can bind directly to methylated-DNA [11], [18]. The structure of the SUVH5 SRA domain bound to methylated DNA has been solved revealing that two SRA domains bind independently to each strand of the DNA duplex at either a fully or hemimethylated site [19]. These data support a model where regions rich in DNA methylation serve as binding platforms for KYP, SUVH5 and/or SUVH6, leading to H3K9 methylation. Histone methylation would then provide a binding site for CMT3 via its chromodomain, leading to CHG methylation, and thus creating a purely epigenetic self-reinforcing feedback loop for the maintenance of DNA and histone methylation, which explains the stability of epigenetic silent states and their self perpetuating nature [11]. The link between H3K9 methylation and DNA methylation is further supported by the strong genome-wide correlation between heterochromatic H3K9me2 and DNA methylation [20]. In addition, kyp mutants show decreased levels of both H3K9me2 and cytosine methylation [14], [21], [22], which are even further reduced in higher order suvh mutants [16], [17]. Moreover, loss of DNA methylation in met1 mutants correlates with a global loss of H3K9me2 [22]. In this report we show that Arabidopsis SU (VAR) 3–9 RELATED 5 (SUVR5), which lacks the SRA domain present in its SUVH counterparts, is able to recognize specific DNA sequences through a DNA binding domain that contains three zinc fingers, and induce silencing through DNA-methylation independent H3K9me2 deposition, possibly acting as part of a histone modifier multimeric complex. We propose that SUVR5 mediates a mechanism for heterochromatin formation that is distinct from the self-perpetuating loop existing between H3K9me2 and DNA methylation, and that this lack of perpetuation allows for the increased plasticity needed in response to environmental or developmental cues during an organism' s life. To test the role of Arabidopsis SU (VAR) 3–9 RELATED genes in plant development we screened T-DNA mutants in all five suvr single mutants and higher order combinations for visible morphological defects. We found that the suvr5-1 mutation produces a delay in flowering time that was not further enhanced in the quintuple suvr1 suvr2 suvr3 suvr4 suvr5 mutants (Figure S1). These observations were consistent with results from earlier analysis of a suvr5 mutant [23] and suggested a role for SUVR5 (but not the other SUVR family members) in flowering time. SUVR5 differs from the other SUVR family members in that it contains a set of three C2H2 zinc fingers in tandem in the central part of the protein (Figure 1a). SUVR5 homologs with a similar domain architecture (zinc fingers plus a C-terminal SET domain) are found in all plant species analyzed suggesting that it is widely conserved in the plant kingdom (Figure S2). We hypothesized that the zinc fingers have a DNA-binding function and may direct SUVR5 epigenetic activity to sequence-specific regions of the genome. To test this, we used the Systematic Evolution of Ligands by Exponential Enrichment (SELEX) technique with the recombinant SUVR5 zinc fingers domain to analyze binding to oligonucleotides that included a 15 base-pair (bp) random sequence (Figures S3 and S4). We identified an 8-nucleotide motif favored by SUVR5 binding (Figure 1b, upper panel). Next, we repeated the experiment using 100 bp fragmented Arabidopsis wild-type Col-0 genomic DNA (genomic SELEX, gSELEX) to identify naturally occuring SUVR5 binding sequences (Figure S5). We identified almost the exact same binding motif “TACTAGTA” (Figure 1b, lower panel) —a palindromic octamer that is consistent with the 9-nucleotide that is the maximum expected size of a sequence recognized by three zinc fingers in tandem, since each zinc finger repeat has a predicted alpha-helical core that binds to 3 nucleotides in the major groove of DNA [24]. The binding and its specificity were confirmed by electromobility shift assays (EMSAs) (Figure 1d, Figure S6). The high throughput sequencing results from the genomic SELEX experiment allowed us to map the identified SUVR5 binding regions to the Arabidopsis genome. Metaplot analysis showed that these regions mapped preferentially to the area immediately upstream of transcriptional start sites of protein coding genes (Figure 1c). Given the SUVR5 SET domain homology to Drosophila SU (VAR) 3–9 we hypothesized that SUVR5 is an active methyltransferase. Consistent with this, SUVR5 bound to the methyl-group donor SAM (Figure S7) and its SET domain contains all of the crucial residues required for histone methyltransferase activity in the HΦΦNHSC motif. However, we were unable to demonstrate in vitro histone methytransferase activity against various histone substrates. This could indicate that other binding partners are necessary for SUVR5 enzymatic activity, similar to other histone methyltransferase complexes such as those containing Enhancer of Zeste [25], or that SUVR5 biochemical activity is dependent on a particular chromatin context [26]. We directly tested for the role of SUVR5 on H3K9me2 levels in vivo by utilizing chromatin immunoprecipitation followed by microarray analysis (ChIP-chip) experiments in mature leaves of wild-type Col-0 and suvr5-1 mutants. The suvr5 mutants showed an overall decrease in H3K9me2 accumulation on pericentromeric heterochromatin (Figure 2a, Figure S8) and transposable elements (TEs) (Figure 2b), although these effects were relatively minor. Heterochromatic H3K9me2 is known to be mostly maintained by KYP, SUVH5 and SUVH6 [14], [16], [17], [21], [22], and ChIP-chip data with the kyp suvh5 suvh6 triple mutants showed a much more dramatic decrease in H3K9me2 levels than with suvr5 (Figure 2a and 2b). These data confirm that KYP, SUVH5, and SUVH6 are the major H3K9m2 enzymes in heterochromatin, but also suggest a minor role for SUVR5. H3K9me2 is correlated with DNA methylation in Arabidopsis on a genome wide level [20]. The loss of H3K9me2 in kyp mutants produces a decrease in DNA methylation [14], [21], [22] that is enhanced in the kyp suvh5 or kyp suvh6 double mutants and in the kyp suvh5 suvh6 triple mutant [16], [17]. Importantly, in the case of suvr5 mutants, we did not detect a decrease in DNA methylation at pericentromeric heterochromatin (Figure 2c, Figure S9) or TEs (Figure 2d, Figure S10), suggesting that SUVR5 functions differently than the SUVH proteins. We could also detect regions within the arms of the chromosomes with a decrease in H3K9me2 levels in the suvr5 mutants. Although the majority of these regions overlapped with regions dependent on KYP/SUVH5/SUVH6, over 20% were specific to suvr5 (Figure 3a and 3b). These suvr5-specific regions consisted of discrete patches of H3K9me2 that were solely dependent on SUVR5 (Figure 3d), and were characterized by very low levels of cytosine DNA methylation, and these levels of DNA methylation were not altered by the loss of SUVR5 (Figure 3e). These results suggest that, in those specific locations, SUVR5 is controlling H3K9me2 deposition in a DNA-methylation-independent manner that is not perpetuated by the KYP/CMT3 epigenetic loop. We could also find a small number of transposons in the chromosome arms whose H3K9me2 decrease was specific for suvr5 mutants and independent of kyp/suvh5/suvh6, and these tended to be smaller transposons with lower levels of DNA methylation (Figure S11). We analyzed for the presence of SUVR5 binding motifs within the sequence of these 423 TEs that show decreased levels of H3K9me2 specifically in suvr5 mutants ±2 Kb and 8. 5% of them contain the motif TACTAGTA. To determine if there is a correlation between H3K9me2 levels and SUVR5 binding, we analyzed H3K9me2 levels in the set of genes that were shown to bind the SUVR5 zinc fingers (i. e. with signal 3 Kb upstream of their transcription start site) in the gSELEX experiment. In that specific set of genes, we found a significant decrease of H3K9me2 when comparing suvr5 mutants to wild-type (Figure 3c). This decrease was significant for both of the ChIP-chip replicates analyzed (Figure S12). Analysis of all the genes that show a H3K9me2 decrease in suvr5 mutants compared to wild type showed that around 27% of them have gSELEX signal in their proximal promoter (1 Kb upstream their TSS). Interestingly, when we analyze not only euchromatic regions, but all decreased H3K9me2 regions including those in pericentromeric heterochromatin, only 5. 4% of them overlap with the gSELEX signal. This suggests that targeting of SUVR5 to pericentromeric heterochromatin may be mediated by another unknown mechanism, which is likely responsible for the redundancy of SUVR5 with KYP/SUVH5/SUVH6. To measure the effects of SUVR5 on gene expression, we performed mRNA sequencing (mRNA-Seq) experiments to analyze the transcriptome of suvr5-1 mutants. We observed a large number of genes that were signficantly upregulated, the majority of which were located in the euchromatic chromosome arms (Table S1, Figure S13). Although many of these genes are likely to be indirect targets to SUVR5,11% of these genes were among those that showed decreased H3K9m2 levels, and 69. 5% of these genes contained at least one significant SUVR5 binding motif in their promoter. Examples of genes with a decrease in H3K9me2 levels and upregulated expression in two different alleles of suvr5 mutants can be found in Figure S14 (See Figure S15 for suvr5-2 mutant allele characterization). Consistent with the slight decrease of H3K9me2 levels that occurred in suvr5 at TEs, very few transposons were reactivated in the mutant (Table S2). To identify the biological processes that SUVR5 may regulate, we applied gene ontology (GO) term analysis to the genes upregulated in the suvr5 mutant (over 4 fold, p-value<0. 01). Of the three broad GO term categories significantly over-represented in this set of genes, the most significantly enriched was “response to stimulus” (Figure S16). This category includes subcategories such as defense response, response to biotic stimuli like bacterium, and response to endogenous stimuli like the plant hormone auxin, which were strongly and significantly enriched (p_value<0. 01; Figure 4a). Auxin plays a key role in many plant developmental processes [27], [28]. For example auxin plays a central role in elaborating root architecture because of its role in endogenous developmental programs as well as its mediation of environmental stimuli responses [29]. We hypothesized that the overexpression of auxin inducible genes in suvr5 mutants might generate a partially constitutive auxin-response in the abscence of the hormone. Auxin causes inhibition of root growth by reduction of cell division and elongation, and a constitutive response could explain the defects in root growth earlier reported for suvr5 mutants [30], which we also observed here for both of the suvr5 alleles tested (Figure 4b and c). To examine this, we analyzed the expression of three examples of genes annotated as “auxin-responsive” and that have significant SUVR5 binding sites in their promoters (Figure S17). These genes are annotated as a PINOID (PID) -binding protein (At5g54490), an auxin-responsive GH3 family protein (At5G13320), and a SAUR-like auxin-responsive family protein (At3g12830). We found that these genes were indeed upregulated upon auxin treatment (Figure 4d) and that in the suvr5 mutants, these genes also showed increased expression, even in the absence of the hormone (Figure 4d). These data are consistent with a model whereby a stimulus such as auxin treatment overcomes the repression established by SUVR5, activating the genes and thus guaranteeing an appropriate response to environmental and developmental cues. The majority of chromatin modifiers characterized in higher organisms are present in large multi-protein complexes. SUVR5 was shown to interact in vitro with the Arabidopsis homolog of LYSINE-SPECIFIC DEMETHYLASE (LSD), termed LSD-LIKE 1 (LDL1) [23], an H3K4 demethylase partially redundant with its paralog LDL2 [23]. We tested for the existence of this complex in vivo by generating a transgenic line that expressed a FLAG tagged version of LDL1 under its own promoter, which was shown to complement the late flowering phenotype of the ldl1 ldl2 mutant (Figure 5a). Using affinity purification coupled with mass spectrometry (IP-Mass Spec [19]) (Figure 5b) we indeed identified an in vivo complex including both SUVR5 and LDL1. We also generated plants carrying a tagged version of SUVR5 expressed under the control of its own promoter, however the very poor expression levels of the tagged protein rendered our purification attempts unsuccessful. The physical interaction between SUVR5 and LDL1 suggests that their H3K9 methyl transferase and H3K4 demethylase activities may work together in collaboration to repress gene expression. To analyze the genetic interaction between SUVR5 and LDL1 we generated the suvr5 ldl1 ldl2 triple mutant and analyzed the effect on flowering time. Flowering time was as late in the triple mutant as in the single suvr5 or double ldl1 ldl2 mutants, indicating an epistatic relationship between SUVR5 and LDL (Figure 5c and 5d). mRNA-Seq in the double and triple mutants revealed 270 genes that were affected by both suvr5 and by ldl1 ldl2 mutations, which is more than 30% of the genes controlled by suvr5 alone. This suggests that SUVR5 and LDLs share a broad regulatory function. Furthermore, the GO category “response to stimulus” was also the most significantly enriched in ldl1 ldl2 mutants when analyzing their upregulated genes, supporting the idea that LDL1 and SUVR5 co-regulate a diverse set of targets involved in environmental responses (for the list of genes, see Table S3, for GO term analysis, see Figure S18). The 270 genes co-regulated by SUVR5 and LDL1 had very low expression levels in wild-type Col-0, and their degree of upregulation in the triple suvr5 ldl1 ldl2 mutant was the same as in the single suvr5 or double ldl1 ldl2 mutants (Figure 5e). This confirms that the relationship between the genes is indeed epistatic, with likely their H3K9 methylation and H3K4 demethylation activities acting together to repress gene expression for a large number of genes with common biological functions. Consistent with this, the most significantly over-represented GO term for the common 270 genes was again “response to stimulus”, which supports a common role for SUVR5 and LDLs in environmental adaptation (for the list of genes, see Table S3, for GO term analysis, see Figure S19). The ability of eukaryotic cells to respond to external stimuli and adapt to their environment depends on the coordinated activation and repression of specific subsets of genes. In order to facilitate this, repressive and permissive chromatin states must be readily altered in response to those stimuli. Our data are consistent with a model in which SUVR5 is part of a multimeric complex including LDL1 (and perhaps also other chromatin modifying enzymes) that recognizes genes with the sequence TACTAGTA (or related sequences) in their promoters and, in the absence of stimuli, represses their expression by altering epigenetic histone marks. This represents a unique form of epigenetic control via H3K9me2 that is independent from DNA methylation, and not perpetuated by the KYP/CMT3 loop, which potentially makes it more adaptable and dynamic for responding to environmental changes (Figure 6). One possibility is that SUVR5 mediated repression acts to modulate responses to various environmental signals as well as to provide an epigenetic memory of transcriptional states. The functioning of SUVR5 has analogies with some repressive chromatin modifiers characterized in animals that are also present in large multiprotein complexes. One example is the mammalian silencing transcription factor REST that is important in neural differentiation. It binds to the conserved RE1 motif through its 8 Krüppel zinc finger motifs and represses many neuronal genes in non-neuronal cells [31]. This transcriptional regulation is achieved by the recruitment by REST of histone deacetylases (like HDAC1/2) [32], [33], [34], [35], demethylases (like LSD1) [36], and methyltransferases (like G9a) [37], in a similar way to the proposed SUVR5 mode of action [30]. Another example is that of PR proteins. PR (PRDI-BF1 and RIZ homology) domain proteins (PRDMs) represent a distinct and unique branch of metazoan proteins that contain a PR domain, which at the amino acid level is 20–30% identical to the SET domain found in many histone lysine methyltransferases (HMTs) [38]. The PR domain is not present in fungi or plant genomes having originated in invertebrates [39], and is almost always accompanied by C2H2-like zinc finger motifs. PRDMs act as specific transcriptional regulators catalyzing histone methylation and/or recruiting interaction partners to modify the epigenetic regulation of target genes [38]. A common feature of PRDM proteins is their ability to act as transcriptional repressors by binding both to G9a and class I histone deacetylase enzymes such as HDAC1–3 [38]. In conclusion, multisubunit complexes containing different histone modifying enzymes targeted by specific DNA binding proteins appears to be a phenomenon conserved in plant and animals and may play a greater role in gene regulation than previously appreciated. The wild-type control in this study was the Columbia 0 ecotype (Col-0). suvr5-1 [23] and suvr5-2 are T-DNA insertion lines obtained from the SALK Institute Genomic Analysis Laboratory (SALK_026224 and SALK_085717 respectively). The kyp suvh5 suvh6 line was described in [40]. The ldl1–2 ldl2 line was described in [41]. The GST fusion protein used for SELEX and EMSA experiments was made by cloning the SUVR5 zinc finger domain (aminoacids 720 to 866) using the Gateway cloning system with pDEST15 as the final destination vector. For the SAM binding assay, the SET domain was cloned (aminoacids 1078 to 1376) also in pDEST15. Protein expression and purification was performed as previously described [11] plus the addition of 100 µM ZnSO4 to the cell culture at the time of protein expression induction (in the case of the Zinc finger domain) and avoiding the use of EDTA during the protein purification. The basic protocol for SELEX experiments described in [42] was followed with some minor modifications. For details, see Text S1. Sequencing data for the genomic SELEX experiment have been deposited at Gene Expression Omnibus (GEO) (accession number GSE39405). The protocol described in [11] was followed with slight modifications to the binding buffer composition (12% glycerol, 20 mM Tris-HCl pH7. 5,50 mM KCl, 1 mM MgCl2,1 mM DTT). For info on the primers used to test the protein binding, see Text S1. H3K9me2 ChIP experiments were performed using 3 week old leaves of wild type Col-0 and suvr5-1 plants, as previously described [43]. The ChIP-chip was performed as described in [20], the results show a comparison of the abundance of DNA pulled down with the anti-H3K9me2 antibody (#1220, monoclonal anti-H3K9m2 antibody, Abcam) versus INPUT. For info on the primers used to validate the ChIP-chip results by ChIP-qPCR, see Text S1. Each probe in the array was normalized by taking the log2 ratio of H3K9m2 to INPUT intensities, and the scores were scaled so that the average score across the arrays were zero. H3K9me2 hypomethylated regions were defined by tiling the genome into 500 bp bins (250 bp overlap), and computing the log2 ratios of the scores of suvr5 vs Col-0, and Z-score transformed. A Z<−3 cutoff was applied, and regions within 2. 5 kb were merged. Data have been deposited at Gene Expression Omnibus (GEO) (accession number GSE39405). DNA from leaves of 3 week old plants was extracted using a standard CTAB protocol. We performed sodium bisulfite treatment using EZ DNA Methylation Gold (Zymo Research) following the manufacturer' s instructions, amplified specific fragments using the primers described in Text S1 and cloned the resulting PCR fragments into pCR2. 1-TOPO (Invitrogen) to sequence and analyze around 20 clones per sample. To compare the converted clones to the original unconverted sequence, we used the sequence alignment tool of CLC Workbench software. We counted the converted/unconverted cytosines at each site manually and subsequently calculated the percent of methylation. BS-Seq was performed as previously described [44]. Sequencing data have been deposited at Gene Expression Omnibus (GEO) (accession number GSE39405). Leaves from wild type Col-0, suvr5-1, ldl1–2 ldl2 and suvr5-1 ldl1–2 ldl2 3 week-old plants were used for RNA extraction using Trizol (Invitrogen) following the manufacturer instructions. 10 µg of total RNA was treated with DNaseI (Roche), and cleaned up with RNeasy columns (Qiagen). Poly (A) was purified using the Dynabeads mRNA Purification Kit (Invitrogen) and used to generate the mRNA-seq libraries following the manufacturer instructions (Illumina). The libraries were sequenced using an Illumina Genome Analyzer. Gene and transposon expression in the RNA-seq data was measured by calculating reads per kilobase per million mapped (RPKM). P-values to detect differential expression were calculated by Fisher' s exact test and Benjamini-Hochberg corrected for multiple testing. Genes differentially expressed in wild-type and mutants were defined as those that have log2 (suvr5/wild-type) >4 and P<0. 01. Sequencing data have been deposited at Gene Expression Omnibus (GEO) (accession number GSE39405). For affinity purification of LDL1-3xFLAG ∼15 g of inflorescence tissue from transgenic and Col-0 plants was ground in liquid nitrogen, and resuspended in 75 ml of lysis buffer (50 mM Tris pH 7. 5,300 mM NaCl, 5 mM MgCl2,5% glycerol v/v 0. 02% NP-40 v/v, 0. 5 mM DTT, 1 mg/mL pepstatin, 1 mM PMSF and 1 protease inhibitor cocktail tablet (Roche, 14696200) ). Mass spectrometry analyses were performed as described in [19]. The identities of proteins co-purifying with LDL1 in Figure 5b are shown for those proteins appearing in two replicate purifications, and present at levels equivalent to at least 1% of the level of LDL1. Wild type Col-0, suvr5-1 and suvr5-2 plants were either grown for 13 days in vertical MS plates (CONTROL) or grown in vertical MS plates for 5 days before being transferred to MS+0. 5 µM NAA (Sigma) plates for 7 additional days. The web-based tool agriGO was used for the gene ontology analysis [45]. SUVR5 information is available in The Arabidopsis Information Resource under accession number AT2G23740.

The ability of eukaryotic cells to respond to external stimuli depends on the coordinated activation and repression of specific subsets of genes, often relying on chromatin structure modification. Here, we have characterized a locus-specific mechanism to repress gene expression by the action of an Arabidopsis thaliana SET domain protein, SUVR5, the first example of sequence-dependent heterochromatin initiator in the plant kingdom. Our results suggest that SUVR5 establishes the heterochromatic state by H3K9me2 deposition in a DNA methylation–independent manner that is not perpetuated and thus allows for changes in response to the environment or developmental cues.

Abstract Introduction Results Discussion Materials and Methods

molecular cell biology cell biology gene expression genetics plant genetics epigenetics biology dna modification plant cell biology molecular biology genetics and genomics histone modification

2012

The SET-Domain Protein SUVR5 Mediates H3K9me2 Deposition and Silencing at Stimulus Response Genes in a DNA Methylation–Independent Manner

Typhoid and paratyphoid fever are endemic in Hongta District and their prevalence, at 113 per 100,000 individuals, remains the highest in China. However, the exact sources of the disease and its main epidemiological characteristics have not yet been clearly identified. Numbers of typhoid and paratyphoid cases per day during the period 2006 to 2010 were obtained from the Chinese Center of Disease Control (CDC). A number of suspected disease determinants (or their proxies), were considered for use in spatiotemporal analysis: these included locations of discharge canals and food markets, as well as socio-economic and environmental factors. Results showed that disease prevalence was spatially clustered with clusters decreasing with increasing distance from markets and discharge canals. More than half of the spatial variance could be explained by a combination of economic conditions and availability of health facilities. Temporal prevalence fluctuations were positively associated with the monthly precipitation series. Polluted hospital and residential wastewater was being discharged into rainwater canals. Salmonella bacteria were found in canal water, on farmland and on vegetables sold in markets. Disease transmission in Hongta district is driven principally by two spatiotemporally coupled cycles: one involving seasonal variations and the other the distribution of polluted farmland (where vegetables are grown and sold in markets). Disease transmission was exacerbated by the fact that rainwater canals were being used for disposal of polluted waste from hospitals and residential areas. Social factors and their interactions also played a significant role in disease transmission. Typhoid and paratyphoid fever are serious infections, particularly in low-income countries, causing about 16 million cases and 600,000 deaths annually worldwide [1]. These diseases are notorious for their high infection rate, long duration, and heavy health burden. In China, typhoid and paratyphoid have been recorded daily since 2004 by the National Infectious Diseases Reporting Information System (NIDRIS) of the Chinese CDC, which enables all health care institutes across the country to report individual case information of important infectious diseases in real time using the Internet. Typhoid and paratyphoid fever are now regarded as being under control nationwide, and during 2010 the national disease incidence was 1. 2 per 100,000. However, prevalence, varied considerably from place to place with the highest incidence of 113 per 100,000, occurringinHongta district of Yunnan province in southwestern China [2], [3]. Water and food sanitation and environmental awareness generally can effectively reduce food and water borne diseases. A total of 3538 cases of typhoid and paratyphoid fever were reported in Hongta district during the period 2006–2010. Sanitation conditions in general are better for Hongta residents than for people in other parts of the province: a previous investigation showed that 50% of the population did not drink unboiled water, 91% washed their hands before dinner and 79% washed fruit before eating it. The case-control study [4], which included80 pairs of cases and controls, showed that adding fresh mint (OR = 2. 17,95%CL: 1. 04–4. 54) to breakfast, eating uncooked vegetables (OR = 2. 29,95%CL: 1. 24–4. 24) at restaurants or roadside food sites, and eating flavoring that contained fresh caraway and mint (OR = 2. 38,95%CL: 1. 00–5. 69) are all risk factors for typhoid and paratyphoid fever in the Hongta district. The incidence reaches its highest peak during the June–October period, and then decreases, reaching its lowest value during February. This pattern is consistent with the seasonal characteristics of other intestinal infectious diseases. During the high incidence period the weather is hot and humid, allowing more rapid bacteria reproduction; also during this weather people often eat more raw and cold food. The cycle has been repeated for many years and the situation has become chronic. Furthermore, in the Hongta district, infection is often left untreated and only 42. 6%–47. 9% of carriers receive medical care. This leads to a large number of infection sources and carriers, and contributes significantly to the persistence of these diseases in the region. In recent years, intervention has become a crucial determinant of disease transmission [5]–[7]. During 2008, the typhoid and paratyphoid fever control program (2008–2010) in the Hongta district focused on the restriction of fresh caraway and mint eating in restaurants and roadside food sites, the surveillance and hospital treatment of patients and carriers using appropriate antibiotics, improvement of the environment, health education, food hygiene (especially for fresh vegetables), drinking water disinfection and vaccination. As a result, the morbidity in the Hongta district dropped from 232 per100,000 during 2000–2007 to 113 per 100,000 in 2010, a decrease of about 51%, but still high compared with other regions of China. Intervention would be much more efficient if the disease sources, transmission paths and factors were accurately identified. According to other studies [8]–[16], and our present work, the following disease transmission hypotheses are proposed: (1) Polluted water from hospitals and residential sewage discharged into canals is an important disease determinant. (2) Polluted vegetables sold in the markets constitute another determinant. (3) Social and environmental factors interact together to influence disease transmission. The environmental health processes and variables involved in the above hypotheses are distributed across space and time [17]. Understanding this pattern should provide valuable clues about disease sources and determinants. The disease pattern can be rigorously addressed using stochastic spatiotemporal analysis [18]–[20] and powerful GIS technology [21]–[25] available at a relatively lower cost than the mainstream epidemiological techniques. A non-random disease distribution will always display spatial disease clusters [22]. The concordance between the spatial pattern of a disease and that of a contributing factor or determinant usually indicates the power of that determinant [23], [24]. Integration of GIS with statistical models of disease distribution forms a nationwide web-based automated system for early disease outbreak detection and rapid response in China [25]. The dataset and tools have already been used in the analysis of typhoid transmission [26]–[28]. The main objective of the present study is to identify the sources, transmission processes, and determinants of typhoid and paratyphoid fever in the Hongta District, based on the space-time analysis of the available epidemiological and notification data. The Hongta district is located in the center of the Yunnan province, with coordinates 24°08′N–24°32′N, 102°17′E–102°41′E, at an elevation of 1630 m, (see Figure 1). The district' s climate is subtropical and sub-humid. The total area is 1004 km2,85% of which is mountains and the rest is irrigated farmland. The total population is 420,553, of which 268,635 are rural people. There are 11 towns and 81 administrative villages. The per capita income of the farmers is 4432 CNY, which lies in the middle of the range of farmers' income country-wide. Data on incidence included both probable and laboratory-confirmed cases as defined by the national Salmonella typhi and paratyphoid standards. Illness caused by S. typhi is often characterized by insidious onset of sustained fever, headache, malaise, anorexia, relative bradycardia, constipation or diarrhea, and non-productive cough. Laboratory criteria for diagnosis are based on positive isolation of S. typhi or paratyphi in blood, stool, or other clinical specimen. A confirmed case is defined as a clinically compatible case that is laboratory-confirmed. It was estimated that the proportion of unreported cases was below 5%. These cases generally concerned mild illnesses identified in both hospitals and community. Daily numbers of confirmed typhoid and paratyphoid cases during the period 2006–2010 were obtained from the national infectious diseases reporting information system of the Chinese CDC. The system covers 95% of the country' s population and provides certain patient information, including personal identification number (ID), specimen used for testing paratyphoid pathogen, and family address defined by GPS. Typhoid and paratyphoid cases were diagnosed according to clinical symptoms and blood culture Salmonella-positive test (the national diagnostic criteria for typhoid and paratyphoid fever were used [29]). All study data were stripped of personal information. Demographic and socio-economic data were obtained from the local Statistics Bureau and included GDP per capita, number of medical institutions, population, percentage of farmers in the population, and geographical distribution of the villages. Annual precipitation and other meteorological data were available from the China meteorological data-sharing network, and were interpolated using the spatial kriging technique to cover the entire study area. Geomorphic data were obtained from the China geomorphy map (1∶1,000,000 scale), and seven categories were defined: plain, terrace, hill, low-relief mountain, middle-relief mountain, high-relief mountain and extremely high-relief mountain. Geomorphic data were provided by the state key Laboratory of Resources and Environmental Information Systems from the Institute of Geographic Sciences and Natural Resources Research, Chinese Academy of Science. The Normalized Difference Vegetation Index (NDVI), acquired by satellite remote sensing imaging, reflects greenness of vegetation, photosynthetic rate, and vegetation metabolic intensity and measures seasonal and inter-annual changes. This makes NDVI an efficient indicator of vegetation and eco-environment in the locations where typhoid/paratyphoid Salmonellae occur. We used monthly NDVI values with spatial resolution 1 km by 1 km. The spatial distribution of canals, markets, water wells, streets, and landuse in Zhoucheng town (Hongta district) were provided by the Yuxi city Bureau of Mapping and Survey. Field surveying showed that hospital disposal and residential sewage are connected and discharged into the rainwater canal system. Water and vegetable samples were collected from markets and farmland, canals irrigating the farmland, hospital disposal products and residential sewage. The samples were tested by pulsed field gel electrophoresis, a technique used for the separation of and identification of Salmonella DNA. The data were organized either into excel tables or handled by means of ENVI and ArcGIS 10. 1 software. Figure 2 outlines the study design. The observed yearly disease incidence for towns located in the Hongta district is shown in the top part of the diagram. The bottom part of the diagram depicts the infected patient and the route taken by the patient' s unsterilized excretion products to water sources that are then used to irrigate vegetables grown on the farmland. Transmission to susceptible patients may occur either directly via flies or polluted water or via vegetables grown on the polluted farmland. Infected individuals may recover and become susceptible, and then become infected again. Untreated individuals may recover but remain carriers of the disease. As noted earlier, disease transmission mechanisms depicted in the bottom part of diagram and observations shown along the top, have suggested certain study hypotheses involving disease sources and factors (x), transmission (y∼x), and spatiotemporal patterns (y). These hypotheses can be investigated by means of rigorous spatiotemporal analysis and laboratory testing. Cases and suspected determinant data were collected for this purpose. Spatial modeling goes beyond standard data analysis by providing new dimensions that can process additional information, discriminate factors that may not be perceivable (in simple time series analysis for example), and locate sources and factors for accurate intervention. We examined clustered and geographically correlated cases, spatial patterns linked to suspected sources, and case distribution-disease determinant consistencies suggesting transmission mechanisms. These were analyzed by means of, local indicator of spatial association (LISA), spatial buffering, and geographical detector techniques [24] respectively. Multivariate analysis was performed by the interaction detector technique, measuring the spatial consistency between disease distribution and a distribution formed by overlaying the distributions of the multivariate of the determinants. Multivariate regression was not suitable because either the sample size was too small (9 towns in Hongta district) or the associated explanatory variables were not available (in villages and streets). Local Moran' s I, i. e. LISA measures the spatial autocorrelation of an attribute as [22]where xi is the attribute value in the spatial cell i; c is the matrix of spatial weights, e. g. , cij denotes the weight of relationship between cells i and j; and s2 denote the mean and variance of x, respectively. A positive Ii implies that the cell i and its neighbors are similar; e. g. , high-high or low-low cells. A negative Ii implies that the cell i and its neighbors are dissimilar; e. g. , high-low or low-high cells. LISA was performed using the publically available software GeoDA software. An object' s buffer refers to a zone covering a specified distance around it. For a single object a series of buffers may be considered at increasing distances. If the object is a point (e. g. a market), its buffers are a series of circles centered at that point. If the object is a line (e. g. , a canal), its buffers are a series of belts parallel to the line. If the prevalence in the buffers exhibits a trend with distance from an object, it implies that the specified object is associated with disease transmission. Buffer statistics were calculated using ARCGIS software. The geographical detector [24] is used to assess environmental risks to human population. The method is different from conventional mathematical correlation or regression, which require that the determinants or their proxies are quantitative. It is assumed that the disease will exhibit a spatial distribution similar to that of an environmental factor if the environment contributes to disease transmission, as measured by the Power of Determinant (PD), where ℜ and σ2 denote the area and the dispersion variance of disease incidence of the study area, respectively. The study area is stratified into L stratums, h = 1, …, L [23] according to spatial heterogeneity of a suspected determinant or its proxy of the disease. Note that spatial heterogeneity is defined as an attribute whose statistical properties (e. g. , mean and standard deviation) change in space. PD ∈[0,1], where 1 indicates that the determinant completely controls the disease, and 0 indicates that the determinant is completely unrelated to the disease. In other words, PD expresses the extent to which a determinant explains disease incidence. In addition, disease transmission is often determined by multiple factors. The PD, combined with spatial overlay techniques of GIS and set theory, yields a geographical detector that can effectively assess the relationship between multiple factors: Are these factors independent or interacting? Are they enhancing or weakening each other? Is their interaction linear or nonlinear? In the present study, the geographical detector was implemented using the software www. sssampling. org/geogdetector (publicly available). The overall disease distribution as a function of time is shown in Table 1. Figure 3 shows that the disease incidence is in inverse phase with NDVI (Figure 3a) and in the same phase as precipitation (Figure 3b). Figure 4 displays yearly incidence in 9 towns of the Hongta district during the period 2006–2010, with the exception of 2009, together with population densities and landuse. Towns are categorized as having low (L), medium (M) and high (H) population densities. Zhoucheng town is located at the south-eastern and central area of Hongta district and is made up of 51 villages. It has the highest population density and much higher incidence than towns in the western and northern hill woodland, which is why it was made the focus of this study. LISA statistic (P = 0. 01) in Figure 5, shows that low incidence villages are close to other low incidence villages in the outskirts, whereas villages forming high incidence clusters are located close to the town centre. Higher disease incidence occurs in areas with a larger proportion of farmers in the population. Data relating to distribution of canals and markets are also available. Figure 6 displays the associations between disease incidence and suspected determinants in Zhoucheng town. The x-axis represents the distance between sewer and confluence (in Figure 6a) and between market and well (in Figure 6b), shown with their respective buffer zones. The y-axis denotes years (2006–2010), and the z-axis the population incidence. Figure 6a shows that the population incidence increases with decreasing distance from the canals. Interestingly, the incidence reduces rapidly away from confluence canals (note: blue bars rapidly shorten along buffer-axis), whereas it reduces rather slowly away from sewage canals (red bars slowly shorten along buffer-axis). Figure 6b shows that disease incidence reduces with distance from markets and groundwater wells, the decline being more rapid from the markets (red bars rapidly shorten along buffer-axis) than from the wells (blue bars slowly shorten along buffer-axis). Figures 6a and b, show that the abnormally high incidence during 2009 occurred within a 500 m buffer zone of sewers and confluences and an 800 m buffer zone of markets and wells, suggesting that the suspected disease sources are not only a cause of the observed spatial disease distribution but they were also responsible for the 2009 disease outbreak. After being standardized as (y−min) / (max−min), the time series of incidence, precipitation and NDVI were analyzed. The typhoid fever and the paratyphoid fever both fluctuated in time, experiencing two peaks per year; the first peak from April to June and the second from September to November. The incidence was positively correlated with the precipitation time series and usually peaked about one month earlier or one month later than the precipitation peak. NDVI is an index reflecting biological mass, and its time series appears to be negatively correlated with disease incidence. Laboratory testing indicated that the Salmonella strains found in water samples from hospitals, canals, farmland, and market vegetables belonged to closely related clones of S. typhi. Vegetables sold in the markets came from farmland irrigated by water contaminated by hospital and residential sewage. The findings of our spatiotemporal analysis, that hospital and residential sewage discharge and food markets are the most probable routes of disease transmission, support previous surveys carried out in this area. In addition to these direct disease sources, other environmental factors are believed to exert an influence on disease transmission. These include factors such as economic and social status, types of disease intervention, climate-biology types and physical conditions. However, these factors are rather difficult to measure in a straightforward manner. Instead, they are represented by their proxies: GDP per capita (economic determinant), proportion of farmers in total population (social determinant), number of medical institutions (intervention), NDVI (climate-biology type), and geomorphic types (environmental conditions). Using the geographical detectors it was found that social factors have the highest disease impact of all those measured (Table 2). The combined impact of two factors A and B on disease incidence generally differs from the simple linear summation of the separate A and B impacts, i. e. , A ∩ B≠A+B. The spatial incidence distribution is highly correlated with that of farmer proportion and population density. The proportions of spatial prevalence variation explained by these two factors reach PD = 78. 3% (for A) and 73. 8% (for B). The farmer proportion and population density together explain 84. 7% of the variation of disease incidence, higher than that explained by any single factor. The incidence is weakly associated with the economic condition (GDP per capita) and the climate-biological type (NDVI), with PD values of 9. 7% and 1. 8%, respectively. However, the incidence is more strongly linked with the spatial combination of GDP and NDVI. More than half (53. 8%) of the spatial variation of the incidence can be explained by the spatial distribution of economic condition (GDP per capita) combined with availability of medical intervention (number of medical institutions). The individual determinant impact of GDP and NDVI are clearly low (8. 7% and 18. 7% respectively). Geomorphology explains only a small part of spatial disease transmission, but its impact increases when it is combined with population density or GDP per capita. These findings jointly imply that a policy instrument based on analysis of a combination of the most relevant determinants, (in this case, population density and farmer proportion), would be highly efficient in controlling spatial disease transmission in a region. Based on the results of rigorous spatiotemporal analysis and GIS technology, disease transmission processes can be identified, modeled, and used for population health management purposes. The findings of the present Hongta district study include the following: Unsterilized water from hospital disposal and residential sewage are discharged into the rainwater canal system to irrigate the region' s farmlands where local vegetables are grown. Polluted vegetables are sold in the markets and consumed without being adequately washed. Infection is magnified during the rainy seasons, when more contaminated water irrigates the farmlands and the polluted water, mixed with additional garbage spreads over a larger area of the inhabited region. Over time, the combination of the farmers' poor hygiene conditions and low cure rates lead to considerable pathogen reserves among the population. The general seasonal-intestinal infection cycle, driven mainly by annual temperature fluctuations and varying volume of polluted canal water in the irrigation system driven mainly by the rainy seasons, combine to cause the persistent high endemic transmission of the disease in the Hongta district. The availability of spatiotemporal data and the development of health information systems facilitate the GIS-spatiotemporal exploration of public health issues. The physical and social determinants of health are usually spatially distributed: surface and subsurface contaminated water, polluted water emission from workplaces, disease prevention methods, and nutrition and food habits. More importantly, the susceptible and the infected populations are always geographically distributed. Health determinants may be detected when the disease incidence shares similar spatial features with environmental features suspected of being associated with the disease. When disease cases and environmental features do not share similar spatial features, however, this does not necessarily mean that they are unrelated. Instead, the geographically dispersed disease cases may have the same casual sources as features which may be elucidated by alternative techniques such as molecular tracking in food or serum samples, and food electronic label tracking systems. This study identified some of the disease sources, determinants, and transmission cycles, which affect the distribution of typhoid and paratyphoid cases in the Hongta district which currently exhibits the highest incidence in China. This knowledge can be used to design intervention measures to reduce and hopefully eradicate the disease in this area. In addition to the measures common to the control of enteric infectious disease intervention, such as washing hands before eating and environmental sanitation, particular attention should be paid to the sterilization of hospital waste prior to discharge into canals, especially during the rainy season.

Typhoid and paratyphoid epidemics are serious events in low-income countries; these diseases are notorious for their high infection rate, long duration, and heavy health burden. In China, typhoid and paratyphoid are considered to be under control, although the situation varies considerably from place to place. During 2010 the disease incidence was 1. 2 per 100,000 at the national level. The highest incidence, 113 per 100,000, occurred in the Hongta District of Yunnan province, in southwestern China. We used quantitative spatiotemporal analysis to explore the relationship between the incidence of these diseases and a number of factors suspected of influencing their occurrence. We found that cases tended to occur near discharge canals and polluted farmland where vegetables are grown for sale in local markets. The spatial characteristics of disease transmission were associated with the seasonal variations common to all intestinal infectious diseases. The findings of this work could inform local public health planners and the health directorate and help to improve public health intervention programs in regions with the highest incidence of these diseases.

Abstract Introduction Methods Results Discussion

medicine public health and epidemiology epidemiology environmental epidemiology spatial epidemiology epidemiological methods

2013

Spatiotemporal Transmission and Determinants of Typhoid and Paratyphoid Fever in Hongta District, Yunnan Province, China

The vector-borne disease leishmaniasis is transmitted to humans by infected female sand flies, which transmits Leishmania parasites together with saliva during blood feeding. In Iran, cutaneous leishmaniasis (CL) is caused by Leishmania (L.) major and L. tropica, and their main vectors are Phlebotomus (Ph.) papatasi and Ph. sergenti, respectively. Previous studies have demonstrated that mice immunized with the salivary gland homogenate (SGH) of Ph. papatasi or subjected to bites from uninfected sand flies are protected against L. major infection. In this work we tested the immune response in BALB/c mice to 14 different plasmids coding for the most abundant salivary proteins of Ph. sergenti. The plasmid coding for the salivary protein PsSP9 induced a DTH response in the presence of a significant increase of IFN-γ expression in draining lymph nodes (dLN) as compared to control plasmid and no detectable PsSP9 antibody response. Animals immunized with whole Ph. sergenti SGH developed only a saliva-specific antibody response and no DTH response. Mice immunized with whole Ph. sergenti saliva and challenged intradermally with L. tropica plus Ph. sergenti SGH in their ears, exhibited no protective effect. In contrast, PsSP9-immunized mice showed protection against L. tropica infection resulting in a reduction in nodule size, disease burden and parasite burden compared to controls. Two months post infection, protection was associated with a significant increase in the ratio of IFN-γ to IL-5 expression in the dLN compared to controls. This study demonstrates that while immunity to the whole Ph. sergenti saliva does not induce a protective response against cutaneous leishmaniasis in BALB/c mice, PsSP9, a member of the PpSP15 family of Ph. sergenti salivary proteins, provides protection against L. tropica infection. These results suggest that this family of proteins in Ph. sergenti, Ph. duboscqi and Ph. papatasi may have similar immunogenic and protective properties against different Leishmania species. Indeed, this anti-saliva immunity may act as an adjuvant to accelerate the cell-mediated immune response to co-administered Leishmania antigens, or even cause the activation of infected macrophages to remove parasites more efficiently. These findings highlight the idea of applying arthropod saliva components in vaccination approaches for diseases caused by vector-borne pathogens. As a vector-borne parasitic and tropical disease, leishmaniasis has been reported in 98 countries (in 4 continents) with an annual incidence level of 0. 9–1. 6 million individuals, and 350 million individuals are at the risk of this infection [1,2]. The Leishmania parasite, as the causative agent of this disease, is delivered to the host during blood-feeding by infected female sand flies, namely, the Phlebotomus species in the Old World (Asia, Africa and Europe) and the Lutzomyia species in the New World (Central and South America). Various clinical symptoms, in visceral, mucocutaneus and cutaneous forms have been reported for leishmaniasis. The most commonly observed type of this disease is CL, which results in ulcers and permanent scars on the affected regions of the body and serious disabilities for the patient. Clinical forms of leishmaniasis are geographically distributed depending on the availability of the pathogenic species of Leishmania and their vectors. For example, CL has been mainly observed in Afghanistan, Brazil, Iran, Iraq and Syrian Arab Republic [1]. The fact that infection with certain species, including L. major or exposure to live Leishmania (leishmanization) results in long-term protection, in addition to the high costs of current treatments, drug toxicity and parasite resistance emergence intensify the need for producing a vaccine for this disease [3–5]. However, despite numerous investigations, no human vaccine for Leishmania is yet available [6]. Studies on development of Leishmania vaccines have focused on Leishmania antigens. In the past two decades, researchers started to examine sand fly salivary proteins [7], which are transmitted along with the parasite during blood feeding, as an alternative or a component of a vaccine against leishmaniasis. Different molecules are present in sand fly saliva which alter the hemostatic, inflammatory and immune response of the host, thereby facilitating parasite infection [8]. Simultaneous inoculation of Leishmania parasite and sand fly saliva has been reported to increase the parasite burden and cause exacerbation of the generated lesions [7,9–12]. However, immunization of animals with salivary components [13–15] or salivary gland homogenate (SGH) [12,16,17] from sand flies or exposing them to the bites of uninfected sand flies [18–25] resulted in protection against infection with Leishmania [26]. A cell-mediated immune response (CIR), as a form of a DTH response to sand fly salivary proteins, was shown to provide protection against Leishmania infection. This anti-salivary immunity appears to be important, considering that the parasite is unavoidably naturally injected along with salivary proteins into the biting site. Hence, similar to an adjuvant, a Th1 anti-saliva immunity can accelerate the induction of a protective Th1 immunity against Leishmania [27]. In Iran, zoonotic CL is caused by L. major, which is usually transmitted by the Ph. papatasi vector. On the other hand, anthroponotic CL is caused by L. tropica mainly transmitted by the Ph. sergenti sand fly [28]. According to the literature, exposing mice to the bites of uninfected sand flies or immunizing them with Ph. papatasi SGH can protect them against L. major infection caused by needle inoculation [7] or the bites of infected sand flies [13]. Immunization with PpSP15, a salivary protein from Ph. papatasi, provides protection in mice against L. major infection [20,21,29,30]. The immunity induced by Ph. sergenti salivary proteins and their effects on L. tropica infection have not been previously reported. Therefore, in this work, the immunogenicity of Ph. sergenti salivary components were assessed and we tested whether immunizing BALB/c mice with Ph. sergenti SGH or DNA plasmids coding for Ph. sergenti salivary proteins can lead to an appropriate immunity to control L. tropica infection. Apyrogenic deionized water (Milli-Q System, Millipore, Molshem, France) was used to prepare all needed solutions. The Endo-Free Plasmid Mega kit, RNeasy Mini Kit, Quanti Nova SYBR Green Master Mix and Anti-His antibody were purchased from QIAGEN, Germany. TRIzol Reagent and SuperScript III First-Strand Synthesis system were from Thermo Fisher Scientific (Invitrogen Company, USA). The materials needed for PCR reaction, agarose gel electrophoresis and enzymatic digestion were all provided by Roche Applied Sciences, Germany. Diaminobenzidine (DAB) powder, Bovine Serum Albumin (BSA) and acrylamide were provided by Merck, Germany. Horseradish peroxidase-conjugated goat anti-mouse IgG, Urea, Ponceau-S, Sodium dodecyl sulfate (SDS), Tris-base, Tris-HCL, M199 medium, RPMI-1640, DMEM, gentamicin, kanamycin, L-glutamine, hemin, HEPES, adenosine and Ficoll-400 were purchased from Sigma, Germany. Fetal Calf Serum (FCS) and Schneider insect media were provided by Gibco (Life Technologies, Germany). Goat anti-mouse IgG1-HPR and IgG2-HPR were obtained from Southern Biotech, Canada. Peroxidase Substrate System as an ELISA substrate was from KPL (ABTS, USA). Linear Polyethylenimine was purchased from Polyscience, Germany. Protran Nitrocellulose Transfer Membranes was from Schleicher & Schuell BioScience, Germany. GF-1 Tissue DNA Extraction Kit was purchased from Vivantis Technologies, Malaysia. The Agilent RNA 6000 Nano reagent kit were purchased from Agilent Technologies, USA. The Agilent 2100 Bioanalyzer instrument (Agilent Technologies, USA), ELISA reader (Tecan, USA), NanoDrop (Nanodrop, ND-1000, USA) and Speed Vac (Thermo Scientific, USA) devices were also used in this investigation. Ph. sergenti were kept in the insectary at the Laboratory of Malaria and Vector Research, National Institutes of Health, NIH (Rockville, MD, USA). The salivary glands were dissected from 5- to 7-day-old and non-blood fed female sand flies and then transferred to PBS for subsequent storage at −70 °C. After disruption by ultra-sonication and centrifugation, the produced supernatant was collected and dried in a Speed Vac device and reconstituted before use. Female BALB/c mice, (6–8 weeks old) with weight range of 18–20 g were obtained from Pasteur Institute of Iran. The animals under investigation were maintained, handled, anesthetized and euthanized under the approval of Institutional Animal Care and Research Advisory Committee of Pasteur Institute of Iran (ethical code: IR. RII. REC. 1394. 0201. 6417, dated 2015). All experiments were designed and carried out according to the Specific National Ethical Guidelines for Biochemical Research (2005) by the Research and Technology Deputy of Ministry of Health and Medicinal Education (MOHM) of Iran. Mice were euthanized through cervical dislocation method. Animals were anesthetized via intraperitoneal (i. p.) administration of Xylazine/Ketamine anesthetizing cocktail [31] in order to minimize suffering animals under investigation. The mammalian codon optimized nucleotides encoding the N-terminus to the stop codon of the most abundant secreted proteins from Ph. sergenti salivary gland proteins were cloned in a modified mammalian expression plasmid (VR1020-TOPO) through T/A cloning strategy and topoisomerase technology and then transformed into the DH5α strain. The VR1020-TOPO plasmid has features such as a CMV promoter, the signal-secretory peptide of tissue plasminogen activator (TPA), replacing the sand fly specific-secretory signal peptide and a 6×His-tag, downstream from the target insert. This modified vector enables efficient production of the secreted proteins in animal tissues and other mammalian-based expression systems. The fourteen cloned transcripts used in this study are PsSP7 (HM560864, D7-related proteins), PsSP9 (HM569364, PpSP15-like protein), PsSP14 (HM560870, PpSP15-like protein), PsSP15 (HM560868, PpSP15-like protein), PsSP20 (HM560866, yellow-related proteins), PsSP26 (HM569362, yellow-related proteins), PsSP40 (HM560860, apyrase), PsSP41 (HM560862, apyrase), PsSP42 (HM560861, apyrase), PsSP44 (HM569368, PpSP32-like protein), PsSP52 (HM537134, antigen 5-related proteins), PsSP54 (HM569365, PpSP15-like protein), PsSP73 (HM569367, unknown), and PsSP98 (HM569366, unknown). In order to purify all recombinant plasmids and the empty control plasmid (VR1020), an Endo-Free Plasmid Mega kit was used according to the manufacturer’s protocol. In order to confirm the expression of the 14 plasmids harboring each Ph. sergenti salivary protein, COS-7 cells (ATCC CRL-1651) were used as an expression host. Briefly, COS-7 cells were cultured in six-well plates (Greiner) in complete RPMI medium supplemented with 10% FCS at 37 °C in the presence of 5% CO2. For cell transfection, we used PEI/DNA complexes, prepared through mixing Linear Polyethylenimine (LINPEI, MW = 25 kDa, 10 μM) and 5 μg of each recombinant plasmid or VR1020 (as the control) [32]. Expression of Ph. sergenti salivary proteins in transfected COS-7 cells were confirmed by western blot analysis. In brief, forty eight hours after transfection, the supernatant of transfected COS-7 cells were harvested, then mixed with SDS-PAGE sample buffer and boiled for 5 min, and run on a 12. 5% or 15% SDS-PAGE gel. Proteins were then transferred from gel to nitrocellulose membranes by electro-blotting. Free binding sites on nitrocellulose membrane were blocked with blocking solution (PBS with 0. 05% Tween 20 and 2. 5% BSA) for 2 hours. After three washes, the membrane was incubated with HRP-conjugated goat anti-mouse IgG (1: 2000) for 2 hours at room temperature and visualized using the 3,30-diaminobenzidine substrate (DAB). In order to assess the immunogenic characteristics of DNA plasmids coding for Ph. sergenti salivary proteins, 6–8 weeks old BALB/c mice (6 mice in each experimental group) were i. d. immunized three times at two-week intervals using 30-gauge needle in the right ear. For this purpose, we used 10 μg of plasmid (either the empty plasmid control or a recombinant plasmid encoding a Ph. sergenti salivary protein), an equivalent of 0. 5 Ph. sergenti salivary gland pair or PBS; all in a total volume of ~10 μl with PBS. ELISA microplates were coated with 100 μl of SGH diluted to two pairs of SGHs/ml in coating buffer (Na2CO3 0. 02 M, NaHCO3 0. 45 M, pH 9. 6) overnight at 4 °C. Following rinsing with PBS-0. 05% Tween, wells were blocked using 100 μl of 1% BSA in PBS for 2 h at 37 °C. Wells were incubated for 3 h with sera from mice immunized with recombinant or control plasmids obtained two weeks after the last immunization and diluted (1: 50) in PBS-0. 05% Tween-1% BSA. After one more washing step, wells were incubated with Horseradish peroxidase-conjugate goat anti-mouse IgG diluted (1: 5000) in PBS-0. 05% Tween-1% BSA for 2 h at 37 °C. Following another washing step, plates were incubated with Peroxidase Substrate System (KPL) as the substrate for 30 min at 37 °C. After stopping the reactions with 1% SDS, the absorbance was measured at 405 nm using an ELISA reader. The cut off value was determined by measuring anti-saliva IgG of plasmid control mice group (mean + 3 SD). Two weeks after the last step of immunization, the animals were inoculated i. d. into the left ear dermis with Ph. sergenti SGH (0. 5 salivary gland pair per mouse) using a 30-gauge needle. Forty eight hours later, we measured ear thickness (DTH response) using a digital caliper (with a resolution of 0. 01 mm). For histopathological assessment at this time point, after fixing the dissected ears in 10% phosphate-buffered formalin, they were processed, embedded in paraffin, and then the 5-μm sections prepared by microtome were stained using hematoxylin and eosin (H & E) and analyzed using light microscopy. For morphometric analyses, inflammatory cells were counted in three fields/section using a 400× magnification, covering a total area of 710 mm2. After screening the Th1 DTH response in BALB/c mice which was triggered by the saliva proteins of Ph. Sergenti, the animals (10 mice per group) were immunized three times with two-week intervals using the selected plasmid encoding Ph. sergenti salivary protein, SGH, empty plasmid or PBS in the right ear using a 30-gauge needle. Two weeks after the last immunization, animals were challenged i. d. into the left ear dermis with 107 metacyclic L. tropica parasites plus Ph. sergenti SGH (0. 5 salivary gland pair) using a 30-gauge needle in an almost 10 μl total volume. The L. tropica parasite named as MOHM/IR/09/Khamesipour-Mashhad was isolated from patient in city of Mashhad, Iran, in 2009 (provided as gift by Dr. Ali Khamesipour). L. tropica promastigotes were cultured in M199 medium supplemented with 10% hi-FCS. The Ficoll-400 step-gradient was used to isolate the L. tropica metacyclic promastigotes [33]. To mimic the natural model of infection and examine whether immunity to the salivary proteins of the sand fly can protect the animals against CL, mice were infected by injection of parasites together with Ph. sergenti SGH into their ear dermis. The ear thickness was monitored and measured weekly using a digital caliper. To measure the disease burden (area under the curves, AUC), the ear thickness of each individual immunized mouse was recorded once per week. A disease course curve for each mouse in the experimental and control groups was separately obtained. Prism (GraphPad Software) was used to calculate AUC. Cytokine profiles were analyzed at two time points: once after Ph. sergenti SGH inoculation (to screen the Th1 DTH response against each DNA plasmid), and once after infectious challenge with L. tropica plus Ph. sergenti SGH. For this purpose, total RNA was extracted from the mouse dLN using TRIzol reagent and then RNeasy Mini Kit. The quality of extracted RNA was confirmed using an Agilent RNA 6000 Nano reagent kit. For first-strand cDNA synthesis, approximately 2 μg of RNA reverse-transcribed in a total volume of 20 μl using SuperScript III reverse transcriptase according to the manufacturer’s instructions. For quantification of gene expression, the 1: 10 diluted cDNA was subjected to the reaction containing 5 pmol of each forward and reverse primer and 12. 5 μl Quanti Nova SYBR Green Master Mix in a 25 μl total volume. Real-time PCR reactions were performed in duplicates on an Applied Biosystems 7500 instrument. Thermal cycles with an initial incubation step at 95 °C for 5 min followed by 45 cycles at 95 °C for 10 s, at 60 °C for 15 s, and at 72 °C for 35 s. The mRNA levels of each target gene were normalized to that of HPRT. The results are shown in fold change compared to the PBS control. Gene expression was analyzed based on the comparative method. The cycle threshold (Ct) values for cytokines were normalized to the expression of HPRT based on the following formulation: ΔCt = Ct (target gene) −Ct (HPRT gene). We obtained the fold change using 2−ΔΔCt, in which ΔΔCt = ΔCt (test) −ΔCt (control) [34]. We used the following primers for real-time PCR: HPRT (Forward: 5′-GTCCCAGCGTCGTGATTAG-3′; Reverse: 5′-GAGCAAGTCTTTCAGTCCTGTC-3′); IFN-γ (Forward: 5′-TCTGAGACAATGAACGCTACAC-3′; Reverse: 5′-CTTCCACATCTATGCCACTTGAG-3′); IL-5 (Forward: 5′-TGACAAGCAATGAGACGATGAG-3′; Reverse: 5′-CTCCAATGCATAGCTGGTGA-3′). Quantification of the parasites in the infected ear of animals in different groups was performed using Real-time PCR at one and two months after challenge. After euthanizing animals in each group (5 mice per group) the genomic DNA was extracted from each infected ear using GF-1Tissue DNA Extraction Kit according to manufacturer’s instruction. DNA concentration was measured by a NanoDrop device. The following primer set was used to target a part of L. tropica kinetoplastid minicircle DNA: (KDNA1F: (5' -GGGTAGGGGCGTTCTGC-3' ) and KDNA1R (5' -TACACCAACCCCCAGTTTGC-3' ) ) [35,36]. The absolute copy number corresponding to the target sequence was measured on an Applied Biosystems 7500 real time PCR system. Standard L. tropica genomic DNA was used in 10-fold dilution corresponding to 2×108 to 2×101 parasites for drawing the standard curve. To quantify the parasites in tissues, 50 ng of DNA was applied to a reaction containing 5 pmol of each of the forward and reverse primers and 12. 5 μl of Quanti Nova SYBR Green Master Mix in a 25 μl total volume. The PCR program was as the following: at 95 °C for 5 min; 40 cycles at 95 °C for 10 s, at 60 °C for 15 s, and at 72 °C for 35 s. All reactions were performed in duplicate. We replicated each measurement twice and averaged the obtained results. Shapiro-Wilk test was used to check the distribution of DTH response, Antibody response, IFN-γ, IL-5 and ratio of IFN-γ to IL-5 mRNA expression in dLN as well as parasite burden and disease burden (area under the curves, AUC). Due to the non-normality of all data, non-parametric van der Waerden’s normal score test and Dunn’s multiple comparison tests were used to compare the distribution of these variables between experimental and control group. To assess the effect of variables on ear thickness, we used Linear Mixed Models for repeated measured data. The significant interaction between time and group factors implied that the experimental groups were changing with time, with different manners. A statistically significant interaction effect may indicate that the overall patterns of differences at the level of main effects are not likely to be consistent across all groups. The p values below 0. 05 were considered as significant. Statistical analyses were carried out using Stata (14. 0) (StataCorp. 2015. Stata Statistical Software: Release 14. College Station, TX: StataCorp LP) and R 3. 4. 3 (R Core Team (2017). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria) using the Rfit, PMCMR, nparcomp and multcomp packages. Before screening the 14 different plasmids coding for Ph. sergenti salivary proteins in animals, we tested the protein expressions of these DNA constructs in COS-7 cells. The construct has a signal secretory peptide and we added to each transcript a histidine tag. We observed expression of protein in COS-7 cells from all plasmids (Fig 1). The 14 different DNA plasmids coding for Ph. sergenti salivary proteins were screened to select a plasmid that can induce a Th1 cellular immune response. We also compared these plasmids to Ph. sergenti salivary gland homogenate. The parameters for selection were 1) The presence of a distinct DTH response associated with mononuclear cell infiltration in the ear and 2) increased levels of IFN-γ and low levels of IL-5 expression in dLN in comparison with the empty plasmid control group. Surprisingly, immunization with the whole Ph. sergenti salivary gland homogenate (SGH) did not induce a detectable DTH response (Fig 2A). In contrast, seven plasmids induced a positive DTH and showed greater ear thickness than the control plasmid that produced a median of 0. 18 mm (Fig 2A). The median ranks of DTH responses were in the following descending order: PsSP40 (0. 26 mm), resulting in the greatest measurable response in animal skin; PsSP52 (0. 24 mm), PsSP44 (0. 23 mm), PsSP9 (0. 23 mm), PsSP26 (0. 22 mm), PsSP41 (0. 22 mm) and PsSP42 (0. 21 mm), as the smallest detectable levels. Groups immunized with PsSP40 (p <0. 01), PsSP52 (p <0. 01), PsSP9 (p <0. 01), PsSP44 (p <0. 01), PsSP26 (p <0. 01), PsSP41 (p = 0. 04) and PsSP42 (p = 0. 03) showed significantly higher DTH responses compared to the control plasmid (Fig 2A and S1 Table). Specific total IgG antibody responses against Ph. sergenti SGH in the sera of all groups was measured. In comparison with the control groups (PBS and VR1020), the Ph. sergenti SGH-immunized mice group produced higher levels of anti-saliva-IgG than the cut off value (median at OD405 nm = 0. 40, S1 Table and Fig 2B). Moreover, the group immunized with PsSP26-encoding plasmid had the highest level of total IgG antibody production against Ph. sergenti SGH in comparison with the control plasmid group (median at OD405 nm = 1. 53, p<0. 01). In addition, the two other groups encoding the PsSP7 and PsSP44 proteins also produced higher levels of total IgG than the cut off value (PsSP44 with median OD405 nm = 0. 30, p = 0. 04 and PsSP7 with median OD405 nm = 0. 45, p = 0. 07) as demonstrated in Fig 2B and S1 Table. Based on histological analysis 48 h after Ph. sergenti SGH injection in the ear dermis of plasmid-immunized animals, PsSP9- and PsSP40-immunized groups were characterized by a robust mononuclear infiltration mainly containing macrophages, lymphocytes and a lower number of neutrophils compared to the PBS- and plasmid- control groups (Fig 3). The number of inflammatory cells recruited in the Ph. sergenti SGH-immunized mice group was moderate and there were fewer cells than in the PsSP9- and PsSP40-immunized groups (Fig 3). Furthermore, the number of inflammatory cells recruited in PsSP41- and PsSP52-immunized mice was similar to that of Ph. sergenti SGH-immunized mice. No detectable increase in the number of inflammatory cells recruited in PsSP42-immunized mice was observed, compared to the PBS- and plasmid- control groups (Fig 3). Based on the outcome of the DTH response, the antibody response (as shown in Table 1) and histological analysis against various salivary proteins, PsSP9, PsSP40, PsSP41, PsSP42, and PsSP52 plasmids were selected for further evaluation of the cellular immune response in BALB/c mice. Forty-eight hours after Ph. sergenti SGH injection, mice immunized with the above-mentioned plasmids were euthanized and IFN-γ and IL-5 cytokine expression was evaluated in their dLN by Real-time PCR (Fig 4). PsSP9 immunized mice was the only group that induced a statistically significant increase in IFN-γ mRNA expression, compared to the control plasmid group (median of fold change = 22. 30, p = 0. 04, Fig 4A, S2 Table), and was the group exhibiting the lowest level of IL-5 (Fig 4B, S2 Table), which translated to the highest ratio of IFN-γ to IL-5 expression (median of fold change = 17. 12, p = 0. 16, Fig 4C, S2 Table). In PsSP52-, PsSP41- and PsSP42-immunized mice, IL-5 expression was significantly higher compared to the control plasmid group (p<0. 01 in PsSP52, p = 0. 01 in PsSP41 and p = 0. 03 in PsSP42, Fig 4B, S2 Table). Furthermore, neither the IFN-γ expression nor the ratio of IFN-γ/IL-5 expression in these groups was significantly different from the control plasmid group (Fig 4A and 4C, S2 Table). The data also revealed that there was no significant difference in terms of IL-5 and IFN-γ expression between the PsSP40-immunized mice and the control plasmid group (Fig 4A and 4B, S2 Table). The exact p values and median (Q1-Q3) for the six tested samples are presented in S2 and S3 Tables. Based on the findings that the PsSP9-immunized group induced a Th1 immune response, we examined whether immunization with DNA coding for PsSP9 salivary protein can lead to protection of animals against L. tropica infection. BALB/c mice were immunized intradermally in their ears three times (with two-week intervals) with either Ph. sergenti SGH, plasmid encoding a PsSP9 salivary protein, PBS, or empty plasmid. Two weeks after the last immunization, animals were challenged with metacyclic forms of L. tropica plus Ph. sergenti SGH. PsSP9-immunized mice had significantly smaller nodules, indicated by ear thickness measurements, compared to the control plasmid group which showed a significantly greater ear thickness (Fig 5A, S4 and S5 Tables). The disease burden was determined based on the area under the curves (AUC) as shown in Fig 5B. The data indicated that there was a significant reduction in the disease burden after PsSP9 immunization, in comparison with the plasmid control group (p = 0. 02, Fig 5B, S6 Table). One-month post-infection, the parasite load in the ear of PsSP9-immunized mice was significantly lower than the control plasmid group (7. 2×105 parasites in the control plasmid group and 1. 4×105 parasites in the PsSP9-immunized group, p = 0. 02, Fig 5C, S7 Table). At two months post challenge, the parasite load in the ear in PsSP9-immunized mice remained lower than that in the plasmid control mice group, but the difference was not statistically significant (1. 0×106 parasites in control plasmid group and 2. 7×105 parasites in PsSP9-immunized group, p = 0. 36, Fig 5D). The reduced parasite load correlated with the observed lower ear thicknesses in PsSP9-immunized mice. SGH-immunized mice showed no significant differences in ear thickness or disease burden compared to the control groups (Fig 5A and 5B, S4–S6 Tables). In addition, at one and two month post-infection, there were no statistically significant differences in the ear parasite burden in SGH-immunized mice compared with the control groups (Fig 5C and 5D, S7 Table). The cytokine profile expression in the dLN was assessed at one and two months post-challenge. At one-month post-challenge, no significant differences in IFN-γ and IL-5 expression levels or in the ratio of IFN-γ to IL-5 expression were observed in any of the groups (Fig 6A–6C, S8 Table). At two months post-infection, IFN-γ expression was higher in the dLN of PsSP9-immunized mice than in the control plasmid group, although the difference was not statistically significant (median of fold change = 13. 29, p = 0. 05, Fig 6D, S9 Table). At this time point, the expression of IL-5 in dLN of PsSP9-immunized mice was not different from that in the control plasmid group (Fig 6E, S9 Table). Importantly, a significant increase in the ratio of IFN-γ to IL-5 expression was observed in PsSP9-immunized mice compared with the control plasmid group (median of fold change = 17. 06, p = 0. 04, Fig 6F, S9 Table). Exact p values and median (Q1, Q3) for the 5 studied samples are presented in S4–S9 Tables. It has been previously reported that bites of Ph. papatasi [13] and Ph. duboscqi [15,24], as well as the injection of Ph. papatasi [7], Lu. longipalpis [17], and Lu. whitmani SGH [16], protected rodents against leishmaniasis, manifested by reduced lesion size and decreased parasite burden. Moreover, exposure to sand fly SGH has been reported to result in a DTH response which is associated with the enhanced production of IFN-γ and IL-12 [13,37] or a higher IFN-γ to IL-4 ratio [7]. The current study demonstrated that immunization with whole Ph. sergenti SGH did not confer protection against L. tropica infection. Indeed, the ear thickness, disease burden and parasite load in the Ph. sergenti SGH-immunized mice were similar to those of the mice in the control groups. The lack of protection in the Ph. sergenti SGH-immunized mice against infection correlates with a low ratio of IFN-γ to IL-5 production in the dLN during infection. The current results were similar to the findings of Moura et al. , who reported that the immunization of BALB/c mice with Lu. intermedia SGH resulted in a non-protective Th2 immunity [38]. In contrast to whole SGH immunization, a single sand fly salivary protein was shown to influence the outcome of leishmaniasis. Moura et al. observed that DNA immunization with Lu. intermedia salivary protein Linb-11 resulted in an intense protective cell-mediated immunity against infection with L. braziliensis [23] in contrast to the whole Lu. intermedia saliva that exacerbated L. braziliensis infection [38]. The current results from our work also suggest that distinct Ph. sergenti salivary proteins induce different immune responses. Immunization with 14 plasmids coding for Ph. sergenti secreted salivary proteins resulted in the identification of 9 salivary proteins which produced either a positive DTH response, an antibody response, or both responses in BALB/c mice (Table 1). The triggering of a humoral immunity by such proteins is not necessary for anti-leishmaniasis protection. As demonstrated by Gomes et al. , LJM19 was the only saliva protein that induced strong DTH in hamster without any detectable antibodies and DNA immunization with this plasmid protected the animal against VL [19]. Salivary proteins that result in a Th1 type cellular immune response can be considered as suitable candidate for producing an anti-leishmaniasis vaccine. In this study, the type of immune response generated by the protective plasmid PsSp9 was characterized by a DTH response with no detectable antibody response and a significantly high ratio of IFN-γ to IL-5 expression in the dLN compared to the control group. The current study reveals that DNA immunization with the Ph. sergenti salivary protein PsSP7 can induce an antibody response, but no detectable DTH response. Moreover, DNA immunization with PsSP41, PsSP42 and PsSP52 produced a DTH response but induced high IL-5 expression in the dLN after Ph. sergenti SGH inoculation. Therefore, DNA immunization with these proteins shifted the immune response to a Th2 type. Hence, application of such salivary proteins of Ph. sergenti (i. e. PsSP7, PsSP26, PsSP41, PsSP42, PsSP44 and PsSP52) may exacerbate L. tropica infection or cause no protective impact. Interestingly, DNA immunization with Ph. sergenti salivary protein PsSP9 (among the 14 tested Ph. sergenti salivary proteins) produced a DTH with high mononuclear infiltration in the ear, a high ratio of IFN-γ to IL-5 expression in the dLN and no detectable antibodies 48 h after Ph. sergenti SGH inoculation. Therefore, this protein directed the immunity toward a Th1 response and was chosen as a candidate for producing an experimental vaccine against L. tropica. The mice immunized with PsSP9 displayed a smaller ear thickness, lower disease burden, and a lower parasite load in the ear in addition to a high IFN-γ to IL-5 expression ratio in the dLN compared with the empty control plasmid group two months after challenge. The draining lymph node is a part of the immune system, where the induction of specific immunity against an antigen occurs and where effector cells migrate to the skin and contribute to protection [39]. The protection observed in the PsSP9-immunized mice against CL can be explained by the anti-PsSP9 immunity at the parasite transmission site in the ear dermis, which may facilitate direct parasite killing by macrophage activation through IFN-γ. In fact, IFN-γ acts to restrict Leishmania growth in macrophages of mice and humans as well as the progression of leishmaniasis [40]. Vinhas et al. [41] demonstrated that the peripheral blood mononuclear cells (PBMC) of individuals who were subjected to uninfected Lu. longipalpis sand fly bites, exhibited IFN-γ expression after SGH stimulation. The production of IFN-γ was also associated with L. chagasi parasite killing in a macrophage-lymphocyte autologous culture [41]. This implies that after PsSP9 DNA immunization, the effector cells (IFN-γ+) might have migrated to the site of L. tropica and Ph. sergenti SGH inoculation, activating the macrophages and causing parasite killing which led to smaller lesions and reduced parasite load. The presence of an immune response against PsSP9 at the site of the parasite plus SGH injection in the dermis of the ear may act as an adjuvant to accelerate the triggering of a proper host immune response against Leishmania. Macrophage activation induces inflammatory responses and prevents the growth of Leishmania [42] by increasing the production of ROS, antigen presentation and maturation of phagosome [43]. After taking the L. tropica parasite and transporting the antigen, macrophages and Langerhans cells migrate and then differentiate into the APCs within the LNs and trigger a DTH response by CD4 expressing T cells. According to the findings of Oliveira et al. , the immune response caused by the salivary protein induces a protective immunity against Leishmania. The DNA immunization of mice with PpSP15 resulted in a Th1 immunity against L. major [20]. In the current study, it was also hypothesized that DNA immunization with PsSP9 may act similarly. Only two critical cytokines in the dLN of immunized mice were analyzed (i. e. IFN-γ and IL-5). Further investigation of early time points in the dLN and the ear skin at the site of parasite challenge are needed to determine precisely the contribution of anti-saliva immunity in direct parasite killing and its indirect role in accelerating the induction of immunity against Leishmania. It is noteworthy that as a PpSP15-like protein family member, the PsSP9 is related to a salivary protein (14 kDa) which has no structural or sequential homology to identified proteins in humans or other organisms. PsSP9 protein is closely similar to PpSP12 (70% identity), PdSP14 (69% identity), PpSP14 (51% identity) and PpSP15 (46% identity) salivary proteins. PpSP15 was the first salivary protein considered as a candidate for an anti-leishmaniasis vaccine [21]. Oliveira et al. showed that DNA immunization of mice with PpSP15 triggered a DTH response and induced high levels of IL-12 and IFN-γ expression 2h post exposure to Ph. papatasi bites [20,21]. This immune response was associated with the accelerated induction of an immunity and protection against Leishmania [20,21]. Recently, DNA immunization with PdSP15 from Ph. duboscqi salivary protein, as a member of the PpSP15 family, was demonstrated to protect non-human primates against L. major which was delivered through the bite of an infected sand fly [24]. Furthermore, PdSP15 has been reported to exhibit immunogenic characteristics in humans, considering that PdSP15 was recognized by PBMC and sera in individuals exposed to the bites of Ph. duboscqi [24]. In this regard, it is suggested that among different Ph. sergenti salivary proteins, PsSP9 would be a better inducer of a Th1 protective immune response in a mouse model. Further investigations are still required to assess the immunogenic characteristics of PsSP9 as a recombinant protein vaccine for mice and whether it exhibits immunogenicity in human subjects, so as to be nominated as a potential candidate or as a component of a human anti-leishmaniasis vaccine. In summary, the current study demonstrated that while the whole Ph. sergenti SGH did not induce any protective immune response against L. tropica infection, a particular salivary protein from Ph. sergenti, PsSP9, induced a powerful protection against L. tropica in BALB/c mice. These findings highlight the application of specific salivary proteins from sand flies in anti-leishmaniasis vaccination approaches, which would be improved when used in combination with a Leishmania antigen and/or another prophylactic agent.

Leishmaniasis is a vector-borne disease transmitted to humans by an infected sand fly bite, through which Leishmania parasites and saliva are co-delivered into the host skin. Despite the numerous studies performed in this area, no vaccine is yet available to control this neglected disease in humans. During the past two decades, saliva of sand flies has been tested for possible application as a vaccine against leishmaniasis. Exposure to specific salivary proteins or sand fly bites can induce a protective cell-mediated immunity. Immunization with Ph. papatasi saliva or recombinant PpSP15 has been previously reported to provide protection against L. major infection. In this study, the efficiency of immunization with Ph. sergenti saliva or plasmid coding for Ph. sergenti salivary proteins in protecting the BALB/c mice against L. tropica was explored. Here we show that although immunization with whole saliva induces a humoral response, this immune response is unable to protect mice against infection; in contrast, immunization with a plasmid coding for Ph. sergenti PsSP9 salivary protein induces a Th1 immune response characterized by a strong DTH response, no detectable antibody response, and a high expression ratio of IFN-γ to IL-5 in lymph nodes. The Th1 induced immunity in PsSP9-immunized mice correlates with the protection observed against infection with L. tropica, which was associated with a significant reduction in ear thickness and decrease in parasite load in the ear of protected animals in comparison to control group. Our data suggest that instead of using the whole sand fly salivary proteins, a more effective approach is the use of a single Ph. sergenti salivary protein.

Abstract Introduction Materials and methods Results Discussion

medicine and health sciences ears immunology sand flies parasitic diseases parasitic protozoans otology ear infections protozoans leishmania insect vectors antibody response digestive system infectious diseases exocrine glands head otorhinolaryngology disease vectors immune response eukaryota anatomy salivary glands biology and life sciences species interactions organisms

2019

DNA plasmid coding for Phlebotomus sergenti salivary protein PsSP9, a member of the SP15 family of proteins, protects against Leishmania tropica

Expansion of the CGG•CCG-repeat tract in the 5′ UTR of the FMR1 gene to >200 repeats leads to heterochromatinization of the promoter and gene silencing. This results in Fragile X syndrome (FXS), the most common heritable form of mental retardation. The mechanism of gene silencing is unknown. We report here that a Class III histone deacetylase, SIRT1, plays an important role in this silencing process and show that the inhibition of this enzyme produces significant gene reactivation. This contrasts with the much smaller effect of inhibitors like trichostatin A (TSA) that inhibit Class I, II and IV histone deacetylases. Reactivation of silenced FMR1 alleles was accompanied by an increase in histone H3 lysine 9 acetylation as well as an increase in the amount of histone H4 that is acetylated at lysine 16 (H4K16) by the histone acetyltransferase, hMOF. DNA methylation, on the other hand, is unaffected. We also demonstrate that deacetylation of H4K16 is a key downstream consequence of DNA methylation. However, since DNA methylation inhibitors require DNA replication in order to be effective, SIRT1 inhibitors may be more useful for FMR1 gene reactivation in post-mitotic cells like neurons where the effect of the gene silencing is most obvious. The most common cause of Fragile X mental retardation syndrome (FXS) is the silencing of the FMR1 gene that occurs when the number of CGG•CCG-repeats in its 5′ untranslated region (5′ UTR) exceeds 200 [1], [2]. The net result is a deficiency in the FMR1 gene product, FMRP, a protein that regulates the translation of mRNAs important for learning and memory in neurons. How repeats of this length cause silencing is unknown. However, since the sequence of the promoter and open reading frame of these alleles is unchanged, the potential exists to ameliorate the symptoms of FXS by reversing the gene silencing. The extent of silencing is related to the extent of methylation of the 5′ end of the gene [3], [4], [5]. Treatment of patient cells with 5-aza-dC, a DNA methyltransferase inhibitor, decreases DNA methylation and this is accompanied by partial gene reactivation [4], [5]. However, this compound has 2 major drawbacks: it is extremely toxic and it requires DNA replication to be effective. This would clearly limit its usefulness in vivo, particularly in post-mitotic neurons where the FMRP deficiency is most apparent. It also leaves open the question of whether DNA demethylation is necessary for gene reactivation to occur, a situation that for the reasons just mentioned, would severely limit the likelihood that gene reactivation would ever be a viable approach to treating FXS. While the silenced gene is associated with overall H3 and H4 hypoacetylation, lysine 4 and 9 of histone H3 are the only 2 specific modifiable sites that have been examined thus far. In individuals with FXS, the levels of histone H3 acetylated at K9 (H3K9Ac) and H3 dimethylated at K4 (H3K4Me2) are decreased relative to the normal gene while the level of H3K9 dimethylation (H3K9Me2) is increased [5], [6], [7]. By analogy with other genes that have been studied more extensively, we would expect that there are a number of other histone residues that are differentially methylated or acetylated, when the FMR1 gene is aberrantly silenced. The acetylation state of the histones associated with a particular genomic region is thought to play a critical role in regulating gene expression. The level of acetylation is dependent on the dynamic interplay of histone acetyltransferases (HATs) and histone deacetylases (HDACs). HDACs are sometimes divided into 4 functional classes based on sequence similarity. Class I (HDAC1,2, 3, and 8) and class II (HDAC4,5, 6,7, 9, and 10) HDACs remove acetyl groups through zinc-mediated hydrolysis. Class III HDACs, which includes SIRT1, catalyze the deacetylation of acetyl-lysine residues by a mechanism in which NAD+ is cleaved and nicotinamide, which acts as an end product inhibitor, is released. Class IV HDACs are HDAC11-related enzymes that are thought to be mechanistically related to the Class I and II HDACs. To date, only inhibitors of Class I, II and IV HDACs have been tested for their ability to reactivate the FMR1 gene in FXS cells [4], [6], [8]. These HDAC inhibitors (HDIs), which include TSA and short-chain fatty acids like phenylbutyrate, have a much smaller effect on FMR1 gene reactivation than 5-aza-dC when used alone, although some synergistic effect was noted when these compounds were used in conjunction with 5-aza-dC [5], [6], [7], [9]. Recently, it has become apparent that not only do some HDACs act preferentially on specific lysines on different histones, but they also target certain genes for deacetylation [10]. Thus the available data did not rule out a role for HDACs, specifically Class III HDACs, in gene silencing in FXS. We show here that SIRT1, a member of the Class III HDAC family, plays an important role in silencing of FMR1 in the cells of Fragile X patients acting downstream of DNA methylation. Furthermore we show that SIRT1 inhibitors result in increased FMR1 transcription. This increase is associated with an increase in H4K16Ac and H3K9Ac but does not involve DNA demethylation or an increase in H3K4 dimethylation. Nicotinamide (Vitamin B3), an end product inhibitor of NAD+-dependent enzymes like the Class III HDACs [11], increased FMR1 expression of a lymphoblastoid cell line from a Fragile X patient with a partially methylated FMR1 gene (GM06897) [12], [13]. Fifteen millimolar nicotinamide increased FMR1 mRNA levels by ∼3-fold while having little or no effect on the amount of FMR1 mRNA produced in normal cells (Figure 1A). A much smaller effect was seen in GM03200B cells in which the FMR1 gene is more heavily methylated [12], [13] and makes much less FMR1 mRNA (too small to see on the scale of the graphs shown in Figure 1A). Splitomicin, a compound with a saturated six-membered lactone ring, is a more specific inhibitor of Class III HDACs and is thought to have a mechanism distinct from that of nicotinamide, inhibiting these enzymes by competing for binding of the acetylated substrate [14]. Splitomicin not only increased FMR1 mRNA levels in GM06897, but it produced a 200–600-fold increase in the amount of FMR1 mRNA in cell lines like GM03200B that were only minimally responsive to 15 mM nicotinamide (Figure 1B). This corresponded to a final FMR1 expression level that was ∼15–25% of normal, depending on which normal cell line was used for comparison. This level of activation was comparable to that achieved with 10 µM 5-aza-dC, an inhibitor of DNA methylation and much higher than the level of activation seen with TSA (Figure 2). The extent of activation was impressive given the low potency of splitomicin (in the micromolar range) and its relative instability (it has a half-life of 30 minutes at neutral pH [14]). A much smaller level of reactivation was seen with GM09145 and GM04025, lymphoblastoid cell lines that are more heavily methylated [12], [13] and that make less FMR1 than GM03200B (Figure 1C). A similar low level of reactivation was seen for 2 fibroblast cell lines that make very little FMR1 mRNA in the absence of splitomicin (Figure 1D). The simplest interpretation of these data is that a class III HDAC is involved in downregulating FMR1 expression from full mutation alleles. As has been reported for 5-aza-dC, the extent of reactivation is inversely related to the extent of silencing [6]. Whether the failure to completely reactivate the FMR1 gene with either drug reflects a suboptimal dosing strategy or the limits of what these classes of compounds can accomplish remains to be seen. The ∼2-fold increase in FMR1 mRNA seen in GM06897 treated with 300 µM splitomicin is accompanied by a ∼2-fold increase in FMRP (Figure 2B and 2C). However, for cell lines where the FMR1 gene is more heavily methylated and that make no detectable FMRP, splitomicin did not result in the production of detectable levels of the FMR1 gene product (Figure 2B). The cell lines GM03200B, GM09145 and GM04025 are not only more heavily silenced than GM06897 but they also have more repeats (GM06897 has 477 repeats compared to 530 and 645 for GM03200B and GM04025 respectively). The failure to detect FMRP in these cells may reflect some combination of the low level of gene reactivation with the difficulty translating long CGG-repeat tracts previously reported for lymphocytes and lymphoblastoid cells [15], [16], [17], [18], [19]. Of the known class III HDACs, only SIRT1 is predominantly nuclear [20]. In order to assess whether SIRT1 was involved in FMR1 gene silencing, we transfected plasmids encoding a human SIRT1 protein and a dominant negative version of this construct (dnSIRT1) [21] into fibroblast cells from 3 different males, 1 who was unaffected and 2 with FXS. Fibroblasts were chosen because of the relative efficiency of transfection compared to lymphoblastoid cells. Transfection of the FXS fibroblasts (GM05131 and GM05848) with the normal SIRT1 construct led to a decrease in FMR1 expression from the low level seen in untransfected cells. In contrast a large increase in FMR1 expression was seen when the dnSIRT1 construct was used (Figure 3). This is consistent with a negative effect of SIRT1 on FMR1 transcription. Overexpression of these constructs only had a small effect on the level of FMR1 expression in unaffected individuals analogous to what was seen with nicotinamide and splitomicin. To examine whether the effect of SIRT1 was direct or indirect, we carried out ChIP assays using an anti-HA antibody on a FXS cell line transfected with a construct encoding the HA-tagged SIRT1 [21]. The HA-tagged SIRT1 was enriched on the FMR1 allele in FXS cells compared to normal alleles (Figure 4). SIRT1 binding to the promoter would be consistent with a role of this deacetylase in modification of the chromatin associated with the FMR1 gene in FXS cells. We therefore investigated the chromatin changes caused by splitomicin treatment using ChIP with antibodies to H3K9Ac and H4K16Ac since these are the major residues deacetylated by SIRT1 in vitro [22]. We also examined the levels of H3K4Me2, which is a mark of active chromatin that has been shown to increase when FXS alleles are reactivated with 5-aza-dC [7]. We examined the region upstream of the start of transcription and a region of exon 1 downstream of the repeat, with and without, splitomicin treatment. To better understand the differences between gene reactivation mediated by splitomicin and that mediated by 5-aza-dC we also examined the same histone modifications in these cells after 5-aza-dC treatment. Both the promoter and exon 1 from a normal allele had higher levels of H3K9Ac and H3K4Me2 than the heavily silenced FMR1 full mutation allele, consistent with previous reports (Figure 5A and 5B, left and center panels). In unaffected cells splitomicin had little, if any, effect on the level of H3K9Ac on either the promoter or exon 1 (Figure 5A and 5B, left panel). However, splitomicin treatment of FXS cells increased H3K9Ac on ∼2-fold on the promoter and on ∼15-fold on exon 1. The net result of this increase is that H3K9Ac levels in FXS cells treated with splitomicin are very similar to that seen in normal cells. This suggests that SIRT1 is responsible for the hypoacetylation of H3K9 seen on FXS alleles, consistent with the observed in vitro properties of SIRT1 [22]. In contrast, 5-aza-dC had no effect on H3K9Ac in this region. The opposite situation was seen with H3K4Me2, in that splitomicin had no effect while 5-aza-dC caused a large increase in H3K4Me2 levels on exon 1 of the FXS allele (Figure 5B, center panel). However, both splitomicin and 5-aza-dC increased the levels of H4K16Ac on both the promoter and exon 1 of the FXS allele (Figure 5A and 5B, right panel). This suggests that DNA methylation and SIRT1 may act in the same or overlapping pathways and that this modification may play a key role in FMR1 gene silencing. To assess the contribution of H4K16 acetylation to splitomicin-mediated FMR1 gene reactivation, we examined the effect of hMOF, a histone acetyltransferase that specifically targets H4K16 [23], on splitomicin-treated patient cells. As can be seen in Figure 6, transfection of patient fibroblasts with a dominant negative version of hMOF completely blocked the splitomicin-mediated increase in FMR1 mRNA, confirming the importance of H4K16 acetylation in FMR1 gene reactivation. To examine the contribution of DNA demethylation to splitomicin-mediated gene reactivation we used an assay that monitors a region containing 8 CpG residues that is located just upstream of the CGG•CCG-repeat in the FMR1 gene [24]. Demethylation of a single cytosine produces a 0. 5°C drop in the Tm of the PCR product obtained after bisulfite treatment. Reactivation with splitomicin did not change the Tm of the PCR product (Figure 7), suggesting that little, if any, demethylation occurred in this region. DNA demethylation-independent gene reactivation by splitomicin has also been seen in certain tumor suppressor genes aberrantly silenced in cancer cells [25]. In contrast, when these cells are treated with 5-aza-dC the Tm of the PCR product was indistinguishable from the results obtained from unaffected individuals (Figure 7). This is consistent with previous reports of the almost complete demethylation of the promoter by this treatment [4], [6], [9], [26]. We have shown that SIRT1, a class III HDAC, is involved in repeat-mediated FMR1 gene silencing via the deacetylation of H3K9 and H4K16. Our data suggests that deacetylation of H4K16 is also one of the major downstream consequences of DNA methylation. Since SIRT1 inhibition is able to reactivate the gene without affecting DNA demethylation, DNA methylation is not dominant over chromatin modifications like H4K16Ac with regard to gene expression. Furthermore, it demonstrates that DNA demethylation is not necessary for relieving gene silencing. This resembles the situation in Friedreich ataxia, another Repeat Expansion Disease, in which expanded alleles that are also aberrantly methylated at the DNA level [27], can be reactivated using an HDI alone [28]. The increased acetylation of H4K16 seen after treatment with both 5-aza-dC and splitomicin is important since the H4K16 acetylation status is thought to be a key determinant of chromatin accessibility [29]. However, the outcomes of the 2 treatments are not completely equivalent. DNA demethylation by 5-aza-dC is accompanied by an increase in H3K4Me2 that is not seen with splitomicin treatment. In contrast, splitomicin, but not 5-aza-dC, causes acetylation of H3K9. One interpretation of our data is that silenced alleles are associated with a methyl-binding protein or protein complex (MeBP) that binds to the methylated promoter and recruits SIRT1 (Figure 8). SIRT1 in turn deacetylates H3K9, H4K16 and potentially other residues as well. DNA demethylation causes the dissociation of the MeBP-SIRT1 complex from the promoter and creates conditions that favor the recruitment of H3K4 methylases and hMOF which specifically acetylates H4K16, but does not facilitate recruitment of a HAT that uses H3K9 as a substrate (Figure 8A). Splitomicin treatment, on the other hand, inhibits SIRT1 while leaving the promoter methylated. This helps generate a chromatin context conducive to recruiting both hMOF and an H3K9 HAT, but not an H3K4 methyltransferase (Figure 8B). Despite the differences in the final histone modification profile, the extent of gene reactivation resulting from the use of these compounds is similar and they show little additive effect when used in combination (data not shown). This raises the possibility that the most significant action of both compounds is exerted via the acetylation of H4K16 with both H3K4Me2 and H3K9Ac having little direct effect on gene expression. Since the effect of splitomicin is not dependent on DNA replication, SIRT1 inhibitors may be more useful than 5-aza-dC for reversing FMR1 gene silencing in neurons which no longer divide and where the absence of FMRP is most debilitating. However, there are significant barriers to using SIRT1 inhibitors to treat FXS. Firstly, Sir2p, the yeast homolog of SIRT1, plays a role in the extension of lifespan in yeast [30] raising the possibility that SIRT1 inhibition may reduce lifespan in humans. However, there is some evidence that SIRT1 actually limits lifespan in mammals, at least in response to chronic genotoxic stress [31]. Furthermore, SIRT1 inhibition sensitizes cancer cells to apoptosis while sparing normal cells, making HDAC III inhibitors promising anti-cancer drugs [32]. It could also be argued that inhibition of HDACs could lead to inappropriate expression of other genes, which could be deleterious. However several HDIs are already approved for use in humans including dihydrocoumarin, an FDA approved food additive and valproate, a broad spectrum HDI, that has been used for decades in the treatment of epilepsy and is also an effective mood stabilizer. Today Valproate is one of the most highly prescribed antiepileptic drugs [33] and is already used in Fragile X patients to treat seizures, aggression and depression [34]. The fact that RNA with long CGG-repeat tracts is thought to be responsible for the Fragile X associated tremor and ataxia syndrome, a late onset neurodegenerative disorder seen in carriers of FMR1 premutation alleles [35], is a more general problem applicable to any gene reactivation approach for treating FXS. However, some HDIs have actually been shown to be neuroprotective [36], [37] and to expedite the recovery of learning and memory lost as a result of induced neurodegeneration [38]. Thus the beneficial effects of HDIs may help offset the negative effect of the expression of long CGG-repeat tracts. The final impediment to gene reactivation approaches is the difficulty translating FMR1 transcripts with long CGG-tracts that has been seen in cells like lymphocytes and lymphoblasts [15], [16], [17], [18], [19]. However, there is reason to think that the translation difficulties do not affect all cells equally. For example, in Fragile X embryonic stem cells where the repeat is still unmethylated, both FMR1 mRNA and FMRP are made [39]. Furthermore we have shown that the negative effect of the repeats on translation is more severe in some parts of the mouse brain than others [16]. This is consistent with the fact that individuals with unmethylated full mutations show only mild symptoms of FXS [40], [41], [42]. It could thus be argued that when the FMR1 gene is not silenced, translation occurs at adequate levels in those parts of the brain critical for learning and memory. Even in lymphocytes and lymphoblastoid cells with ∼400 repeats some FMRP is made without treatment ([43] and this manuscript). The fact that even the GM06897 lymphoblastoid cell line, which has 477 repeats, makes some residual FMRP and that FMRP levels increase when the cells are treated with splitomicin, raises the possibility that increased RNA production may lead to increased FMRP production in the ∼40% of individuals with FXS who have repeats of <500 (Sally Nolin, personal communication). Even in lymphoblastoid cells there have been reports of FMRP production in cell lines with >800 repeats after reactivation with 5-aza-dC [4]. New SIRT1 inhibitors with higher stability, selectivity or potency [44] may allow the level of FMR1 transcription from previously silenced alleles to approach that seen in carriers of unsilenced full mutations. Since HDIs do not require DNA replication to be effective, this class of compounds may thus have therapeutic potential at least in that subset of individuals with repeat numbers that do not preclude translation. Lymphoblastoid cells (GM02168, GM06895) and fibroblasts (GM00357) from unaffected males and lymphoblastoid cells (GM03200B, GM04025, GM09145) and fibroblasts (GM05131 and GM05848) from males with FXS were obtained from the Coriell Cell Repository (Camden, NJ). The antibodies used in this study were obtained from the following sources: anti-acetyl-Histone H4 (Lys 16) (Cat. #: ab1762) and anti-HA-tag (Cat. #: ab9110) were purchased from Abcam (Cambridge, MA); anti-acetyl-Histone H3 (Lys9) (Cat. #: 07-352), anti-dimethyl-Histone H3 (Lys 4) (Cat. #: 07-030) and anti-rabbit Ig were purchased from Millipore (Temecula, CA). Splitomicin and TSA were obtained from Tocris (Ellisville, MO). Nicotinamide and 5-aza-dC were obtained from Sigma (St. Louis, MO). The mutant human MOF (hMOF) construct in pcDNA3 was a kind gift of Arun Gupta (Washington University School of Medicine, St. Louis, MO). The pCRUZ-HA vector, pCRUZ-HA-SIRT1 and a dominant negative version of this construct were kindly provided by Toren Finkel (NHLBI, NIH, Bethesda, MD). Lymphoblastoid cells were cultured in RPMI medium supplemented with 10% fetal bovine serum and 100 units each of penicillin and streptomycin (Invitrogen, Gaithersburg, MD). Fibroblasts were cultured in Minimum Essential Medium supplemented with 1% Glutamax, 10% fetal bovine serum and 100 units of penicillin and streptomycin (Invitrogen). All cells were grown at 37°C in 5% CO2. Cells were treated where indicated with either 300 µM or 700 µM splitomicin, 15 mM nicotinamide, or 3 µM TSA for 24 hours or 10 µM 5-aza-dC for 72 hours. Transfection of fibroblasts was carried out using Fugene 6 (Roche USA, Nutley, NJ) according to the supplier' s instructions. Total RNA was isolated from the cell lines using Trizol (Invitrogen) and reverse transcribed using SuperScript™ III RT First Strand Synthesis system for RT-PCR (Invitrogen), as per the manufacturer' s instructions. Real time PCR was carried out using an ABI 7500 FAST PCR system (Applied Biosystems, Foster City, CA) using TaqMan™ Universal PCR master mix and FMR1 and GUS Taqman probe primer mixes (Applied Biosystems). For quantitation the comparative threshold (Ct) method was used with normalizing to GUS. The fold change was calculated by comparing the normalized treated versus untreated Ct values. The ChIP assay kit from Upstate was used according to the manufacturer' s instructions with slight modifications as previously described [27]. The amount of FMR1 promoter and exon 1 DNA immunoprecipitated with each antibody was determined using quantitative real time PCR as described below. Real time PCR was carried out using an ABI 7500 FAST PCR system and the Power SYBR™ Green PCR kit (Applied Biosystems). For amplification of the promoter region Promoter-F (5′-ACAGTGGAATGTAAAGGGTTG-3′) and Promoter-R (5′-GTGTTAAGCACTTGAGGTTCAT-3′) were used. This primer pair amplifies the 140 bp region from 146800256–146800396 of the human genome sequence (March, 2006 assembly, http: //genome. ucsc. edu/cgi-bin/hgBlat) which terminates 736 bp upstream of the 3′ most transcription start site. For amplification of exon 1, the primer pair Exon1-F (5′-CGCTAGCAGGGCTGAAGAGAA-3′) and Exon1-R (5′-GTACCTTGTAGAAAGCGCCATTGGAG-3′) was used. This primer pair amplifies the region 146801368–146801444 of the human genome sequence that corresponds to the region in exon 1 236–311 bp downstream of the transcription start site. All experiments were done in triplicate. The ChIP experiments were performed in triplicate and each PCR reaction was done in duplicate. The immunoprecipitated DNA was expressed relative to the amount of input DNA that constituted 10% of the original material. GAPDH was used for normalization using hs_GAPDH exon1F1 primer (5′-TCGACAGTCAGCCGCATCT-3′) and hs_GAPDH intron1R1 (5′-CTAGCCTCCCGGGTTTCTCT-3′). Genomic DNA from cell lines was bisulphite modified according to standard procedures except that the bisulphite treatment was carried out overnight at 55°C. The methylation status of the promoter was determined as previously described [24]. SDS protein gel electrophoresis and Western blotting of protein extracts was carried out using standard procedures. Anti-FMRP antibody (MAB2160, Millipore) was used to detect FMRP. Anti-β-actin antibody (Abcam) was used to normalize the FMRP levels for variations in protein loading. Detection of antibody binding was carried using an ECL™ kit (Amersham, Buckinghamshire, UK) according to the manufacturer' s instructions. The amount of FMRP and β-actin were determined by standard densitometry. The increase in FMRP was calculated based on the average of 3 independent experiments.

Fragile X syndrome is the leading cause of heritable intellectual disability. The affected gene, FMR1, encodes FMRP, a protein that regulates the synthesis of a number of important neuronal proteins. The causative mutation is an increase in the number of CGG•CCG-repeats found at the beginning of the FMR1 gene. Alleles with >200 repeats are silenced. The silencing process involves DNA methylation as well as modifications to the histone proteins around which the DNA is wrapped in vivo. Treatment with 5-azadeoxycytidine, a DNA methyltransferase inhibitor, reactivates the gene. However, this reagent is toxic and since no DNA demethylase has been found in humans, methylation inhibitors are not useful in cells like neurons that no longer divide. We show here that splitomicin is also able to reactivate the Fragile X allele. It does so by inhibiting a protein deacetylase, SIRT1, thus favoring the action of another enzyme, hMOF that reverses the SIRT1 modification. We also found that 5-azadeoxycytidine acts, at least in part, by reversing the effect of SIRT1. However, since splitomicin reactivation occurred without DNA demethylation, DNA replication is not necessary for its efficacy. Thus, unlike DNA methylation inhibitors, SIRT1 inhibitors may be able to reactivate Fragile X alleles in neurons.

Abstract Introduction Results Discussion Materials and Methods

molecular biology/histone modification molecular biology/transcription initiation and activation molecular biology/dna methylation neurological disorders/developmental and pediatric neurology genetics and genomics/epigenetics

2008

SIRT1 Inhibition Alleviates Gene Silencing in Fragile X Mental Retardation Syndrome

The coordination of chromosome segregation with cell growth is fundamental to the proliferation of any organism. In most unicellular bacteria, chromosome segregation is strictly coordinated with cell division and involves ParA that moves the ParB nucleoprotein complexes bi- or unidirectionally toward the cell pole (s). However, the chromosome organization in multiploid, apically extending and branching Streptomyces hyphae challenges the known mechanisms of bacterial chromosome segregation. The complex Streptomyces life cycle involves two stages: vegetative growth and sporulation. In the latter stage, multiple cell divisions accompanied by chromosome compaction and ParAB assisted segregation turn multigenomic hyphal cell into a chain of unigenomic spores. However, the requirement for active chromosome segregation is unclear in the absence of canonical cell division during vegetative growth except in the process of branch formation. The mechanism by which chromosomes are targeted to new hyphae in streptomycete vegetative growth has remained unknown until now. Here, we address the question of whether active chromosome segregation occurs at this stage. Applied for the first time in Streptomyces, labelling of the chromosomal replication initiation region (oriC) and time-lapse microscopy, revealed that in vegetative hyphae every copy of the chromosome is complexed with ParB, whereas ParA, through interaction with the apical protein complex (polarisome), tightly anchors only one chromosome at the hyphal tip. The anchor is maintained during replication, when ParA captures one of the daughter oriCs. During spore germination and branching, ParA targets one of the multiple chromosomal copies to the new hyphal tip, enabling efficient elongation of hyphal tube. Thus, our studies reveal a novel role for ParAB proteins during hyphal tip establishment and extension. Chromosome segregation in unicellular bacteria is strictly coordinated with the cell cycle and chromosomes are segregated during their replication and prior to cell division. However, the spatial chromosome organization in bacteria, determined by the position of the origin of replication (oriC) in the cell, differs with respect to the morphology and growth strategy of the organism [1–3]. These differences in chromosome organization are reflected in the specific, cell cycle-tuned mechanism of chromosome segregation [4,5]. In most bacterial species (with the exception of Escherichia coli and some γ-proteobacteria) efficient chromosome segregation relies on the activity of two proteins, ParA and ParB [4,6]. By binding parS sites clustered around oriC, ParB assembles this region of the chromosome into a large nucleoprotein complex. Soon after the initiation of replication, ParB complexes are segregated into specific locations of the cell due to interaction with the ATPase, ParA [6–8]. Species-specific differences in spatial chromosome organization are linked with variations in the ParA and ParB choreography. For instance, during vegetative growth of Bacillus subtilis, ParB complexes are segregated bi-directionally to opposite cell poles, while in Caulobacter crescentus and in the case of Vibrio cholerae chromosome I, only one of the two ParB/oriC nucleoprotein complexes is moved toward the opposite pole by the ParA assembly [2]. The interaction of ParA with proteins localized at the cell pole, such as PopZ and TipN in C. crescentus or HubP in V. cholerae, translate the cell polarity to asymmetric chromosome segregation [9–11]. The chromosome arrangement and mechanism of segregation remains unexplored in multigenomic bacteria such as the filamentous actinobacteria including Streptomyces. Streptomyces differ remarkably from other bacteria with their hyphal growth that is reminiscent of filamentous fungi [12]. Moreover, Streptomyces elongated hyphal cells contain multiple copies of linear chromosomes. During colony development, two types of hyphal cells are produced: branching vegetative hyphae that form a dense mycelial network and sporogenic hyphae, that are converted into chains of spores by multiple cell divisions. In contrast to most bacteria that extend along the lateral cell wall, Streptomyces as other actinobacteria, grow by cell extension at the poles (tips) [13,14]. The apical synthesis of peptidoglycan in Actinobacteria is linked to the activity of the essential coiled-coil protein, DivIVA that localizes at the cell poles [15–17]. What is unique to Streptomyces growth, is the unidirectional cell extension at the hyphal tips. In Streptomyces, polar growth is directed by a protein complex localized at the hyphal tip (‘polarisome’ or tip-organizing complex, TIPOC), which includes DivIVA and another coiled-coil protein, Scy [16–18]. Branching is initiated by assembly of the polarisome at a site on the lateral wall distant from the extending tip [14,19]. A similar mechanism of branch formation was observed in the filamentous fungus Neurospora crassa, suggesting that it is a feature shared between bacteria and eukaryote [20]. The two stages of Streptomyces development, vegetative growth and sporulation, differ with respect to the cellular organization and cell cycle events. During sporulation, multiple, synchronized divisions of elongated sporogenic cells are accompanied by condensation and segregation of numerous chromosomal copies [12]. Our earlier studies showed that the segregation proteins ParA and ParB uniformly distribute chromosomes along the long sporogenic cell at the time of its septation [21,22]. During vegetative growth, multigenomic hyphal cells, named hyphal compartments, do not undergo typical cell division. Widely spaced cross walls delimit, but do not separate, hyphal compartments which remain adjacent in long hyphae [23,24]. Very little is known about chromosome organization in vegetative hyphae. Several copies of the chromosomes remain uncondensed and visibly unseparated when visualized by DNA staining in hyphal compartments. FISH (fluorescence in situ hybridization) experiments indicated that the ends of linear chromosomes are spatially close and the chromosomes are unevenly distributed in vegetative hyphae [25]. In addition, replisome labeling demonstrated that chromosomes replicate asynchronously within the compartments and follow the extending tip [27,28]. Localization of segregation proteins in vegetative hyphae is significantly different from their localization in sporulating hyphae. ParB was visualized as multiple, irregularly spaced complexes, with a distinct focus located at a constant distance from the hyphal tip [26]. Meanwhile, ParA in vegetative hyphae, localizes exclusively at the hyphal tips (not along the cell as in sporogenic cells), where it interacts with Scy [29]. Even though the localization of the segregation proteins suggested their engagement in the organization of the apical chromosome, chromosome distribution and the role of ParA and ParB during growth of multigenomic vegetative hyphae remains unknown in the absence of chromosomal locus-specific labelling tools. To understand how hyphal tip extension and branching are coordinated with distribution and segregation of multiple chromosomal copies during vegetative growth of S. coelicolor, we took advantage of oriC labeling and time-lapse fluorescent microscopy. We show that in multigenomic hyphal cells ParA anchors the single apical chromosome and unidirectionally segregates one of the newly replicated oriC regions at the tip. During establishment of the new hyphal tip, ParA-mediated apical oriC anchorage targets a chromosome from multigenomic cell to a new branch or a germ tube. Our study reveals a unique mechanism for bacterial chromosome segregation that is adjusted to accommodate hyphal growth and branching. To address the question how chromosomes are distributed in S. coelicolor apically extending and branching vegetative hyphae we constructed a fluorescent reporter-operator system (FROS) to mark chromosomal oriC regions. A FROS cassette that contained an array of 120 tandem tetO repeats [30] was integrated into the S. coelicolor chromosome approximately 29 kb from the oriC region by in vitro transposition and intergenic conjugation [31] (Fig 1A, S1A Fig) resulting in the strain EJTH31. Subsequently the tetR-mcherry gene was integrated into the chromosome of this strain on pMS83mCherry resulting in the FROS strain, DJ-NL102 (S1A Fig, strains were verified as shown in S1B and S1C Fig). Analysis of DJ-NL102 hyphae revealed irregularly distributed mCherry foci (the mean distance between foci was 2. 0 ± 1. 2 μm, S2A and S2B Fig). The foci disappeared upon addition of anhydrotetracycline (aTc) to the culture medium, presumably due to the relief of TetR binding to tetO (S2A Fig) [30]. However, in contrast to E. coli [30], we did not detect any growth impairment or disturbed replication of FROS strain (s) in the absence of aTc (S2C, S2D, S2E and S2F Fig). This suggests that, at least in vegetative hyphae, the binding of TetR-mCherry to the tetO array in the FROS cassette did not cause serious replication roadblocks detrimental to growth. We expected that in hyphal cells ParB-EGFP should co-localize with at least some oriCs of the multiple chromosomal copies. To check this, we analyzed a strain AK113 with the FROS cassette and tetR-mcherry expressed in a parB-egfp background. In AK113 hyphae, 97% of ParB-EGFP foci and mCherry-FROS foci overlapped (distance between foci less than 1 μm) (Fig 1B and 1C). We did not observe any FROS foci unaccompanied by the ParB-EGFP complex at the hyphal tips, suggesting that the tip-proximal chromosome is constantly bound by ParB. Thus, the FROS labeling of oriC regions in S. coelicolor confirmed irregular distribution of the multiple copies of chromosomes and indicated binding of ParB to each chromosomal oriC region in the multigenomic hyphae. Earlier studies of chromosome replication in S. coelicolor vegetative hyphae showed replisome trafficking and suggested that chromosomes follow the extending tip [27]. The constant distance between the FROS/ParB complex and the tip, observed in snapshot analysis, suggests that the tip-proximal chromosome is anchored to the tip during hyphal extension. The application of time-lapse microscopy and the FROS strains (DJ-NL102 and AK113) allowed us to examine the chromosome distribution during hyphal growth (Fig 2, S3A, S3B, S3C and S3D Fig). The earlier observations showed that replication starts before spore germination [27,28], and the germinating spores showed multiple FROS foci, as expected. In spore germ tubes (Fig 2A top panel, S3A and S3D Fig top panel, S1 Movie) and extending vegetative hyphae of 24 hours old colonies (Fig 2A bottom panel, S3D Fig, bottom panel), the distance between the hyphal tip and the first, tip-proximal FROS complex was constant (1. 4 ± 0. 4 μm), demonstrating that the first oriC follows the elongating tip (Fig 2B). Interestingly, the distance between the tip and the FROS complexes located further from the tip was more variable during hyphal extension (Fig 2B, inset). We compared the tip-proximal (oriC 1) and tip-distal complexes (oriC 2—oriC 8) by analyzing the correlation between their movement and tip growth rates (Fig 2C). The correlation was high for the first chromosome and decreased with an increasing distance between the oriCs and the extending tip. Next, we measured the distance between the hyphal tip and the first FROS signal in extending hyphae of FROS, parB-egfp strain (AK113) and plotted it against the distance between the ParB complex and the tip (Fig 1D). This analysis confirmed that the positions of the ParB and FROS signals at the hyphal tip were highly correlated and average distance between both complexes and the tip was 1. 5–2. 0 μm. Thus, the time-lapse analysis confirmed that chromosomes follow the extending hyphal tip. Markedly, only the first oriC remains tightly associated with the extending hyphal tip and maintains a constant distance to it. ParB binds the oriCs of each chromosome along the hyphae, but only the first one maintains a constant distance and follows the extending hyphal tip. As ParA in vegetative hyphae is localized at the tip [21,29] (S4 Fig), we hypothesized that the presence of ParA and its interaction with the tip-proximal ParB complex are critical for the localization of the oriC/ParB complex at the constant distance from the tip. Analysis of FROS-marked oriC regions in the ΔparB and ΔparA background allowed us to verify this. The snapshots analysis revealed that in the ΔparB and ΔparA (AK114 and AK115) mutant strains, the tip-proximal FROS complex was further away from the hyphal tip than in the wild type FROS strain. The distances between tip-proximal FROS complexes and the tip were 2. 5 ± 1. 4 μm in ΔparB, 2. 3 ± 1. 0 μm in ΔparA and 1. 4 ± 0. 4 μm in “wild type” strain (Fig 3A and 3B, differences between mutants and the wild type strain were statistically significant, p<0. 001, verified with ANOVA and a post hoc Games-Howell test). Moreover, the position of the apical FROS complex exhibited higher variation in ΔparB and ΔparA strains than in the wild type strain (demonstrated with an F-test; the ratio of variances was 6. 3 ± 1. 2 for ΔparA and 11. 9 ± 2. 3 for ΔparB in relation to the wild type strain (p<2. 2e-16) ). We also observed that in the ΔparA strain, but surprisingly not in the ΔparB strain, the distance between the edge of nucleoid and the hyphal tip was increased, when compared to the wild type FROS strain (S5 Fig, only the difference between the ΔparA and the wild type was statistically significant, p<0. 001 verified with ANOVA and post hoc Tukey’s test). The increased distance of the tip-proximal oriC to the extending hyphal tip in the ΔparA (AK115) and ΔparB (AK114) strains was confirmed by time-lapse analysis of FROS complex dynamics (S6 Fig). Plotting the correlation between the hyphal extension rate and oriC movement showed that the association of the first oriC with the hyphal tip was visibly decreased in the mutant strains (Fig 3C). Although chromosome trafficking in the hyphae of the ΔparA and ΔparB strains was maintained, we noted a slightly increased variation of the distances between oriC 2 or oriC 3 and the tip (S6 Fig). Our analyses show that in the ΔparA and ΔparB mutant strains, chromosome trafficking in hyphae was maintained albeit slightly disturbed. Moreover, both segregation proteins ParA and ParB were essential for anchorage of the tip-proximal chromosome oriC region at the tip of the extending hyphae. Having established that both segregation proteins are required for the tip-proximal oriC localization, we sought confirmation that ParA anchors and organizes the first ParB complex. To address this question, we compared the localization of the ParB complex in a set of strains with different parA modifications: ΔparA (J3318), a parA overexpression strain (DJ532) and a strain with a mutation that abolishes the interaction with ParB and Scy—parAmut (DJ598) [29]. In parA mutant strains, the tip-proximal ParB complex was delocalized and positioned further away from the tip than in the wild type strain. The distance between the ParB-EGFP complex and the tip was 2. 2 ± 1. 4 μm in the ΔparA strain, 3. 0 ± 1. 5 μm in the strain overexpressing parA and 2. 5 ± 1. 1 μm in the parAmut strain in comparison to 1. 7 ± 0. 8 μm in the control parB–egfp strain (Fig 4A and 4D, the differences between the mutants and the wild type strain were statistically significant, p<0. 001, calculated with ANOVA and post hoc Games-Howell test). The measured distance of ParB-EGFP complex to the tip somewhat differs from earlier reports [26] and from oriC–tip distance (1. 4 ± 0. 4 μm, see above). This is possibly due to different sample processing and data analysis (use of cell wall staining instead of transmitted light images (Figs 1,2 and 3) presumably affects the analysis of the hyphal tip position). The variation of the tip-proximal ParB complex positioning in the parA mutant strains was very high (shown with an F-test, the ratio of variances 3. 0 ± 0. 8 for ΔparA, 3. 2 ± 0. 7 for parA overproduction and 1. 7 ± 0. 4 for parAmut strain (p < 0. 001) ) (Fig 4D). Overexpression of parA resulted in the most notable change of the ParB complex position. Immunostaining of ParA in the wild type parB-egfp strain revealed that the ParB complex was positioned at the edge of the ParA signal (Fig 4B). The overproduction of ParA led to a huge accumulation of ParA at the hyphal tips and shifted the ParB complex away from the tip. We showed before that tip localization of ParA is dependent on Scy and ParAmut protein is mislocalized [29]. However, since the ParAmut protein does not interact with both, ParB and Scy, on the basis of parAmut strain analysis, we cannot conclude that the Scy dependent localization of ParA determines localization of ParB complex. The analysis of ParB-EGFP foci in the Δscy, parB-egfp strain (BD05) showed that in the absence of Scy, the distance between the tip and the apical ParB-EGFP focus in the newly formed branches was much more varied than in the “wild type” strain J3310 (S7 Fig). This reinforced the notion that Scy-ParA interaction is required to establish anchorage for oriC/ParB complex soon after branch emergence. Finally, we examined whether ParA influences the overall organization of the tip chromosome through the ParB complex and oriC anchorage. To address this question, we analyzed the localization of ParB complexes in conjunction with DNA staining (Fig 4C). In the wild type parB-egfp strain, the ParB complex was found at the tip-proximal edge of the stained nucleoid, whereas in the ΔparA strain, the ParB complex was located further away from the tip than the edge of the nucleoid. Measurement and plotting of the ParB fluorescence and DNA staining intensities confirmed that in the absence of ParA, the tip-proximal orientation of the oriC/ParB complex at the edge of the nucleoid was lost (Fig 4E). To recapitulate, analyses of the parA mutant strains revealed that the distance between the first ParB complex and the tip is dependent on ParA. ParA interactions with Scy and ParB are presumably required to establish the chromosomal tip anchorage. Furthermore, the ParB complex is positioned at the edge of the tip ParA assembly, and the interaction between the ParB complex and ParA orientates the nucleoid with the oriC region toward the tip of hyphae. We have demonstrated that ParA, presumably through interaction with Scy, anchors the oriC/ParB complex at the tip, assuring that it follows the extending hyphal tip. This raises the question how this anchorage is established after oriC duplication. To answer this question, we used a strain with the EGFP tagged replisomes (DnaN-EGFP, J3337, [28]) as the parent strain for the introduction of the FROS system, to monitor duplication of the tip proximal oriC by time-lapse analysis. We set out to examine how daughter oriCs become anchored after replication. Possible scenarios included a loss of connection with the tip during oriC replication; a close association of both newly replicated oriCs with the tip, followed by a loss of anchorage of one of the duplicated oriCs; or anchorage of only one of the newly replicated oriCs to the tip. By analyzing the time-lapse images of the FROS strain expressing dnaN-egfp (strain AK122), we observed the appearance of a DnaN focus close to a tip-proximal FROS focus (Fig 5A top panel, B, S8 Fig, top panel, S2 Movie). 10 min after the replisome appearance, we detected, both newly replicated oriCs separated by a distance of around 0. 8 μm in almost 15% of the hyphae. Twenty minutes later, duplicated oriCs separated by a distance of roughly 1. 3 μm were visible in approximately 70% of hyphae (Fig 5B, 5D and 5E). However, the distance between the tip and tip-proximal oriC was the same 20–10 minutes before and 10–20 minutes after oriC duplication (Fig 5C). Thus, after duplication, the tip-proximal daughter oriC did not move toward the tip, but maintained a constant distance to the tip (Fig 5B and 5C). Interestingly, the second of the newly replicated oriCs remained co-localized with the replisome in the subsequent time-lapse images (Fig 5A top panel, Fig 5B, S8 Fig, top panel). The distance between the tip and tip-distal daughter oriCs increased as hyphae extended (Fig 5B and 5C). These results indicate that while tip the proximal oriC follows the tip, the second oriC stays behind the extending tip. In the ΔparA FROS dnaN-egfp strain (AK123), both daughter oriCs remained closely associated after duplication (Fig 5A bottom panel, S8 Fig, bottom panel, S3 Movie). In the ΔparA strain (AK123), we detected double FROS foci later after replisome appearance than observed in the “wild type” AK122. Daughter oriCs separated by a distance of approximately 0. 9 μm were visible 20 minutes after replisome appearance in only 12% of the hyphae (Fig 5A bottom panel, Fig 5B right panel, D, E). Approximately 30 minutes after replisome detection, the average distance between the duplicated oriCs reached 1 μm in about 35% of hyphae, but it did not exceed this value (for comparison, in the wild type hyphae at this time point the distance was 1. 3 μm in 70% of hyphae). Interestingly, in the ΔparA strain, both oriCs and replisomes still followed the extending hyphal tip, but the distance between the first oriC and the hyphal tip remained greater than that found for the “wild type” AK122 strain (Fig 5C). This observation indicates that, although both newly duplicated oriCs still move behind the extending tip independently of ParA (in ΔparA strain), their separation is less efficient and the first oriC is not attached to the hyphal tip. Since ParB complexes form at each oriC in the multigenomic hyphal compartment we expected that shortly after replication, both daughter oriCs should be bound by ParB. To test this we used the strain expressing parB-egfp and dnaN-mcherry (AK101). Time-lapse analysis showed that ParB complex duplication, as oriC duplication, was detected 30 min after replisome appearance at the tip in almost 70% of hyphae (S9 Fig). In order to confirm that segregation of duplicated oriC in hyphae is efficient in the presence of ParA we also checked how the newly replicated tip-distal oriCs are separated in the hyphal stem. The analysis of the distances between FROS foci after their duplication showed that the separation of tip-distal foci is much less efficient than separation of tip-proximal foci which occurs in presence of ParA (S10 Fig). In summary, during replication the position of the tip-proximal daughter oriC does not change in relation to the tip, while the other oriC co-localizes with the replisome and gradually falls behind the extending tip. The tip-distal oriC are not separated as efficiently after duplication as tip-proximal foci. In the ΔparA strain, the tip anchorage of oriC is impaired and separation of daughter tip-proximal oriCs is inefficient. Thus, although both daughter oriCs are bound by ParB shortly after their duplication, immediately after replication ParA captures one of the apical ParB/oriC complexes and maintains its constant distance to the tip. The elimination of ParA affects chromosome segregation during sporulation, but it has not been reported that ParA is required for vegetative growth. As our experiments indicated that ParA anchors the oriC/ParB complex of the tip-proximal chromosome in the vegetative hyphal tips, we expected that its elimination should affect the population of the new hyphal tips with the chromosomes. Thus, using time-lapse microscopy analysis, we re-investigated the influence of ParA and ParB on vegetative growth with a particular focus on spore germination and branching. Microscopy analysis of spore germination showed that a parA deletion decreased the efficiency of germ tube formation. The number of germinating spores during the 16-h time-lapse microscopy observation dropped from approximately 61% in the “wild type” FROS strain (DJ-NL102) to approximately 32% in the ΔparA strain (AK115). Interestingly, the spores of the ΔparB strain (AK114) germinated with a similar efficiency (58%) as the “wild type” strain. According to our hypothesis, the germination of the ΔparA strain may be impaired due to the less efficient population of the germ tube by the chromosomes. To test this hypothesis, we used the FROS strains (control strain DJ-NL102, ΔparA-AK115 and ΔparB-AK114) to analyze the influence of parA and parB deletion on the length of the germ tube at the time of the oriC signal appearance (Fig 6A top panel, S11 Fig, top panel). We found that in the germinating spores of the wild type FROS strain (DJ-NL102), the FROS signal was detected when the germ tube reached the 2 μm length, but in ΔparA and ΔparB strains, the germ tubes were longer (2. 5 μm and 2. 7 μm, respectively) at the time when the FROS signal appeared (Fig 6B, the differences were statistically significant p<0. 01, verified by ANOVA and post hoc Games-Howells test). This suggests that elimination of ParA and ParB delayed chromosome migration from the spore to the elongating germ tube. The chromosome anchorage should not only be important during formation of the germ tube but also during new branch formation. Indeed, the time-lapse experiment and analysis of the length of emerging new branches at the time of FROS signal appearance showed that, while the wild type FROS signal could be detected in hyphal branches when they were 2 μm on average (similarly to germ tubes). In ΔparA and ΔparB strains (AK115 and AK114) the hyphae were longer than in wild type at the time of FROS signal detection, approximately 3. 6 μm and 3. 8 μm long, respectively, (Fig 6A, bottom panel and Fig 6C, S11 Fig, bottom panel, S4 and S5 Movies). This suggests that ParA and ParB elimination impairs also the migration of chromosomes from the hyphal stem to the newly formed branch. Interestingly, we noted that chromosome targeting to the new hyphal branch does not need to follow oriC duplication in hyphal stem. In 60% of observed new hyphal branches, we could not detect oriC duplication during 30 minutes proceeding oriC targeting to hyphal branch (Fig 6A). Thus, targeting of the chromosome to new hyphal branches is not accompanying the chromosome replication. The impaired chromosome population of new branches might also affect their extension. Thus, we investigated the extension of the new branches in “wild type” (DJ-NL102), ΔparA (AK115) and ΔparB (AK114) FROS strains. In the “wild type” strain, only a small fraction of branches (8%) stopped extension, mostly before they reached a length of 3. 5 μm (Fig 6D). Interestingly, in the ΔparA and ΔparB strains, the percentages of stalled short branches were markedly increased in comparison to the wild type strain. In both strains, approximately 18% of the branches stopped extension when they were still shorter than 3. 5 μm, whereas some were stalled when their length was within the range between 3. 5 μm and 7. 0 μm (7% in ΔparA and 5. 8% in in ΔparB, in contrast to 1% in wild type) or even longer than 7 μm (7% in ΔparA, in contrast to 0% in wild type and ΔparB). Most of the stalled branches did not show the FROS signal; however, some of the branches that stopped growth contained chromosomes (4% in wild type, 17. 1% ΔparA and 6. 8% in ΔparB), as judged by the detection of the FROS signal. This observation indicated that elimination of either ParA or ParB affects the extension of short branches, presumably through their impaired population with chromosomes. Interestingly, the lack of ParA (but not ParB) also disturbs the extension of branches that received chromosomes, possibly due to the lack of chromosome tip anchorage. In summary, the analyses of germination and branching revealed that ParA tip anchorage of the chromosome is crucial for efficient formation and extension of germ tubes and new branches. Our data indicate that ParA interacting with the ParB complex at one of chromosomal copies directs it to populate the new hyphal tip. In apically extending and branching Streptomyces vegetative hyphae that do not undergo canonical bacterial cell division, multiple copies of the chromosome replicate asynchronously and remain visibly unseparated [24,28]. Due to this mode of growth and a lack of chromosome-locus specific labeling tools available for Streptomyces until now, the distribution of nucleoids in vegetative hyphae remained elusive. Furthermore, little was known about the function of segregation proteins during vegetative growth. In contrast to sporulating hyphae, in which ParA extends along the whole cell and accompanies the array or regularly spaced ParB complexes, in vegetative hyphae ParA is exclusively associated with the hyphal tip [21,29]. We noted before the presence of irregularly spaced ParB complexes of varying intensity along the hyphae and the tip-proximal ParB complexes which exhibit the highest fluorescence intensity [26]. That observation is in agreement with the finding that ParA is required for efficient assembly of the ParB nucleoprotein complex [22]. Therefore, we hypothesized that the apical ParB complex assembled in the presence of ParA assures organization of the tip-proximal chromosome during apical extension. OriC labeling using the FROS system revealed that chromosomes follow the extending hyphal tip. The distance between the tip and oriC of the apical chromosome is constant. Immediately after oriC duplication, ParB complexes are reestablished at both daughter oriCs, but only one of them is captured by an apically localized ParA complex to renew the tip anchorage of the apical chromosome. A distinctive feature of Streptomyces tip-proximal replication is that one of the daughter oriCs is abandoned by the segregation machinery. We suggest that the presence of ParA, exclusively at the tip, most likely explains segregation and anchoring of the tip-proximal oriC alone. Thus, at the tips of Streptomyces vegetative hyphae, oriC segregation is asymmetric. This mode of chromosome segregation is somewhat similar to classical chromosome segregation described in other bacteria such as C. crescentus in which ParA segregates one of the daughter oriCs unidirectionally. However, even though in Streptomyces ParA is required for separation of the duplicated tip-proximal oriCs, the oriC does not move toward the tip and the distance between apical ParB/oriC complex and the tip remains constant during segregation. Unlike in C. crescentus and V. cholerae, ParA assemblies in Streptomyces were not observed to retract to the cell pole (tip). Those observations, and the increase of the distance between apical ParB/oriC complex and the tip during ParA overexpression, suggests a lower degree of ParA assembly dynamics in Streptomyces compared to rod-shaped bacteria. Thus, in Streptomyces ParA dependent oriCs separation is not a result of the active movement of one of them but rather is directly dependent on the tip extension, which suggests the anchorage model. It was shown earlier that in Streptomyces, ParA interacts with the tip-associated, coiled-coil Scy protein, which together with DivIVA and FilP forms a polarisome complex (TIPOC) [18,29,32]. Until now, it was believed that the function of the polar complex was to maintain the rigidity of the extending tip and to establish the cell wall synthesis machinery [14,33]. We have revealed an additional function, which is to provide anchorage for the oriC of the apical chromosome. Presumably densely-packed protein complexes occupy the tip-proximal space occluding the chromosome from migrating to the very tip. The interaction between the polarisome, ParA and ParB complex assembled at oriC ensures that this chromosome closely follows the extending hyphae but also orientates the apical chromosome with its oriC toward the tip of the hyphae. Elimination of ParB moved the oriC region, but not the nucleoid, away from the tip, whereas the absence of ParA resulted in shifting of both the nucleoid and ParB complex away from the tip. This may be explained by potential interactions of tip-localized ParA with DNA (in this case, the tip proximal chromosome only), such as detected in other bacteria [34]. The apical anchorage of the segregation machinery is not unique to Streptomyces. In C. crescentus, polar proteins TipN and PopZ, mediate chromosome segregation via interactions with ParA and ParB [9,35–37]. In V. cholerae oriCI targeting to the cell poles is executed by the interaction between the polar protein HubP and ParA [10]. During B. subtilis sporulation, the origin-binding protein RacA is recruited by apically localized DivIVA, and during vegetative growth DivIVA may be involved in an indirect interaction (via MinD) with segregation proteins and oriC positioning [38,39]. Among the Actinobacteria, DivIVA interacts with ParA in M. smegmatis; however, the oriC is positioned close to center of cell, at least in optimal growth conditions [40], and the biological role of this interaction has not yet been described. In contrast to M. smegmatis, in C. glutamicum oriC is also anchored at the cell pole by the interaction between the ParB and the pole associated DivIVA [41,42]. Interestingly, in S. coelicolor we could never detect the direct interaction between ParB or ParA and DivIVA. In fact, the distance between ParB and the tip-localized DivIVA is likely to exclude this possibility. The subapical localization of oriC in S. coelicolor, is reminiscent of the oriC localization in Myxococcus xanthus, however the positioning of the oriC is presumably dependent on different mechanisms. At this point it cannot be fully determined if the interaction of ParA with Scy in Streptomyces may be playing a similar role to the sequestration of monomeric ParA by TipN or PopZ in C. crescentus. Scy does not bind the ParAK44E mutant (ATP binding and dimerization abolished), and it promotes dissociation of higher order ParA assembly [29]. Microscopy analysis showed that ParA, at least transiently, co-localizes with Scy in the DNA-free region at the hyphal tip. A distinctive features of Streptomyces ParA is its ability to form higher order structures in the absence of DNA [29]. On the basis of these observations we speculate that ParA may extend from the Scy complex at the tip forming a higher order structure that interacts with ParB and anchors oriC. However, due to its higher stability, upon contact with the segrosome, ParA assemblies, do not retract and instead serve as the anchor for the apical ParB complex. The lack of direct interaction between ParB and Scy or DivIVA excludes the immobilization of the ParB complex independently of ParA. Moreover, we have shown here that in the strains with parA mutations, including a mutation that abolishes interaction with Scy and ParB, or in absence of Scy, the connection between the tip and the first chromosome was lost. Thus, the unique features of Streptomyces ParA, which presumably have evolved as an adaptation to hyphal growth, contribute to a less dynamic mechanism of chromosome segregation, involving the anchorage of a single oriC region exclusively. During new branch formation, ParA recruited to the polarisome captures one of the multiple chromosomal oriCs complexed with ParB and directs it to the new hyphal tip. It was shown earlier that, DivIVA is required to establish a new hyphal branch and Scy organizes the new polarity center [17,18,43]. We have demonstrated that in strains lacking ParA and ParB, the branch or germ tube extension is frequently abolished. It was shown earlier that abortive branches are not populated by replisomes [27]. Here, we demonstrated that, in the parA and parB mutant strains, the length of the branch when it is populated with oriC is significantly increased and greater than the mean distance between oriC and the tip. This suggests that parA or parB deletion does not simply shift the oriC away from the tip but rather breaks its anchorage. We suggest that impaired new branch extension is the result of a delayed population of the new branch with the nucleoid (s). The eventual appearance of chromosomes in the empty branches is likely to be the result of diffusion from the crowded stems. The fact that oriC targeting to the branch does not need to directly follow oriC duplication proves that ParA mediated anchorage is not dependent on post-replicational segregation. It is tempting to speculate that the role of ParB complexes at the chromosomes along the hyphae is to facilitate targeting of chromosomes to the newly forming branches by interacting with ParA. This would represent a new function of ParA and polarisome complexes during germination and formation of new branches. The establishment of new hyphal tips is yet another feature of Streptomyces growth that resembles filamentous fungi. Hyphal growth in fungi is driven by Spitzenkörper, the polar structure which assembles secretory vesicles delivered to the apical region by cytoskeletal tracks [44]. In Aspergillus nidulans, it was shown that although branch initiation is independent of the presence of a nucleus, the population of the extending branch with the nuclei is very efficient and dependent on the coiled-coil proteins ApsB and ApsA, which are also responsible for microtubule organization and nuclei migration [45–47]. Interestingly, in Arabidopsis, an increase in the distance between the nucleus and the root hair apex stops cell growth [48]. Considering that all apically growing organisms, including plant root hairs, pollen tubes, fungal hyphae and filamentous bacteria, require the assembly of cell wall building blocks at one cell pole (Spitzenkörper in fungi) [49,50], we conclude that there is a need for a machinery that assures the delivery of genetic material into the elongating cell and suggest that similar mechanisms permitting polar growth have evolved in both eukaryotic and bacterial cells. The distribution of subapical chromosomes is rather random and we speculate that in the absence of cell division there is no requirement for their active segregation. In fact, the newly duplicated oriC in the hyphal stem are not separated after replication as efficiently as apical oriCs, as expected in the absence of ParA. Interestingly, we observed a flow of chromosomes following the extending tip, which indicated that although only the apical oriC is tightly anchored, the other chromosomes, also follow the extending hyphae. The flow of the replisomes in hyphae has been observed in Streptomyces before [27]; however, the mechanism of the movement of chromosomes remains unknown. It is possible that molecular crowding and viscoelastic properties of the environment of nucleoids and/or internucleoid linkages provide the cytoplasmic flow that pulls the chromosomes behind the tip in the apically extending cell. The nucleoid flow in extending hyphae is reminiscent of nuclear migration associated with hyphal growth in filamentous fungi. In polarized cells of filamentous fungi, the nuclei distribution is dependent on motor proteins but movement of the nuclei is regarded to be also partially passive and driven by cytoplasmic flow [45,51,52]. In addition, in apically extending plant root hairs, nuclei follow the tip at a constant distance to the cortex [48]. It is likely that hyphal growth may impose a similar pattern of chromosome migration in Streptomyces. In conclusion, our observations support a chromosomal anchor model (Fig 7) in which ParA interacts with a polarisome and binds one of the multiple oriCs associated with ParB. Soon after the initiation of apical chromosome replication, ParA captures one, tip-proximal, daughter oriC and maintains its constant distance to the tip. Remarkably the other daughter oriC is abandoned by the extending tip and allows that chromosome to act as the template for tip distal replication and chromosome population of branches. ParA also imposes the apical oriC orientation of the first chromosome. During the new hyphal tip establishment, ParA serves as a tip-anchor that captures one of the ParB-oriC complexes from multigenomic cellular compartments. Targeting the chromosome to the new hyphal tube permits its efficient extension. Thus, although the interaction of ParA with polar proteins as part of the chromosome segregation mechanism is shared by a number of bacterial species, in Streptomyces, this interaction provides a unique tip-anchor essential during spore germination and branching. The E. coli and S. coelicolor strains used are listed in S1 Table. DNA manipulations were carried out by standard protocols [53]. Culture conditions, antibiotic concentrations, transformation and conjugation methods followed general procedures for E. coli [53] and Streptomyces [54]. The oriC region of S. coelicolor M145 was labelled with Tn5341 carrying 120 tandemly arrayed copies of tetO and the apramycin resistance gene. Following in vitro transposition of cosmid SCH18 with Tn5431 [31,55], an obtained cosmid (EJTH31A) carrying the transposon at a site located in an intergenic region ~29 kb from the oriC region was selected. Introduction of the EJTH31A cosmid into S. coelicolor M145 generated the strain, EJTH31. This strain was further modified by introduction of the integrating plasmid pMS83-mCherry to express tetRmcherry fusion yielding the “wild type” FROS strain, DJ-NL102 (S1A Fig) (for more details, see Supplementary Information). To avoid potential selection for chromosomal rearrangements or deletions of the tetO cassette, the FROS strain was always cultured in the presence of 0. 1 μg ml-1 anhydrotetracycline, except for microscopy analyses of the oriC localization. A detailed description of other strains construction is presented in the Supplementary Information. For the fixed microscopy specimen preparation, spores were inoculated in the acute-angled junction of coverslips inserted at 45° in MM agar containing 1% mannitol [54] and cultured for 21–24 h. For ParA induction, the strain DJ532 was grown in the presence of thiostrepton (5 μg ml-1). Staining procedures were as described previously [21,56]. Briefly, samples were fixed for 10 min with paraformaldehyde/glutaraldehyde mixture, digested 2 min with 1 mg ml-1 lysozyme, washed with PBS and blocked with 2% BSA. For immunostaining, samples were incubated with antibody against ParA (1: 5000 dilution) overnight, washed six times with PBS and then incubated for 1 h with secondary antibody (anti-rabbit) conjugated with Alexa Fluor546. For DNA staining, samples were incubated with 1–10 μg ml-1 propidium iodide, and for cell wall visualization with 1–10 μg ml-1 WGA-Texas red or Alexa Fluor350 conjugate (Molecular Probes). After five washes with PBS, coverslips were mounted in 50% glycerol in PBS buffer. Florescence microscopy was carried out using a Zeiss AxioImager M1 or Zeiss Observer Z1 with camera AxioCam MRm Rev. The images were analyzed by AxioVision Rel. 4. 5 Software equipped with AutoMeasure module or FIJI Software. R analysis tool was used for foci detection [57]. The focus was identified when the fluorescence intensity was above the threshold, which was set as 50% of the highest signal intensity in the particular hyphae. For time-lapse imaging, spore dilutions were spotted onto cellophane membranes on MM solid medium supplemented with 1% mannitol and cultured for 2 h (for germination analysis) or 24 h (for vegetative growth analysis) before the start of the experiment. The cellophane membrane was transferred to a μ-dish (∅35 mm, Ibidi) and covered with a block of agar. Fluorescence microscopy was carried out at 30°C, using a Delta Vision Elite inverted microscope equipped with a 100x oil-immersion objective, ultimate focus and Olympus IX71 camera. Images were acquired every 10 minutes using DIC (differential interference contrast) and EGFP or cherry filter set with the exposition time of 50 and 100 or 200 ms, respectively. Images were analyzed using FIJI software. Data analysis was performed using R software. The cell contour was determined manually in DIC images. After background signal subtraction, the fluorescence along the hyphae was averaged using Fiji software and subsequently R package Peaks was used to find and localize foci [58]. For every time-point, a fluorescence intensity profile for the whole hyphae was generated. Based on the raw data a smoothed intensity profile was calculated using R package Peaks with a Markov chain method. All maxima indicated by the algorithm were manually checked and false positives were discarded. This approach allowed us to determine the exact position of all foci in the hyphae. If two maxima were observed that could be distinguished by the Peaks package, then we assumed that two foci were present in the hyphae (S3A, S3B and S3C Fig). Density (used for distribution analysis) was calculated using a kernel density estimate in R program [57]. ANOVA, Games-Howell test (Figs 3,4 and 6), and chi-squared test (Fig 6) were applied for statistical analysis. Differences were considered significant when p-values were lower than 0. 05.

To proliferate, cells synchronize growth and division with chromosome segregation. In unicellular bacteria, chromosomes segregate during replication by active movement of nucleoprotein complexes toward the cell pole (s). Here, we asked the question how active chromosome segregation occurs in the absence of cell division, during hyphal growth and branching of the filamentous bacterium, Streptomyces coelicolor. We show that in multigenomic Streptomyces hyphae, the bacterial segregation machinery anchors a single chromosome at the hyphal tip. Through chromosomal anchorage, segregation proteins facilitate chromosome targeting to the newly formed germ tubes or branches. Thus, being adapted for apical growth, in Streptomyces hyphae the bacterial segregation machinery imposes a chromosome distribution that is reminiscent of nuclear distribution in apically growing eukaryotic cells such as filamentous fungi.

Abstract Introduction Results Discussion Materials and Methods

bacteriology cell physiology fluorescence imaging medicine and health sciences pathology and laboratory medicine fungal spore germination pathogens cell cycle and cell division cell processes microbiology cell polarity streptomyces mathematics statistics (mathematics) bacterial genetics microbial genetics fungal reproduction fungal pathogens microbial genomics research and analysis methods bacterial genomics mycology imaging techniques chromosome biology medical microbiology microbial pathogens confidence intervals cell biology genetics biology and life sciences physical sciences genomics chromosomes

2016

Unique Function of the Bacterial Chromosome Segregation Machinery in Apically Growing Streptomyces - Targeting the Chromosome to New Hyphal Tubes and its Anchorage at the Tips

Gram-negative bacteria secrete virulence factors and assemble fibre structures on their cell surface using specialized secretion systems. Three of these, T2SS, T3SS and T4PS, are characterized by large outer membrane channels formed by proteins called secretins. Usually, a cognate lipoprotein pilot is essential for the assembly of the secretin in the outer membrane. The structures of the pilotins of the T3SS and T4PS have been described. However in the T2SS, the molecular mechanism of this process is poorly understood and its structural basis is unknown. Here we report the crystal structure of the pilotin of the T2SS that comprises an arrangement of four α-helices profoundly different from previously solved pilotins from the T3SS and T4P and known four α-helix bundles. The architecture can be described as the insertion of one α-helical hairpin into a second open α-helical hairpin with bent final helix. NMR, CD and fluorescence spectroscopy show that the pilotin binds tightly to 18 residues close to the C-terminus of the secretin. These residues, unstructured before binding to the pilotin, become helical on binding. Data collected from crystals of the complex suggests how the secretin peptide binds to the pilotin and further experiments confirm the importance of these C-terminal residues in vivo. The secretins are an important group of bacterial membrane proteins whose function is to facilitate the transport of secreted proteins and macromolecular complexes across the outer membrane [1]. They are essential components of the type II and type III secretion systems (T2SS and T3SS respectively) and play a key role in the assembly of type IV pili (T4P) and release of filamentous bacteriophages. Determination of the structure of secretins has been confined to low-resolution transmission electron microscopy and cryo EM studies [2]–[5] which show membrane penetrating ring structures with 12–14 fold rotational symmetry. A specialized class of small lipoprotein pilotins bind their cognate secretins and facilitate oligomerization, insertion and proper assembly in the outer bacterial membrane. In this paper we explore the structure and function of the pilotin from D. dadantii (OutS), several other pilot proteins have been described [6]–[12]. Pilotins whose structures have been determined are MxiM (PDB code: 1Y9L) of the T3SS of Shigella flexneri and PilW/PilF (PDB codes: 2VQ2) /2FI7 and 2HO1) of the T4P of Neisseria meningitidis or Pseudomonas aeruginosa [13], [14]. The cracked β-barrel structure of MxiM has been solved in complex with an 18 residue peptide from the cognate secretin MxiD (PDB code: 2JW1) and the authors propose a model for the way MxiM assists MxiD assembly [10], [15]. The other known pilot structure, PilW/PilF, appears to perform a broadly similar function to MxiM, ensuring multimerization of the secretin PilQ into the outer membrane, but has a different architecture comprising six serial α-helical tetratricopeptide repeats [9], [12]. A third auxiliary secretin-binding protein has been characterised structurally, PilP, which is also involved in the assembly or stability of the secretin PilQ of Pseudomonas aeruginosa [16], the structure comprises a sandwich of two sheets each with three anti-parallel β-strands. The type II secretion system spans both the inner and outer bacterial membranes [17], [18]. It consists of an inner membrane subcomplex, periplasmic pseudopilins and the outer membrane secretin [19]. There have been considerable recent advances in our understanding of the T2SS secretin. First the structure of the N-terminal periplasmic domains N0, N1, and N2 in complex with a nanobody [20] was elucidated (PDB code 3EZJ) and secondly a cryo EM reconstruction of the secretin itself has been described [3]. In the absence of pilotin the D. dadantii secretin (OutD) mislocates to the inner membrane [21]. The pilotin possesses at its N-terminus the characteristic lipoprotein signal sequence (LAAC), with the signal peptidase LspA cleaving site just before the cysteine to which the lipid is covalently attached. The pilotin (OutS) binds to the C-terminal 62 residues of the secretin (OutD) [11], [19]. Here we elucidate the structure of the T2SS pilotin and show that it binds tightly to 18 residues close to the C-terminus of the secretin subunit causing this unstructured region to become helical on forming the complex. To ensure authentic folding and production of soluble Dickeya dadantii pilotin (OutS) in the E. coli periplasm, the PelB secretion sequence was substituted in place of the N-terminal lipidation sequence thereby preventing lipidation of the pilotin. This substitution facilitated protein production and crystallization without compromising secretin-binding [11]. Cleavage of the secretion signal accompanies transport in to the periplasm. The crystal structure of the pilotin was determined using the anomalous scattering from a potassium tetrachloroplatinate derivative and the structure refined at 1. 65 Å resolution (Table 1). The two copies of the pilotin subunit in the asymmetric unit of the crystal are virtually identical in structure (root mean square deviation of 94 α-carbon atoms, residues 38 to 132, is 0. 243 Å). The structure is clearly defined in the electron density map except for the N-terminal residues preceding Val 38 which presumably form a flexible linker to the lipidation site. The architecture of the pilotin is the remarkable insertion of one α-helical hairpin into a second open α-helical hairpin with bent final helix (Figure 1); this is unlike the two other pilotin structures solved and is profoundly different from any previously described four helix bundle. The first helix of five turns (residues 40–60), is connected to the second of four turns (residues 69–82) by an 8 residue loop. The second loop of 10 residues connects to the third helix of four turns (residues 93–106) which packs against helix one. A short three residue loop which connects helices three and four and the disulfide bridge, between Cys 115 in the second turn of helix four and Cys 61 the first residue of the helix one to helix two loop (Figure 1B), sets the scene for the packing of helix four (residues 111–130). The pronounced bend of 65° in helix four is important for the architecture; the helix has three large hydrophobic residues which are at least partially buried by interactions with hydrophobic residues on the other three helices: Phe 118, Met 122, and Phe 125. The requirement to pack conserved Phe 125 appears to dictate the severe bend of this helical element. In the crystal the pilotin subunits form a dimer, with Arg 63 and Asn 119 (Figure 1B) involved in an electrostatic interface, between subunits, however there is currently no evidence that dimerization occurs in solution or in vivo. The majority of the 13 absolutely conserved residues in the sequence alignment (Figure S1A) appear to be of structural rather than of direct functional significance. The two highly conserved cysteine residues, 61 and 115, form the disulfide bridge between helices α1 and α4 that stabilizes the correct nested α-helical protein fold of the pilotin is functionally relevant. When a reducing agent was used in pull-down assays or in bacterial two-hybrid tests, the pilotin was unable to bind the cognate secretin (Figure S6). Presumably this is because the disulfide is reduced and the pilotin did not fold correctly in the cytoplasm. Interestingly, the previous mutagenesis analysis revealed several structurally or functionally relevant residues of pilotin (OutS), notably conserved Leu 57, Arg 63 and Ser 97 [11]. Substitution each of these prevented secretin (OutD) targeting to the outer membrane. Conserved Cys 21 covalently attaches to the lipid and this residue is essential for targeting the pilotin to the outer membrane. Conserved Gln 46, on the solvent exposed surface of helix 1 (Figure 1), must also be of functional rather than structural significance and maps to the extremity of the concave surface of the pilotin formed by helices α1, α3 and α4; it is plausible that this concave surface is the binding site for the secretin (Figure 1B). Conserved residues including: Gln 46 and Leu 104 and semi-conserved Leu/Val/Ile 50, Phe/Leu 118 are also in this region. Residues 49,52 and 53 on the solvent exposed surface of helix α1 are Ser/Ala for the former two and Ala/Gly for the latter, respectively (Figure S1A). This conservation of small residues at these positions is consistent with this region being important in binding as there is no structural reason why larger residues could not be accommodated at these sites unless the secretin binds tight up against the first helix (Figure 1). A DaliLite database search revealed that P40 nucleoprotein has a similar arrangement of α-helices to that of the pilotin. The Dali score was 6. 0 and sequence identity 5%. The P40 nucleoprotein domain architecture is however substantially more complex with seven helices instead of the pilotin' s four. The concave surface of the corresponding helices of P40 nucleoprotein is occupied by a helix supporting the view that this is the binding site for an α-helix. Mutations of the pilotin binding surface confirm the importance for binding of some of the residues decorating the concave surface (Table S1). Mutating Ser 49 to Arg has a profound effect on binding and the mutants Leu 96 Ala, Leu 100 Ala and Gln 114 Ala have a substantial effect on binding as expected if this is the binding surface (Figure 1B). It had previously been shown that the C-terminal 62 residues of the secretin bound to the pilotin [11]. This secretin peptide was produced with 15N-label as a fusion to GST and then released with PreScision protease. The backbone amide protons were poorly dispersed in the 1H-15N HSQC spectra revealing the C-terminal 62 residues are unstructured in solution (Figure S4). NMR cross-titration studies revealed that only peaks corresponding to residues 691–708 of the secretin peptide were shifted on addition of unlabelled pilotin (Figure S2; with assignment of secretin peptide shown in Figure S3). When the pilotin was 15N-labelled, good dispersion of the backbone amides protons was observed as expected given its folded structure (Figure 2). Titration of the unlabelled 62 residue secretin peptide into the 15N-labelled pilotin produced a large number of peak shifts (Figure 2A). Shift perturbations are extremely sensitive indicators of structural changes and the extent of the changes observed is compatible with the secretin peptide decorating the surface of the pilotin and causing subtle structural rearrangements, perhaps in packing interactions in the hydrophobic core, reflected in chemical shift changes across much of the structure. The secretin peptide binds to the pilotin in a 1∶1 stoichiometric ratio as determined from the NMR titration where the intensity of the shifted peaks of the complex saturate at an equimolar ratio of OutD to OutS. Since only residues 691–708 of the secretin were affected by interaction with pilotin (Figure S2), a synthetic 18 residue peptide corresponding in sequence to these residues was assessed. The pattern of shifts in 1H-15N HSQC pilotin spectrum observed using this synthetic 18 residue peptide was identical to that using the 62 residue secretin peptide (Figure 2B, C) confirming that it is these 18 residues that are those principally involved in binding to the pilotin. Circular dichroism measurements also showed the unstructured nature of the secretin peptide and provided evidence that the peptide becomes helical on binding (Figure 3A). The signal saturates at a stoichiometric ratio of secretin peptide to pilotin. The CD spectra of the pilotin and secretin peptide together correspond to more helical structure than the spectra of the pilotin and secretin peptide summed. The additional helical content can be quantified as 12 residues assuming all secretin and pilotin molecules are in complex, a reasonable assumption given the high affinity of complex formation (see below). The most plausible explanation for this is that 12 residues of the secretin peptide become helical on binding to the pilotin. The helical propensity of the 18 residue secretin peptide was apparent from secondary structure predictions (Jpred [22] and shown on Figure S1B). To estimate the binding affinity of pilotin for the secretin peptide, fluorescence spectroscopy was used. Since the 18 residue secretin peptide has no tryptophan residues, quenching of the fluorescence signal from the single tryptophan residue in the pilotin on addition of the secretin peptide, was used to determine the affinity of binding. The 1∶1 stoichiometric binding ratio can be seen from the saturation of the fluorescence quenching of OutS by an equimolar quantity of the secretin peptide (Figure 3B). The binding of the 18 residue peptide is tight with Kd of 55±20 nM (Figure 3B) and is comparable to that of the T3SS pilotin-secretin complex [10], [15]. 3JHNHA spectra of the complex showed peaks coupled by less than 5 Hz (Figure S5) confirming that at least four residues of the secretin peptide become helical on binding to the pilotin, there may be more, but they are hidden by overlapping peaks. Crystals of the pilotin/secretin complex were grown, they belong to space group P65 but are twinned (Table S2) and have solvent channels with disordered density within. Nevertheless four pilotin subunits can be found by molecular replacement and density for peptide can be seen occupying the concave surface of the pilotin. The evidence that suggests residues 694 to 704 of the secretin adopt a helical conformation in the electron density map, a part of the simulated annealed omit map is shown in Figure 3C. Ten residues are in helical conformation and the hydrophobic surface of the amphipatic helix interacts with the hydrophobic surface of the pilotin (Figure 3D, E). In this model the methyl groups of T692 and V697 interact with L100 and L104, I701 interacts with L50 and M122; F704 (with L47, V79 and F125) and F694 (with F118, L104 and Q114) interactions occur either side of these central interactions. D107 is the N-terminal helix capping residue, stabilizing the helix dipole of the bound secretin peptide. The peptide binds tight up against Q46, S49, A52 and A53, providing an explanation for their conservation or presence only as small residues in the case of the latter three. Interestingly, the same dimer as described for the non-complexed secretin is seen in these crystals too suggesting that the interaction may have some biological relevance. The quality of the refinement is relatively poor because of the disordered protein and for that reason the structure is being referred to as a model of the secretin peptide/pilotin interaction. Intrinsically disordered regions of proteins such as these C-terminal residues of the secretin subunit facilitate binding by increasing their capture radius for cognate partner, the so-called fly-casting mechanism [23]. Initial weak binding may draw the secretin and pilotin together and as the secretin peptide folds on the pilotin surface the binding becomes tighter, locking the two together. A series of in vivo experiments were used to test the proposed model of secretin binding to the pilotin. The 62 residue C-terminus of secretin possesses three putative α-helices (Figure 4A) [22]. If the region consisting of the first two C-terminal helices were deleted, the truncated OutDΔC1 behaves like wild type secretin. Firstly, OutDΔC1 was barely detectable in the absence but was well produced in the presence of pilotin OutS (Figure 4B). Secondly, with pilotin, the mutant secretin was mainly recovered in the outer membrane fractions (Figure 4C). Consistent with the outer membrane location, expression of OutDΔC1 in the presence of OutS results in a rather low level of pspA induction. Phage shock protein (psp) response helps to maintain proton motive force in cells under pmf-dissipating stress and is indicative of mislocalization of the secretins in the inner membrane [8], [11]. These results are consistent with the NMR experiments demonstrating that it is the 18 residue C-terminus of OutD which binds tightly to the pilotin. Despite its outer membrane location, OutDΔC1 was unable to restore pectinase secretion in D. dadantii ΔoutD A3559 strain (data not shown) indicating an important functional relevance of the deleted region. Deletion of the extreme C-terminus of secretin resulted in partial stabilization of the truncated OutDΔC2 secretin as judged from the quantity produced in the absence of the pilotin, but prevented its correct targeting to the outer membrane (Figure 4B, C). In the presence of pilotin, the amount of OutDΔC2 was increased indicating that the pilotin can still stabilize and hence interact with the truncated secretin but is not able to target it to the outer membrane. In agreement with this, expression of OutDΔC2 strongly induced pspA even in the presence of pilotin. Deletion of the full 62 residue C-terminus of OutD resulted in neither stabilization nor the correct targeting of the truncated secretin OutDΔC3 as judged by low protein content but high pspA level (Figure 4B). The type II secretin/pilotin complex from Klebsiella oxytoca has been imaged by cryo electron microscopy at modest resolution [24]. The secretin subunits form a dodecameric ring with relatively weak radial arms that the authors tentatively assign as the pilotin or pilotin bound to the secretin C-terminus [4], [24]. Comparing the envelope of the complex with that determined by Reichow et al. (2010) of the secretin only [3], the radial arms are located either in the periplasm or in the inner leaflet of the outer membrane. This is the position anticipated from the observation in this work that the pilotin interacts tightly with 18 residues close to the C-terminus of the secretin subunit, so it is entirely plausible that the radial arms seen in the cryo-EM map correspond to the unusual four helix bundle of the type II secretion pilotin bound to the induced C-terminal helix of the secretin subunit. Recent structural studies have revealed striking structural similarities between components from distant secretion systems and other bacterial cell machines. Besides the expected structural homologies between several conserved components of ancestrally related T2SS and T4P [25]–[27] several other remarkable structural similarities should be mentioned. Notably, the extreme N-terminal N0 domain is shared by secretins from the T2SS, T3SS and T4P but is also structurally related to a domain of lipoprotein DotD from the T4SS, a domain of VgrG from the T6SS and from a TonB dependent receptor FpvA [20], [28]–[30]. Similarly the N1/N2 domains of secretins show a significant structural homology with several ring-forming proteins from the T3SS [20], [31], [32]. It is therefore becoming common to attribute similar function to the proteins or domains of a related structure. Given this background it is remarkable then that the T2SS pilotin described here is profoundly different in architecture to the T3SS pilotin but has similar function. Both the T2SS and the T3SS pilotin bind the extreme C-terminal region of their cognate secretins and this previously unstructured part of the secretin becomes an ordered α-helix on binding. It is therefore remarkable that the corresponding pilotins are different in architecture, one an open β-barrel (T3SS; MxiM) the other an unusual helical bundle (T2SS; OutS). These striking structural differences show that in these systems the pilotins have been evolved independently to play similar roles. The pET-20b (+) plasmid expressing non-lipidated OutS (residues 26 to 132) fused to N-terminal PelB signal peptide has been constructed previously [11] at a manner as the sequence of mature non-lipidated OutS after cleavage by the signal peptidase LepB is: MDP26VKNT etc. To fuse a C-terminal 6His tag to non-lipidated OutS, SalI site was introduced at the end of outS sequence by using the primer (5′-ctt gac gcc atg cgc acc gtc gac tga ggg gga agc aac tgc) and the reverse complementary one (mutated bases are underlined). Then, by SalI/XhoI digestion the sequence coding for non-lipidated OutS was fused with that coding for 6His. Mutants of OutS were made using Strategene QuikChange and confirmed by sequencing. To generate OutD truncated derivatives, an Eco47III site and V678A substitution were introduced using the primer (5′-gcgcggcgaaggcaacggagcgctggataacaacaccctgc) and the reverse complementary one. This site and naturally existing NruI and PsiI sites were used to generate OutDΔC1 (Δ650–678), OutDΔC2 (Δ679–705) and OutDΔC3 (Δ650–705) derivatives. To fuse the C-terminal segments of OutD to GST, the corresponding gene fragments were subcloned from pTdB-OD plasmids expressing either OutD, or OutDΔC1, or OutDΔC2, or OutDΔC3 into pGEX-6P-3 or pGEX-3X vectors in frame with the GST coding sequence. E. coli BL21 (DE3) strain (Stratagene) was used to produce non-lipidated pilotin (OutS) and GST-secretin (OutD) derivatives. Non-lipidated OutS was released from the periplasm by osmotic shock as described previously [33] and purified by size-exclusion chromatography Superdex S75 10/300 GL (GE Healthcare). The OutD peptide was purified and then released from GST-OutD fusion as described previously [34]. For NMR spectroscopy, uniformly 15N- and 13C-labeled pilotin and secretin peptides were produced by growing cell cultures in M9 minimal medium that contained 15N-ammonium chloride and 13C-D-glucose (Cambridge Isotope Laboratories Inc.) as the sources of nitrogen and carbon, respectively. The 18 residue synthetic secretin peptide (residues 691 to 708 inclusive) was purchased from Generon. Cell membrane fractionation by sucrose density gradient centrifugation was performed as described previously [11] with steady-state cultures of E. coli NM522 (Stratagene) expressing OutD derivatives from pTdB-OD and OutS from pACT-S plasmid. The location of outer membrane porins was determined by staining with Coomassie G-250. The position of inner membrane fractions was estimated by immunoblotting with TolA antibodies and NADH oxidase activity. E. coli MC3 strain carrying a pspA-lacZ fusion [35] was used to estimate miss location of OutD derivatives. To assess functionality of OutD derivatives, complementation assays with D. dadantii ΔoutD A3559 strain were used as previously described [36]. SDS-PAGE and immunoblotting were performed as previously [19]. Anti-OutD rabbit serum was raised against entire OutD purified from a recombinant E. coli strain. Anti-TolA serum was kindly provided by J. C. Lazzaroni. Hampton Research sparse matrix screen was used to search for crystallization conditions. Crystals were grown using hanging drop vapour equilibration using 10 mg/ml OutS and a reservoir of 2 M ammonium sulphate, 2% PEG 400 and 0. 1 M HEPES pH 7. 5. Vitrification of crystals in liquid nitrogen was achieved using the reservoir solution with 2. 1 M ammonium sulphate and augmented with 15–25% glycerol. Data were collected at ESRF ID23-1 and processed using MOSFLM [37] and scaled using SCALA [38]. SAD data were collected from a crystal soaked in 25 mM potassium tetrachloroplatinate (K2PtCl4) for 4 days. The structure was solved using PHENIX [39] and COOT [40] and refined using the native data at 1. 65 Å resolution and non-crystallographic symmetry restraints. The final model comprises 188 amino acid residues and 208 water molecules. DALILITE [41] was used to search for similar structures, CLUSTALW [42] for sequence alignment and JPRED for secondary structure predictions [22]. Crystals of the complex were grown using a 1. 0∶1. 1 molar ratio of pilotin: secretin peptide and crystallized using a reservoir of 2 M ammonium sulphate, 0. 1 M Tris, pH 8. 5. Around 100 complex crystals were screened before a well-diffracting reasonable ordered crystal was found. The pilotin/secretin complex was solved by molecular replacement using data collected at DLS I02 and CCP4/PHENIX/COOT. These crystals appear to be P6522 but are most likely twinned P65 with four pilotin molecules in the asymmetric unit. The packing of the molecules is such that there are large solvent channels running through the crystal lattice. These solvent channels appear to have disordered protein present the modelling of which hampers refinement. The disordered regions do not gain clarity if the lower symmetry space group P32 is used (for more details see Table S2). Samples of 0. 05 to 0. 5 mM labelled proteins in 90% H2O, 10% 2H2O containing 20 mM Tris (pH 7. 0) and 150 mM NaCl. All NMR spectra were acquired at 15°C using Bruker Avance 700- and 600-MHz spectrometers. Assignment of 1H, 15N, and 13C resonances of the backbone was achieved by analysis of HNCACB, CBCA (CO) NH triple resonance experiments [43]. Far-UV CD measurements were made using a Jasco J-715 spectropolarimeter equipped with a PTC-348WI temperature controller. Spectra were recorded in 20 mM Tris, 150 mM NaCl (pH 7. 0) at 15°C using 1 mm path length fused silica cuvettes. The spectra are presented as differential absorbance after baseline subtraction. Calculations employed CONTIN, SELCON, and CDSSTR [44]. Fluorescence data were collected using a Jasco FP-6300 Spectrofluorometer. To avoid exciting tyrosyl side chains, an excitation wavelength of 290 nm was used. Emission spectra were recorded at 15°C in steps of 2 nm from 310 to 400 nm. The fluorescence signal at 340 nm was plotted to calculate Kd. Pilotin spectra was measured at 1 µM. 50 µM of secretin peptide prepared in 1 µM pilotin was titrated into pilotin solution (for more information see also Table S1 and reference [45]).

Pathogenic bacteria deliver toxins and virulence proteins into host cells and tissues using specialised secretion systems such as the type II and type III secretion systems. These secretion systems have a pore formed by secretin protein subunits through which the disease causing protein effectors and toxins pass. The secretin must be targeted to and assembled in the outer-membrane and a pilotin protein facilitates this process. In the absence of the pilotin the secretin is degraded or mislocates to the inner-membrane, in either case the secretion system in non-functional and the bacterium cannot cause disease. Here we show how the pilotin and the secretin of the type II secretion system interact, these insights may be useful for the development of antibacterial compounds to interfere with secretin targeting and assembly and defeat pathogenic bacteria such as Vibrio cholerae and enterotoxigenic E. coli which infect man and Dickeya dadantii which threatens food security.

Abstract Introduction Results/Discussion Materials and Methods

bacteriology biochemistry biology microbiology biophysics bacterial pathogens

2012

Structural and Functional Insights into the Pilotin-Secretin Complex of the Type II Secretion System

To generate complex bilateral motor patterns such as those underlying birdsong, neural activity must be highly coordinated across the two cerebral hemispheres. However, it remains largely elusive how this coordination is achieved given that interhemispheric communication between song-control areas in the avian cerebrum is restricted to projections received from bilaterally connecting areas in the mid- and hindbrain. By electrically stimulating cerebral premotor areas in zebra finches, we find that behavioral effectiveness of stimulation rapidly switches between hemispheres. In time intervals in which stimulation in one hemisphere tends to distort songs, stimulation in the other hemisphere is mostly ineffective, revealing an idiosyncratic form of motor dominance that bounces back and forth between hemispheres like a virtual ping-pong ball. The intervals of lateralized effectiveness are broadly distributed and are unrelated to simple spectral and temporal song features. Such interhemispheric switching could be an important dynamical aspect of neural coordination that may have evolved from simpler pattern generator circuits. Owing to its complexity and high precision, birdsong has provided an important animal model for studies of motor control. Adult zebra finch songs are formed by repetitions of a highly stereotyped motif that is composed of two to eight syllables and is acquired from a tutor during a critical sensorimotor period [1]. Because the stereotypy of birdsong is sustained after removal of auditory feedback, birdsong has been thought to be organized by a “central motor program” [2–4]. The main cerebral brain areas for vocal production are the robust nucleus of the arcopallium (RA), HVC (used as a proper name), and the lateral magnocellular nucleus of the anterior nidopallium (LMAN), the latter of which forms the output of an avian basal-ganglia pathway [5]. Song-related neural activity in premotor brain areas is precisely coordinated across hemispheres, because both hemispheres contribute to the production of one unique and highly stereotyped song. This precise coordination is illustrated by the strong synchronization of multiunit activity in left and right HVC during singing [6]. A useful method to probe the functional roles of premotor brain areas is electrical stimulation. In general, electrical stimulation during motor production leads to specific behavioral distortions that depend on the location of stimulation electrodes [7,8] as well as on the stimulation time (or phase) within ongoing motor patterns [9–12]. For example, in LMAN, which is involved in modulating birdsong by social context [13,14], unilateral electrical stimulation induces small transient effects on sound amplitude or sound pitch, depending on the precise stimulation time within the ongoing song motif [10]. In HVC, which generates adult song by means of ultrasparsely firing “clockwork” neurons [15,16], unilateral electrical stimulation also leads to transient song degradations such as syllable distortions and syllable truncations [17]. More importantly, both LMAN and HVC stimulation sometimes induce nontransient effects such as song stoppings or early song restarts [4]. During such restarting events caused by HVC stimulation, ongoing premotor activity in the contralateral HVC is reset within a few tens of milliseconds [18]. Given that there are no direct interhemispheric connections between cerebral song-control areas, interhemispheric synchronization and resetting must rely on common inputs to the song-control system from interhemispherically connected mid- and hindbrain areas [19–23] (Figure 1). To explore the mechanisms of interhemispheric coordination and the dependence of song distortions on stimulation time, we chronically implanted HVC in adult male zebra finches with bipolar stimulation electrodes. We trained an artificial neural network to reliably detect the earliest possible note in a song motif in real time and stimulated either right or left HVC with a brief 0. 4-ms biphasic (0. 2 ms/phase) current pulse at random time lags after detection. We frequently interleaved stimulation trials by catch trials in which no stimulation was delivered. We also explored temporally modulated effectiveness of LMAN stimulation by using suitable multipulse current trains delivered to bipolar stimulation electrodes implanted in LMAN [10]. In line with earlier work, we found that unilateral HVC and LMAN stimulation distorted songs at the levels of song syllables and song motifs (Figure 2A) [4,10,17,18]. By definition, syllable-level effects were restricted to the stimulated or the subsequent syllable and consisted of either syllable distortions or syllable truncations. On the other hand, motif-level effects were manifest in longer time windows after stimulation and consisted of sudden song stopping or early motif restarts (see Materials and Methods for exact definition of effects). The prevalence of syllable- and motif-level distortions caused by HVC and LMAN stimulation is reported for all birds in Table S1. When hundreds of stimulated motifs were reordered by stimulation time, a temporal contiguity of stimulation effects became apparent in which nearby stimulation times led to qualitatively similar song distortions (Figure 2B). Hence, song distortions were not random, but were often deterministically linked with stimulation time, possibly caused by strong perturbation of stereotyped premotor activity. A more detailed analysis revealed that song distortions most frequently occurred on the syllable level within several tens of milliseconds after stimulation. The probability of sound-amplitude distortions sharply increased 20 ms after stimulation, peaked roughly 50 ms after stimulation, and decayed thereafter (Figure 2C). This sharp rise agrees with measurements of air sac pressure deviations, the average onset of which lags HVC stimulation by 15–20 ms [17], whereas the late decay suggests that some perturbations of neural activity were transient and affected only a subpopulation of neurons. Interestingly, on a fine time scale, not all distortions were locked to stimulation time. We occasionally observed syllables that were truncated, not with a fixed delay to stimulation, but during a fixed time point with respect to the unperturbed motif (Figure 2D and 2E). In these cases, stimulation needed to occur within some time interval before a particular note in order to truncate that note, revealing that the motor program exhibits time points of high perturbation sensitivity. And, more interestingly, sometimes stimulation effects such as early motif restarts occurred neither after a fixed latency to stimulation nor at a fixed time point of the unperturbed motif, but at some intermediate time (Figure 2D), further demonstrating nonlinear timing aspects of the song motor program. We automated the inspection of song distortions by analysis of sound amplitudes. We were mostly interested in motif-level effects because these seemed to arise from wide-spread and irreversible perturbation of premotor activity. For each stimulation time, we computed a late-effect (LE) value, defined as the fraction of 3. 9-ms time bins in a 78–312-ms window after stimulation in which sound amplitudes were significantly different from amplitudes recorded during catch trials (see Materials and Methods). LE curves as a function of stimulation time had many sharp peaks that corresponded to different motif-level effects, separated by troughs in which stimulation was rather ineffective (on the motif level). When we increased the stimulation currents, the set of effective stimulation times grew, as revealed by LE peaks that grew in height and width (Figure 2F). At the extreme of very high currents on the order of 0. 5–1 mA, birds always stopped singing, and significant LEs were seen for all stimulation times (n = 3 birds, unpublished data). In this study, our experimental strategy was to rapidly tune stimulation currents in order to observe highly modulated LE curves with coexistence of very large and close to zero values, a task that typically was achieved within 2 d. At the current intensities chosen, LE curves displayed diverse peaks (the mean peak width at the effectiveness threshold was 20 ms, median 8 ms, range 4 to 160 ms, n = 20 HVC stimulation sites in 10 birds). This wide range of peak widths in LE curves indicates that HVC stimulation perturbed neural activity on multiple time scales. The strong modulation of LE curves suggests rapid waxing and waning of the ipsilateral HVC drive, raising the question about modulation in the contralateral hemisphere. To probe evidence of lateralized stimulation effectiveness, we implanted birds with stimulation electrodes in both left and right HVC, and performed unilateral stimulation in randomly chosen hemispheres and at random time lags after note detection. After sorting all trials recorded over 1–3 d by hemisphere and stimulation time, a remarkable complementarity became apparent: For most stimulation times, stimulation effects were seen either for right- or left-side stimulation, but not for both (Figure 3A; see Figure S1 for all birds used in our study). LE curves associated with left and right HVC stimulation were strongly modulated, but in an alternating fashion. We quantified the interhemispheric complementarity of stimulation effectiveness by the correlation coefficient (CC; see Materials and Methods) between right and left LE curves, and found that negative correlations prevailed (average CC −0. 36, range −0. 68 to −0. 01, n = 10 birds). To assess the significance of these anticorrelations, in three birds we implanted two pairs of stimulation electrodes in right HVC (in a cross arrangement). By running the same experimental protocol on the two ipsilateral stimulation sites in HVC, we found that CCs between corresponding LE curves were positive (average CC 0. 36, range 0. 25 to 0. 46, n = 3 birds), illustrating that the dependence of stimulation effects on electrode position within HVC is weak and demonstrating that the anticorrelation of stimulation effectiveness in bilateral stimulation experiments was highly significant. Moreover, in two birds, we implanted stimulation electrodes in right HVC and right LMAN, and also found positive CCs between corresponding LE curves (0. 65 and 0. 51, Figure 3B). The CCs in all birds are depicted in Figure 3C (see Table S2 for additional characterizations of the complementarity of right and left LE curves). We interpret this complementarity as evidence that interhemispheric motor coordination involves temporally alternating neural mechanisms. We were interested in determining whether the events at which stimulation effectiveness switched from one hemisphere to the other were locked to salient song features and whether the resulting switching intervals obeyed any regularity. Visually, the effectiveness of electrical stimuli appeared to switch several times from one hemisphere to the other within a song motif, but often there was no obvious relationship between the discrete switching events and the sound spectrum produced at these times (inset of Figure 3A). When we assessed the events at which the effectiveness of electrical stimuli switched from one hemisphere to the other in terms of onsets of contralateral effectiveness (LE values larger than baseline), the mean switching interval was 35 ms (median 28 ms, range 4 to 150 ms). By contrast, when the switching events were defined by joint occurrence of ipsilateral ineffectiveness and onsets of contralateral effectiveness, the mean interval was 64 ms (median 44 ms, range 4 to 240 ms). Hence, on average, stimulus effectiveness switched back and forth between hemispheres within a few tens of milliseconds. However, our estimates of lateralized effectiveness and switching intervals must be interpreted with caution because of the aforementioned dependence of LE peak widths on stimulus current, implying that switching intervals depend (nontrivially) on stimulus current. Nevertheless, because we found broadly distributed switching intervals both across all birds and within single birds, there is little evidence of periodicity in this interhemispheric switching process. We further explored whether effective stimuli and their lateralization were related to specific sound features. Zebra finches mostly expire during syllables and inspire during syllable gaps [24]. Both expiratory and inspiratory nuclei in the brainstem project bilaterally and therefore may be involved in controlling effectiveness switching. Because we did not measure bronchial air flow, here we inferred respiratory patterns from sound pitch curves using the simplifying assumption that zero pitch during syllable gaps corresponded to inspiration and nonzero pitch to expiration. We defined a rhythm curve as being equal to one during expiration and zero during inspiration. There was no significant coherence between this rhythm curve and either the right or left LE curves (see Materials and Methods). These results were unchanged when we defined expiratory patterns in terms of pitch values in the limited range 20–5,000 Hz (thereby assuming that some high-pitched notes are generated during inspiration). Similarly, there was no significant coherence between right/left LE curves and each of the following: sound-amplitude curves, pitch curves (see Materials and Methods), syllable onset curves, and syllable offset curves (the latter were binary curves in which a pulse of variable width was set at the transitions between inspiration and expiration as assessed by the rhythm curve). Thus, the evidence for a consistent relationship between stimulation effectiveness and simple sound features is rather weak. Notice though that all our conclusions were reached from just a few seconds of effect-curve data (15 birds) and that it would be worthwhile to reinvestigate the relation between stimulation effectiveness and song features in the future provided a larger body of interhemispheric stimulation data will be available. We have demonstrated an interhemispheric switching process for vocal production. In this process, the motor program exhibits perturbation sensitivity that rapidly alternates between hemispheres. Such alternation is surprising given that HVC activity is highly synchronized across hemispheres during singing, and suggests that motor dominance rapidly switches back and forth between hemispheres. Possibly, the apparent alternation of dominance is related to birds' ability to independently control the two halves of their vocal organ [25,26]. However, alternation is not synonymous with independent control as it represents a restriction on independence. It is difficult to ascertain which hemisphere is dominating at any time in this switching process, because we were not able to find a simple relationship in zebra finches between stimulation effectiveness and either song features or song rhythm. On the one hand, one could argue that stimulation should be more effective in a dominant hemisphere because this hemisphere is being perturbed while generating a song in both syringeal halves. On the other hand, one could argue that stimulation should be less effective in a dominant hemisphere because the perturbation is not strong enough to overrule the ongoing activity there. In the following, we discuss the evidence for these two interpretations, as well as for interpretations on whether stimulations perturb activity in local or in distributed networks. From existing data, we cannot infer whether or not the motor apparatus necessitates continuous and simultaneous drive from both cerebral hemispheres: adult birds do not sing normally after unilateral RA lesions [19], but these data do not exclude the possibility that at any time, the effective motor program resides in just a single hemisphere and bounces back and forth between hemispheres during singing. For example, if singing at all times is based on activity in just a single hemisphere and the drive provided by premotor activity in the other hemisphere is temporarily gated off, then we would conclude that the dominant hemisphere is the one in which low-intensity stimulation is effective. In this view, stimulation of the nondominant hemisphere above a given current threshold would also be able to distort songs, because strong perturbations might ultimately find their way to the dominant side (past the gate) where they could interrupt the ongoing motor program. However, if normal song production at all times requires simultaneous contributions from both hemispheres, then high stimulation effectiveness might be an attribute of the nondominant hemisphere, because this hemisphere can be perturbed at lower stimulation currents. On the dominant side then, low-intensity stimulation would be corrected by redundant neural mechanisms that were not sufficiently perturbed by the stimulation. Not only the dominance question is difficult to address, but it is similarly difficult to tell whether song disruptions were entirely due to perturbation of local ongoing HVC activity or of a larger distributed network. For example, the number of spiking RA-projecting HVC neurons might drift randomly up and down during the song motif (with some inertia). Such random drifts could be associated with a compensatory increase in the number of spiking neurons in the contralateral HVC and thus to alternation of dominance. A compensatory process could be regulated during song development (e. g. , by neurogenesis [27] and programmed cell death), and therefore alternating dominance would not have to rely on real-time interhemispheric communication. According to this interpretation, LMAN and ipsilateral HVC stimulation lead to similar song distortions because LMAN stimulation perturbs RA-projecting HVC neurons, for example, via RA [28]. Although at this stage we cannot rule out this scenario, it is unclear why compensatory mechanisms would act across hemispheres, but not within the same hemisphere. Furthermore, it is difficult to reconcile this scenario with observations of interhemispheric synchronization of HVC activity and with some stimulation effects such as early song restarts. The more likely scenario within which our observations can be explained is that LMAN and HVC stimulation induce similar song distortions because of widespread perturbation of subpallial structures via RA. Because we observed a wide range of switching intervals, we found little support for the idea that switching times are determined by fixed signal propagation times (for example as reverberating activity in closed synaptic loops) or by the fixed period of a simple pattern generator circuit. Rather, some switching events may arise from detection of specific premotor patterns in one hemisphere that are subsequently relayed to the contralateral hemisphere. Interhemispheric switching processes in relation to motor production have been reported also in mammals, for example during the preparation of vocal production in humans, in which effectiveness of transcranial magnetic stimulation (TMS) of motor cortex alternates between hemispheres [12]. Interhemispheric switching has also recently been shown to exist during perceptual rivalry, as evidenced by the hemispheric dependence of magnetic and calorimetric stimulation [29]. Interhemispheric switching may thus be a fundamental mechanism by which sensory and motor-related activity is coordinated across hemispheres. In mammals, interhemispheric coordination seems to be mainly mediated by corticocortical projections [30,31]. However, during saccadic eye movements of split-brain monkeys, activity in the two hemispheres has been shown to remain coordinated despite the lack of cerebral commissures, suggesting that subcortical pathways can subserve coordination also in the mammalian brain [32], and suggesting that similarities may exist between interhemispheric coordination in avian and mammalian brains. Based on networks models, switching has been proposed to depend on competitive interactions [33] mediated by inhibition [34]. Evidence for interhemispheric inhibition has been found in TMS studies of human motor cortex [35,36]. We speculate that interhemispheric switching in songbirds could also rely on inhibitory mechanisms. A possible function of such inhibition could be to suppress mirror-symmetrical movements, which are thought to represent one of the default operation modes of bilateral motor systems [37]. In this sense, interhemispheric inhibition would coexist with more cooperative (excitatory) interactions between hemispheres. Inhibitory gating mechanisms could be mediated, for example, via tonically spiking Uva projection neurons (see also Figure 1) [38], and excitatory mechanisms could be relayed by respiratory nuclei, known to generate mirror-symmetrical respiratory patterns [39]. The reported interhemispheric switching process is reminiscent of one of the most prominent motor programs with left–right alternating dynamics, which is locomotion. In vertebrates, locomotion is subserved by central pattern generators in the spinal cord, which can display sustained rhythmic activity with left–right alterations even in in vitro preparations [40]. Because locomotion is much older than birdsong on an evolutionary time scale, phase-alternating neural circuits must have existed long before birds started to sing. Possibly, principles of limb coordination in locomotor circuits have been replicated by evolution for the more recent advent of birdsong. Some support for this idea comes from the conservation of bilateral projection patterns in brainstem nuclei of songbirds and non-songbirds [41], suggesting that old brain circuits have evolved to support new functions. Adult (>90 d old) male zebra finches (Taeniopygia guttata) were used for experiments. Birds were selected on the basis of singing frequency and song complexity, and were isolated in a sound-attenuating chamber. To maximize singing frequency, birds had visual contact to one or more female zebra finches through the glass door of the chamber. A total of 15 birds were used; data in one bird were discarded because HVC stimulation did not reliably produce motif-level effects. At the end of experiments, electrolytic lesions were performed at the stimulation sites by DC current injections (15 μA for 20 s), birds were killed by overdose of Nembutal, and stimulation sites were verified in histological brain sections. All procedures were approved by the Veterinary Office of the Kanton of Zurich. We delivered electrical stimuli with uniformly distributed probability over the time span of song motifs using custom written Labview software (National Instruments Corporation). With probability 0. 35, detection triggered microstimulation at site A, with probability 0. 35 at site B; and with probability 0. 3, no stimulation was delivered (catch trials). Electrodes were made of 50-μm stainless steel wire. Electrical stimuli in HVC consisted of a single 0. 4-ms biphasic (0. 2 ms/phase) current pulse of amplitude between 100 μA and 1 mA. In LMAN, electrical stimuli consisted of trains of ten biphasic current pulses at 400 Hz (0. 4 ms/phase; train duration 23. 3 ms) and amplitudes in the range 10–100 μA. The current threshold at which single-pulse stimulation in LMAN induced motif-level effects (song suspensions) was high (typically >1 mA). For this reason and to adhere to previous stimulation studies [4,10], we chose a multipulse paradigm in LMAN in which we stimulated for ten pulses at low currents (10 ∼ 100 μA per pulse). We distinguished among different syllable and motif-level effects as follows: Syllable truncations. First, we measured baseline distributions of syllable lengths from data of selected catch trials (only complete motifs). Stimulated syllables were then classified as truncated if their duration was within the lowest percentile of the baseline distribution. We searched for truncations only in a time window up to 156 ms (corresponding to 40 time bins of 3. 9 ms or 128 sound samples each) after stimulation. Syllable distortions. In each time bin after note detection, we calculated the baseline distribution of sound amplitudes during selected catch trials (no spontaneous song stopping). We then counted the number of 3. 9-ms time bins up to 78 ms post stimulation time in which the stimulation-related sound amplitudes were significantly different from baseline (percentile p < 0. 025 or p > 0. 975). If this number was large enough (binomial test, alpha = 0. 05), then we classified this stimulation effect as a syllable distortion. Distortions and truncations were not mutually exclusive. Motif stoppings and restarts. For each bird, we chose a sound-amplitude threshold slightly above cage-noise level (we found that a threshold of 20% into the 1–99th percentile interval worked well for all birds). For all stimulation trials, under visual supervision, we then used this threshold to mark the offset time of every prematurely stopped motif and the successive restart time of the following note (independently of whether this note come from a song syllable, an introductory note, or a call). If the offset time fell into a window from 0 to 156 ms after stimulation and there was no restart until 312 ms, we then classified the stimulation effect as a stopping event. If, on the other hand, there was a restart after a premature offset within 312 ms after stimulation, then the stimulation effect was a restart. Hence, restarts and stoppings were mutually exclusive (however, song stoppings and syllable truncations were not). All songs (stimulation and catch trials) were aligned by detection time. For each stimulation site, we sorted the trials by stimulation time and grouped them into 9. 75-ms sets with centers separated by 3. 9 ms from each other. With a mean stimulation range of approximately 500 ms and typical detection of 800–2,000 song motifs per day, we obtained roughly three to eight stimulation trials per set per day. Typically, we collected a mean of 10–20 trials per set and then tested for each set whether the sound amplitudes in 3. 9-ms bins after stimulation were different from amplitudes in matched time bins during catch trials using the Kolmogorov-Smirnov (KS) test (p < 0. 01). For each set, we quantified the stimulation effect by the fraction of time bins in which significant differences were detected. LE curves were based on bins ranging from 78 to 312 ms after stimulation (bins 21 to 80). Early-effect (EE) curves were based on bins ranging from 0 to 78 ms after stimulation (bins 1 to 20). To assess the time scales of song perturbations, we computed the peak widths in LE curves at the effectiveness threshold, defined by the baseline LE value during catch trials (binomial test, p < 0. 01). Our results did not depend critically on the EE and LE time windows in which syllable-level and motif-level effects were assessed. We chose the offset of the LE window (312 ms) as a compromise between being large enough to yield high sensitivity and small enough to not extend too far beyond the motif end where songs became highly variable. We set the onset of the LE window (or offset of the EE window, 78 ms) so as to exceed the peak time of stimulation effectiveness (Figure 2D), which was within 70 ms of stimulation (in agreement with previous reports [4]). Small changes in the LE window onsets (from 58. 5 to 117 ms) and LE window offsets (from 234 to 390 ms) did not affect our findings of interhemispheric switching in any way. By experimental design, our results were robust to variability in sound amplitudes caused by movements of the bird' s head relative to the microphone. That is, head-position variability must have had identical influences on sound amplitudes recorded during catch trials and during stimulation trials because we randomly chose all stimulation parameters right after each detection event (i. e. , whether and where to stimulate, and the stimulation time). Hence, by design there were no correlations between head position and stimulation parameters. We assessed the similarity between effect curves x and y associated with different stimulation sites by the (Pearson) CC: , where Cov (x, y) is the covariance between x and y. Because effect curves were nonnegative, stimulation times for which both x and y were ineffective (compared to sound-amplitude variability before stimulation, binomial test at 99% significance level) imposed a bias toward positive correlations. To avoid this bias, we ignored bilaterally ineffective stimulation times when calculating the CC (for LE curves, these were 32% of all stimulation times). Note that our conclusions were unchanged when CCs were calculated over the full set of stimulation times (thereby imposing a positive bias): the difference between average CCs in unilateral and bilateral stimulation experiments was highly significant in either case (p < 0. 001, Wilcoxon rank sum test). The relationship between right/left LE curves y and the rhythm curve z was investigated by the coherence, where p (yz) is the cross-spectral density, and p (yy) and p (zz) are the power spectral densities of LE and rhythm curves, respectively. We chose the coherence function because its phase insensitivity allowed us to detect significant correlations irrespective of their time lag. We assessed the significance of coherence peaks by testing whether these exceeded two jackknife estimates of standard deviation (corresponding to 95% confidence). The ten jackknifes were defined by leaving out each of the ten birds from the analysis. By visual inspection, stimulation effectiveness at the syllable level showed weaker interhemispheric complementarity than effectiveness at the motif-level. Yet, EE curves associated with bilateral stimulation (average CC −0. 13, range −0. 68 to 0. 43, n = 10) showed significantly lower correlations (p = 0. 019, Wilcoxon rank sum test) than EE curves associated with unilateral stimulation (average CC 0. 35, range 0. 04 to 0. 80, n = 5 birds). As before, to compute these CCs, we only considered stimulation times that were associated with effectiveness in at least one hemisphere, thereby omitting 25% of stimulation times (compared to omission of 32% for LE curves). In conclusion, alternating effectiveness was seen most clearly for stimuli that disrupted normal singing, but also for stimuli that induced immediate amplitude distortions. We investigated the possibility that pitch differences exist between times at which right and left HVC stimulation is effective. The coherence between the sound pitch curve and either right or left LE curve was not significant, neither when we considered the full pitch curve nor when we clamped the pitch curve to zero below either 2 or 5 kHz. Similarly, the median pitch during right-effective stimulation was not statistically different from the median pitch during left-effective stimulation (Wilcoxon rank sum test, p = 0. 4). We also tested whether pitch differences were seen at a particular time lag after effective stimulation times. We found that the median pitch 40 ms after left-effective stimulation was significantly higher than 40 ms after right-effective stimulation (p = 0. 031, n = 10 birds). However, when we excluded any one of two particular birds from the analysis, then significance broke down (p > 0. 1). Significance also broke down when assessed using a shuffle predictor of pitch differences in songs of randomly shuffled syllables and gaps from different birds (Monte Carlo simulations, p > 0. 05).

As for all vertebrates, the songbird cerebrum has two halves (or hemispheres), each of which controls mainly the muscles in one half of the body. Many motor behaviors such as singing rely on high coordination of activity in both hemispheres, yet little is known about the neural mechanisms of this coordination. By using electrical stimuli to briefly perturb the activity of neurons in the motor pathway during song production, we study their involvement in generating the different elements of a song in zebra finches. We find mostly disjoint time intervals in which stimulation of either the right or left hemisphere is effective in distorting a song. This interhemispheric switching of stimulation effectiveness is evidence of a novel form of ping-pong–like motor coordination. Because left–right alternation is the basis of many motor patterns such as swimming and walking, we speculate that interhemispheric switching in songbirds has its evolutionary roots in older circuit principles invented for locomotion.

Abstract Introduction Results Discussion Materials and Methods

neuroscience

2008

Rapid Interhemispheric Switching during Vocal Production in a Songbird

LXR (Liver X Receptors) act as “sensor” proteins that regulate cholesterol uptake, storage, and efflux. LXR signaling is known to influence proliferation of different cell types including human prostatic carcinoma (PCa) cell lines. This study shows that deletion of LXR in mouse fed a high-cholesterol diet recapitulates initial steps of PCa development. Elevation of circulating cholesterol in Lxrαβ-/- double knockout mice results in aberrant cholesterol ester accumulation and prostatic intra-epithelial neoplasia. This phenotype is linked to increased expression of the histone methyl transferase EZH2 (Enhancer of Zeste Homolog 2), which results in the down-regulation of the tumor suppressors Msmb and Nkx3. 1 through increased methylation of lysine 27 of histone H3 (H3K27) on their promoter regions. Altogether, our data provide a novel link between LXR, cholesterol homeostasis, and epigenetic control of tumor suppressor gene expression. The Liver X Receptors (LXRα, encoded by the gene Nr1h3, and LXRβ, encoded by the gene Nr1h2) belong to the nuclear receptor superfamily and bind to naturally occurring oxidized forms of cholesterol, known as oxysterols [1]–[3]. These receptors heterodimerize with RXR (Retinoid X Receptor) and stimulate various target genes expression, among which, genes encoding proteins in charge of cholesterol efflux, storage and uptake. Deletion of these receptors in mouse has been previously associated with the development of benign prostatic hyperplasia (BPH) lesions in ventral prostates [4], [5]. These findings enlighten the role of LXR in prostate homeostasis. However, BPH and prostate cancer (PCa) appear in distinct regions of the prostate and have distinct etiologies. Therefore, not much is known about PCa and LXR in vivo. Consistent with a potential role in prostate tumor formation, LXR have been reported to modulate proliferation [6], [7] and survival [8] of human prostatic cells in culture and in xenograft models. In these models, inhibition of proliferation through LXR activation was inversely correlated with expression of the ATP-binding cassette A1 (ABCA1) and G1 (ABCG1), two known target genes of LXR, which are involved in cholesterol efflux [9]. These observations suggest that the tumor suppressive activity of LXR on human PCa cell lines could result from their capacity to limit intracellular cholesterol concentration. This notion was supported in vivo by exposure of the transgenic adenocarcinoma of the mouse prostate (TRAMP) model, which carries a transgene encoding the SV40 large T antigen driven by the probasin promoter, to a high cholesterol diet. In TRAMP mice, this diet led to an acceleration of prostate tumor development [10]. A similar diet also increased aggressiveness of tumors generated by LNCaP cells in xenograft experiments [11]. On the basis of these observations, we hypothesized that LXR, through control of cholesterol metabolism, could act as “gatekeeper” preventing prostate tumor development. Thus we investigated the consequence of LXR ablation in the dorsal prostates of mice fed a high cholesterol diet. Under a standard diet, dorsolateral prostates of Lxrαβ-/- double knockout mice (Lxr-/-) were histologically indistinguishable from their wild-type (WT) counterparts, as shown by H&E staining (Figure 1Aa and e) and Ki67 IHC (Figure 1Ab and f). In order to increase circulating cholesterol levels, WT and knockout mice were fed a standard or a hypercholesterolemic diet, as previously described [11], [12]. This cholesterol surge had no effect on the gross histology of WT dorsolateral prostates (Figure 1Ac). In contrast, analysis of LXR mutant prostates revealed a disorganization of the epithelial layer, which was reminiscent of PIN grade II [13] (Figure 1Ag), characterized by the formation of cribriform and tufting patterns. Nuclei were enlarged and displayed prominent nucleoli (Figure 1Ai). The PIN status of the lesions was confirmed by an increased proliferation as demonstrated by Ki67 staining (Figure 1Ah, 1B) and Cyclin D1 and D2 overexpression (Figure 1C). The PIN phenotype was restricted to the dorsolateral prostate (Figure S1A, S1B) and was dependent on the ablation of both Lxrα and Lxrβ. Indeed, single knockout prostates were comparable with WT glands in terms of histology and proliferation (Figure S1C, S1D). The identity of proliferative cells was determined by immunofluorescence analyses using markers for prostatic cells subtypes. To identify proliferative cells within the different prostatic compartments, we performed double staining for PCNA and CK18 (luminal cells), p63 (basal cells) or SMA (stromal smooth muscle cells). Most PCNA+ cells were positive for CK18 (Figure 2Aa, b, and c) and were surrounded by p63+ epithelial basal cells (Figure 2Ad, e and f). Occasionally, p63+; PCNA+ cells were observed (data not shown), indicating that all the epithelial lineage could be targeted by proliferation in LXR null mice fed a high cholesterol diet. PCNA+ cells were exclusively localized inside the epithelium delineated by smooth muscle actin (SMA) staining (Figure 2Ag, h and i). PCNA+ or Ki67+ cells were not observed in the stroma (data not shown). Altogether, these results indicated that proliferation was restricted to the epithelial compartment. This was consistent with previous observations in the ventral prostate lobes of LXR mutant mice [4]. Presence of abnormal proliferation in the epithelium suggested that cell renewal could be deregulated. TUNEL staining showed increased apoptosis in the epithelium (Figure S2A, S2B) and identified delaminating apoptotic cells inside the lumen (Figure 2B). BrdU+ cells were also present inside prostatic ducts, suggesting that proliferative cells could detach into the lumen (Figure 2B). The increase of apoptosis could be the result from cholesterol cytotoxicity as shown in cholesterol-overloaded foam cells in atherosclerosis [14]. However, a similar cell death surge has been reported in a PTEN-deficient mouse prostates [15], [16]. In prostate of Lxr-/- mice under high cholesterol condition, it could therefore be a consequence of pathological development. Altogether, these observations suggested that the epithelium of LXR null mice presented both increased proliferation and apoptosis that resulted in an alteration of cell turnover. LXR are essential regulators of lipid metabolism. However, there was no major difference in circulating cholesterol levels in LXR knockout mice when compared with WT, irrespective of the diet (Figure 3A). Therefore, we speculated that the PIN phenotype resulted from deregulated lipid metabolism within the prostate. Indeed LXR knockout prostates accumulated large amounts of Oil-Red-O staining under high cholesterol condition, consistent with neutral lipid accumulation (Figure 3B). Quantitative analyses revealed a significant accumulation of cholesterol esters in LXR mutant mice fed a standard diet, which was largely amplified when mice were fed the hypercholesterolemic diet (Figure 3C). This phenotype was also associated with an increase in free cholesterol. Intra-prostatic triglycerides concentration was not altered and expression of genes involved in lipogenesis was even inhibited in LXR knockout prostates compared with WT (Figure 3C, 3D). This suggested that the accumulation of neutral lipids in the prostate of LXR knockout mice resulted from a deregulation of cholesterol transport in prostatic cells. Indeed, expression of Abca1, the transporter in charge of cholesterol efflux, was decreased both at the mRNA and protein levels in LXR knockout prostates (Figure 3E, 3F). Conversely, LDLR protein accumulation was increased by LXR ablation (Figure 3F, white arrow), even though Ldlr mRNA accumulation was decreased (Figure 3E). This was correlated with a decreased expression of the LXR target gene Idol (Figure 3E), which catalyzes the ubiquitination and subsequent degradation of LDLR [17]. Therefore, aberrant cholesterol ester accumulation in LXR deficient prostatic cells results from both increased uptake and decreased efflux. Our data showed that control of cholesterol homeostasis by LXR is crucial to restrain epithelial cell proliferation in the prostate. In order to determine key molecular events resulting from elevation of cholesterol in the prostate, we designed microarray experiments. We compared prostatic gene expression of WT and LXR mutant mice in normal and high dietary cholesterol conditions (Figure 4A). The list of up- and down-regulated genes has been established on the basis of signal intensity, Log ratio and p-value (Figure S3). The highest number of deregulated genes was observed when WT and LXR knockout mice were exposed to high circulating cholesterol levels, again emphasizing the central role of cholesterol in the establishment of the phenotype (Figure 4A). In order to determine gene expression signature of the PIN phenotype in LXR mutant mice fed a high cholesterol diet and to identify relevant molecular events, we have restricted the gene list using Venn analysis. We selected common deregulated genes associated with the PIN phenotype and eliminated those that were sensitive to diet and/or LXR ablation alone. Therefore, we focused on the genes involved in the establishment of the PIN phenotype by selecting genes that were deregulated in both arrays 3 (lxr-/- normal vs. lxr-/- high chol.) and 4 (+/+ high chol. vs. lxr-/- high chol.) and by subtracting genes that were deregulated in both arrays 1 (+/+ normal vs. +/+ high chol.) and 2 (lxr-/- normal vs. +/+ normal). This resulted in a list of 463 genes (Dataset S1), 253 up and 210 down (Figure 4B). Ingenuity Pathway Analysis (IPA) was used to investigate potential biological processes that underlay the PIN phenotype of LXR mutant mice (Figure S4). The second most significantly enriched gene-category was ‘cancer’, which was associated with a large list of 146 genes (Dataset S2). More than 50% of these 146 genes were also deregulated in a mouse model of prostate cancer resulting from PTEN deletion in prostatic epithelium [18] (data not shown). This strongly suggested that the PIN lesions observed in LXR knockout mice in the high cholesterol condition were genuine pre-cancerous alterations. Interestingly, this analysis showed down-regulation of two well described prostatic tumor suppressor genes Nkx3. 1 and Msmb (Dataset S2, highlighted in red), which was further confirmed by qPCR analysis (Figure 5A, Figure S5). These two genes were specifically found in gene categories such as tumor development, cell proliferation and prostate organogenesis (Dataset S3, highlighted in red). Nkx3. 1 and Msmb promoters have recently been demonstrated to be targets of the histone methyl transferase EZH2 that represses gene expression through H3K27 trimethylation. qPCR and western blot analyses showed that Ezh2 was specifically overexpressed in LXR knockout prostates when animals were fed a high cholesterol diet (Figure 5A, 5B). Immunohistochemistry further confirmed overaccumulation of EZH2 in proliferative PCNA+ cells in LXR knockout prostates, when animals fed a high cholesterol condition (Figure 5C). This suggested that the effect of cholesterol on the development of PIN was dependent on down-regulation of Nkx3. 1 and Msmb, resulting from EZH2-mediated modification of their promoter chromatin. Indeed, ChIP analyses confirmed that nucleosomes at both Nkx3. 1 and Msmb promoters were significantly trimethylated on H3K27 in the prostates of LXR null-mice fed a high cholesterol diet (Figure 6A, 6B). Interestingly, Msmb expression was increased by a high cholesterol diet in WT mice. This was independent of Ezh2, whose expression was unaltered (Figure 5A). Such observation indicates that other mechanisms are involved in the regulation of this tumor suppressor gene expression and that it is highly sensitive to metabolic changes in prostate tissue. To further confirm the potential link between LXR and EZH2 expression, we performed a retrospective study of publicly available DNA microarray data of human PCa cohorts, using Oncomine. These analyses showed that LXRβ expression was significantly down-regulated in prostate carcinomas compared to normal tissue and that this down-regulation was associated with increased EZH2 expression (Figure 6C). Interestingly, careful analysis of normal prostate gland as well as metastasis heat maps revealed that levels of LXRβ, EZH2 and MSMB were tightly coordinated between each other (Figure S8). The expression pattern of NKX3. 1 present no significant modification. Therefore, the connection between LXR, cholesterol homeostasis, EZH2 and MSMB expression that we uncovered in mouse could also be relevant in human PCa. Previous analyses of LXR null mice have shown the development of a BPH-like phenotype in the ventral lobe of the prostate [4], [5]. However in patients, BPH arises in the periurethral and transition zones distinct from the peripheral zone from which cancer emerges. Therefore, to date, the role of LXR in PCa had been postulated on the basis of studies performed in tumor cell lines [6]–[9]. Here we show for the first time that LXR ablation results in the development of PIN in the dorsal prostate in mouse, which is the most similar lobe to human peripheral prostate, the area from which the majority of cancerous lesions occurs in human [18]. Consistent with previously published data [4], this phenotype is not observed under normal dietary conditions. Indeed, in our model, PIN development is associated with a high cholesterol diet, which results in prominent intra-prostatic accumulation of cholesterol ester. Cholesterol has been extensively associated with prostate malignancy [19]. We therefore hypothesize that increased cholesterol ester storage is a major contributor to the appearance of the PIN phenotype. Interestingly, abnormal cholesterol storage was also observed in LXR mutant mice fed a standard diet, albeit to a lesser extent. Absence of PIN development under this condition, even in 18 month-old animals (data not shown) suggests that cholesterol accumulation needs to be tipped over a threshold to become deleterious. It is therefore tempting to speculate that in patients, the combination of metabolic disease and/or high cholesterol diet with abnormal LXR activity may favor prostate cancer development, by increasing cholesterol accumulation beyond this threshold. Consistent with this idea, we show decreased expression of LXRβ in prostatic carcinomas compared with normal prostate (Figure 6C) [20], [21]. Numerous in vivo and ex vivo studies have shown the sensitivity of already transformed tumor cells to variations in cholesterol supply and de novo synthesis [8], [11], [19], [22]. Our data goes one step further by showing that LXR ablation and the subsequent accumulation of cholesterol may in fact initiate neoplastic development in the prostate. The molecular mechanism by which LXR control cell cycle in human prostatic tumor cell lines is still poorly understood. LXR activation has been shown to slow down the cell cycle through accumulation of the p27 cell cycle inhibitor and downregulation of SKP2 in LNCaP cells [6]. RNA interference demonstrated that part of this antiproliferative effect was supported by LXR themselves [23]. Interestingly, aberrant proliferation observed in LXR null mice fed a high cholesterol diet was found in only 24% of the acini (Figure 1B). These findings indicate that the cellular context of one particular epithelial cell plays an essential role in cell cycle deregulation and in the development of PIN lesions. It is therefore very likely that the prostatic phenotype of LXR-null mice is not only dependent on an epithelial cell-autonomous effect of LXR ablation. This hypothesis is supported by our previous observation that LXR were required to establish a cellular dialogue between stromal and epithelial compartments in ventral prostate [5]. One interesting observation of our study is the correlation between increased cholesterol accumulation and increased expression of Ezh2. Overexpression of EZH2 is associated with aggressive prostate carcinomas in patients [24] and has been shown to control prostate cell proliferation through epigenetic silencing of the tumor suppressors NKX3. 1 and MSMB [25], [26]. Here, we show that the combination of LXR ablation and high cholesterol diet is associated with decreased Nkx3. 1 and Msmb expression, which is correlated with an increase in the H3K27me3 mark on their promoter regions. It is therefore tempting to speculate that some of the oncogenic effects of cholesterol accumulation in the context of LXR ablation are mediated by up-regulation of EZH2 and the conscutive gene silencing. How this is achieved is still unclear. However two scenarios could account for such a mechanism. In the first scenario, deregulation of Ezh2 expression could be triggered in an epithelial cell-autonomous fashion as lipids (PUFA) have been already identified in such a process [27]. However, the underlying molecular mechanisms remain unknown as the promoter sequences of Ezh2 are still poorly characterized [25]. In the second scenario, Ezh2 overexpression could result from an accumulation of a specific epithelial cell compartment. EZH2 is not a canonical stem/progenitor marker in the prostate but has been involved in cancer stem cell maintenance in various diseases [28], [29]. In human prostate, a minor subgroup of “stem” cells (CD44+, Oct4+) expresses EZH2 and has been proposed to represent a cell reservoir for prostatic adenocarcinoma initiation [30]. Consequently, increased expression of Ezh2 in LXR null mice could result from expansion of a progenitor epithelial cell population. The effect of LXR ablation and cholesterol accumulation on epigenetic processes is likely to extend beyond EZH2. Indeed, we show increased expression of Uhrf1 in correlation with Ezh2 accumulation in LXR mutant mouse prostates, under high cholesterol condition (Figure S6). This is consistent with reports of a positive correlation between these two factors in human prostate tumors. UHRF1 acts with Suv39H1 and DNA methyltransferases to alter histone H3K9 methylation, acetylation and DNA methylation to epigenetically repress target genes. Furthermore, UHRF1 and EZH2 have been proposed to synergistically promote inactivation of oncosuppressor genes, among which Nkx3. 1 and Msmb [31], in tumor cells. Consistent with the idea that Ezh2 deregulation results from interactions between different cell compartments of the prostate and thus from expansion of Ezh2-positive cells, LXR activation or knockdown did not change EZH2 accumulation in prostatic culture cell lines (data not shown). Another intriguing observation regards the upregulation of Msmb in WT mouse prostate under high cholesterol condition (Figure 5A). Transcriptional regulation of Msmb is poorly characterized beyond the role of EZH2 and androgens [26], [32]. Since levels of androgen target genes, as Nkx3. 1 [33], [34], were unchanged (data not shown), we hypothesized that androgen amount was stable irrespective of the diet. Thus we concluded that upregulation of Msmb expression was not due to a higher level of androgens. It was also unlikely be dependent on EZH2, whose expression was unaltered in response to cholesterol in WT mouse prostate (Figure 5). Taken together, these observations suggest that Msmb is sensitive to prostate metabolic status and that an unknown mechanism yet is involved. Given the role of Msmb repression as a maker of prostate cancer progression and a bona fide tumor suppressor gene [35]–[37], we speculate that Msmb overexpression in WT mice prostates represents a defensive molecular mechanism against the metabolic stress induced by a high cholesterol diet. Among canonical LXR functions, primum movens leading to PIN phenotype in prostate of Lxr-null mice could originate from deregulation of inflammatory response in prostate tissue as suggested by gene ontology (Dataset S3). Indeed, inflammation has been widely associated with prostate cancer development. Even though there was no clear CD45+ staining Lxr-/- in dorsal prostate in high cholesterol condition (Figure S7A), Cd45 expression measured by qPCR was 2-fold increased compared to WT (Figure S7B). Moreover, analysis by hierarchical clustering comparing array 1 and array 4 of inflammation-associated genes expressions (Figure S7C) showed that mouse prostate displayed a specific gene signature. While a high cholesterol diet in prostate of WT mice induces expression of inflammatory genes without leading to an in vivo phenotype, some of these genes failed to be upregulated in LXR mutant mice (Figure S7C, compared group 1 and 2). Conversely, genes that were insensitive to a high cholesterol diet in WT mice, showed a massive deregulation in LXR mutant mice in similar diet conditions (Figure S7C, group 3). Altogether, prostate of LXR mutant mice exhibits a specific gene expression signature that revealed a deregulation of the inflammatory network. This raises the question of LXR-dependent regulation of inflammation in prostate tissue and its impact on the PIN development. Human dataset analysis pointed out that LXRβ but not LXRα expression could be linked to EZH2 expression while both isoforms need to be invalidated to induce a PIN occurrence in mice (Figure S8). Absence of any change in LXRα expression could explain the lack of a clear deregulation of some LXR target genes in Oncomine datasets (data not shown). Moreover, both LXRα and LXRβ have been demonstrated to be expressed and functional in human PCa cells [8], [38]. These observations suggest that EZH2 deregulation could be linked to a mechanism specifically depending on LXRβ. Such specificity has already been shown in human, particularly in a study on preeclampsia providing a LXRβ-dependent risk in this pathology [39]. Another point emphasized by the human dataset is the absence of NKX3. 1 expression changes between normal prostate, carcinoma and metastasis group in both examined cohorts (Figure S8). NKX3. 1 expression profiles are somehow unexpected, as this gene has been largely reported as a tumor suppressor gene in the prostate. Nevertheless, various mechanisms have been demonstrated to repress NKX3. 1 during carcinogenesis and these observations suggest that filtrating analysis of human datasets based on association with identified oncogenic alterations, such as PTEN inactivation [40], should me more informative. Altogether, our results show that LXR act as “gate keeper” in mouse prostate to prevent cholesterol accumulation and subsequent PIN development. Our findings further suggest that the metabolic status of the prostate can govern epigenetic processes involved in prostate cancer progression. Lxrα and lxrβ double knockout mice and their wild-type controls [41], [42], [43] were maintained on a mixed strain background (C57BL/6: 129Sv) and housed according to local ethical regulations. Mice were fed ad libitum a normal mouse chow (Global-diet 2016S) until 5 months of age. Mice were then fed either a normal or hypercholesterolemic diet (Teklad diet number 88051; Harlan, Gannat, France) for 5 weeks. Animals were sacrificed, blood plasma was collected and prostates were dissected. For histological analysis, prostates were either embedded in NEG 50 (Thermo Scientific, Kalamagoo, MI, USA) or fixed in an alcohol/formaldehyde 37% and acetic acid mixture (7. 5∶2∶0. 5; v/v) before embedding in paraffin for histological analysis. For lipid, protein and RNA extractions, prostates were snap-frozen in liquid nitrogen. All animals were maintained in a controlled environment and animal care was conducted in compliance with the national standards and policies (C 63 014. 19). The Regional Ethics Committee approved all experiments (CE 74-12 S) (Text S1). Prostate tissues were fixed overnight in 4% paraformaldehyde, paraffin-embedded, sectioned and stained with hematoxylin and eosin according to a standard protocol. For immunochemistry, paraffin sections were dewaxed, rehydrated, unmasked using 0. 1M citrate buffer (pH 6. 0) and then incubated with primary antibodies overnight at 4°C in a humidified chamber. Primary antibodies were: PCNA (FL-261) sc-7907 (Santa Cruz Biotechnology, Santa Cruz, CA), EZH2 (AC22) #3147 (Cell signaling, Montigny-Le-Bretonneux, France), BrdU (Roche diagnostic, Meylan, France), p63: 69241A (BD Pharmigen, San Diego, CA, USA), Cytokeratin 18 (H-80) sc-28264 (Santa Cruz Biotechnology, Santa Cruz, CA), Actin A2066 (Sigma-Aldrich). Detections were performed alternatively using the NovaRED substrate kit for peroxidase (Vector Laboratories, Burlingame, CA) or Alexa 488 conjugated anti-mouse IgG/Alexa 555 conjugated anti-rabbit IgG (Invitrogen). Cell nuclei were stained using Hoechst 33342 (Sigma-Aldrich) at 1 mg/ml. Apoptotic nuclei were visualized through a TUNEL reaction relying on terminal deoxynucleotidyl transferase (TdT; Euromedex, Souffelwegersheim, France) and biotin-11-dUTP (Euromedex), dATP (Promega, Charbonnière, France). Positive nuclei were revealed by addition of extravidin-coupled alkaline phosphatase and FastRed TR/Naphthol AS-MX substrate (Sigma-Aldrich). Nuclei were counterstained with Mayer hematoxylin solution. Cross-sectional areas of the prostate were photographed using a Zeiss Axioplan fluorescence microscope and the Axiovision 4. 2 software (Carl Zeiss Vision GmbH, Le Pecq, France). Lipid stainings were performed on cryosections with Oil-Red-O (Sigma-Aldrich) as previously described [44]. Microarray study is detailed in Text S1. Briefly, mRNA samples were analyzed using Agilent 44K Whole Mouse Genome microarrays (Agilent Technologies, Palo Alto, CA). For each microarray, log ratio, fold-change and p-value were determined using the Rosetta Resolver Gene Expression Analysis System and these criteria were used for Venn analysis by threshold method. Microarrays results were deposited in the EBI MIAME-compliant database (E-MTAB-546). Total RNAs were isolated using NucleoSpin RNA II column kit (Macherey-Nagel, Hoerd, France). cDNAs were synthesized with Moloney Murine Leukemia Virus Reverse Transcriptase (Promega) and random hexamer primers (Promega) according to the manufacturer' s instructions. cDNA templates were amplified by MESA GREEN MasterMix Plus for SYBR Assay (Eurogentec, Seraing, Belgium) using an iCycler (Bio-Rad, Marnes-la-Coquette, France). Primer sequences are listed in Text S1. qPCR results were normalized alternatively using 36b4 or 18S as a standard. Blood concentrations of circulating cholesterol were determined on an automated clinical chemistry analyzer (Roche Diagnostics) according to manufacturer' s instructions. Lipid samples from prostate tissues were extracted by the Folch method as previously described [8] and analyzed on high-performance thin layer chromatography (TLC) plates. Proteins were extracted in Hepes 20 mM, NaCl 0. 42 M, MgCl2 1. 5 mM, EDTA 0. 2 mM and NP40 1% supplemented with PMSF 1 mM (Sigma-Aldrich), Complete 1X (Roche Molecular Biochemicals, Meylan, France), NaF 0. 1 mM and Na2VO3 0. 1 mM (Sigma-Aldrich). For western blot, 40 µg of protein lysates were separated by SDS PAGE and were incubated with antibodies against Actin A2066 (Sigma-Aldrich), ABCA1 NB400-105 (Novus, Littletown, CO), EZH2 (AC22) #3147 (Cell Signaling) and LDLR 10007665 (Cayman Chemical). Chromatin preparation from dorsolateral prostate and for immunoprecipitation has been described previously (3). Immunoprecipitation was performed using Anti-trimethyl Histone H3 (Lys27) #ABE44 (Millipore, Billerica, MA) and negative control IgG #Kch-504-250 (Diagenode, Liège, Belgium). Primers used for qPCR analysis are listed in Text S1. qPCR data, lipids assays and Ki67-staining parameters are expressed as mean ± standard deviation. Statistical analysis was performed with a two-tailed Student' s t test.

Cholesterol is one of the major metabolic molecules required for a broad range of cellular processes. Recent advances in prostate cancer research have demonstrated that tumor cells need to increase their supply of cholesterol to sustain membrane building, proliferation, and survival capacities. Liver X receptors, which belong to the nuclear receptor superfamily, are central mediators of cholesterol homeostasis. Indeed, they regulate the expression of many genes involved in cholesterol uptake storage and efflux. Here, we show that genetic ablation of LXRs in mice results in the formation of precancerous lesions in the prostate, called prostatic intra-epithelial neoplasia. These are only observed when mice are fed a high-cholesterol diet. Hence, LXRs regulate cholesterol homeostasis in the prostate and protect cells from abnormal proliferation when exposed to high dietary cholesterol.

Abstract Introduction Results Discussion Methods

animal models model organisms reproductive system biology anatomy and physiology mouse

2013

Liver X Receptors Protect from Development of Prostatic Intra-Epithelial Neoplasia in Mice

Attention is a core cognitive mechanism that allows the brain to allocate limited resources depending on current task demands. A number of frontal and posterior parietal cortical areas, referred to collectively as the fronto-parietal attentional control network, are engaged during attentional allocation in both humans and non-human primates. Numerous studies have examined this network in the human brain using various neuroimaging and scalp electrophysiological techniques. However, little is known about how these frontal and parietal areas interact dynamically to produce behavior on a fine temporal (sub-second) and spatial (sub-centimeter) scale. We addressed how human fronto-parietal regions control visuospatial attention on a fine spatiotemporal scale by recording electrocorticography (ECoG) signals measured directly from subdural electrode arrays that were implanted in patients undergoing intracranial monitoring for localization of epileptic foci. Subjects (n = 8) performed a spatial-cuing task, in which they allocated visuospatial attention to either the right or left visual field and detected the appearance of a target. We found increases in high gamma (HG) power (70–250 Hz) time-locked to trial onset that remained elevated throughout the attentional allocation period over frontal, parietal, and visual areas. These HG power increases were modulated by the phase of the ongoing delta/theta (2–5 Hz) oscillation during attentional allocation. Critically, we found that the strength of this delta/theta phase-HG amplitude coupling predicted reaction times to detected targets on a trial-by-trial basis. These results highlight the role of delta/theta phase-HG amplitude coupling as a mechanism for sub-second facilitation and coordination within human fronto-parietal cortex that is guided by momentary attentional demands. Attention, a critical component of perception and goal-directed behavior, allows the brain to allocate its limited resources depending on current task demands. A number of areas located in frontal and posterior parietal cortex (PPC), often referred to collectively as the fronto-parietal attention network, are crucial for controlling the attentional selection process in both humans and non-human primates [1]–[5]. These areas include the intraparietal sulcus (IPS) and superior parietal lobule (SPL) in PPC as well as portions of superior-lateral precentral cortex, also known as the frontal eye fields (FEF), and dorsal medial frontal cortex, also known as the supplementary eye field (SEF) [1], [4]. In the human brain, the temporal-parietal junction (TPJ) and portions of the inferior, middle, and superior frontal gyri are also involved during attentional allocation [1]. Attentional control is often investigated in these regions using functional magnetic resonance imaging (fMRI) methods (i. e. , investigating BOLD time courses in each region) or electrophysiological methods (single-unit, multi-unit, or local field potential [LFP] recordings in macaques or with electroencephalography [EEG] in humans). fMRI has excellent spatial resolution and can identify functionally active networks, but is constrained by the low temporal resolution of the blood-oxygen-level dependent (BOLD) signal. Electrophysiological recordings in monkeys provide excellent temporal resolution, but cannot easily be used to simultaneously investigate an entire network of areas. Scalp EEG studies (see [6], [7] for reviews) have provided important insights into the understanding of attention, but EEG has spatial limitations due to the inverse problem. A handful of studies have attempted to solve the inverse problem using EEG or magnetoencephalography (MEG) combined with source localization techniques to investigate the spatial attention network in greater detail [8], [9], but relatively little knowledge exists about how fronto-parietal areas support attentional behavior on a fine spatiotemporal scale in the human brain. Populations of neurons oscillate together and synchronize their firing and post-synaptic potentials in a rhythmic fashion [10]. These oscillations emerge in multiple frequency bands: delta (1–4 Hz), theta (4–8 Hz), alpha (8–12 Hz), beta (12–30 Hz), and low gamma (30–60 Hz). It is also possible to capture neuronal activity by sampling power fluctuations in a broadband high-frequency range, known as high gamma (HG; 60–250 Hz), as studies have shown that HG activity correlates with local spiking activity [11], [12]. Interactions within or between these bands and/or broadband HG have been proposed to serve as mechanisms for coordination within and between brain networks engaged in cognitive processing [13]–[15]. Phase-amplitude cross-frequency coupling (PAC) is a mechanism that has been proposed to coordinate the timing of neuronal firing within local neural networks (see [16] for a review). PAC, the statistical dependence between the phase of a low-frequency rhythm and the amplitude (or power) of the high-frequency component of electrical brain activity, is proposed to operate through a physiologically plausible mechanism: low-frequency phase controls neuronal excitability through fluctuations in membrane potentials, while increases in high-frequency power reflect increases in local neuronal spiking activity [11], [12]. Previous studies have demonstrated that the probability of neuronal spiking changes with the phase of the lower frequency rhythm and can define time windows during which neurons are more or less likely to fire [16]–[18]. Thus, it has been hypothesized that these lower-frequency oscillations (i. e. , in the delta, theta, alpha, or beta bands) coordinate information among areas by modulating the excitability within local neuronal ensembles [14], [17], [18]. Canolty and colleagues [19] first demonstrated that the phase of the ongoing theta oscillation is coupled to increases in HG (>70 Hz) power across a range of motor and language tasks in humans. PAC has since been reported between numerous different frequency bands within human cortex and subcortical structures [20]–[23]. To our knowledge, no study to date has investigated if PAC may serve as a mechanism for visuospatial attention control within human fronto-parietal cortex. We measured the neural dynamics and frequency-band interactions within regions of the human fronto-parietal attention network during allocation of spatial attention using a dynamic spatial-cuing task [24]. We collected electrocorticography (ECoG) recordings, measured directly from subdural electrode arrays implanted in patients undergoing intracranial monitoring for localization of epileptic foci, in order to examine the role of PAC as a mechanism for coordination within the fronto-parietal network, which adjusts parameters on a millisecond basis depending on momentary attentional demands. To determine which electrodes were responsive to spatial attention, we first investigated changes in power across a wide range of frequencies in response to attentional allocation to either the contralateral or the ipsilateral visual field (i. e. , the visual field contralateral or ipsilateral to the implanted hemisphere). In a number of electrodes (263 out of 875 electrodes; 30. 06%) across eight subjects, we found significant increases in broadband HG power (70–250 Hz; all p<0. 05, corrected) that were time-locked to trial onset (following each cue). HG power remained elevated throughout the period when subjects were allocating their spatial attention in wait for the target appearance (0–1000 ms post-trial onset). These HG power increases were observed in electrodes covering PPC, lateral, and medial frontal cortex, occipital cortex, and posterior ventral temporal cortex (see Figure 2 for event-related spectral perturbations [ERSPs] and vertically stacked single-trial HG traces in two example electrodes across two separate subjects [S4, S7]). A subset of these electrodes (47 of 263; 17. 87%; in seven subjects) exhibited greater HG power increases when subjects allocated attention to the contralateral than to the ipsilateral visual field (all p<0. 05, corrected). In comparison, only three out of 263 electrodes (1. 14%; in one subject) exhibited greater HG power increases when subjects allocated attention to the ipsilateral than to the contralateral visual field (all p<0. 05, corrected). Increases in HG power were sometimes accompanied by power increases in lower frequencies (delta/theta: 2–5 Hz, all p<0. 05, corrected; Figure 2a) and sometimes not (Figure 2b). 59. 32% (156 of 263; in eight subjects) of the electrodes with significant HG power increases also had significant power increases in this lower frequency range. Of the 156 electrodes with significant delta/theta power increases, 17 (10. 90%) overlapped with the set of 47 electrodes that showed visual field-dependent attentional increases in HG power. A number of electrodes (119 out of 875 electrodes; 13. 60%; in eight subjects) also exhibited significant decreases in alpha and beta power (10–30 Hz; all p<0. 05, corrected) compared to baseline. This alpha and beta suppression was sustained for 500 ms (e. g. , Figure 2a) to 1000 ms or longer (e. g. , Figure 2b) following trial onset. All electrodes with sustained alpha/beta suppression were located either in parietal or visual cortex. A subset of these electrodes (21 of 119; 17. 65%; in seven subjects) exhibited larger decreases in alpha/beta power when subjects allocated attention to the contralateral than to the ipsilateral visual field (all p<0. 05, corrected). In comparison, 2 out of 119 electrodes (1. 68%; in one subject) exhibited larger decreases in alpha/beta power when subjects allocated attention to the ipsilateral than to the contralateral visual field (all p<0. 05, corrected). These results are consistent with the notion that decreases in alpha power, which are strongest in the hemisphere contralateral to a stimulus, represent a release from functional inhibition within the visual system [26]. Taken together, the results suggest that power increases in broadband HG and the delta/theta bands, as well as power decreases in the alpha/beta bands, track the allocation of visuospatial attention in human frontal, posterior parietal, and visual cortex. Because previous studies have demonstrated that frequency bands can interact in behaviorally meaningful ways [19], [27], we assessed whether any relationships existed among the different frequency bands (delta, theta, HG) during attentional allocation. We examined PAC between a wide range of frequencies for phase data (2–20 Hz) and for amplitude data (5–250 Hz) (Figure 3a, b) in all of the electrodes that had significant increases in HG power (70–250 Hz) following attentional allocation compared to baseline. In many of these electrodes (123 of 273; 45. 05%; in eight subjects), we found significant coupling between the phase of the delta/theta signal (2–5 Hz) and HG amplitude (80–250 Hz) during attention to the cued visual field while awaiting target appearance (0–1000 ms post-trial onset). The coupling values averaged across all attention trials, as measured by the phase-locking value (PLV), ranged between 0. 16 and 0. 57 for all electrodes showing this significant PAC. We observed delta/theta phase-HG amplitude coupling in electrodes over PPC, especially those surrounding the IPS, lateral and medial frontal cortex, occipital cortex, and posterior ventral temporal cortex (see Figure 3 for examples from three individual subjects (S3–S5) and Figure S1 for the group results plotted on the Montreal Neurological Institute [MNI] brain). Furthermore, this PAC remained significant after controlling for eye movements (S8 was eye-tracked during recording and all trials during which a saccade of 0. 50° or larger was made were excluded from further analyses; see Materials and Methods for details). Although a higher percentage of RH electrodes (51. 56%; 66 out of 128) exhibited significant delta/theta-HG PAC during attentional allocation than LH electrodes (39. 31%; 57 out of 145), this difference was not significant between hemispheres (U = 3. 00, p>0. 20). However, interhemispheric differences in electrode coverage and density preclude a strong conclusion regarding left vs. right differences. Many of the electrodes with significant delta/theta-HG PAC (55 of 123; 44. 71%; in seven subjects) had significantly higher coupling when subjects were attending to the contralateral, as compared to the ipsilateral, visual field (all p<0. 01, corrected). Figure 3a, b shows examples of four electrodes (two surrounding IPS and two over lateral frontal cortex) that have higher PAC in contralateral than in ipsilateral trials. Eighteen of the 55 electrodes with higher PAC in contralateral than in ipsilateral trials also had higher HG power in contralateral than in ipsilateral trials. Although the magnitude of the PAC decreased when subjects attended ipsilaterally, it did not disappear completely, which is likely due to the bilateral receptive field properties of the neuronal populations over which we recorded [28]. None of the electrodes exhibited higher PAC when subjects allocated attention to the ipsilateral than to the contralateral visual field (all p>0. 10, corrected). Because power increases, and thus the signal-to-noise ratio, can affect PLVs [29], we compared the magnitudes of delta/theta power between attend contralateral and attend ipsilateral trials during the same time period analyzed above. Out of the 55 electrodes in which higher PAC was present when subjects attended contralaterally than when they attended ipsilaterally, only two electrodes in one subject had significantly higher delta/theta power in attend contralateral than attend ipsilateral trials (p<0. 05, corrected). Thus, power differences between conditions cannot account for the PAC differences that we observed between these same conditions. To address the possibility that the observed PAC was being driven by an event-related potential (ERP) at trial onset, we repeated the analyses using only the longer trials (i. e. , target appearance 1500–2000 ms after trial onset). Only the last 1000 ms prior to the target onset were analyzed, so the first 500 ms, when the ERP was generated, was excluded from every trial. This analysis produced similar PAC results: 107 of 273 electrodes (39. 19%; in eight subjects) had significant coupling between delta/theta phase and HG amplitude (Figure S2). All 107 of these electrodes were in the original set of 123 electrodes showing significant delta/theta-HG PAC. This demonstrates that the observed PAC was sustained throughout the trial, rather than driven by a single cycle created by a transient ERP. Additionally, a number of these electrodes maintained significantly higher delta/theta-HG PAC (42 of 123; 34. 15%; in seven subjects) when subjects were attending to the contralateral, as compared to the ipsilateral, visual field (all p<0. 01, corrected) and none of these electrodes showed significant delta/theta power differences between attend contralateral and attend ipsilateral trials (all p>0. 10, corrected), further demonstrating that the observed laterality effects were not solely driven by transient ERPs. We next examined to which part of the delta/theta phase signal the HG amplitude was locking by calculating the trough-locked spectrogram for each electrode with significant delta/theta phase-HG amplitude coupling. In a large percentage of electrodes (117 of 123; 95. 12%; in eight subjects), the HG power was locked to the trough of the delta/theta signal (see Figure 3c for an example from one electrode in S4). Our results are in agreement with previous studies that found HG locks to the trough of the theta signal [19], [27]. Furthermore, this trough-locking of the HG power was significant for at least three phase cycles of the delta/theta signal (∼600–1500 ms; see Figure 3c for example of three cycles), providing additional evidence that the observed PAC was not solely driven by ERPs. Critically, changes in delta/theta phase-HG amplitude coupling were correlated with attentional behavior. Twenty-eight of the 123 electrodes with significant delta/theta-HG PAC also demonstrated significant correlations of PAC with behavior across trials, in which faster RTs were associated with stronger delta/theta-HG coupling, and slower RTs with weaker delta/theta-HG coupling, during the attentional delay period (examples from two subjects, S3 and S5, are shown in Figure 4). The electrodes with significant PAC-behavioral correlations were dispersed throughout frontal (42. 86%), parietal (32. 14%), and visual cortex (25. 00%) across seven subjects. Only the PAC values from trials when subjects attended to the visual field contralateral to electrode implantation correlated with behavior (all p<0. 05, corrected; Figure 4, S3: r (41) = −0. 51, p<0. 01, corrected; S5: r (47) = −0. 37, p<0. 05, corrected), whereas the PAC values from trials when subjects attended to the ipsilateral visual field did not reach significance in any of the electrodes (all p>0. 10, corrected; Figure 4, S3: r (40) = −0. 11, p>0. 6, corrected; S5: r (47) = −0. 10, p>0. 7, corrected). RTs also did not significantly correlate with mean amplitude of the delta/theta signal (all p>0. 05, corrected) or with mean amplitude of the HG signal (all p>0. 10, corrected) across the same time period for either attentional condition in any of these electrodes, suggesting that amplitude changes alone are not driving the results. These findings provide evidence that momentary changes in the strength of delta/theta-HG PAC predict attentional behavior in humans. We next examined the relationship between visually evoked ERPs and delta/theta-HG PAC in electrodes across frontal, parietal, and visual cortex, as previous studies have suggested a complex relationship between ERPs and the broadband HG response over visual cortex [30]. We found ERP responses time-locked to the beginning of each trial in 77 of 875 electrodes (8. 80%; in eight subjects) and time-locked to the target onset of each trial in 29 electrodes (3. 31%; in seven subjects). Each of these ERP responses consisted of the classic visual P1-N1 complex and subsequent P3b [31]–[34]. The P1, N1, and P3b components peaked around 100 ms, 200 ms, and 300 ms post-trial onset, respectively, which is consistent with previous reports of visual responses from ECoG recordings [35], [36]. Figure 5 provides an example of the visual ERPs averaged across trials (attended/cued vs. unattended/uncued; Figure 5a, top) and single-trial ERPs (attended trials only; Figure 5a, bottom) time-locked to trial onset (Figure 5a, left) and target onset (Figure 5a, right) from one electrode in a single subject (S8). The majority of electrodes with visually evoked ERP responses were located over visual cortex and PPC (Figure 6 depicts the electrodes with significant delta/theta-HG PAC [circled in red], ERP responses [circled in black], or both [circled in yellow] for each subject). Thirty-four electrodes across the group of eight patients had both ERP responses at trial onset and significant delta/theta-HG PAC (Figure 6, electrodes circled in yellow). Thus, 44. 16% (34/77) of electrodes with an ERP response also had significant increases in delta/theta-HG PAC, while 27. 64% (34/123) of electrodes with significant increases in delta/theta-HG PAC also had an ERP response. Electrodes with a visually evoked ERP time-locked to trial onset were more likely to also have significant PAC throughout the attentional allocation period, while electrodes with significant PAC throughout the attentional allocation period were not as likely to be accompanied by ERP responses at trial onset. The small percentage of parietal and extrastriate electrodes with both an ERP and significant delta/theta-HG PAC throughout the attentional allocation period provides further evidence that the PAC results were not solely driven by trial onset ERPs. Next, we averaged the signals from all electrodes that exhibited individual ERPs at trial onset (n = 77) and target onset (n = 29). There were no significant differences between the attended and unattended conditions in the P1, N1, or P3b components of the ERP time-locked to trial onset across the group (all p>0. 10, corrected; Figure 5b, top left). The P1 and N1 components of the ERP time-locked to target onset were not significantly different between the attended and unattended conditions across the group, although attended trials evoked a marginally significantly higher P1 than unattended trials (all p>0. 05, corrected; Figure 5b, top right). However, the P3b amplitude was significantly higher in the attended condition than the unattended condition 280–580 ms following target appearance (all p<0. 05, corrected; Figure 5b, top right, blue shaded area), which is consistent with previous studies [31], [34]. The individual electrode example reflects the group results (Figure 5a, top right). These results suggest that the ERPs generated at trial onset were due to the onset of the display, and did not differentiate between stimulus conditions, while the target responses were modulated by spatial attention. We next examined the ERPs from the same conditions (attended/cued vs. unattended/uncued, time-locked to trial and target onsets) averaged across all of the electrodes with significant delta/theta-HG PAC (n = 123; Figure 5b, bottom). These ERPs were diminished (in the case of the trial onset ERP) or nearly absent (in the case of the target onset ERP). The lack of an ERP at target onset in these electrodes suggests that the significant increases in delta/theta-HG PAC are not always accompanied by the presence of a visually evoked ERP, indicating that the target onset ERP does not contribute to the observed PAC increases in the pre-target interval. We also examined separate ERP responses for trials when attention was directed to either the contralateral or ipsilateral visual field averaged across all electrodes that exhibited individual trial-locked (n = 77) or stimulus-locked (n = 29) ERPs. Again, there were no significant differences between the contralateral and ipsilateral attentional conditions for any of the ERP components time-locked to trial onset (all p>0. 05, corrected; Figure 7, left), while N1 and P3b amplitudes were enhanced when the attended target appeared in the contralateral visual field in comparison to when it appeared in the ipsilateral visual field (all p<0. 05, corrected; Figure 7, right). These results are consistent with previous studies showing enhanced N1 and P3 amplitudes following allocation of spatial attention to the contralateral visual field [33], [34] and provide electrophysiological evidence that subjects were allocating spatial attention during the task. Taken together, these results suggest that the relationship between the visually evoked ERPs that are generated over PPC and visual cortex and the observed increases in PAC over these same areas is complex. Many electrodes had significant PAC without an ERP and vice versa, providing evidence that PAC and ERPs result from distinct cortical mechanisms, with ERPs reflecting post-synaptic potentials and HG power reflecting neuronal firing rates and the synaptic currents induced by this firing [12], [37]. We used ECoG recordings in humans to provide insight into the neural mechanisms supporting visual attention. We found increases in broadband HG power (70–250 Hz) time-locked to trial onset that remained elevated throughout the attentional allocation period over frontal, parietal, and visual areas. Attention modulated broadband HG power in two distinct patterns. First, ∼20% of these electrodes exhibited visual field-dependent attentional increases in HG (stronger in the contralateral than ipsilateral visual field), which is consistent with visual field-dependent attentional increases in single-unit activity observed in primates [2], [3]. Second, the remaining electrodes exhibited task-dependent HG increases that were independent of attended visual field, which is consistent with the broadly tuned, bilateral receptive field properties of the neurons over which we recorded [28]. The phase of the ongoing delta/theta (2–5 Hz) oscillation modulated these HG power changes during attentional allocation and, importantly, the strength of the delta/theta phase-HG amplitude coupling predicted RTs to detected targets on a trial-by-trial basis. The current study provides evidence that coupling between HG amplitude and the phase of the delta/theta signal serves as a mechanism to facilitate processing within frontal, parietal, and visual areas during allocation of visuospatial attention. Based upon evidence from animal neurophysiology, we hypothesize that the PAC identified in the current study may serve to coordinate spiking activity in local regions of the network [38]. Phase coherence, in comparison, has been hypothesized as a potential mechanism through which distant brain areas are engaged across task-relevant networks. This synchronization of neuronal oscillations between areas may provide a mechanism for attentional selection of relevant sensory stimuli [14]. Numerous studies in humans and macaque monkeys have reported coherence increases across a wide range of band-limited frequencies, including delta, theta, alpha, beta, and low gamma, within and between frontal, parietal, and visual cortex during visual attentional allocation [6], [7], [9], [23], [39]–[46]. Thus, we hypothesize delta/theta-HG PAC increases that we observe in local regions of the fronto-parietal attention network may be modulated by neural synchrony across the network using one or more of these frequency bands [47], perhaps through a series of nested oscillations [48]. Our results are compatible with those reported by Lakatos and colleagues [43], who investigated PAC in V1 while monkeys were trained to attend to either visual or auditory stimuli presented in a rhythmic stream. They found that neural activity entrained to the stimulus rhythm (1. 55 Hz) and that the amplitude of the multi-unit responses, and LFPs were systemically related to the phase of this delta oscillation during attention to both modalities. A study using a similar rhythmic attentional paradigm found coupling between delta phase (1–3 Hz) and alpha amplitude in human auditory cortex using ECoG, but amplitudes at higher frequencies were not significantly coupled to delta phase [23]. Several recent studies have also reported increases in coupling between gamma power and alpha phase in visual areas V4 and TEO while monkeys performed a flanker task [46], and between broadband HG power and beta phase over occipital cortex while humans performed a visual search task with free viewing [30]. A number of factors may contribute to differences in the observed results among the aforementioned studies. First, it is not clear whether attention operates using a mechanism at similar frequencies across modalities. For instance, oscillatory mechanisms underlying auditory attention (i. e. , [23]) might be different from those underlying visual attention. Second, it is not clear whether PAC would operate at similar frequencies over fronto-parietal and visual cortex in the monkey and the human during attentional allocation. Although gamma oscillations are associated with attentive aspects of visual processing in monkey [39], [40] and human [49]–[51] cortex, considerably less evidence exists to support functional similarities among lower frequency oscillations between species. Given the size difference between the monkey and the human brain, interactions within and between lower frequency bands may differ between species. Third, each of these studies investigated attention using different types of tasks (detection vs. flanker vs. spatial cuing) and different trial structures. Many previous studies were not designed to investigate lower-frequency rhythms (i. e. , 2–5 Hz), since the cue-target time (the attentional allocation period) was not sufficiently long enough to include 2–3 cycles. We purposely designed our experiment with a long interval between cue and target, so that we could investigate effects at lower frequencies. Furthermore, some experiments have investigated synchronization and PAC using entrainment to rhythmically presented stimuli [23], [43], [45] or using a fixed interval [30], while others, including the current study, have used temporally jittered target presentations. Thus, many of the previous studies have focused on how networks can be entrained by the cadence of a task, whereas the slower rhythms that we report here appear to be endogenous and are most likely determined by the local neurophysiology. It was previously suggested that lower-frequency oscillations might be suppressed when attention is utilized to detect unpredictable target appearances, since this requires a mode of operation with continuously high excitability and vigilance [52]. The results from the current study suggest that this is not the case; lower-frequency oscillations are utilized for sensory processing and attentional control, even in the absence of any obvious task-relevant temporal structure. Finally, the recording locations are different between studies. Most of these studies have investigated neuronal oscillations over sensory cortex during attentional allocation. The current study investigates how frontal and parietal association cortices utilize PAC during attentional allocation in the human brain. In summary, it is evident based upon these studies, that no single frequency band is solely responsible for attentional processes, in accord with the notion that attention is not a unitary function [53]. ERP responses at trial or target onset cannot explain these PAC results. First, there were few electrodes that had both visually evoked ERPs time-locked to trial onset and increases in delta/theta-HG PAC during the attentional allocation period (Figure 6). Many of the electrodes that were responsive to the Starry Night Test showed independent ERP responses or PAC increases. This suggests that ERPs and PAC convey different types of informational content and are dependent upon distinct underlying cortical dynamics. These results support previous studies that have reported a separation between ERPs, broadband HG, and PAC across multiple cortical regions [30], [54], [55]. Secondly, we did not observe any systematic signal changes immediately preceding target appearance in the single trial ERPs (Figure 5a) or in the average ERPs, in any electrode. Thus, it is unlikely that ERPs at target onset were responsible for the observed PAC increases in the pre-target interval. Lastly, delta/theta-HG PAC remained significantly elevated in nearly all electrodes after eliminating the initial 500 ms (when the ERP was generated) from the longest (>1500 ms) trials (Figure S2). Together, these results suggest that the observed increases in delta/theta-HG PAC are due to the allocation of visuospatial attention, rather than an artifact of analysis procedures. A recent study in macaque monkeys reported that microsaccades occur at a frequency of ∼3. 3 Hz and are followed by characteristic increases and decreases in gamma band synchronization [56]. We cannot completely rule out the possibility that the increases in delta/theta-HG PAC were a result of microsaccadic activity, since our eye-tracker did not have sufficient sensitivity to measure microsaccades and hospital protocols did not allow patient head stabilization in the epilepsy ICU environment. We think this case is unlikely, since microsaccades induce bilateral cortical effects [57], while many of the electrodes exhibited strongly lateralized increases in delta/theta-HG PAC. Furthermore, microsaccades were shown to modulate spike-field coherence at lower gamma frequencies, 40–60 Hz [56], while our PAC results were strongest above 70 Hz. Microsaccades are present during most tasks that require sustained fixation [58] and several studies have indicated a close relationship between microsaccades and shifts in visuospatial attention [59], [60]. Several recent studies have provided behavioral evidence that attentional switching occurs between visual field locations or objects at ∼4 Hz rhythm [61], [62]. We propose that delta/theta-HG PAC may reflect an underlying neural mechanism that regulates the timing of such rhythmic sampling across the visual field or across objects. The current study demonstrates a functional relationship between PAC and behavioral performance in the SNT: the strength of delta/theta phase-HG amplitude coupling predicted RTs on a trial-by-trial basis, wherein high PAC predicted faster RTs and low PAC predicted slower RTs during attentional allocation. Similar coupling between theta phase and gamma amplitude has been tied to learning in the rodent hippocampus [63]: the strength of theta-gamma PAC increases as a rodent' s performance increases, both plateauing together, over time. These studies are important, since they provide evidence that PAC serves a functional role in the brain. The current study demonstrates that delta/theta phase-HG amplitude coupling also serves a functional role by indexing the overall level of engagement within human frontal and parietal cortex during the allocation of visuospatial attention. Eight subjects (S1–S8), undergoing pre-surgical epilepsy evaluation, provided written and oral informed consent to participate in the study. The institutional review board of each institution approved the research that was conducted at each respective location. Anti-epileptic medications were discontinued for 2–3 days beforehand, and patients were seizure free for at least five hours before testing. Subjects had normal or corrected-to-normal vision. Subjects were implanted with 74–128 electrodes (1 cm spacing; grids and strips), covering extensive portions of lateral and medial frontal, parietal, occipital, and temporal cortex in the RH and LH (see Figure 1a for overlap of electrodes from all subjects and Table 1 for specific coverage information for each subject). Visual displays were generated on a Dell Precision M4600 laptop (Dell, Inc.) using EPrime software (Psychology Software Tools, Inc.). Subjects were seated 50–60 cm from the computer screen. Electrophysiological and the peripheral (photodiode) channels were acquired using a 128-channel Stellate Harmonie recording system (Natus Medical, Inc. ; 1000 Hz digitization) at Johns Hopkins, a 128-channel Nihon Kohden recording system (Nihon Kohden Corporation; 1000 Hz digitization) at Children' s Hospital, and 128-channel Tucker Davis Technologies recording system at Stanford (3052 Hz digitization). Data were recorded using a subdural electrode reference and a scalp ground. Postoperative computed tomography (CT) scans were aligned to the preoperative MRI anatomical brain volume [64]. Electrodes were visualized on the 3D cortical surface using MATLAB (MathWorks, Inc. ; Figures 2–5). Brains and electrodes were transformed into MNI space across subjects (Figure 1a, Figure S1). A neurologist manually inspected all ECoG channels to identify those with interictal or ictal epileptiform activity and artifacts. Channels and epochs contaminated by epileptiform activity or abnormal signal (e. g. , poor contact, excess drift, high frequency noise) and those located over tissue that was later resected were removed from analysis. Data processing used custom functions written in MATLAB and the EEGLAB toolbox [65]. Raw, continuous data were filtered with a 60 Hz notch filter and re-referenced to a common average reference (mean of remaining channels). Single channels of this ECoG data are referred to as “raw signal. ” Trials were classified based upon attention condition (attended target in cued visual field; or unattended target appearing in the uncued visual field). Contralateral and ipsilateral attention conditions refer to the cued visual field with respect to the implanted ECoG electrodes. All trials were designated by response type (hits, misses, correct rejections, and false alarms). Correctly attended trials were designated hits and correctly unattended trials were designated rejections.

The frontal and parietal areas of the cortex control the ability to focus visuospatial attention, and damage to these areas results in profound attentional disturbances. Although much research has concentrated on where these areas are located, little is known about how these areas may function in humans. Previous studies have demonstrated that neuronal spiking is more likely to occur in specific time windows based upon the phase of lower frequency neural oscillations – rhythmic or repetitive neuronal activity. These low-frequency rhythms are hypothesized to coordinate the timing of neuronal firing within local and across network regions. Here, we investigated how human frontal and parietal cortices use neural oscillations to control visuospatial attention. We identified a high-frequency component of electrical brain activity, broadband high gamma (70–250 Hz) amplitude, that became phase-locked to a slower rhythm, delta/theta (2–5 Hz), over frontal, parietal, and visual areas while the study subjects paid attention to the peripheral visual field. Changes in the strength of the coupling between delta/theta phase and high gamma amplitude predicted the attentional behavior of the subjects across single trials. From these results, we conclude that coupling between delta/theta phase and high gamma amplitude serves to coordinate information within – and perhaps between – frontal and parietal areas during allocation of visuospatial attention.

Abstract Introduction Results Discussion Materials and Methods

cognitive neuroscience cognitive psychology neural networks biology and life sciences vision brain electrophysiology evoked potentials sensory perception cognitive science neuroscience neurophysiology attention

2014

Dynamic Changes in Phase-Amplitude Coupling Facilitate Spatial Attention Control in Fronto-Parietal Cortex

Systems as diverse as the interacting species in a community, alleles at a genetic locus, and companies in a market are characterized by competition (over resources, space, capital, etc) and adaptation. Neutral theory, built around the hypothesis that individual performance is independent of group membership, has found utility across the disciplines of ecology, population genetics, and economics, both because of the success of the neutral hypothesis in predicting system properties and because deviations from these predictions provide information about the underlying dynamics. However, most tests of neutrality are weak, based on static system properties such as species-abundance distributions or the number of singletons in a sample. Time-series data provide a window onto a system’s dynamics, and should furnish tests of the neutral hypothesis that are more powerful to detect deviations from neutrality and more informative about to the type of competitive asymmetry that drives the deviation. Here, we present a neutrality test for time-series data. We apply this test to several microbial time-series and financial time-series and find that most of these systems are not neutral. Our test isolates the covariance structure of neutral competition, thus facilitating further exploration of the nature of asymmetry in the covariance structure of competitive systems. Much like neutrality tests from population genetics that use relative abundance distributions have enabled researchers to scan entire genomes for genes under selection, we anticipate our time-series test will be useful for quick significance tests of neutrality across a range of ecological, economic, and sociological systems for which time-series data are available. Future work can use our test to categorize and compare the dynamic fingerprints of particular competitive asymmetries (frequency dependence, volatility smiles, etc) to improve forecasting and management of complex adaptive systems. Adaptive evolution requires that rivalrous goods are consumed by agents, those agents have heritable variation in how they acquire and consume the rivalrous goods, and the fitness of agents increases with the amount of goods consumed [2]. By Lewontin’s listing of the necessary conditions of evolution, a variety of systems can be seen as evolving. Genes in populations, species in a community, companies in a market, and political groups in a society all satisfy Lewontin’s axioms [2]. Canopy space is a rivalrous resource in the multi-species closed-canopy forests. If nothing else intervenes, a competitively superior tree species will dominate the canopy just like a competitively superior gene will become fixed in a population. In economic systems, companies compete over capital, customers, and labor, and a company well-adapted to a market will increase its share of the resources. In social systems, political groups compete over votes and occupied positions of power, and political groups with superior recruitment compared to other groups—either by persuasion, coercion, aggression, or reproduction with vertical transmission of culture—will increase the votes it receives and/or its representation in various positions of power. These generalized competitive systems are examples of “complex adaptive systems” [3,4, 5] and understanding how they evolve can provide insight into the drivers of adaptive evolution [6], diversity maintenance in human and natural systems [7], portfolio construction in a market [8], and problems of recognition and representation in multicultural societies [9]. Much literature has explored the stochastic fluctuations of individual populations (e. g. [10,11]) or asset prices [12] in these systems, and accurate models of the stochastic time-evolution of multi-species systems can enable calculations of the risk of extinction [13], the dynamics of diversity (such as the entropy or evenness of a system), portfolio analysis, and other features of interest. A common stochastic model in which all groups are functionally equivalent, termed “Neutral Theory” in ecology and population genetics, has been used across many systems [14,15,16,17,18]. By “functionally equivalent”, we mean that every agent’s performance in acquiring the rivalrous resource is independent of their group membership. In other words, an organism’s species identity, a company’s strategy or sector, a citizen’s political identity, or a political party’s platform have no impact on their ability to hold or acquire new rivalrous resources. Neutrality is a parsimonious starting point for community modeling because it is based on first principles of random birth and death or acquisition and release of resources that are appropriate for many competitive systems, and, because neutrality does not assume particular traits that distinguish groups and complex interactions between groups, it is invariant to grouping: the populations of neutral species can be aggregated into larger groups whose competition is also neutral. Neutrality is often posed as a null model for multi-species systems because it can be parsimonious to assume, initially, that all species are equivalent. The mathematical tractability of neutral systems has allowed for useful calculations [19,20] that can sometimes accurately describe features of the system. However, despite the mathematical ease, calculations for features such as extinction time or the dynamics of portfolio diversity based on neutrality may be inaccurate for systems with non-neutral dynamics such as positive or negative frequency-dependent selection. Thus, there is a need for powerful and informative tests of neutrality to assess whether or not the dynamics of the competitive system are neutral. In population genetics, tests of neutrality [21,22] have facilitated rapid conceptual and empirical advancements [23], allowing researchers to scan entire genomes for neutral loci and identify loci that have been under selection. Neutrality tests developed in ecological and sociological systems test features of rank-abundance and frequency-abundance distributions [16,24,25]. Many of these existing neutrality tests utilize snapshots of a competitive system, but time-series contain a tremendous amount of data and can enable stronger tests of neutrality. Some work has been done developing and utilizing tools to test whether or not population dynamics are consistent with Neutral Theory [10,26,11]. These tests rely on a particular description of Neutral Theory as the one-step process [27] posed by Kimura and Hubbell, but the proof [28] that the non-zero-sum volatility-stabilized market models [17] converge to neutral drift in relative abundances motivates a broader definition and more general tests of neutrality. Since neutrality is the per capita equivalence between species, it is necessarily relative, not absolute; a population is not neutral per se, but can only be neutral relative to another population. To the best of our knowledge, the existing time-series tests have all analyzed whether or not the variance in jumps in abundance increase linearly or quadratically with the population size prior to the jump, and none have examined the covariance structure of fluctuations in relative abundance. Here, we present, to our knowledge, the first covariance-based neutrality test for time-series data. Our test determines if the covariance structure of a time-series dataset is compatable with a Wright-Fisher process, thus yielding more insight into the nature of non-neutrality than traditional tests of rank-abundance distributions and the volatility of individual populations. Our test utilizes the grouping invariance of neutral systems to isolate and test the covariance between changes is species’ relative abundances over small time intervals, allowing a rejection of neutrality for the entire community considered. We apply our test to 6 metagenomic time-series [29], a time-series of breeding birds across North America [30], and a time-series of market capitalization of companies in the S&P 500 from 2000–2005. We show that even some systems whose rank-abundance distributions appear neutral can exhibit significantly non-neutral covariances between species as detected by our test. Furthermore, our test, based on random groupings of species, illustrates how to analyze the volatility of randomly formed groups to reveal state-dependent volatility that differs from neutrality. The non-neutral state-dependent covariance structure uncovered here can be incorporated to improve our models of community dynamics and calculations of species’ extinction times, portfolio risk, and more. Large neutral communities are well-approximated by a Wright-Fisher Process (WFP) [31,32]. The convergence of discrete neutral communities to the continuous diffusion model of the WFP is covered in [33], and some numerical methods used for parameter estimation and simulation have been produced by [34]. The WFP is a continuous-state, continuous-time approximation of Kimura’s theory [14], it is an approximation of Hubbell’s neutral theory [15] when speciation rates are negligible over the timescale of interest, and it describes the dynamics of relative abundances of non-zero-sum volatility-stabilized market models [17,28]. Using the WFP as a continuous approximation of large, finite communities simplifies the covariance in the jumps between species’ relative abundances, thereby permitting the analysis below. The WFP models the stochastic time-evolution of relative abundances of n species. Let X t = (X t 1, . . . , X t n) be the vector of relative abundances, the WFP is defined by the Ito SDE d X t = λ ρ - X t d t + σ X t d W t (1) where λ > 0 and ρ > 0. The covariation between relative abundances of different species is given by the elements of Σ = σσT/2, where The deterministic motion, or drift, of the WFP—λ (ρ−Xt) —yields exponential mean reversion like many dynamical systems reverting to a nearby stable equilibrium. The quadratic covariation of the WFP, Σ, captures the stochastic fingerprint of neutrality; it arises from randomly drawing a resource to be freed from its agent followed by randomly drawing one of the remaining agents to acquire that resource with a probability proportional to the agent’s current resource ownership. The family of Wright-Fisher Processes is closed to grouping, meaning that if a multi-species community’s dynamics are governed by a WFP, species can be grouped (e. g. collecting species into genera or higher taxonomic levels) and the dynamics of the resulting, re-grouped community will also be governed by a WFP. We developed a test that is intentionally sensitive to the state-dependent noise of the WFP, allowing researches to test the underlying stochastic model of the random turnover of resources at the heart of neutral competition. Developing a strong test of the state-dependent covariance is not trivial, though, because direct measurement of the covariance of jumps conditioned on the state of the system, Cov [ΔXt|Xt], would require many replicate time points each with the same initial state, Xt, and, even with multiple time points at the same state, the sparsity of the high-dimensional data challenges the accurate estimation and significance testing of the covariance matrix. To circumvent these problems of replicate time points and high dimensionality and develop a strong test the state-dependent noise of the WFP, we find a variance-stabilizing transformation for the WFP that allows a regression-based heteroskedasticity test [35]. To be precise, we are looking for a real-valued function f (Xt) such that for Xt obeying the WFP law in eq (1), lim Δ t → 0 Var f X t + Δ t - f X t Δ t = c o n s t. (3) This approach is conceptually similar to the variance-stabilizing tools used for population fluctuation analyses [10,26,11] which stabilize the variance in jumps of a population size, except our function must stabilize the covariance of jumps between populations, not just the variance. In particular, to have a constant volatility, our function f must satisfy the Hamilton-Jacobi equation, ∇ f T Σ ∇ f = c o n s t. (4) (see S1 Text, part 2, for more details). The grouping invariance of the WFP can be used to intuit and show that there are at least 2n different variance-stabilizing transformations of the WFP, parametrized by a vector a: f a X t = arcsin a 1 X t 1 +. . . + a n X t n, (5) where ai = ±1 for all i. After transforming the data with fa, we need to perform a constant-volatility test. To test the constant volatility of f, we test the homoskedasticity of standardized jumps, ν t = f t + Δ t - f t / Δ t, (6) following regression on ft to eliminate the state-dependent drift. A homoskedasticity test of νt for a single CVT is a test of neutrality for time-series data. However, with 2n different CVTs, one can perform multiple hypothesis tests. For any multiple-hypothesis tests, if the null hypothesis is true, the distribution of P-values is uniform. In this paper, we test the uniformity of the distribution of resultant P-values from homoskedasticity tests of νt for a number of randomly drawn CVTs. Fig 1 illustrates this test for 2,000 randomly drawn CVTs and shows the successful rejection of the WFP for the relative abundances of a mean-reverting geometric Brownian motion. The P-values arising from homoskedasticity tests of different CVTs are not independent. Consequently, a Kolmogorov-Smirnov test of the P-value distribution against a uniform distribution would have a high false-positive rate. To reduce the false-positive rate, we perform a perturbation analysis to generate conservative estimates of cutoffs for the KS statistic at 0. 05 and 0. 005 significance levels. Details of the sensitivity analysis are provided in S1 Text, part 3. Our method is proven to work and can correctly identify a community of uncorrelated geometric Brownian motions as being non-neutral (Fig 1). Some may claim that neutrality of geometric Brownian motions is more easily rejected by a super-linear relationship between mean and variance in population size over short time intervals, or by analysis of how the variance in a measure of fluctuation of population size Nt, D Δ t = (N t + Δ t - N t) / N t, changes with the size of the time-window Δt [11]. We applied existing tests of variance- and fluctuation-scaling to the volatility-stabilized market model (VSM), d N t i = δ i 2 S t d t + N t i S t d W t i (7) where N t i is the absolute abundance of species i, S t = ∑ i N t i is the community size, and δi > 0. Pal [28] proved that the market shares or relative abundances of the VSM follow a WFP with λ = ∑i δi and ρi = δi/λ, and the grouping invariance of the VSM can be shown. Both the scaling of variance in population size with expected population size and scaling of the variance of DΔt indicate would incorrectly reject the proven neutrality of the VSM, whereas our test does not reject neutrality in the relative abundances of the volatility-stabilized market model (Fig 2). Applying existing tests which analyze the fluctuations of individual populations to the volatility stabilized market model reinforces our claim that neutrality is necessarily relative and that fluctuations in absolute abundance are dubious tests of neutrality. Our test of the fluctuations of relative abundances tests the grouping invariance at the heart of neutrality, regardless of the manner of fluctuations of community size, and correctly identifies the volatility-stabilized market model as neutral where existing tests suggest otherwise. We apply our neutrality test to 8 different datasets. Six of these datasets are sequence-count data of microbial communities [29] from three body sites on two individuals. One dataset is survey of breeding birds across North America from 1966–2014 [30], and one dataset is financial data, obtained from the Center for Research in Security Prices, of the day-end market shares and market capitalization of 451 companies in the S&P 500 from January 1,2000 to January 1,2005. These datasets are long, time-series datasets, many of which have rank-abundance distributions that are decently fit by neutral theory’s expected rank-abundance distribution (see S1 Text for a detailed description of the datasets and fits of neutral theory’s rank-abundance distributions). Our test relies on multiple groups of species. To group the species, we randomly selected ai = ±1, for 4,000 independent groups, we then calculated νt as in eq (6), and used a White test [36] to test the homoskedasticity of νt. The White test performed auxiliary regression with a generalized linear model with a log-normal link function of the form ϵ ^ 2 = β 0 + β 1 f t + β 2 f t 2 + γ (8) where ϵ ^ 2 are the squared residuals from similar quadratic regression of ft on νt. With a 0. 005 significance threshold, neutrality was rejected in all but one dataset (Fig 3). Qualitatively, there is a major difference in the neutrality of gut, tongue, and palm microbial communities. The gut and tongue microbial communities are clearly non-neutral, whereas the palm microbial communities, likely subjected to more frequent disturbances and colonizations, both appeared close-to-neutral, with one of the palm microbial communities having a P-value above the significance threshold. These qualitative differences suggest different probabilistic laws governing the colonization and turnover of these communities, and that some microbial communities may be closer to neutral than others. Rejecting neutrality for these competitive systems motivates further investigation on whether the rejection of neutrality stems from sampling error or from true competitive asymmetries in the system. For the financial datasets, there is no sampling error—the reported values of day-end prices are the true values. For the metagenomic datasets, sampling error for sequence count data could be driving apparent non-neutrality. Fig 4 examines the competitive asymmetry in the male tongue’s bacterial community. Scatter plots of νt vs. ft reveal a downward trend indicative of mean reversion—jumps in ft are positive when ft is above its mean and negative when ft is above its mean (Fig 4A). The mean reversion is regressed out prior to our auxiliary regression, but such strong mean reversion is not apparent in the simulated neutral community of Fig 1. The discrepancy between the mean reversion in the data and the simulated neutral community may be due to a stronger and/or non-linear mean reversion in the microbial system, or it may be due to the relatively long time between time points in the data relative to the turnover rate of the community. Such sparse time-sampling could affect the accuracy of our test, which relies on the assumption that Δt in eq 3 is small. A neutral community may still have mean reversion due to migration from a metacommunity or mutation/conversion rates between classes of agents, and sparsely sampling a time-series of such a neutral community may yield the same downward trend on plots of νt versus ft. To examine if the long time between time points accounts for the perceived non-neutrality from our analysis of Fig 3, we produced surrogate data by simulating neutral communities with similarly sparse time points. Parameter estimation of λ, ρ and WFP simulation is described in S1 Text, part 4. For one particular CVT, we simulated 4,000 independent trajectories to allow the superposition of the νt vs ft scatter plots from the male tongue data over the points from the surrogate data. Much of the strong mean reversion in the data can in fact be accounted for by the sparsity of time points (Fig 4A), but the P-value distribution from constant-volatility tests of 16,000 randomly drawn surrogate CVT simulations is much more uniform than the same distribution from the male tongue, which has many small P-values indicative of consistent state-dependent volatility of ft (Fig 4b). Thus, the non-neutrality of the male tongue dataset is not due to long sampling intervals. The male tongue microbiome deviates from neutrality by having a significantly more β2 > 0 than β2 < 0 for those auxiliary regressions yielding significant heteroskedasticity (Fig 4C). β2 > 0 indicates that the volatility increases farther away from the mean and plotting the residuals, ϵ ^ 2 against the state variable, ft, reveals the form of heteroskedasticity (Fig 4D). A similar significant hyper-abundance of β2 > 0 exists for all datasets considered here (Fig 4E). Evolution, driven by competition over rivalrous goods or limiting resources, is a phenomenon common to ecology, economics and sociology, and accurate statistical models of how competitive systems evolve can allow us to forecast, manage, and invest in them [2,3, 8,1, 34]. Neutral Theory is a null model of competition which assumes that all players are equal—that a canopy tree fills a gap in the canopy independent of its species’ identity, a dollar finds its way to another dollar independent of who owns the dollar, and a seat in congress is filled by someone independent of the racial, cultural or political traits of the successor or predecessor. It’s been hypothesized that neutrality could arise naturally as a result of competitively inferior species going extinct [37], and thus systems would tend towards neutrality over long periods of time, but the accuracy and generality of Neutral Theory as a dynamical model for a range of competitive systems was unclear. Previous tests of neutrality have relied on tests of the fluctuations of population size. However, the concept of neutrality is necessarily relative: organisms can only be neutral relative to one-another, and so testing the fluctuations of population size against a “neutral” signature only tests very particular models of neutrality with particular fluctuations in community size. Changes in population size are equal to the product of changes in relative abundance and changes in community size, and the essence of neutrality is in the changes in relative abundance, not changes in population size. Neutrality tests based on fluctuations in population size require assumptions about fluctuations in community size, fluctuations which may not be known from the ecological data, data which may be best viewed as compositional data. There are many ways to maintain neutrality in finite-community-size models while modifying fluctuations in community size: at random points in time, there could be deterministic forcing, geometric increases or decreases, or any other fluctuations in community size independent of community composition, followed by a random draw of organisms to die, migrate and reproduce fill the empty spaces in the community; such populations can be neutral despite markedly different probability laws for the fluctuations in population size. In all cases, however, the relative abundances will be neutral, and with neutrality comes a particular signature of competitive symmetry that can be tested. In this paper, we have provided a time-series test of neutral covariance structure yielding a P-value indicative of the compatibility of the data with neutral dynamics. Applying our test to real datasets has revealed a common feature of non-neutrality across a range of ecological and economic systems. Our test is based on the grouping invariance of neutral communities, and this grouping invariance is maintained by a particular covariance structure of volatilities, namely where the volatility, v, of a group, X t i, is v (X t i) = c X t i (1 - X t i) for some constant, c, which can only be invariant to grouping if the covariance between groups i, and j is Σ i, j = - c X t i X t j. We test neutrality by randomly grouping species and testing if the volatility of those random groups is of the form of v (x). A deviation from v can indicate non-neutral covariance structure. If, for instance, instead of a quadratic curve of v, the volatility of random groups in the data follows a bell-shaped curve with positive curvature at the end-points, it could indicate that there is a positive covariance between rare species, possibly due to kill-the-winner effects or due to the relative abundances being driven by fluctuations in the largest populations. Conversely, excessive negative curvature could indicate strong negative covariances between rare species, possibly due to relatively constant populations of abundant species. Future work analyzing the volatility surfaces of random groups can improve our test and allow researchers to quickly isolate particular forms and fingerprints of common, non-neutral competitive asymmetries. Our test assumes stationary migration and volatility. Non-stationarity in migration, such a daily time-series with a large pulse of migrant birds arriving to their breeding grounds, would very likely lead to a rejection of neutrality. However, the same system, sampled over longer time-intervals, may have stationary distributions of migration or volatility that could lead to a failure to reject neutrality. The results presented here are limited to the particular choice of species (the taxonomic scale), resource (trophic scale), and time scale. However, an analysis restricted to a particular scale means that our test can also allow researchers to probe multiple taxonomic, trophic and time-scales to see if there are patterns in which scales are most/least neutral in their dynamics. An alternative grouping of these original OTUs by genera may reveal different results by conditioning the groupings of species on a particular sub-set of possible groups, such as grouping monophyletic clades of OTUs or grouping OTUS with shared metabolic traits, and consequently this test could serve as a tool for evaluating competition at multiple taxonomic scales. The results are also limited to the choice of resource: companies within a sector governed by trust-busting policies which break the neutral symmetry in their market capitalization dynamics may still be neutral in their competition over the ethnic or cultural composition of their labor force. Much like the example of breeding birds presented above, it’s possible that microbes in the gut are not neutral in their short-term fluctuations over the course of a year, but perhaps are neutral over longer time-scales that average out short-term fluctuations in diet and physiological state that are known to have predictable effects on microbial communities [38,39]. The results from a neutrality test also depend on the number of time points sampled. Ultimately, we use a P value as a quantification of the incompatibility of the data with neutral covariance structure. With P values as our measure of incompatibility, the true measure of relative distance from neutrality to any given competitive system, the P value we obtain in the limit of infinte data, will be 0 or 1 (neutral or not). Most, if not all, systems are non-neutral, much like there is no perfect square in the universe, for all square-like objects are not exactly square. Nonetheless, there are many competitive systems in nature, from ecosystems to economies, where neutrality and the grouping invariance it entails is a compelling null model and scientists pursuing hypothetico-deductive progress can use the P value as a means to test the essence of neutrality and, where they reject neutrality, move on to constructing the next-best model. Others may be interested in comparing the strength of non-neutrality across systems, and for those cases the most rigorous comparisons through P values are those comparisons made between systems with the same number of time points. Our test can be applied to any community of competing agents classified into discrete groups for which time-series of relative abundances (or market share, etc) of the groups are available. This test is most effective when the time-series is long and the spacing between samples is short relative to the turnover rate of the underlying resources (trees, dollars, congressional seats). There are many ways to build on our method. Explicit calculations of the drift and volatility over long time-intervals can improve our method for datasets with sparse time points. The dependence of the CVTs may be calculated as copulas allowing the implementation of a more exact goodness of fit test [40], or a perturbation analysis of the Kolmogorov-Smirnov test-statistic under different numbers of species, timepoints, and CVTs can provide quantitative estimation of the P-values, thus allowing more precise comparison of the degrees of non-neutrality across systems. There may be more CVTs that solve the Hamilton-Jacobi equation, and different CVTs might be more specialized at detecting different asymmetries in competitive systems. Neutrality tests of time-series data can help us understand the stochastic time-evolution of competitive systems and facilitate better prediction and management. For bacteria in the gut, for instance, understanding the important non-neutral forces governing the dynamics could allow progress towards the stochastic pharmacokinetics of probiotics [41]. Understanding the predictability of invasions at different taxonomic scales can tell us whether to evaluate metagenomic or ecological datasets based on the species, or whether other taxonomic levels will yield a more informative analysis—perhaps grouping tropical trees into the family Fabaceae, the family Melastomataceae, the genus Cecropia, and all other trees reveals trophic structure of tropical forests and competitive asymmetries that are drowned out by analyses at the species level. Demonstrating that the non-neutrality of portfolios is consistent with rare-species advantages in the Atlas model [8] would have major implications for portfolio design. In all cases, the first step of empirically demonstrating the existence of competitive asymmetries in time-series data can now be done with the test provided here.

From fisheries and forestries to game parks and gut microbes, managing a community of organisms is much like managing a portfolio. Managers care about diversity, and calculations of risk—for extinction or financial ruin—require accurate models of the covariance between the parts of the portfolio. To model the covariances in portfolios or communities which may have some direct or diffuse competition over limiting resources, it helps to start simple with a null model assuming the equivalence of species or companies relative to one another (termed “neutrality”) and then testing whether or not the data suggest otherwise. Researchers in biology and finance have independently entertained and tested neutral models, but the existing tests have used snapshots of communities or the variance of fluctuations of individual populations, whereas tests of the covariances between species can better inform the development of alternative models. We develop a covariance-based neutrality test for time-series data and use it to show that the human microbiome, North American birds, and companies in the S&P 500 all have a similar deviation from neutrality. Understanding and incorporating this non-neutral covariance structure can yield more accurate alternative models of community dynamics which can improve our management of “portfolios” of multi-species systems.

Abstract Introduction Materials and Methods Results Discussion

species colonization medicine and health sciences ecology and environmental sciences invasive species neutral theory population genetics social sciences random variables covariance mathematics tongue population biology digestive system ecological economics population ecology economics probability theory community ecology finance mouth anatomy ecology genetics biology and life sciences physical sciences evolutionary biology

2016

Novel Covariance-Based Neutrality Test of Time-Series Data Reveals Asymmetries in Ecological and Economic Systems

Yaws is a treponemal infection that was almost eradicated fifty years ago; however, the disease has re-emerged in a number of countries including Ghana. A single-dose of intramuscular benzathine penicillin has been the mainstay of treatment for yaws. However, intramuscular injections are painful and pose safety and logistical constraints in the poor areas where yaws occurs. A single center randomized control trial (RCT) carried out in Papua New Guinea in 2012 demonstrated the efficacy of a single-dose of oral azithromycin for the treatment of yaws. In this study, we also compared the efficacy of a single oral dose of azithromycin as an alternative to intramuscular benzathine penicillin for the treatment of the disease in another geographic setting. We conducted an open-label, randomized non-inferiority trial in three neighboring yaws-endemic districts in Southern Ghana. Children aged 1–15 years with yaws lesions were assigned to receive either 30mg/kg of oral azithromycin or 50,000 units/kg of intramuscular benzathine penicillin. The primary end point was clinical cure rate, defined as a complete or partial resolution of lesions 3 weeks after treatment. The secondary endpoint was serological cure, defined as at least a 4-fold decline in baseline RPR titre 6 months after treatment. Non- inferiority of azithromycin treatment was determined if the upper bound limit of a 2 sided 95% CI was less than 10%. The mean age of participants was 9. 5 years (S. D. 3. 1, range: 1–15 years), 247 (70%) were males. The clinical cure rates were 98. 2% (95% CI: 96. 2–100) in the azithromycin group and 96. 9% (95% CI: 94. 1–99. 6) in the benzathine penicillin group. The serological cure rates at 6 months were 57. 4% (95% CI: 49. 9–64. 9) in the azithromycin group and 49. 1% (95% CI: 41. 2–56. 9) in the benzathine penicillin group, thus achieving the specified criteria for non-inferiority. A single oral dose of azithromycin, at a dosage of 30mg/kg, was non-inferior to a single dose of intramuscular benzathine penicillin for the treatment of early yaws among Ghanaian patients, and provides additional support for the WHO policy for use of oral azithromycin for the eradication of yaws in resource-poor settings. Pan African Clinical Trials Registry PACTR2013030005181 http: //www. pactr. org/ Yaws is a relapsing non-venereal treponematosis caused by Treponema pallidum subspecies pertenue. The disease mainly affects the skin, but if untreated, can also affect bone, joints and cartilage. Yaws may persist for many years as a chronic infection, and late stage disease may lead to crippling disfigurement [1,2]. The bacterium that causes yaws is closely related to T. pallidum ssp. pallidum, the causative organism of venereal syphilis, however T. pallidum ssp. pertenue is thought to be less virulent [3]. Transmission of yaws occurs from person to person through direct skin to skin contact, involving transfer of infectious exudates from the early skin lesions of infected individuals to micro- or macro-abrasions of the skin of siblings/playmates. [4–7]. Yaws was previously widespread throughout the tropics but a global eradication campaign between 1952 and 1964 resulted in a 95% reduction in disease prevalence worldwide [8]. Following this initial success, yaws control was integrated into national primary healthcare systems. Unfortunately, this integration resulted in a weakening of yaws surveillance in many countries and the re-emergence of the disease by the 1970s [9]. Several reports have recently documented a resurgence of yaws in parts of West and Central Africa, South East Asia and several Western Pacific Islands [10–12]. Yaws is an important public health problem in Ghana and affects the poor rural communities. The disease is reported in all the ten regions of the country. A prevalence study in three purposively selected districts in Southern Ghana in 2008 showed an overall prevalence of 1. 92% among children in primary schools; however individual school prevalence rates ranged from 0% to a high of 19. 5% [13]. Previous mass treatment campaigns to eradicate yaws in Ghana were based on the use of a single dose intramuscular (IM) injection of long-acting penicillin. The advantages of single-dose treatment with penicillin are its low cost, adherence and the absence of antimicrobial resistance despite extensive use. However, treatment with intramuscular penicillin has several disadvantages such as pain, anaphylaxis, and potential for transmission of other blood-borne infections. Azithromycin has previously been shown to be an effective agent in the treatment of venereal syphilis [14]and is also the cornerstone of the strategy for the elimination of trachoma [15]. Azithromycin offers a number of advantages as an agent for the treatment of yaws, including its low cost, oral route of administration, excellent safety profile and negligible risk of anaphylaxis. As such, the drug is well suited for use in community mass treatment campaigns for yaws eradication [16]. In 2012, a single centre study conducted in Papua New Guinea, showed that a single oral dose of 30mg/kg azithromycin was non-inferior to a single injection of benzathine penicillin in the treatment of the disease, Clinical cure rate was 85. 4% (95% CI: 78. 2–90. 6) in patients treated with azithromycin compared to 86. 5% (95% CI: 79. 5–91. 5) in those treated with benzathine penicillin [17]. These findings led to recommendations incorporated into the WHO’s Morges yaws eradication strategy [18]. Here, we report a similar study carried out in Ghana, West Africa in order to confirm the efficacy and suitability of azithromycin for treatment of yaws in Sub-Saharan Africa. This trial was conducted according to the principles of the Declaration of Helsinki. The study was approved by the Ghana Health Service Ethical Review Committee (Ref: GHS-ERC: 13/11/10). Written informed consent was obtained from parents/guardians of all participants; those aged 12 years and above also signed an assent certificate. The study was conducted between 25th May 2011 and 31st December 2012 in three neighbouring yaws endemic districts in southern Ghana: Ga South, Awutu Senya and West Akim districts (Fig 1). The three study districts have a total population of 787,747 with a distribution of 316,091,274,584 and 197,072 respectively. Children under the age of 15 years, known to bear the bulk of yaws infections, represent an estimated 38. 3% of these populations. There are more than 600 rural communities in these three districts. The districts have a total of 8 government and private hospitals, 10 government health centres, 81 private health facilities made up of small clinics and maternity homes and 16 Community Based Health Planning and Services (CHPS) compounds. Only 20% of these health facilities are located in the rural parts of the district where yaws occurs. The doctor population ratio in the study area is 1: 15,754. The main occupation of the people in the study area is subsistence farming. There are a total of 1750 primary schools and kindergartens in the study area with a total enrollment of 174,536 made up of 51% males and 49% females. Yaws surveillance in the study area is mainly by passive detection at the health facilities. Based on routine reports of presentation of clinical cases to health facilities and a small number of active case searches in schools, the yaws case notification rate has been estimated to be between 87–241 per 100,000 people. Between 2009 and 2011,2674 clinical cases of yaws were reported in the 3 study districts combined. Participants were recruited based on clinical suspicion by health workers at home or in school, and all subjects with suspected yaws were assessed for eligibility. Clinical examination involved inspection of the skin and scalp for signs of early yaws lesions. A suspected case of primary yaws was defined as an ulcer with raised edges and a dirty crusty base or a papilloma that appeared as a firm yellowish skin lesion with a dark tip, on any part of the body. A suspected case of secondary yaws was defined as any of the following skin lesions: multiple ulcerative or papillomatous skin lesions; a palmar or plantar hyperkeratosis or a macular, papular or maculopapular skin lesion. A confirmed case of yaws was defined as a suspected case with a positive TPHA test and an RPR titre of at least 1: 4. Photographs of the lesions were taken before treatment and subsequently at follow-up review visits. A 5mL sample of venous blood was collected from those with lesions clinically suggestive of yaws and analyzed at the National Public Health Reference Laboratory in Accra by qualitative Treponema pallidum haemaglutination assay (TPHA) testing (Debens Diagnostics Ltd. , Ipswich, UK.). In parallel, a second set of frozen sera were sent to the Komfo Anokye Teaching Hospital serology laboratory in Kumasi for qualitative and quantitative rapid plasma reagin (RPR) testing (Immutrep RPR test kit, Omega Diagnostics, Alva UK). Inclusion criteria for enrolment were individuals aged 1–15 years with suspected primary or secondary yaws, a reactive TPHA test and a RPR titre of at least 1: 4. Exclusion criteria were a negative qualitative TPHA test, a baseline RPR titre of less than 1: 4, allergy to penicillin and/or a macrolide antibiotic, a medical condition that would impair drug absorption, recent ingestion of a broad spectrum antibiotic (30 days prior to the day of randomisation) and patients who were not willing to give informed consent or who would not be available for follow up visits. In this study azithromycin manufactured by Pfizer in the strength of 250mg tablets was supplied by Gokals Laborex, Ghana. Benzathine penicillin, 1. 2 million units per vial, manufactured by Troge Medicals, Hamburg was supplied by Ghana Health Service Central Medical Stores. Participants who met the inclusion criteria were randomized to receive treatment with either a single dose oral azithromycin administered at a dose of 30mg/kg (maximum of 2g) or a single dose of benzathine penicillin administered as an intramuscular injection of 1. 2 million units for subjects 10-15years, and 0. 6 million units for those below 10 years of age. Participants in both arms were directly observed receiving treatment and for two hours thereafter. The allocation sequence was based on a computer generated block randomization scheme, stratified according to district. Participants were allocated in a ratio of 1: 1 to treatment with azithromycin or penicillin. Treatment allocation was concealed from investigators through the use of sequentially numbered, opaque, sealed envelopes that were kept in a safe and were opened at the point of treatment by the treatment team. Owing to the obvious differences between the mode of administration of the two drugs, investigators and study participants could not be blinded to the treatment received, however individuals assessing study outcomes were masked to treatment allocation. Serological tests were conducted in separate laboratories by laboratory technicians masked to clinical data. Participants were followed up at 3 weeks, 3 months and 6 months. Skin lesions were re-examined at 3 weeks post treatment. At 3 months and 6 months, skin lesions were examined and blood collected for repeat quantitative RPR testing. Individuals with lesions that had not healed at follow-up were re-treated with benzathine penicillin. Health workers and community-based surveillance volunteers monitored and documented all adverse events up to 72 hours after treatment. Parents and teachers also were counselled on possible adverse events after the field team had departed and the need to report to the nearest health facility if necessary. Adverse events that occurred within 2 hours of treatment were documented and managed by the field teams. Our primary outcome was clinical cure defined as a total or partial resolution of yaws skin lesions 3 weeks after treatment. The secondary outcome was serological cure defined as at least a 4-fold drop in baseline RPR titre within 6 months of treatment. Treatment was considered to have failed if there was no resolution of yaws skin lesions (complete or partial) 3 weeks after treatment. This trial was designed to assess if azithromycin was non-inferior to benzathine penicillin for the treatment of yaws. With an expected efficacy of penicillin of 95%, a type 1 error of 0. 05, and a non-inferiority margin of 10% and assuming that 10% would be lost to follow-up, a sample size of 310 children (155 per arm) would give a statistical power of 90% to test the hypothesis. Analysis of the primary endpoint of clinical cure was estimated by the two-sided 95% confidence interval for the difference in cure rates between the penicillin and azithromycin groups. Secondary outcome analyses were done using similar methods. Subgroup analyses were performed with stratification by baseline RPR titre, household exposure to yaws and stage of clinical yaws. A two-sided test at a significance level of 0. 05 was used in the comparison of baseline characteristics of the two treatment groups. All statistical analyses for this study were carried out in STATA 11. 1 (Statacorp, Texas, USA). The per-protocol (PP) analysis included all subjects who completed all study procedures at 6 months. The intention-to-treat (ITT) analysis included all eligible participants who were randomised and treated. Individuals with missing data were considered treatment failures for the purposes of the intention-to-treat analysis. The trial profile is shown in Fig 2. From May 2011 to December 2012, four hundred and three subjects with suspected primary or secondary yaws lesions were assessed for eligibility; 50 were found to be ineligible (39 were either TPHA negative or had a RPR titre below 1: 4,11 declined to participate). Therefore 353 eligible participants were randomly assigned to receive either a single-dose oral azithromycin or a single intramuscular injection of long-acting penicillin. Of the 353 subjects randomised, 25 participants (7. 0%) were lost to follow up. Six participants in the azithromycin group relocated and 1 refused to continue participation in the study. In the penicillin group, 15 participants relocated, 2 refused to continue participation and 1 patient died 5 months after treatment, a verbal autopsy concluded cause of death as malaria and National Yaws Eradication Programme was notified. The remaining 328 participants (169 in the azithromycin group and 159 in the penicillin group) completed the study and were analysed in the per protocol analysis. Demographic and clinical characteristics did not vary between the two treatments groups (Table 1). The mean age of study participants across both groups was 9. 5 years (SD: 3. 1, range: 1 to 15 years); 274 (70%) were male. Primary yaws was present in 187 cases (53%), 13 participants (3. 7%) had fever at presentation, 17 (4. 8%) had arthralgia, and 33 (9. 4%) had one or more other skin lesions in addition to those of yaws. One hundred and fifty five (43. 9%) participants had a baseline RPR titre between 1: 4 and 1: 16, and 198 (56. 1%) had titres between 1: 32 and 1: 128. One hundred and seventy one participants (48. 6%) lived in houses with at least one other individual who had been diagnosed with active yaws within the past one month. The most frequent clinical lesions were ulcers (167,47. 3%) followed by papillomas (101,28. 6%), hyperkeratosis of the palms and soles (25,7. 1%), while the rest were macules, papules and maculopapular lesions. Three individuals had ulcers with sabre tibia. Similar cure rates were recorded between the two treatment groups (Table 2). For the primary outcome of clinical cure defined as complete or partial healing of yaws lesion, 166 out of 169 participants (98. 2%) in the azithromycin group and 155 out of 159 participants (96. 9%) in the penicillin group exhibited complete or partial resolution (Figs 3 and 4) ) 3 weeks after treatment (risk difference: -1. 3%, (-4. 7 to 2. 0). For the secondary outcome (serological cure) 97 out of 169 (57. 4%) participants in the azithromycin group showed a 4- fold or greater decline in baseline RPR titres by 6 months after treatment compared to 78 out of 159 participants (49. 1%) in the penicillin group, (risk difference: -8. 3 (-19. 1 to 2. 4). Azithromycin therefore met the criteria for non-inferiority in both the primary and secondary outcomes. Three (1. 8%) participants in the azithromycin group and 4 (2. 5%) in the penicillin group with ulcerative yaws lesions did not resolve 3 weeks after treatment and were classified clinically as “treatment failures”. All participants considered as “treatment failures” were re-treated with intramuscular penicillin and all ulcers subsequently resolved within 2 weeks. Five subjects with ulcerative lesions which had healed at 3 weeks and had achieved serological cure were found to have recurred during the 3 month follow up period. Of these 3 (1. 7%) occurred in the azithromycin group and 2 (1. 3%) in the penicillin group. However due to the fact that subjects had achieved serological cure, lesions were likely of a different aetiology. Azithromycin treatment also proved to be non-inferior to penicillin therapy in subgroup analyses of primary outcome (clinical cure at 3 weeks) by clinical stage of yaws, baseline RPR titre and household exposure (Table 3). Analysis of serological cure by baseline RPR titres are shown in Table 4. Cure rates were low (41. 8% in the azithromycin group and 31. 7% in the penicillin group) in patients with low RPR titres of 1: 4–1: 16 compared to higher cure rates in patients with high RPR titres of 1: 32–1: 128 (69. 5% in the azithromycin group and 60. 4% in the penicillin group). No serious adverse effects related to the treatment drugs were reported in this trial. Minor adverse effects were reported by 4 participants (2. 4%) in the azithromycin group, most commonly gastrointestinal upset, and 8 participants (5%) in the penicillin group, most commonly pain at the injection site. This study shows that a single oral dose of azithromycin given at a dosage of 30mg/kg is non-inferior to a single intramuscular dose of benzathine penicillin in the treatment of yaws in children in Ghana. Indeed the cure rate for the primary outcome at 3 weeks was slightly in favour of azithromycin. For the primary outcome of clinical cure, 166 out of 169 participants (98. 2%) in the azithromycin group and 155 out of 159 participants (96. 9%) in the penicillin group showed a complete or partial resolution of yaws lesions 3 weeks after treatment. Our results were consistent with a previous study conducted in Papua New Guinea (17). We also showed that azithromycin was non-inferior to IM penicillin in all subgroup analyses, confirming the robustness of this conclusion. Azithromycin was well tolerated by participants, with no serious adverse events reported after treatment. The recorded side effects were, in keeping with the known side effects profile of the drug, namely mild to moderate and most commonly gastrointestinal in nature. Serological cure rates at 6 months were higher in the former study in Papua New Guinea (88%) compared to this study (57. 4%), due to enrolment of patients with higher baseline RPR titres (≥1: 16) in the PNG study. In this trial, 42. 6% of participants treated with azithromycin and 50. 9% of those treated with penicillin who were clinically cured did not achieve a decline in baseline RPR titres by at least 2 dilutions (4-fold) by 6 months after treatment. Among patients treated with azithromycin 5. 3% did not achieve a fall in RPR titre, 38. 1% achieved a 2 fold fall, 2. 9% showed a 2 fold increase and 2. 4% had a 4 fold increase in RPR titre 6 months after treatment. In the penicillin group 12. 7% did not show a fall in RPR titre, 33. 5% showed a 2-fold falling titre, 11. 8% showed a 2 fold increase in titre and 0. 6% showed a 4 fold increase in titre 6 months after treatment. Unfortunately, it is clinically and serologically impossible to distinguish treatment failure and relapse from re-infection. Since this study was conducted in endemic communities where exposure to antibiotics is uncommon, it is possible that many of those cases that showed a 4-fold increase in titre reflect re-infection rather than treatment failure. Increase in RPR titre could also be related to performing of initial RPR titre in the early stage in infection before titre reached its peak. The 2-fold increases and decreases in titre that were recorded here largely reflect the relative insensitivity of the quantitative RPR test. The availability of an orally effective treatment for yaws is key if the goal of eradication of yaws is to be attained. In this trial we demonstrate the efficacy of azithromycin in the treatment of yaws in a second yaws-endemic region far removed from the initial trial site in Papua New Guinea, confirming its place in the WHO yaws eradication strategy [19]. Effective treatment of yaws involves treatment of whole communities. Azithromycin is well suited to administration by community health volunteers, even in the poorly-resourced rural communities where yaws occurs. A recent study has demonstrated the impact of a single round of mass treatment with azithromycin in reducing transmission of yaws in Lihir Island, Papua New Guinea [20]. Azithromycin treatment failure among patients with syphilis, caused by a closely-related treponeme T. pallidum ssp. pallidum, has been widely reported in high-resource settings where overuse of antibiotics is common. Treatment failure has been associated with a single amino acid mutation at positions 2058 and 2059 in the 23S rRNAgene [21], which prevents binding with the bacterial 50S ribosomal subunit. In contrast, azithromycin resistance among T. pallidum ssp. pertenue strains has yet to be documented. Although populations in yaws endemic areas are typically not exposed to excessive antibiotic usage, there is a clear need to strengthen surveillance systems and closely investigate possible treatment failures for evidence of emergence of azithromycin resistance. There are a number of limitations to this study. Most notable is the inability to mask treatment assignments. We tried to mitigate this by having outcomes evaluated by independent assessors who were blinded to treatment allocation. We could not determine whether lesions that did not heal at 3 weeks post treatment were true treatment failures, since neither dark field microscopy nor PCR of lesion exudates was performed. As mentioned above, we were also unable to distinguish treatment failure from re-infection. Enrolment of patients with low RPR titres (1: 4–1: 8) may have influenced serological cure in this study, it was impossible to observe a 4-fold drop in RPR in patients with low titres. Finally, participants were followed up for only 6 months, after which no serological or clinical data were collected. This randomized controlled trial has clearly demonstrated that a single oral 30mg/kg dose of azithromycin is non inferior to a single dose of IM benzathine penicillin for the treatment of early yaws in Ghana. There was no significant difference in cure rates between patients treated with azithromycin and those treated with injection benzathine penicillin. Oral treatment with azithromycin overcomes the logistical and operational problems of using intramuscular penicillin in mass treatment campaigns. Our findings lend additional support for the use of a single dose azithromycin as the preferred regimen in yaws eradication programs.

Yaws is a tropical infection caused by a bacterium closely related to that which causes syphilis. It is transmitted from person to person through skin to skin contact and often causes papillomatous and ulcerative skin lesions, usually in young children. Without treatment, it can lead to deformities and disabilities. In the past, treatment of cases and their contacts and mass treatment of whole communities has been conducted using single doses of long acting penicillin. This treatment is inexpensive and does not pose a problem with adherence. However, the injections are painful and make it difficult to gain the cooperation of children. In addition, it requires trained health workers to safely administer treatment in poor-resource settings where yaws commonly occurs. In this study one group of children aged 1–15 years with clinically and serologically confirmed yaws received the standard treatment of a single injection of benzathine penicillin. A second group of children were treated with a single dose of oral azithromycin. The children were followed up at 3 weeks to assess healing of lesions, and at 3 and 6 months respectively to monitor serological indicators of infection. Our conclusion is that single dose oral azithromycin is as effective as a single injection of benzathine penicillin for the treatment of early yaws in Ghana and confirms the findings of a previous study undertaken in Papua New Guinea.

Abstract Introduction Methods Results Discussion

antimicrobials medicine and health sciences pathology and laboratory medicine pathogens drugs tropical diseases microbiology geographical locations treponematoses bacterial diseases signs and symptoms antibiotics ulcers neglected tropical diseases pharmacology africa bacterial pathogens infectious diseases penicillin serology medical microbiology papua new guinea lesions microbial pathogens yaws people and places treponema pallidum ghana diagnostic medicine oceania microbial control biology and life sciences

2017

A Single Dose Oral Azithromycin versus Intramuscular Benzathine Penicillin for the Treatment of Yaws-A Randomized Non Inferiority Trial in Ghana

High cost, poor compliance, and systemic toxicity have limited the use of pentavalent antimony compounds (SbV), the treatment of choice for cutaneous leishmaniasis (CL). Paromomycin (PR) has been developed as an alternative to SbV, but existing data are conflicting. We searched PubMed, Scopus, and Cochrane Central Register of Controlled Trials, without language restriction, through August 2007, to identify randomized controlled trials that compared the efficacy or safety between PR and placebo or SbV. Primary outcome was clinical cure, defined as complete healing, disappearance, or reepithelialization of all lesions. Data were extracted independently by two investigators, and pooled using a random-effects model. Fourteen trials including 1,221 patients were included. In placebo-controlled trials, topical PR appeared to have therapeutic activity against the old world and new world CL, with increased local reactions, when used with methylbenzethonium chloride (MBCL) compared to when used alone (risk ratio [RR] for clinical cure, 2. 58 versus 1. 01: RR for local reactions, 1. 60 versus 1. 07). In SbV-controlled trials, the efficacy of topical PR was not significantly different from that of intralesional SbV in the old world CL (RR, 0. 70; 95% confidence interval, 0. 26–1. 89), whereas topical PR was inferior to parenteral SbV in treating the new world CL (0. 67; 0. 54–0. 82). No significant difference in efficacy was found between parenteral PR and parenteral SbV in the new world CL (0. 88; 0. 56–1. 38). Systemic side effects were fewer with topical or parenteral PR than parenteral SbV. Topical PR with MBCL could be a therapeutic alternative to SbV in selected cases of the old world CL. Development of new formulations with better efficacy and tolerability remains to be an area of future research. More than 12 million people in 88 countries suffer from leishmaniasis, a condition caused by parasites of the genus Leishmania [1]. Annually, two million new cases of leishmaniasis are diagnosed, of which about one quarter present as visceral leishmaniasis, a potentially fatal condition. The rest present as cutaneous leishmaniasis (CL), a non-fatal yet severely disfiguring condition characterized by skin lesions and unsightly scars on the face and extremities. Over the past decade, the worldwide prevalence and geographical distribution of CL have expanded. Pentavalent antimony compounds (SbV), such as sodium stibogluconate (SB) or meglumine antimoniate (MA), have been the mainstay of the treatments [2]. Despite its efficacy, SbV is limited by high cost, poor compliance due to a prolonged course of intramuscular or intravenous injections, and potentially reversible systemic toxicity [3]–[6]. Resistance is also of particular concern [7]. Among various species causing the old world and new world CL, certain species are more likely to self-cure at a slower rate or progress to diffuse or mucocutaneous form than others [8]. Due to such clinical significance, the treatment has been mainly in the form of topical application in the old world CL and systemic in the new world CL. Seeking an alternative to SbV for localized CL has been of particular interest over the past decades. Therapeutic activity of paromomycin (synonymous with aminosidine) (PR) was first reported in the 1960' s [9], [10]. In the 1980' s, El-On et al. demonstrated therapeutic activity of PR in an in vitro study [11]. Epicutaneous administration of PR (topical PR, hereafter) with 12% MBCL (“first-generation formulation”) further showed promising results in animal [12] and human studies [13]. In the early 1990' s, MBCL was replaced with urea to reduce local side effects from MBCL (“second-generation formulation”) [14], [15]. Even though PR has also been administered parenterally [4], [5], topical PR, in particular, has several advantages over SbV, because of its fewer systemic side effects, lower cost, and convenience [3], [16], [17]. Thus, it could be a good therapeutic alternative to SbV. However, clinical trials of topical PR and parenteral PR have showed widely varying results on the efficacy and safety in treating CL. Its cure rate ranges from 4% [18] to 93% [4] and its efficacy compared with SbV has been equivocal [19]. Therefore, we performed a systematic review and meta-analysis to assess the efficacy and safety of various PR regimens as compared to placebo and SbV. We searched PubMed, Scopus, and Cochrane Central Register of Controlled Trials, with no language restriction, from inception through August 2007, to identify all randomized controlled trials evaluating the efficacy and safety of PR for CL, using the following search terms: cutaneous leishmaniasis, paromomycin, aminosidine, and randomized controlled trials. Detailed strategies are described in Appendix S1. The initial search was complemented by a manual search of the reference lists from the retrieved articles and the “Related Articles” function of PubMed. Because various types of PR and SbV regimens were tested, we tried to make clinically meaningful comparisons by pooling the data within similar groups of trials. Reports were excluded according to the following a priori criteria: 1) reviews, meta-analyses, or editorials; 2) case reports or retrospective studies; 3) animal or in-vitro studies; 4) no randomized control group; and 5) no data on efficacy or safety outcomes of PR treatment. When a study originated several reports [5], [20], the report with the largest sample size or the longest follow-up was included [5]. We further excluded two trials that compared different duration or dose of PR without placebo or SbV control group [21], [22]; one trial that compared different MA regimens as an augmentation of the same topical PR regimen [23]; two trials that randomized lesions instead of patients [18], [24] (because certain local treatments may lead to improvement in untreated lesions in the same individual [25], [26]); and four trials that compared PR with second-line treatments such as pentamidine [4], ketoconazole [27], and photodynamic therapy [24], [28]. All the retrieved reports were independently reviewed by two investigators (HJC and DHK) for eligibility and any disagreements were resolved by consensus. Two investigators (HJC and DHK) independently extracted data on participants' characteristics, predominant parasite species, interventions, and outcomes from included reports by using a standardized data collection form. When the parasite species were not reported, we assumed that it was the same as in other trials conducted in the same geographical region [17], [27], [29]. We evaluated the quality of studies using the following criteria: 1) double-blind; 2) concealment of treatment allocation; 3) blinding of outcome assessment; and 4) intention-to-treat analysis. Concealment of treatment allocation was adequate if patients and enrolling investigators could not predict assignment. Outcome assessment was blinded if the investigator who assessed the outcome had no knowledge of treatment assignment. The analysis was performed according to the intention-to-treat principle if all randomized patients were included in the analysis and kept in the originally assigned groups. If there was not enough information to assess the quality, it was assumed inadequate. Any disagreements were resolved by consensus. The main outcome was clinical cure, defined as complete healing, disappearance, or reepithelialization of all lesions. The secondary outcome was clinical improvement, defined as complete or incomplete healing or reepithelialization of the lesion or any reduction in the size. In addition, local and systemic side effects were assessed. Local side effects included pain, burning sensation, pruritus, erythema, edema, and inflammation at the administration site. Systemic side effects included myalgia, generalized symptoms (i. e. fever, malaise, weakness, and anorexia), headache, arthralgia, generalized eruptions, and laboratory abnormalities on blood counts, chemistry, and liver function tests. Trials with placebo control group were analyzed separately from trials with SbV control group. Pooled estimates and 95% confidence intervals (CIs) of risk ratios (RRs) were calculated by using an inverse-variance weighted random-effects model [30] according to the intention-to-treat principle. Between-study heterogeneity was quantified using the χ2 and I2 statistics [31]. A random-effects meta-regression analysis was performed to evaluate whether the heterogeneity among trials with placebo control group was explained by duration of lesion, length of PR treatment, and type of PR regimen. Duration of lesion and length of PR treatment have been proposed as potential explanations for inconsistent results in previous studies [27], [32]. Although different species or type of SbV regimen could have contributed to the heterogeneity, we were not able to examine such factors due to an insufficient number of trials. In addition, the influence of the study quality criteria was evaluated. We chose this approach rather than excluding trials based on a composite quality scale, because disagreement between different scales is common and valuable information may get excluded by the latter approach. We conducted sensitivity analysis by examining the relative influence of each study on the pooled estimate by excluding one study at a time. Finally, a publication bias was examined by the Begg' s test (rank correlation method) and Egger' s test (weighted regression). Statistical significance was defined as P<0. 05 and all statistical analyses were conducted with Stata/SE version 9. 2 (StataCorp, College Station, TX). Fourteen randomized controlled trials with a total of 1,221 patients satisfied our selection criteria [3]–[6], [16], [17], [28], [29], [32]–[37]. The study selection process was summarized in Figure 1. Table 1 summarizes the study characteristics. Eight trials [17], [28], [29], [32]–[34], [36], [37] were conducted in the Middle East and North Africa where CL was caused by L. major (old world CL), and six trials [3]–[6], [16], [35] were conducted in Central and South America where CL was caused by L. braziliensis, L. panamensis, and L. chagasi (new world CL). The mean age ranged from 5 to 24 years and the proportion of male varied between 42% and 100%. The average duration of lesions ranged from 15 to 105 days. Four types of PR regimen were evaluated: topical PR alone [17], [29], [32]–[36]; topical PR with MBCL [3], [6], [16], [28], [37]; topical PR with MBCL and parenteral MA [6]; and parenteral PR [4], [5]. Ointment was used for all topical formulations, except for one trial [29] where a lotion form was used. Three trials [16], [28], [37] used MBCL as a vehicle, whereas four [29], [32], [34], [35] used urea and one [36] used a paraffin and wool fat vehicle. Among SbV-controlled trials, two trials [17], [33] used intralesional regimen and four [3]–[6] used parenteral regimen. After a follow-up period of 27 to 455 days, the efficacy was assessed clinically [3], [6], [16], [17], [35] or in combination with parasitological examination [4], [5], [28], [29], [32]–[34], [36], [37]. Seven [16], [28], [29], [32], [34], [35], [37] were double-blinded and two [3], [6] were only double-blinded with respect to topical treatment. Absolute rate of clinical cure comparing PR regimen versus placebo varied: 13% to 74% versus 10% to 68% for L. major; 4% versus 3% for L. chagasi; and 82% versus 34% for L. braziliensis. Overall, any PR regimen was more effective than placebo to achieve clinical cure (RR: 1. 49; 95% CI: 1. 04–2. 13; P = 0. 031; heterogeneity χ2 = 22. 33, P = 0. 002, I2 = 69%). (Figure 2) The RRs (95% CIs) from six trials of the old world CL and from two trials of the new world CL were 1. 27 (0. 91–1. 77; P = 0. 165; heterogeneity χ2 = 11. 93, P = 0. 036, I2 = 58%) and 2. 34 (1. 48–3. 71; P<0. 001; heterogeneity χ2 = 0. 18, P = 0. 668, I2 = 0%), respectively. The meta-regression analysis suggested the type of PR regimen as a main source of heterogeneity (P = 0. 024), but neither the duration of lesion nor the length of treatment. The heterogeneity disappeared (I2 = 0%), when the data were pooled according to the type of PR regimen. Topical PR was more effective than placebo when it was combined with MBCL (2. 58; 1. 76–3. 76; P<0. 001; heterogeneity χ2 = 0. 37, P = 0. 830, I2 = 0%) compared to when it was used alone (1. 01; 0. 87–1. 18; P = 0. 867; heterogeneity χ2 = 1. 82, P = 0. 769, I2 = 0%). The results for the secondary outcome were similar (data not shown). Table 2 summarizes the existing data on efficacy of PR regimen as compared to controls. In the old world CL, the evidence only addressed the comparison between topical PR and intralesional MA. No trials compared topical PR or parenteral PR with parenteral SbV. In the new world CL, various topical or parenteral PR regimens were compared with parenteral SbV, but no trials compared any PR regimens with intralesional SbV. Data from SbV-controlled trials were pooled by the type of PR and SbV regimens. Absolute rate of clinical cure comparing various PR regimens versus SbV regimens was the following: 17% to 67% versus 42% to 60% for L. major; 59% to 93% versus 88% for L. braziliensis; and 45% to 48% versus 69% to 70% for L. panamensis. Overall, any PR regimen was less effective than any SbV regimen to achieve clinical cure (0. 77; 0. 59–0. 99; P = 0. 043; heterogeneity χ2 = 16. 08, P = 0. 007, I2 = 69%). (Figure 3) In the old world CL, the efficacy of topical PR was not significantly different from that of intralesional MA (0. 70; 0. 26–1. 89; P = 0. 480; heterogeneity χ2 = 6. 06, P = 0. 014, I2 = 84%). In the new world CL, topical PR was less effective than parenteral MA (0. 67; 0. 54–0. 82; P<0. 001; heterogeneity χ2 = 0. 03, P = 0. 856, I2 = 0%), whereas no significant difference was found between parenteral PR and parenteral SbV (0. 88; 0. 56–1. 38; P = 0. 567; heterogeneity χ2 = 3. 52, P = 0. 061, I2 = 72%). Similar results were observed for the secondary outcome (data not shown). Only a small number of trials reported extractable data on side effects. (Table 3) In general, local side effects were more common with topical treatment and systemic side effects were more common with parenteral treatment. Local reactions seem to occur more frequently when topical PR was combined with MBCL (1. 60; 0. 98–2. 61; P = 0. 061; heterogeneity χ2 = 0. 15, P = 0. 701, I2 = 0%), as compared to when topical PR was used alone (1. 07; 0. 52–2. 21; P = 0. 850; heterogeneity χ2 = 3. 36, P = 0. 339, I2 = 11%). Systemic side effects were less frequent with topical or parenteral PR as compared to parenteral SbV [3]–[5]. Laboratory data were not available for extraction; however, no significant difference was reported on blood counts, chemistry, and liver function tests between any topical PR regimen and placebo [19]–[22]. Bone marrow suppression and abnormal liver function tests were reported more often with parenteral SbV as compared to parenteral PR [5]. Concealment of treatment allocation was adequate in six trials [3], [16], [32], [34], [35], [37] and outcome assessment was blinded in five trials [3], [16], [28], [32], [34]. The intention-to-treat analysis was performed in five trials [4]–[6], [17], [35]. Trials that did not meet the study quality criteria tended to slightly exaggerate the efficacy of PR as compared to placebo or SbV. (Figure 4) In addition, the pooled estimates were not significantly changed when an individual trial was omitted. There was no evidence of publication bias based on the Begg' s test and Egger' s test (P = 0. 621 and P = 0. 126, respectively, for placebo-controlled trials; and P = 0. 348 and P = 0. 242, respectively, for SbV-controlled trials). Our analysis suggested that topical PR with MBCL showed therapeutic activity, whereas topical PR with soft paraffin or urea did not. Our finding is further supported by several lines of experimental evidence. MBCL, a quaternary ammonium compound, suppresses the growth of L. major in an in vitro model and increases cutaneous permeability of PR [11], [38]. In vivo studies suggested the synergistic action between PR and MBCL [12], [39]. In a randomized controlled study comparing topical PR with MBCL, topical PR with urea, and parenteral MA [3], Armijos et al. found a non-significant higher cure rate in 12% MBCL group than in 10% urea group (79. 3% vs. 70. 0%). But the study was underpowered for the comparison between the two topical regimens. We also found that local reactions appeared to increase when topical PR was combined with MBCL. It is not clear whether a lower MBCL concentration (i. e. , 5% vs. 12%) can reduce local reactions without compromising efficacy. El-On et al. compared 5% and 12% MBCL as an adjunct to topical PR, and found cure rates of 66. 6% and 76. 6%, respectively [37]. Severe local reactions were observed only in patients treated with 12% MBCL. However, other characteristics of topical formulas, including the composition of vehicle [40], [41] and application methods, such as occlusion [42], also play key roles in determining the efficacy. Although the type of topical PR was responsible for the heterogeneity among placebo-controlled trials, other important clinical factors, such as differences in parasite species and their clinical manifestation (i. e. self-cure rate and types of lesions), length of treatment, and duration of the lesions should be considered for several reasons. The tendency for spontaneous cure or progression to a more severe form of CL varies among the species. Even in the old world CL, spontaneous cure rate at 3 months is 60–70% for L. major, but <1% for L. tropica [8]. Moreover, L. braziliensis infection is associated with a more severe and prolonged course, a higher risk of progression to mucocutaneous form [8], and a lower self-healing rate [43]. Among placebo-controlled trials included in our meta-analysis, the clinical cure rate varied by species: up to 68% for L. major, 3% for L. chagasi, and 34% for L. braziliensis. Limited in vitro and in vivo observations also suggested that the new world CL was more refractory to PR than the old world CL [15], [44]. Although an in vitro [39] and several human studies [3], [16], [43], [45] of topical PR with MBCL demonstrated its efficacy against the new world CL, most clinicians do not use local treatments for L. braziliensis complex infection. Another important characteristic to be considered is the type of lesions. Depending on the stages of infection and species, lesions can vary from small erythema to nodular or ulcerative lesions [8]. Ulcerated lesions are typical of L. major and the new world species, whereas nodular lesions are typical of L. aethiopica and L. donovani and hyperkeratotic lesions of L. tropica. Topical agents may have better absorption in ulcerative lesions than in nodular lesions. Such differences in clinical features were not well-reflected in our analysis due to the limited data and a small number of trials. Lack of significant association between the length of treatment and its efficacy in meta-regression analysis does not necessarily exclude the benefit of a longer treatment course. In fact, none of the included trials involved a direct comparison. Asilian et al. randomly assigned patients with CL caused by L. major to either two-week or four-week PR treatment, and found a significantly better cure rate and reduced need for SbV rescue treatment in the four-week group [22]. Even a small improvement in cure rate can lead to considerable benefit to patients by avoiding serious systemic side effects by SbV treatment. In a self-healing disease like CL, the duration of lesions may have a crucial role on cure rate. Insufficient number and inadequate reporting [6], [17], [34]–[36] of included trials did not allow enough power to detect the trend, but it is possible that the efficacy of PR may diminish, as the lesions get older. An inadequate number of trials did not allow us to examine the efficacy of various PR and SbV regimens for each parasite species. No trials compared the combination of topical PR and MBCL with intralesional or parenteral SbV in the old world CL; or topical PR and intralesional SbV in the new world CL. In general, the old world CL is treated with intralesional SbV, whereas the new world CL is treated with parenteral SbV due to high risk of mucocutaneous involvement [8]. Our study suggests that topical and parenteral PR have lower side effects as compared to intralesional and parenteral SbV. A major limitation to our study is the small number of included trials. There were several comparisons that were based on only two to three trials. This increases the uncertainty of pooled estimates. Certain species, such as L. tropica, have not been examined in the included trials. These limit generalizability of our findings. In addition, overall poor quality in conducting and reporting trials was noted. El-On et al. [37] used a cross-over design which is less desirable in assessing the efficacy of a treatment in a self-limited disease, as criticized in a recent review [46]. Inadequate reporting of demographic characteristics of participants [3], [5], [6], [16], [35], [36], parasite species [3], [17], [29], duration of lesions [6], [17], [34]–[36], and quantitative data on side effects [6], [17], [18], [24], [27], [28] were very common. Standardization of study protocols has been suggested to facilitate between-study comparisons [47]. Furthermore, in our sensitivity analysis, trials that did not meet the study quality criteria tended to slightly exaggerate the efficacy of PR compared with control group. For unbiased and reliable evaluation, investigators should address appropriate quality criteria in design and conduct of trials and strictly follow the reporting standards such as CONSORT [48], [49]. Finally, publication bias cannot be excluded reliably in our meta-analysis, because of low sensitivity of the Begg' s test and Egger' s test in meta-analyses of fewer than 20 trials [50]. However, it has been reported that publication bias did not change the conclusions in most cases [51]. The main findings of our meta-analysis can be summarized as the following: 1) topical PR appears to demonstrate therapeutic activity against the old world and new world CL, with a tendency of increased local reactions, when it was combined with MBCL; 2) in the old world CL, the efficacy of topical PR is not different from that of intralesional SbV; and 3) in the new world CL, the efficacy of topical PR is inferior to that of parenteral SbV, whereas the efficacy of parenteral PR is not different from that of parenteral SbV. Although similar findings have been described in the past [8], [16], [22], [52], a valuable contribution of our meta-analysis is to provide their quantitative dimension. For clinicians, this meta-analysis confirms that the existing evidence does not support topical PR as an acceptable treatment of the new world CL. However, topical PR with MBCL could be a therapeutic alternative for selected cases of old world CL with lower risk of mucocutaneous involvement, due to its lack of serious systemic side effects. An acceptable alternative should demonstrate efficacy as well as local tolerability to ensure compliance. Sustained availability is also an issue. To this end, the efforts are currently made to develop formulations that has equivalent efficacy to that of first-generation formulations and local side effect profile similar to that of second-generation formulations [52]. For instance, a few randomized controlled trials evaluating a new topical PR-based formulation, WR 279396, compared to placebo or pure topical PR in CL caused by L. major are under way. This new formation was found to have therapeutic activity as well as cosmetic effects in an animal model [41]. Future research on topical PR in treatment of the old world CL merits addressing the following issues: examining the efficacy of various topical PR regimens in other species, such as L. tropica; comparison between topical PR with MBCL and topical PR only; evaluation of topical PR with different MBCL concentration for their efficacy and tolerability; and development of new formulations that has similar or superior efficacy and better tolerability than topical PR and MBCL.

Millions of people worldwide are suffering from cutaneous leishmaniasis that is caused by parasites of the genus Leishmania. Although pentavalent antimony compounds are the treatment of choice, their use is limited by high cost, poor compliance, and systemic toxicity. Paromomycin was developed to overcome such limitations. However, there is no consensus on its efficacy. This meta-analysis assessed the efficacy and safety of paromomycin compared with placebo and pentavalent antimony compounds. Fourteen randomized controlled trials, including 1,221 patients, met our selection criteria. Topical paromomycin appeared to have therapeutic activity against the old world and new world cutaneous leishmaniasis, with increased local reactions, when combined with methylbenzethonium chloride. Topical paromomycin was not significantly different from intralesional pentavalent antimony compounds in treating the old world form, whereas it was inferior to parenteral pentavalent antimony compounds in treating the new world form. However, a similar efficacy was found between parenteral paromomycin and pentavalent antimony compounds in treating the new world form. Fewer systemic side effects were observed with topical and parenteral paromomycin than pentavalent antimony compounds. These results suggest that topical paromomycin with methylbenzethonium chloride could be a therapeutic alternative to pentavalent antimony compounds for selected cases of the old world cutaneous leishmaniasis.

Abstract Introduction Methods Results Discussion

infectious diseases/neglected tropical diseases public health and epidemiology/infectious diseases pharmacology dermatology/skin infections

2009

Is Paromomycin an Effective and Safe Treatment against Cutaneous Leishmaniasis? A Meta-Analysis of 14 Randomized Controlled Trials

Structure-specific nucleases play crucial roles in many DNA repair pathways. They must be precisely controlled to ensure optimal repair outcomes; however, mechanisms of their regulation are not fully understood. Here, we report a fission yeast protein, Pxd1, that binds to and regulates two structure-specific nucleases: Rad16XPF-Swi10ERCC1 and Dna2-Cdc24. Strikingly, Pxd1 influences the activities of these two nucleases in opposite ways: It activates the 3′ endonuclease activity of Rad16-Swi10 but inhibits the RPA-mediated activation of the 5′ endonuclease activity of Dna2. Pxd1 is required for Rad16-Swi10 to function in single-strand annealing, mating-type switching, and the removal of Top1-DNA adducts. Meanwhile, Pxd1 attenuates DNA end resection mediated by the Rqh1-Dna2 pathway. Disabling the Dna2-inhibitory activity of Pxd1 results in enhanced use of a break-distal repeat sequence in single-strand annealing and a greater loss of genetic information. We propose that Pxd1 promotes proper DNA repair by differentially regulating two structure-specific nucleases. Structure-specific DNA nucleases contribute to the maintenance of genome stability by processing DNA secondary structures during DNA replication and repair [1], [2]. The activities of these nucleases must be tightly controlled to prevent unintended cleavage; however, the molecular mechanisms underlying the regulation of these nucleases have not been fully elucidated. The roles of several structure-specific nucleases in DNA repair are best understood in the single-strand annealing (SSA) pathway of DNA double-strand break (DSB) repair. SSA is a repair pathway for DSBs occurring between repeat sequences and has been most thoroughly studied in the budding yeast Saccharomyces cerevisiae [3]. SSA relies on the DNA resection process to generate 3′-ended single-stranded DNA (ssDNA) extending from the break to the repeat sequences [4]. Such long-range resection is mediated by two structure-specific nucleases, Exo1 and Dna2, which act in parallel to each other [5]. Upon annealing of the ssDNA of the repeat sequences, the intervening sequence between the repeats, which now becomes 3′ nonhomologous ssDNA tails, is removed by a nuclease complex Rad1-Rad10 in budding yeast (XPF-ERCC1 in mammals and Rad16-Swi10 in the fission yeast Schizosaccharomyces pombe) [6]. The function of Rad1-Rad10 in SSA requires two positive regulators, Saw1 and Slx4 [7]–[10]. Saw1 recruits Rad1-Rad10 to the DNA substrate during SSA [8], [11]; however, the exact role of Slx4 in SSA is not clear. Furthermore, it is not known whether the activities of the resection nucleases are regulated during SSA. Here we show that a novel factor Pxd1 is a key regulator of SSA in fission yeast. It interacts with both the nonhomologous ssDNA cleavage nuclease Rad16XPF and the resection nuclease Dna2, thus influencing different aspects of SSA. Interestingly, Pxd1 regulates these two structure-specific nucleases in opposite ways: it promotes the completion of SSA by activating the nuclease activity of Rad16, while it minimizes genetic information loss by inhibiting RPA-mediated Dna2 activation. A previously uncharacterized fission yeast protein, SPBC409. 16c, has been predicted by PomBase as the ortholog of budding yeast Saw1 [12]. In budding yeast, Saw1 interacts with the Rad1-Rad10 nuclease [8], [11]. In an affinity purification coupled with mass spectrometry (AP-MS) experiment, we found that Rad16 and Swi10, the fission yeast counterparts of budding yeast Rad1 and Rad10, respectively [13], [14], co-purified with SPBC409. 16c (Figure 1A), thus corroborating the PomBase orthology prediction. We will hereafter refer to SPBC409. 16c as Saw1. Intriguingly, Dna2, Cdc24, and an uncharacterized protein SPCC1322. 02 also co-purified with Saw1 (Figure 1A). Dna2 and the fission-yeast-unique protein Cdc24 are known to form a heterodimer and are both required for Okazaki fragment maturation in fission yeast [15]. When SPCC1322. 02 was used as bait for AP-MS analysis, the same six proteins were again isolated together (Figure 1B), suggesting that Rad16-Swi10-Saw1, Dna2-Cdc24, and SPCC1322. 02 co-exist in a protein complex, which we named the PXD (pombe XPF and Dna2) complex. Accordingly, we named SPCC1322. 02 Pxd1. Pxd1 is annotated by PomBase as a “sequence orphan” with no apparent orthologs outside of the fission yeast clade, and it does not contain any known domains. To identify the regions of Pxd1 that participate in its interactions with Rad16-Swi10 and Dna2-Cdc24, we performed truncation analysis and found that its interaction with Rad16-Swi10 is mediated by the middle region of Pxd1 (residues 101–233), whereas its interaction with Dna2-Cdc24 is mediated by the C-terminal region of Pxd1 (residues 227–351) (Figure 1C). Because distinct regions of Pxd1 mediate its interactions with Rad16-Swi10 and Dna2-Cdc24, we hypothesized that Pxd1 may act as a scaffold to bring these two nucleases together. We tested this idea by examining the association of the two nucleases in wild-type and pxd1Δ backgrounds. Cdc24 co-immunoprecipitated with Rad16 in the wild type, but this interaction was abolished in pxd1Δ (Figure 1D). Similarly, the interaction between Saw1 and Cdc24 was abolished in pxd1Δ (Figure 1E). These results suggest that, within the PXD complex, Pxd1 acts as a physical link between the Rad16-Swi10-Saw1 and Dna2-Cdc24 subcomplexes (Figure 1F). To determine where Pxd1 binds on its binding partners, we performed yeast two-hybrid (Y2H) assay, immunoprecipitation using truncated proteins, and cross-linking mass spectrometry (CXMS) (Figure S1). Rad16, Dna2 and Cdc24, but not Swi10, exhibited positive Y2H interactions with Pxd1. An N-terminal fragment of Rad16 (residues 1–451), which contains a helicase-like domain, was sufficient to co-immunoprecipitate Pxd1 in the absence of Swi10. CXMS analysis of a Dna2-Cdc24-Pxd1 (227–351) complex detected cross-links between the K148 residue of Cdc24 and two different residues of Pxd1 (K276 and K351). Consistently, Cdc24 (80–245), which contains the K148 residue, is the smallest fragment of Cdc24 that could robustly co-immunoprecipitate Pxd1. To understand the function of Pxd1, we generated a pxd1 deletion mutant, which exhibited no growth defect (Figure 2A). Thus, Pxd1 is unlikely to be important for the replication function of Dna2-Cdc24, which is essential for viability. We then examined the DNA damage sensitivity of deletion mutants of pxd1 and related nonessential genes. pxd1Δ showed mild sensitivity to ionizing radiation (IR) but displayed no obvious sensitivity to UV, methyl methanesulfonate (MMS), camptothecin (CPT), or hydroxyurea (HU) (Figure 2A). Consistent with the known role of Rad16-Swi10 in nucleotide excision repair (NER), rad16Δ and swi10Δ showed severe sensitivity to UV that was at a level similar to the mutant lacking another NER factor, Rhp14XPA (Figure 2A). These three mutants also showed similar sensitivity to MMS and HU. However, rad16Δ and swi10Δ were more sensitive to IR than rhp14Δ, which most likely reflected the non-NER functions of Rad16-Swi10, such as the removal of the 3′ nonhomologous ssDNA tails during homologous recombination (HR) repair [16], [17]. Surprisingly, saw1Δ displayed no sensitivity to any treatment (Figure 2A). In addition, deletion of saw1 did not enhance the DNA damage sensitivity of pxd1Δ (Figure 2B). To test the epistatic relationship between pxd1Δ, rhp14Δ, and rad16Δ, we examined the sensitivity of their single, double, and triple mutants (Figure 2C). Deletion of pxd1, rhp14, or both in rad16Δ did not enhance the IR sensitivity. In contrast, the pxd1Δ rhp14Δ double mutant showed greater IR sensitivity than either single mutant, reaching a level similar to that of rad16Δ. These results suggest that Pxd1 acts with Rad16-Swi10 in the non-NER repair of IR-induced DNA damage. To further delineate the role of Pxd1 in non-NER repair, we examined whether Pxd1 functions with Rad16-Swi10 in SSA. We constructed a strain in which an HO endonuclease-induced DSB is flanked by two direct repeats (Figure 3A). In such a system, homologous recombination between the two repeats may proceed through either the SSA or BIR mechanisms, but because the two repeats are only about 6 kb apart, SSA is expected to be the predominant pathway [18]. Regardless of which mechanism is used, two 3′ nonhomologous ssDNA tails, one 6,328 nt long and the other 29 nt long, must be removed by a nuclease such as Rad16-Swi10, resulting in the loss of the HO cleavage site and a leu1+ marker (Figure 3A). For simplicity, we will hereafter refer to this repair process as SSA. When wild-type cells harboring the SSA system were shifted from an HO repression (+ thiamine) to an HO induction condition (−thiamine) in liquid media, no obvious growth arrest was observed, but the cells became Leu− (Figure S2A and B), indicating that SSA repair was highly efficient. In contrast, when HO was induced in rad16Δ and swi10Δ cells, their proliferation was retarded for approximately 20 h, suggesting a delay of the repair process (Figure S2A). Eventually most of the rad16Δ and swi10Δ cells survived and became Leu−, most likely due to backup nuclease activities (Figure S2B). On thiamine-free solid media, the repair defect of rad16Δ and swi10Δ also manifested as a growth delay (Figure 3B). pxd1Δ cells showed the same growth delay as rad16Δ and swi10Δ cells (Figure 3B). In addition, the double mutants rad16Δ pxd1Δ and swi10Δ pxd1Δ exhibited the same phenotype as the three single mutants, indicating that Rad16-Swi10 and Pxd1 function in the same process. In this assay, saw1Δ again behaved like the wild type. Moreover, deleting saw1 in pxd1Δ did not exacerbate the phenotype. Thus, unlike its budding yeast ortholog, fission yeast Saw1 does not appear to be important for SSA. To more directly monitor SSA, we examined the elimination of the intervening DNA sequence between the repeats using qPCR (Figure 3C). The rate of DNA elimination in the pxd1Δ and swi10Δ mutants was significantly slower than in the wild type and the saw1Δ mutant (Figure 3C). In addition, we visualized Rad52 nuclear foci, which is an indication of ongoing DNA repair activity. In the wild-type and saw1Δ cells, the level of Rad52 foci transiently increased after HO induction but returned to the pre-induction level within 8 h (Figure S2C and D). In contrast, in pxd1Δ, rad16Δ, and swi10Δ cells, HO-induced Rad52 foci remained at a high level for more than 10 h. Thus, DNA repair in these three mutants failed to efficiently proceed to completion. To test whether the interaction between Pxd1 and Rad16-Swi10 is required for SSA, we examined cells expressing truncated versions of Pxd1. Pxd1 missing either its N-terminal region or C-terminal region could rescue the defect of pxd1Δ, whereas Pxd1 without the middle region failed to rescue the phenotype (Figure 3D). Thus, the region of Pxd1 involved in Rad16-Swi10 binding is required for SSA. During SSA, the role of Rad16-Swi10 is to remove the 3′ nonhomologous ssDNA tails. Given that the interaction between Pxd1 and Rad16-Swi10 is required for SSA, we hypothesized that Pxd1 is involved in the same step. To test this idea, we monitored the level of 3′ ssDNA using a qPCR assay. In this assay, the PCR template was genomic DNA pre-digested with a restriction enzyme, BstUI, that cuts double-stranded but not single-stranded DNA. Thus, the level of the PCR product reflects the amount of ssDNA (Figure 3E). In wild-type and saw1Δ cells, only a transient and small increase (approximately 10%) of ssDNA occurred after HO induction (Figure 3F). In contrast, in pxd1Δ and swi10Δ cells, ssDNA accumulated to a much higher level and persisted (Figure 3F). Thus, 3′ ssDNA removal is defective in pxd1Δ and swi10Δ, but not in saw1Δ, mutants. Rad16 (also known as Swi9) and Swi10 are required for mating-type switching, presumably due to their involvement in resolving recombination intermediates of the HR process triggered by the programmed DSB at the mating type locus (Figure S3A) [19], [20]. To test whether Pxd1 also participates in mating-type switching, we performed an iodine-staining assay on h90 homothallic strains growing on a medium compatible with mating and sporulation (Figure S3B). Dark staining indicates efficient mating-type switching, whereas light or sectored staining indicates defects in mating-type switching. Wild-type and saw1Δ h90 colonies were darkly and homogenously stained (Figure 4A). In contrast, rad16Δ and pxd1Δ colonies showed much weaker and uneven staining patterns. This result suggests that pxd1Δ, like rad16Δ, is defective in mating-type switching. Consistent with the idea that a failure of the HR process underlies the mating-type switching defect of rad16Δ and pxd1Δ, we observed using ChIP-seq that, in heterothallic h− cells, Rad52 accumulated more strongly at the mating type locus in rad16Δ and pxd1Δ than in wild-type cells (Figure 4B). In h− cells, the programmed DSB also triggers an HR process, but the mating type does not switch because only one type of donor sequence is available. When different truncated forms of Pxd1 were tested for their abilities to rescue the mating-type switching defect, the middle region-deleted version of Pxd1 failed to rescue the iodine-staining phenotype of pxd1Δ h90 colonies, suggesting that the interaction between Pxd1 and Rad16-Swi10 is important for mating-type switching (Figure 4C). Covalent Top1-DNA adducts, referred to as Top1 cleavage complexes (Top1cc), arise spontaneously and can jeopardize cell survival if not removed. It was shown recently that Rad16-Swi10 and Tdp1 redundantly remove Top1cc in fission yeast [21]. We, therefore, tested whether Pxd1 also contributes to this process. Tetrad analysis showed that, like swi10Δ, pxd1Δ is synthetic lethal/sick with tdp1Δ, and the synthetic lethality/sickness can be rescued by the deletion of top1 (Figure 4D and Figure S4A). Further analysis showed that the C-terminally truncated version, but not the middle region-deleted version, of Pxd1 could rescue the synthetic lethality/sickness (Figure 4E and Figure S4B). These results suggest that Pxd1 acts with Rad16-Swi10 in the removal of Top1cc (Figure S4C). To understand how Pxd1 acts with Rad16-Swi10, we tested whether its absence affects the nuclease activity of Rad16-Swi10 purified from fission yeast cells. For a positive control, we used a strain expressing C-terminally truncated Pxd1 as the only form of Pxd1, so that Dna2-Cdc24, which also has nuclease activities, does not co-purify with Rad16-Swi10. As described earlier, this truncated form of Pxd1 is sufficient for SSA, mating-type switching, and Top1cc removal. Consistent with the known substrate specificity of XPF family nucleases, Rad16 immunoprecipitated from such a strain showed robust nuclease activity toward 3′ overhang DNA and Y fork DNA but not 5′ overhang DNA (Figure S5A). The nuclease-dead mutant Rad16-D700A immunoprecipitated from the same Pxd1 C-terminal truncation background did not show nuclease activity toward any substrates, demonstrating that the nuclease activity we observed was Rad16-specific (Figure S5A). Rad16 immunoprecipitated from pxd1Δ cells had much weaker nuclease activity than the positive control (Figure 5A and Figure S5B). The expression level and stability of Rad16 were not affected by the loss of Pxd1 (Figure S5C). Thus, Pxd1 is required for a robust nuclease activity of Rad16-Swi10. The middle region of Pxd1 is required for its interaction with Rad16-Swi10 and is needed for SSA, mating-type switching, and the removal of Top1cc. To identify functionally important residues within this region, we mutated the residues conserved between Pxd1 and its homologs in two other fission yeast species and found that a double point mutation, A155D/E172A, significantly weakened the interaction between a recombinant Pxd1 protein purified from E. coli and Rad16-Swi10 immunoprecipitated from pxd1Δ fission yeast cells (Figure 5B). When introduced into the pxd1 gene in fission yeast, this mutation impaired 3′ ssDNA removal during SSA (Figure 5C) and diminished the nuclease activity of Rad16-Swi10 purified from the Pxd1 C-terminal truncation background (Figure 5D). These data strongly suggest that the interaction between Pxd1 and Rad16-Swi10 is needed for Pxd1 to activate Rad16-Swi10. When we added Pxd1 protein purified from E. coli to Rad16-Swi10 immunoprecipitated from pxd1Δ cells, we observed a dose-dependent enhancement of nuclease activity (Figure 5E). As a control, the A155D/E172A mutant form of Pxd1 purified from E. coli failed to activate the nuclease activity (Figure 5E). Thus, recombinant Pxd1 is sufficient for activating Rad16-Swi10. To probe the role of the interaction between Pxd1 and Dna2-Cdc24, we overexpressed a Pxd1 C-terminal fragment, Pxd1 (227–351), which encompasses the Dna2-Cdc24–interacting region. Remarkably, Pxd1 (227–351) overexpression caused severe growth defect, and this defect could be suppressed by co-overexpression of both Dna2 and Cdc24, or Dna2 alone (Figure 6A). Two mutant alleles of the gene encoding the DNA helicase Pfh1 (Pif1 homolog), pfh1-R20 and pfh1-R23, which are suppressors of temperature-sensitive mutants of dna2 and cdc24 [22], [23], also suppressed the growth defect caused by Pxd1 (227–351) overexpression (Figure S6A). Thus, the growth defect is likely due to a down-regulation of the functions of Dna2-Cdc24. To determine whether the interaction between Pxd1 and Dna2-Cdc24 is important for this down-regulation, we performed mutagenesis on the C-terminal region of Pxd1 and found that simultaneously mutating five residues conserved between Pxd1 and its homologs in two other fission yeast species, referred to as the 5A mutation, weakened the interaction between Pxd1 and Dna2-Cdc24 (Figure S6B). The overexpression of Pxd1 (227–351) -5A did not cause any growth defect (Figure 6B), indicating that the Pxd1 (227–351) overexpression phenotype is mediated by an interaction with Dna2-Cdc24. To understand how Pxd1 (227–351) down-regulates the functions of Dna2-Cdc24 when overexpressed, we investigated whether in vitro it influences the nuclease activity of Dna2-Cdc24. We found that Dna2 and Cdc24 co-overexpressed and purified from pxd1Δ cells were able to cleave a 5′ overhang DNA substrate (Figure 6C). The stability of Dna2 and Cdc24 was not affected by pxd1Δ (Figure S6C). Consistent with the results obtained with budding yeast and human Dna2 [24], [25], the addition of RPA markedly stimulated the nuclease activity of Dna2. Recombinant Pxd1 (227–351) purified from E. coli did not affect the basal activity of Dna2; however, it significantly weakened the activation effect of RPA (Figure 6C). Pxd1 (227–351) -5A failed to inhibit the RPA-mediated activation of Dna2 (Figure 6D). Thus, the interaction between Pxd1 and Dna2 impedes the activation of Dna2 by RPA. RPA can enhance the nuclease activity of Dna2 by promoting the binding of Dna2 on ssDNA in budding yeast [24]; therefore, we hypothesized that Pxd1 (227–351) may block RPA-mediated Dna2 binding to DNA substrates. To test this idea, we first investigated the ability of Pxd1 and Dna2-Cdc24 to bind a 5′ overhang DNA using a gel mobility shift assay. In this assay, DNA cleavage was prevented by using a buffer containing 1 mM EDTA and no divalent cations. Dna2-Cdc24 shifted the mobility of the DNA, whereas Pxd1 (227–351) had no effect (Figure 6E, lanes 2–5). The addition of Pxd1 (227–351) with Dna2-Cdc24 led to the formation of a complex that migrated faster than the Dna2-Cdc24-DNA complex (Figure 6E, lanes 6–8 and Figure 6F, lanes 3–5), most likely due to a higher negative charge of the Pxd1-Dna2-Cdc24-DNA complex because the recombinant Pxd1 (227–351) has a low PI of 5. 09. As a control, the addition of Pxd1 (227–351) -5A, which cannot efficiently interact with Dna2-Cdc24, had much weaker ability to shift the Dna2-Cdc24-DNA complex (Figure 6F, lanes 6–8). These results show that, consistent with the lack of effect of Pxd1 on the basal nuclease activity of Dna2, Pxd1 does not appear to affect the ability of Dna2-Cdc24 to bind naked DNA. When RPA was added to the DNA binding reaction with Dna2-Cdc24, a Dna2-Cdc24-RPA-DNA complex that migrated slower than the Dna2-Cdc24-DNA complex and the RPA-DNA complex was detected (Figure 6E, lanes 10–12). Addition of Pxd1 (227–351) interfered with the formation of this higher-order complex and resulted in a form of DNA that appeared to be bound by only RPA (Figure 6E, lanes 14–16 and Figure 6F, lanes 11–13), suggesting that Dna2-Cdc24 was dissociated from the RPA-DNA complex in the presence of Pxd1. In comparison, Pxd1 (227–351) -5A was weaker in its ability to disrupt the higher-order complex (Figure 6F, lanes 14–16). From these results, we conclude that Pxd1 inhibits the RPA-mediated activation of Dna2 by blocking the binding of Dna2-Cdc24 to RPA-coated DNA. The Dna2-inhibitory effect of Pxd1 may influence the actions of Dna2 in either DNA replication or DSB resection. Because Pxd1 is down-regulated during the S phase of the cell cycle (our unpublished observation), we hypothesized that it may mainly regulate the resection function of Dna2. During resection, Dna2 is expected to act with Rqh1, a RecQ family helicase, in a pathway parallel to Exo1 [5]; therefore, in an exo1Δ background, the residual resection activity should be Rqh1- and Dna2-dependent. Using a qPCR-based assay to monitor resection from an irreparable HO-induced DSB (Figure 7A), we found that, as reported [26], the deletion of exo1, but not rqh1, strongly reduced long-range resection (Figure 7B). No obvious difference was found between pxd1Δ and the wild type. However, deletion of pxd1 in exo1Δ partially rescued the resection defect. Thus, consistent with the results of Pxd1 (227–351) overexpression and the in vitro nuclease assay, Pxd1 appears to attenuate the Dna2- and Rqh1-mediated resection activity, at least in the exo1Δ background. Supporting this idea, the deletion of pxd1 did not rescue the DNA resection defect of rqh1Δ exo1Δ cells (Figure 7C). The DNA damage sensitivity of exo1Δ cells was not rescued by pxd1Δ (Figure S6D), probably due to Exo1 also playing nonresection roles in genome maintenance. To determine which region of Pxd1 is involved in resection inhibition, we examined the effect of introducing truncated versions of Pxd1 into an exo1Δ pxd1Δ double mutant. The N-terminal–truncated and middle-region–deleted versions curtailed long-range DNA resection as strongly as the full-length Pxd1. In contrast, a C-terminally truncated version, Pxd1-Δ (302–348), which is defective in binding Dna2, failed to impede resection (Figure 7D). These results suggest that the interaction between Pxd1 and Dna2 is required for the inhibitory effect of Pxd1 on DNA resection. In addition, the C-terminal region of Pxd1 alone can inhibit DNA resection in the exo1Δ pxd1Δ background (Figure 7D). During the SSA repair process, DNA resection is required for rendering the homologous repeats single-stranded [4], [18]. We hypothesized that the C-terminal region of Pxd1 may regulate the homologous partner choice during SSA repair when there are multiple homologous sequences on the same side of the DSB [4], [27]. To test this idea, we constructed an SSA competition system. In this system, one additional homologous sequence was inserted between the two repeats in the original SSA strain (Figure 7E). During SSA repair, the repeat sequence on the left side of the HO site can anneal with either potential homologous partner on the right side of the HO site. If partner1 is used, the postrepair cells will remain Leu+; however, if partner2 is used, cells will become Leu− and suffer a greater loss of genetic information (Figure 7E). We found that the DSB-proximal homologous sequence, partner1, was more frequently used in exo1Δ than in wild-type cells (Figure 7F), presumably because slower resection in exo1Δ cells reduces the chance of partner2 becoming single-stranded before a productive repair using partner1 has occurred. Removing the Dna2-inhibitory region of Pxd1 reversed the effect caused by exo1 deletion (Figure 7F), consistent with the rescue of the resection defect observed using the irreparable HO system. Interestingly, in an exo1+ background, the same Pxd1 truncation enhanced the use of the distal homologous sequence, partner2 (Figure 7F). These results suggest that Pxd1 restricts the use of break-distal homologous sequences during SSA repair to prevent excessive loss of genetic information. In this study, we identified a novel fission yeast protein, Pxd1, which interacts with two structure-specific nucleases, Rad16-Swi10 and Dna2-Cdc24. Our data indicate that Pxd1 can activate the 3′ nuclease activity of Rad16-Swi10, but inhibit the RPA-mediated activation of the 5′ nuclease activity of Dna2-Cdc24. These two capacities of Pxd1 allow it to promote SSA and, at the same time, reduce the negative impact of SSA on genome integrity (Figure 7G). Unlike the situations in budding yeast, in fission yeast, neither saw1Δ nor slx4Δ has an observable SSA defect (Figure 3 and Figure S5D). Among the two functionally important features of S. cerevisiae Saw1 [11], the R19 residue required for Rad1 binding is conserved in S. pombe Saw1, whereas the C-terminal positive amino acid stretch required for DNA binding is missing in S. pombe Saw1 (Figure S7). We suspect that S. pombe Saw1 may have lost its SSA-related function or become redundant. Compared with Slx4 proteins in S. cerevisiae and metazoans, S. pombe Slx4 is much shorter and appears to have lost the region required for the interaction with XPF-ERCC1 [28]. On the other hand, the middle region of Pxd1 (residues 101–233), which mediates Rad16 binding, seems to possess sequence similarity to the XPF-binding region of metazoan Slx4 proteins, which has been referred to as the MLR (MEI9XPF-interaction-Like Region) (Figure S5E) [28]–[30]. Thus, we speculate that during evolution, in the lineage leading to the fission yeast, the ancestor Slx4 protein may have split into two proteins, one becoming Pxd1 and the other evolving into the current-day S. pombe Slx4, which is solely involved in the regulation of the Slx1 nuclease [31]. In budding yeast, CDK1-mediated phosphorylation promotes the resection function of Dna2 [32]. Here we show that the resection activity of fission yeast Dna2 is subject to a negative regulation by Pxd1. Thus, Dna2 appears to be a regulatory target used in diverse organisms for controlling the resection process. Intriguingly, pxd1 C-terminal truncation caused an overt phenotype in the SSA competition assay, but pxd1 deletion did not alter resection in the irreparable HO system, suggesting the possibility that the resection process may be regulated differently depending on whether strand annealing with a homologous partner has occurred. Highly repetitive DNA elements, such as retrotransposons in yeasts and Alu elements in humans, mediate chromosome rearrangements through homologous recombination pathways including SSA [33]–[35]. The results of our SSA competition assay suggest that fine-tuning the resection activities may be a strategy that evolution has exploited to ameliorate the deleterious consequences of repeat-mediated recombination. Are there evolutionary advantages of using one protein to exert opposite controls on two nucleases? One possibility is that Pxd1 may serve as a hub to integrate regulatory signals so that the up-regulation of one nuclease and the down-regulation of the other can be more precisely coordinated. The expression level of Pxd1 appears to decrease in S phase (our unpublished observation), suggesting that cell cycle control of these two nucleases is imposed through Pxd1. Thus, the activity of Dna2 is relieved from inhibition during S phase when it is needed for DNA replication. On the other hand, given that the activation of Rad16 by Pxd1 is important for removing the 3′ nonhomologous ssDNA, the decrease of Pxd1 during S phase may curtail HR repair events involving nonhomologous ssDNA. Further analysis will be needed to assess to what extent such a regulation affects DNA repair pathway choices. The fission yeast strains used in this study are listed in Table S1, and plasmids used in this study are listed in Table S2. Genetic methods for strain construction and the composition of media are as described [36]. To construct an SSA system based on a strain in which an HO cleavage site is inserted at the arg3 locus [37], [38], we first cloned a 1. 2-kb sequence immediately upstream of the arg3 ORF between the EcoRI and ClaI sites in the integrating vector pJK148 [39], resulting in plasmid pDB169. Then, a 0. 6-kb sequence corresponding to cmb1 ORF, which is immediately downstream of arg3, was cloned into the BamHI site in pDB169, resulting in plasmid pDB174. A 0. 3-kb sequence from the intergenic region between arg3 and cmb1 was cloned between the NotI and SacII sites in pDB174, resulting in plasmid pDB176. Integration of XbaI-cut pDB176 into the HO strain DY1012 resulted in the SSA strain DY2392. For monitoring the ssDNA tail removal, a BstUI restriction site was introduced into pDB176, resulting in plasmid pDB459. Integration of pDB459 into the HO strain DY4840 resulted in the SSA strain DY5999. To create the SSA competition system, a 400-bp sequence immediately upstream of the arg3 ORF was inserted into the AatII site in pDB176, resulting in plasmid pDB1637, which was then integrated into an HO strain. Protein overexpression in S. pombe was conducted using pDUAL vectors containing the strong nmt1 promoter [40], [41]. The lysate from 50 OD600 units of cells was prepared by glass bead beating in lysis buffer A (50 mM Tris-HCl, pH 8. 0,0. 1 M NaCl, 10% glycerol, 0. 05% NP-40,1 mM PMSF, 1 mM DTT, 1× Roche Protease Inhibitor Cocktail). TAP-tagged and YFP-tagged proteins were immunoprecipitated with IgG Sepharose beads (GE healthcare) and GFP-Trap beads (Chromotek), respectively. Rad16-YFH and Swi10 were co-overexpressed in an isp6Δ psp3Δ pxd1Δ fission yeast strain. Cells were lysed using a French press in lysis buffer A. YFH-tagged protein was enriched with anti-FLAG M2 affinity gel (Sigma) and eluted with 3× FLAG peptide. Cdc24-YFH and Dna2 were co-overexpressed and purified as above. His6-tagged RPA and Pxd1 were expressed in a BL21 E. coli strain. Cells were lysed using a French press in lysis buffer B (50 mM phosphate buffer, pH 8. 0,0. 3 M NaCl, 10 mM imidazole, 10% glycerol, 1 mM PMSF), and purification was performed using Ni-NTA-agarose (QIAGEN). The eluate was dialyzed with storage buffer (50 mM Tris-HCl, pH 8. 0,0. 1 M NaCl, 10% glycerol, 1 mM DTT) before freezing at −80°C. For yeast two-hybrid analysis, we used the Matchmaker system (Clontech). Bait plasmids were constructed by inserting cDNAs into a modified pGBKT7 vector. Prey plasmids were constructed by inserting cDNAs into a modified pGAD GH vector. Bait and prey plasmids were co-transformed into the AH109 strain, and transformants were selected on the double dropout medium (SD/–Leu/–Trp). The activation of the HIS3 and ADE2 reporter genes was assessed on the quadruple dropout medium (SD/–Ade/–His/–Leu/–Trp). Dna2-Cdc24-Pxd1 (227–351) complex was prepared by incubating anti-FLAG beads bound by Cdc24-YFH and Dna2 from fission yeast with Pxd1 (227–351) from E. coli, washing the beads, and eluting with 3× FLAG peptide. About 12 µg of purified complex in a volume of 20 µl was cross-linked by BS3 or DSS at a final concentration of 0. 5 mM for 1 h at room temperature. The reactions were quenched with 20 mM NH4HCO3. Proteins were precipitated with ice-cold acetone, resuspended in 8 M urea, 100 mM Tris, pH 8. 5. After trypsin digestion, the LC-MS/MS analysis was performed on an Easy-nLC 1000 UHPLC (Thermo Fisher Scientific) coupled to a Q Exactive-Orbitrap mass spectrometer (Thermo Fisher Scientific). Peptides were loaded on a pre-column (75 µm ID, 8 cm long, packed with ODS-AQ 12 nm–10 µm beads from YMC Co. , Ltd.) and separated on an analytical column (75 µm ID, 11 cm long, packed with Luna C18 3 µm 100 Å resin from Phenomenex) using an acetonitrile gradient from 0–25% in 55 min at a flow rate of 200 nl/min. The top 10 most intense precursor ions from each full scan (resolution 70,000) were isolated for HCD MS2 (resolution 17,500; NCE 27) with a dynamic exclusion time of 60 s. Precursors with 1+, 2+, or unassigned charge states were excluded. pLink was used to identified cross-linked peptides with the cutoffs of FDR<5% and E_value<0. 001 [42]. For MMS, CPT, and HU sensitivity analysis, five-fold serial dilutions of cells were spotted onto YES with or without the indicated concentration of the chemical. To measure UV sensitivity, after spotting on YES plates, the cells were exposed to the indicated dose of UV treatment. To measure IR sensitivity, the cells were irradiated in microfuge tubes using a Cesium-137 Gammacell 1000 irradiator and then spotted onto YES. The plates were incubated for 2 or 3 d at 30°C. Genomic DNA was extracted from 3–5 OD600 units of cells collected at different times after HO induction. Five hundred nanograms of genome DNA was digested by 4 U of BstUI for 1. 5 h. The amount of amplifiable DNA was determined by qPCR, using the actin gene, act1, as the normalization control. Primer sequences are listed in Table S3. Genomic DNA was extracted from 3–5 OD600 units of cells collected at different times after HO induction. Five hundred nanograms of genome DNA was digested by 4 U of ApoI for 1. 5 h. The amount of amplifiable DNA was determined by qPCR. Primers located at different distances from the HO site were used, and their sequences are listed in Table S3. The following formula was used to calculate the percentage of DNA that was resected: %resected = (100/2ΔCt−1) /f. ΔCt is the difference in average cycles between digested template and undigested template, and f is the fraction of DNA that has been cut by HO. Oligo461 (5′-CACGCTACCGAATTCTGACTTGCTAGGACATCTTTGCCCACGTTGACCC-3′) and oligo462 (5′-GTCAGAATTCGGTAGCGTG-3′) were used to prepare the 3′ overhang DNA structure. Oligo461 and oligo463 (5′-GGGTCAACGTGGGCAAAG-3′) were used to prepare the 5′ overhang DNA structure. Oligo461 and oligo464 (5′-TCGATAGTCTCTAGATAGCATGTCCTAGCAAGTCAGAATTCGGTAGCGTG-3′) were used to prepare the Y fork DNA structure. The oligos were annealed in 1× annealing buffer (50 mM Tris-HCl, pH 7. 5,100 mM NaCl). For radiolabeled substrates, oligo461 was radiolabeled at its 5′ end. For reactions analyzed with ethidium bromide (EB) staining, 30 pmol of nonradioactive substrate was used per reaction. For reactions analyzed with autoradiography, 30 pmol of nonradioactive substrate mixed with about 50 fmol of radioactive substrate was used per reaction. Anti-TAP immunoprecipitates from 50 OD600 units of cells were incubated with substrate in 50 mM Tris-HCl, pH 7. 5,50 mM NaCl, 1 mM MnCl2,1 mM dithiothreitol, and 0. 1 mg/ml bovine serum albumin (BSA) at 30°C for 1 h. The products were separated in 15% denaturing or 10% native gels. The substrates used for denaturing gel analysis were radiolabeled, whereas the substrates for native gel analysis were not radiolabeled. The native PAGE gels were stained with EB, and the denaturing PAGE gels were analyzed by autoradiography. The reaction mixtures (20 µl) contained 50 mM Tris-HCl, pH 7. 5,50 mM NaCl, 1 mM MgCl2,1 mM dithiothreitol, 0. 1 mg/ml BSA, and 30 pmol of substrate. Reactions were carried out at 30°C for 1 h, and the products were analyzed in a 15% denaturing gel. The assay mixtures (10 µl) contained 50 mM Tris-HCl, pH 7. 5,1 mM EDTA, 1 mM dithiothreitol, 0. 1 mg/ml BSA, 50 mM NaCl, 5% glycerol, and 15 fmol of radioactive 5′ overhang DNA. The assay mixtures were incubated at room temperature for 30 min, and then 2 µl of 6× native loading buffer was added. The products were separated in a 5% PAGE gel in 1× TBE at 3 W for 2 h and analyzed by autoradiography.

Genome stability maintenance relies on DNA repair enzymes, among which are structure-specific nucleases that cleave DNA in a sequence-independent but structure-dependent manner. It is important to understand how the activities of such nucleases are controlled, because either insufficient or excessive cleavage of DNA could jeopardize genome integrity. In this study, we discovered a new regulator of two different structure-specific nucleases in the fission yeast Schizosaccharomyces pombe. The identified protein, which we named Pxd1, promotes the activity of the 3′ endonuclease Rad16, but restrains the activity of the 5′ endonuclease Dna2. In the absence of Pxd1, several Rad16-dependent DNA repair processes become defective. One of these processes is a DNA-repeat–mediated double-strand break repair pathway called single-strand annealing, which causes genomic deletions. When the Dna2-inhibitory activity of Pxd1 is impaired, Dna2-dependent end processing of double-strand breaks is enhanced and a more extensive deletion occurs during single-strand annealing. Thus, Pxd1 facilitates a potentially dangerous DNA repair process, but in the meantime minimizes its deleterious consequences. We propose that a dual-target regulator like Pxd1 is ideally suited for coordinating multiple enzymatic activities during DNA repair.

Abstract Introduction Results Discussion Materials and Methods

dna damage biochemistry genetics biology and life sciences dna dna repair dna recombination

2014

Fission Yeast Pxd1 Promotes Proper DNA Repair by Activating Rad16XPF and Inhibiting Dna2

The eukaryotic cell cycle is the repeated sequence of events that enable the division of a cell into two daughter cells. It is divided into four phases: G1, S, G2, and M. Passage through the cell cycle is strictly regulated by a molecular interaction network, which involves the periodic synthesis and destruction of cyclins that bind and activate cyclin-dependent kinases that are present in nonlimiting amounts. Cyclin-dependent kinase inhibitors contribute to cell cycle control. Budding yeast is an established model organism for cell cycle studies, and several mathematical models have been proposed for its cell cycle. An area of major relevance in cell cycle control is the G1 to S transition. In any given growth condition, it is characterized by the requirement of a specific, critical cell size, PS, to enter S phase. The molecular basis of this control is still under discussion. The authors report a mathematical model of the G1 to S network that newly takes into account nucleo/cytoplasmic localization, the role of the cyclin-dependent kinase Sic1 in facilitating nuclear import of its cognate Cdk1-Clb5, Whi5 control, and carbon source regulation of Sic1 and Sic1-containing complexes. The model was implemented by a set of ordinary differential equations that describe the temporal change of the concentration of the involved proteins and protein complexes. The model was tested by simulation in several genetic and nutritional setups and was found to be neatly consistent with experimental data. To estimate PS, the authors developed a hybrid model including a probabilistic component for firing of DNA replication origins. Sensitivity analysis of PS provides a novel relevant conclusion: PS is an emergent property of the G1 to S network that strongly depends on growth rate. During the life cycle of eukaryotic cells, DNA replication is restricted to a specific time window, called the S phase. Several control mechanisms ensure that each DNA sequence is replicated once, and only once, in the period from one cell division to the next. Following S phase, replicated chromosomes separate during mitosis (M phase) and segregate in two nuclei that eventually will be endowed to each newborn daughter cell at cell division. Two gap phases, called G1 and G2, separate cell birth from S phase and S phase from M phase, respectively. Typical pie chart representation of the cell cycle (Figure 1A) stresses the discontinuous events that have to take place only once per cell cycle (i. e. , S and M phases), but fails to show that proliferating somatic cells are continuously increasing in their mass throughout the cell cycle (Figure 1B). As pointed out as early as 1971 by Mitchinson [1], the “continuous events of the growth cycle” (i. e. , increase in cell mass) and the “discontinuous events of the DNA division cycle” (i. e. , DNA replication, mitosis, and cell division) need to be tightly coordinated in order to maintain cell size homeostasis. It has been proposed that coordination of mass accumulation with cell cycle progression relies on a sizer mechanism, so that DNA replication and/or cell division start only when cells have reached a critical cell size (see [2] for a review). In this way, tiny newborn cells will have to grow more than mother cells before being able to overcome the cell size checkpoint. Conversely, a larger cell will overcome the cell size checkpoint earlier than the “normal, average” cell. As a result, both small and large cells will stabilize cell size to the “normal, average” value (Figure 1C). Although evidence for the occurrence of a sizer mechanism (s) has been collected in different eukaryotes from unicellular microorganisms to flies to mammalian cells [3,4], cell size control has been best studied and come to be generally accepted in unicellular fungi, such as the distantly related fission and budding yeasts. In the fission yeast Schizosaccharomyces pombe, a size control is operative over mitosis in wild-type cells, while a cryptic control over S phase is revealed in wee mutants [5] or under conditions of nutritional limitation [6]. In the budding yeast S. cerevisiae, the major size control is operative over DNA replication and bud emergence (G1/S transition, often referred to as “Start” in this organism). We refer to the critical cell size required for the G1/S transition in budding yeast as PS, which represents the protein content per cell at the onset of DNA replication. Ploidy, nutrients, and growth rate modulate PS that is smaller in slow-growing cells than in fast-growing cells [2,7]. Despite extensive efforts, the molecular basis of the cell sizer mechanism controlling the G1/S transition in budding yeast has long remained elusive. While gene dosage data strongly argued for a role for Cln3 in the mechanism [8,9], physiological data gave apparently contradictory results [10–12]. We propose that a growth-dependent threshold lies at the core of the cell sizer mechanism. In its essence, a threshold entails the interplay of two molecules, one acting as an activator and a second acting as an inhibitor [13,14]. When the number of molecules of activator is low and that of inhibitor is high, the activator is below the threshold; if the activator increases with growth, the threshold is overcome when enough molecules of activator are made to exceed the inhibitor. This mechanism is schematically reported in Figure 1D under the idealized assumption that the equilibrium is totally shifted toward the formation of the complex. We propose that the activator and inhibitor molecules are, respectively, Cln3 and Far1 [14,15]. Understanding of the molecular basis of the critical cell size–controlling onset of S phase (PS) would be of great relevance for cell biology. To this end, we present here a mathematical model of the G1 to S transition that integrates the different regulatory links proposed so far. The model takes into consideration data from the literature and various models (not necessarily mathematical models) that were proposed for the G1 to S transition [2,3, 16], together with our own recent experimental results [15,17]. The model was implemented by a set of ordinary differential equations (ODEs) [18], an approach successfully used to describe the cell cycle control in budding yeast [19,20] and the cellular response of yeast to hyperosmotic shock [21]. These equations describe the temporal change of the concentrations of the involved proteins and complexes. Our model explicitly considers the localization of components in different cellular compartments (cytoplasm or nucleus) and cell growth during the G1 phase. We investigate the dynamics of the G1/S transition in various growth conditions, in several mutants, and in response to different signaling pathways. Sensitivity analysis in the form of time-dependent response coefficients [22] is used to estimate the influence of parameter values on the dynamics of key components and to investigate the relevance of the nucleo/cytoplasmic localization of Sic1 on the regulation of the G1/S transition, which has been neglected by earlier cell cycle models. To arrive at experimentally testable predictions, we developed two extended versions of the model. To compare the model output with data on budding, stochasticity in parameter values was taken into account. To estimate the PS value, we constructed a hybrid model regarding the firing of DNA replication origins where the probabilistic model uses as input the output of the deterministic model, here the nuclear concentration of Cdk1-Clb5,6. Our results newly indicate that PS is an emergent property of the network that depends on growth rate, thereby rationalizing the observed increase of PS at fast growth rates. In an effort to cope with the recently released recommendations for annotation of biochemical models [23], we first list experimental data used for designing the network structure. In Materials and Methods we describe the modeling principles [18,21] and give details on parameter estimation. The full set of equations is given in Table 1, while values of parameters are summarized in Tables 2–5. To study the G1 to S transition network, we assembled all the essential elements into a concise mathematical model that captures the logic of the underlying processes. The main goal is to represent as much complexity as possible through a small number of quantities that have direct experimental interpretation. We drew the model with CellDesigner [44], a structured diagram editor for drawing gene-regulatory and biochemical networks that are stored using the Systems Biology Markup Language (SBML), a standard for representing models of biochemical and gene-regulatory networks. The essential elements we considered are (Figure 2): (1) production and degradation of mRNAs and proteins; (2) formation of dimeric and trimeric protein complexes; (3) nucleo/cytoplasmic localization of the compounds, transport processes being described like reactions, (e. g. , converting Cdk1cyt into Cdk1nuc); (4) cell growth in terms of volume increase with proportional rate constants for growth and protein production; and hence (5) concentration changes in the nuclear and cytoplasmic compartments. We use these elements to develop a mathematical model based on ODEs that describes the dynamics of both how different molecular species transform into each other, and how they traffic between the cytoplasmic and nuclear compartments (Table 1). In its essence, the time course of the network can be summarized as follows (see paragraphs above for references). The first threshold is based on the growth-dependent cyclin Cln3, and on the Cki Far1. Cln3 concentration remains approximately constant during G1, so that its total amount in the cell increases proportional to cell mass. Given that Far1 is mostly endowed to the newborn cell at the end of the previous cycle, the cell sizer threshold will be activated when Cln3 overcomes Far1. The ensuing Cdk1-Cln3–catalyzed release of the inhibitory effect exerted by Whi5—a protein with a role similar to that of pRb in mammalian cell cycle—on SBF/MBF will promotes synthesis of Cln1,2 and Clb5,6. Far1 is degraded after priming by Cdk1-Cln1,2, thus introducing a reinforcing loop in the network that guarantees irreversibility to the transition. The second threshold Cdk1-Clb5,6/Sic1 is overcome when Sic1 gets degraded following Cdk1-Cln1,2–primed phosphorylation. The model was constrained to fit the observed experimental behavior of the G1 to S transition of small, elutriated G1 cells growing on glucose [15,17]. First, we constrained features that play important regulatory roles in this phase of the cell cycle, such as the Cln3/Far1 threshold that controls the release of the SBF and MBF transcription factors from the Whi5 transcriptional repressor, and the complex formation of Sic1 with Cdk1-Clb5,6. To simulate cell growth at a low growth rate (in ethanol medium), we used both parameters' values derived from experimental data (such as the initial levels of Cln3 and Far1 and the growth rate), and a parameter (the binding constant of Sic1 for the Cdk1-Clb5,6 complex) inferred from the experimental dynamics of Clb5 and Sic1 as indicated in the following. A sensitivity analysis was performed for the fine-tuning of the parameters used to simulate the biochemical network. Then, we tested the mathematical model with the dynamics of (1) a large number of mutants, implemented by overexpression or deletion of key regulatory genes; (2) time course of budding obtained both in glucose and in ethanol media; and (3) estimation of the critical cell size PS for wild-type cells grown in both media. Finally, a sensitivity analysis indicates that PS is an emergent property of the G1 to S network. Figure 3 displays the simulated time courses of the concentration of several cell cycle players in small newborn cells growing in glucose using the following input parameters: growth rate characteristic of glucose; cell size; and Cln3 and Far1 levels as detected in small, elutriated cells grown in the same medium [15]. Since we are simulating the G1/S transition and not a full cell cycle with all relevant proteins, time courses of some variables become meaningless after overcoming the second threshold. At the beginning, the Far1, Cln3, and Cdk1 present in the cytoplasm are imported in the nucleus. The Cdk1-Cln3-Far1 complex reaches its maximal value in the nucleus after 30 min, and then it starts to be degraded upon the overcoming of the Cln3/Far1 threshold. Such degradation is mostly dependent on the execution of the first threshold, being strongly delayed in simulated cln3Δ cells (unpublished data). Cdk1-Cln3 starts to build up in the nucleus, the major factor driving accumulation of Cdk1-Cln3 being Cdk1-Cln–primed Far1 degradation (Figure 3A; T1), since little if any Cdk1-Cln3 complex forms in cln1,2Δ cells (unpublished data). At about 50 min, the accumulation of active SBF/MBF reaches its half-maximal value (Figure 3B). Cln and Clb cyclins are produced (Figure 3C), and Cdk1-Cln1,2 is accumulated both in cytoplasm and in nucleus (Figure 3D), while Cdk1-Clb5,6 accumulates preferentially in the nucleus (Figure 3E and 3F). The half-maximal value of Cdk1-Clb5,6 in the nucleus is reached at around 80 min (Figure 3F; T2). Thus, T1 and T2 represent the times when the first and second thresholds, respectively, are overcome. The coherence of this timing with experimental dynamics of cell cycle is discussed below. All data presented above refer to simulation of an idealized single cell. By taking into account biological variability, it is possible to simulate a cohort of synchronous cells that more closely resembles, for instance, a population of newborn elutriated cells placed to grow in a fresh medium. To this end, we modeled a population by a probabilistic approach that simulates cell-to-cell variability through repeated simulations with noisy parameters (see Materials and Methods for further details) (i. e. , all parameters were sampled from a normal distribution with the original model values [Tables 2 and 3] as mean value). Figure 7 shows representative time courses for the Cdk1-Cln1,2cyt and Cdk1-Clb5,6nuc complexes for cells growing in glucose (Figure 7A and 7B) and in ethanol media (Figure 7D and 7E). The simulated time course of budding obtained for a population growing in glucose (black dots) is very closed to the experimentally observed one (gray curve), both in the time required for the onset of budding and in the initial slope (Figure 7C). The same simulation run for ethanol-grown cells (Figure 7F) indicates a close agreement in the timing of the onset of budding and a satisfactory slope for three subsequent points. Although the maximum value of experimental budding is not reached in both simulations, we observe a fairly good correspondence between experimental and simulated behaviors. Taken together, these data show that the model correctly predicts properties of the cells related to the G1 to S transition that have not been taken into account during model construction, thereby offering support to the overall consistency between input data and output performance. Cell viability requires the coordination between cell growth and cell division, which in budding yeast is achieved by the attainment of a nutritionally modulated critical cell size (PS) to trigger budding and DNA replication [2,16]. Since the present model monitors cell growth, it should be possible in principle to estimate PS. To do so, we need to simulate the onset of DNA replication, since operationally PS is defined as the protein content of cells that enter S phase [52]. Therefore we correlate one of the outputs of the model, namely nuclear concentration of Cdk1-Clb5,6, with the onset of DNA replication. To this end, we constructed a hybrid model that uses the time course of Cdk1-Clb5,6nuc as input to a probabilistic model. Given the large number of DNA replication origins present in a yeast nucleus [42,43], and the reported role of the Cdk1-Clb5,6 complex in inducing firing [39], the probability of firing for each DNA replication origin could then be related to the nuclear concentration of the Cdk1-Clb5,6 complex, as explained in more detail in Materials and Methods. We performed simulations of the onset of DNA replication for cells grown in glucose and in ethanol media (Figure 8). In glucose (Figure 8A), activation of DNA replication origins takes place in a coordinated fashion roughly within a period between 70–90 min, in agreement with experimental data. In ethanol (Figure 8B), the Cdk1-Clb5,6 complex should be inefficiently imported into the nucleus, resulting in a longer S phase. This behavior agrees with reported data showing that a very poor carbon source such as ethanol [53], or a nitrogen source limitation, yield elongation of the S phase [54,55]. At this point, we could estimate PS as the cell size when 50% of replication origins were activated in a single cell. The values of PS obtained for various conditions and/or mutants, shown in Table S4 compare well with the data present in literature. In fact, overexpression of FAR1 in glucose-grown cells results in about a doubling of Far1 level and in a modest increase in cell size and PS, while overexpression in ethanol-grown cells yields a larger increase in both Far1 level and cell size (PS) [15]. Appropriate simulations of the modulation of FAR1 overexpression yield PS values that are very consistent with our experimental data [15]. Notably, moderate overexpression of FAR1 in a situation simulating growth in glucose results in a minor increase of PS (Table S4). Having obtained a good agreement between predicted and experimental PS under a set of conditions, we moved to analyze the effects of various parameters of the model on the setting of PS. Sensitivity analysis (Figure 9A) shows that several parameters affect its value: the initial level of Cln3 and Far1, the binding value of Sic1 to the Cdk1-Clb5,6 complex, and the growth rate. These results allow us to reconcile different, apparently conflicting, experimental results that pointed either to the increase of Cln3 concentration in the nucleus, or to the activation of transcription factors SBF/MBF, or to the regulation of Sic1 as the event setting the critical cell size. Besides, they highlight the significant role of the growth rate in determining the PS value, a particularly noteworthy result that could not be obtained by visual inspection of the model reported in Figure 2. First of all, our model considers that the control over the entrance into S phase is distributed over two sequential thresholds that involve Cdk1, cyclins, and two distinct inhibitors: the Ckis Far1, in the first threshold, and Sic1, which acts on the second threshold. While Sic1 has long been recognized as regulating initiation of DNA replication, the involvement of Far1 in the control of the Cdk1-Cln3 activity is not present in earlier models [19,20]. A role for an inhibitory molecule (either a Cki or a phosphatase) in setting of PS has been proposed independently by different groups [10,14]. Such an inhibitor was expected to present a peak in late M phase, a basal, low synthesis during the other phases of the cycle, and have a Cdk1-Cln–dependent degradation [10]. This pattern, initially suggested for Sic1, fits well also for Far1 [31,67]. Since inhibition of Cln3-Cdk1 by Far1 is a major innovative feature of the network proposed in this paper, let us briefly review relevant experimental evidence about this item in addition to that reported in the section describing the construction of the model, regarding physical interaction, biochemical activity, and genetic interaction. Direct identification of the Cdk1-Cln3-Far1 complex in mitotic cells has not yet been reported. The lack of genome-wide two-hybrid or mass spectrometry data to support the interaction of Far1 and Cdk1-Cln3 is not surprising, since the high unreliability of these data is well-known. In fact, comparison between two-hybrid datasets (obtained independently from two laboratories) and mass spectrometry data (also obtained by two different groups) showed a surprisingly small overlap: about 10% and 14%, respectively (as reviewed by Ito et al. [68]). Moreover, false signals have been estimated to be as high as 50% [69]. On the other hand, in α-factor–treated cells, coimmunoprecipitation experiments showed that Far1 became a sizable co-precipitated substrate of Cdk1-Cln3 [70], and biochemical assays indicate that Far1 inhibits the activity of Cdk1-Cln3 in immunoprecipitates after α-factor treatment [71]. Since activation of Fus3 protein kinase determined by pheromone treatment promotes increased interaction of Far1 to Cdk1-Cln3, it would be reasonable to hypothesize that a basal binding between Cdk1-Cln3 and Far1 is present also in the absence of α-factor. It must be stressed that neither the report of Tyers and Futcher [70] nor that of Jeoung et al. [71] were designed to address Cdk1-Cln3-Far1 interaction in mitotic cells, but only in the presence of pheromone. More compelling experimental evidence for the involvement of Far1 in the G1/S transition of mitotic cycles has recently been accumulated. By analyzing mutants in CDC48, a cell cycle gene essential in the protein degradation pathway that acts as an ubiquitin-selective chaperone, Fu et al. [29] showed that blocking of Far1 degradation results in cell cycle arrest in G1 in mitotic cells. Alberghina et al. [15] independently showed that both FAR1 deletion and its overexpression affect the critical cell size PS as expected for a Cln3 inhibitor, and that deletion of either CLN3 or FAR1 in a sic1Δ background has an almost identical effect on nutritional modulation of cell size and PS. These findings are consistent with the notion that Far1 is acting in the same pathway as Cln3 with regard to nutritional modulation of cell size. To prove more formally that Far1 is acting on Cln3 and not on the closely related Cln1,2 cyclins, the effect of Far1 overexpression on the cell size of cln3Δ cells in exponential growth on ethanol-supplemented medium—chosen because these are the growth conditions in which the effect of Far1 overexpression on cell size is larger [15]—was tested and compared with that of wild-type cells. Figure 10 shows that, as previously reported [15], in wild-type cells overexpression of FAR1 leads to a large increase in cell size (66%). Such an increase is highly statistically significant (Student t-test, p < 0. 01). The increase in cell size is much lower (∼15%) in cells devoid of the CLN3 gene. Such an increase in cell size is at the borderline of statistical significance (Student t-test, p = 0. 09). Put in another way, ∼80% of the increase in size brought about by Far1 overexpression is lost in the absence of the CLN3 gene. These data appear even more relevant when one takes into account that the amount of Cln3 in growing yeast cells (which, according to these newly presented data, contributes to the majority—and possibly all—of the effect of Far1 on cell size) is much lower than the amount of Cln1,2 protein [24–27]. Taken together, literature results independently obtained by several laboratories and experimental findings newly presented here indicate that the assumption made in our model (i. e. , that Far1 plays a role in the control of entrance into S phase of mitotic cells by inhibiting Cdk1-Cln3), has a firm base, although our understanding of its mode of action is not yet complete. The effect that the introduction of the Cln3/Far1 threshold has on cell cycle dynamics must be stressed: it allows the cells to detect the reaching of a given size (set by the amount of Far1 present in a cell) when the amount of Cln3, which increases in proportion to cell mass, overcomes Far1. The actual critical cell size is reached when the cells overcome the second threshold dependent on the Cki Sic1. Recent evidence showed that alterations in nucleo/cytoplasmic localization of cyclins [28] and Ckis [17] play a relevant regulatory role in cell cycle progression. Nuclear localization of Cdk1-Cln3 and Cdk1-Clb5 complexes is assumed throughout recent papers by Tyson and coworkers [19,66], but is in fact modeled in a simplified way by multiplying cyclin synthesis rates with the parameter mass. This implementation gives different patterns of accumulation to cyclins as compared with other components. However, compartmentalization involves much more than that, since: (1) proteins are synthesized in the cytoplasm, starting from mRNA migration from the nucleus; (2) both import and export can be independently regulated; and (3) controlled partitioning affects binding equilibria by altering the actual concentration of a given protein available for binding to a given interactor within a subcellular compartment. None of these regulatory issues is addressed in the models of Chen et al. [19,20]. On the contrary, shuttling of proteins and complexes in and out of the nucleus is a major, explicitly modeled feature of our network that allows us to address biological significance of subcompartmentalization of biochemical reactions, avoiding inconsistencies intrinsically present in the Chen et al. model, where “nuclear” Clb proteins are free to interact with “cytoplasmic” Sic1, thus violating a major biochemical consequence of subcellular compartmentalization. It is relevant at this point to mention that it is the explicit modeling of nuclear compartmentalization that allows us to make predictions regarding biochemical properties of the Cdk1-Clb5,6-Sic1 complexes (discussed in the Results section), which have to be verified experimentally in the iterative procedure characteristic of systems biology [72–74]. The suggestion that comes from our simulation analysis indicates that Sic1 should have a higher binding affinity for the complex Cdk1-Clb5,6 in cells grown in glucose in comparison with cells grown in ethanol, and therefore stresses the interest in analyzing the phospho-signature of Sic1 in two conditions. Although we do not want at this stage to be bound to any specific hypothesis, it is also worth remembering that we have shown that a Sic1-derived peptide has a dramatically increased affinity towards a heterologous Cdk-cyclin complex in the CK2-phosphorylated versus the unphosphorylated state [41], as CK2 is a protein kinase whose activity is found to be higher in fast-proliferating cells [75]. The trigger that controls the activation of transcription factors SBF/MBF in Chen et al. [19,20] is modeled according to a zero-order ultrasensitivity switch [76,77], a function that was chosen because it is suitable for the phosphorylation/dephosphorylation reaction assumed to control SBF/MBF activation [19,20]. It was assumed that each transcription factor (both SBF and MBF) exists either in an active or inactive form, and that the transition between the two forms is catalyzed by two competing enzymes: a protein kinase (Cdk1-Cln3) and a protein phosphatase (whose existence was only supposed at the time of the construction of the model), each of which follows Michaelis–Menten kinetics. The transitions are taken to be fast enough to maintain each transcription factor in a steady-state distribution. The kinetic equation describing the behavior of the Goldbeter and Koshland ultrasensitive switch [56] assumes a sharply sigmoidal function when the cell grows through the critical size. The value of cell mass that triggers the SBF/MBF switch has been calculated from a quadratic equation containing half a dozen variables, and the solution has been set to 1. 2 (1 being the cell mass of the newborn cell), the large margin of error being the result of a determination that requires the estimation of a sizable number of kinetic and efficiency parameters. Subsequent experimental work has shown that the activation of SBF/MBF is not due to direct phosphorylation of transcription factors, with a delicate equilibrium between a kinase (Cdk1-Cln3) and a phosphatase (never found) as assumed by the very basic assumptions of the Chen et al. model [20]. Instead, as we modeled, the activation of SBF/MBF is due to the dissociation of an inhibitor (Whi5) that is phosphorylated by Cdk1-Cln3 [32]. The interest of this regulatory link is further stressed by the fact that Whi5 has the same role as the retinoblastoma protein in the control of cell cycle progression in mammalian cells [33]. The predictive ability of the Chen et al. [20] model is due to the built-in control for the onset of S phase given by the estimation of the critical cell size at 1. 2 that fortunately is very close to the experimental one for cells growing in glucose. In fact, it has been common knowledge for many years, derived from quantitative microscopic observations, that cells growing in batch in glucose medium at the onset of budding are only slightly larger than newborn daughter cells [78,79]. A double-tag flow cytometric analysis performed in one of our laboratories on chemostat-grown cells at variable glucose dilution rates has clarified that the ratio of critical cell size at the onset of S phase/cell size of newborn daughters is about 1. 5 at low-growth rates, then decreasing to 1. 24 at fast-growth rates [52]. Therefore, the value 1. 2 is quite close to the actual one if we consider cell growing in glucose, but it should not be used to simulate cell cycles at low-growth rates. Based on recent biochemical and genetic evidence [15,29], here we show that a simpler, better-defined mechanism (i. e. , the Cln3/Far1 threshold) brings about a switch-like accumulation of nuclear SBF and MBF transcription factors not at a prefixed cell mass, but as a result of the dynamics of the regulatory molecules (see Figure 3B), thus effectively coupling cell growth to the onset of DNA replication and budding. The ensuing Cdk1-Cln3 activity phosphorylates Whi5 [32,33], and leads to the activation of transcription factors SBF/MBF, opening up the pathway that leads to the onset of DNA replication. The requirement of a critical cell size to enter into S phase has been known for several decades, but its molecular basis is still under discussion. Cln3 is certainly involved in the setting of the critical cell size, but does not work alone in the mechanism. In fact, while cells growing at faster growth rates are larger than those growing at slower rates [12], in any given medium, cells overexpressing Cln3 are smaller and have a shorter G1 phase than wild-type cells [2,9], a paradox pointed out by Heideman and collaborators [11]. Results of the sensitivity analysis summarized in Figure 9A neatly show that PS does not originate from the properties of a single molecule, but rather is an emergent property of the network structure. Besides the Cln3 and Far1 dosage, the growth rate is a major factor in setting PS. The role of the growth rate in determining the value of PS can be explained as follows. The value of the cell size at the traverse of the first cell sizer threshold is quite similar both in rich and in poor carbon sources (Figure 9B), since the Cln3/Far1 ratio remains almost equimolecular at the various growth rates [15], with both Cln3 [11] and Far1 [15] increasing at faster growth rates. As shown before, a sizable period of time is needed to move from the first to the second threshold both in glucose and in ethanol. Since it is the traverse of the second threshold that actually sets PS, its value clearly will be much larger in cells that grow faster (Figure 9B). We use the term “emergent property” with true systems biology significance: a property that individual components of the G1 to S network do not have but that emerges from their interaction [80]. Of course, since the setting of PS is an emergent property of the entire G1 to S network, other parameters—such as the accumulation rate of Cln3 and/or Far1, or the binding affinity of Sic1 to the Cdk-cyclin complexes—affect PS independently from the growth rate and may become relevant in appropriate growth conditions. In the current Tyson model (s), the onset of DNA replication is controlled by a single event: a cell sizer is taken to be operative only at low-growth rates, while an oscillator mechanism is assumed to be active at fast-growth rates [66]. In our model, a sizer mechanism is operative at all growth rates, and the presence of two distinct—temporally spaced—thresholds acting together to set PS not only introduces a delay, but also makes the delay sensitive to the growth rate (Figure 9B). Taken together, the results presented in this paper offer an example of the usefulness of a systems biology approach for the understanding of complex biological processes. The model comprises equations for production and degradation of mRNA and proteins and for the formation of dimeric and trimeric protein complexes (Figure 2). It accounts for the nucleo/cytoplasmic localization of the compounds. Transport processes are described like reactions, converting, for example, Cdk1cyt into Cdk1nuc. Cell growth is characterized as exponential increase in volume. All concentration changes are dependent on the volume changes of the respective compartment. The full set of equations and parameters can be found in Tables 1–5, and the impact of the chosen parameters on behavior of the simulated system was studied using sensitivity analysis [22]. We explicitly consider two compartments, the nucleus and the cytoplasm, with volumes Vnuc and Vcyt, respectively. Nuclear and cytoplasmic compounds are denoted by the subscripts nuc or cyt, respectively. The dynamics of the concentration of every compound is determined by three different types of processes: (1) biochemical reactions, (2) transport over the nuclear membrane, and (3) change of the volumes. The dynamics are described by sets of ODEs. The temporal evolution of a nuclear compound reads and the temporal evolution of a cytoplasmic compound is given by for i = 1, . . , m, where m is the number of biochemical species with the concentrations ci (either nuclear or cytoplasmic). The quantity r is the number of biochemical reactions with the rates vj, and t1 and t2 are the number of transport steps from cytoplasm to nucleus and vice versa with the rates wj. The quantities nij denote the stoichiometric coefficients of the compounds in the respective reactions or transport steps. The volume changes are given by and where kV, cyt and kV, nuc are the rate constants of volume change. All individual reaction rates vj and transport rates wj are governed by irreversible mass action kinetics: and where kj are the rate constants and the index x runs over all substrates and modifiers of reaction j, and the index y denotes the compound transported in step j. Rate constants for production. The model comprises 54 rate constants. The only relevant transcriptional control implemented in the model relates to transcription of CLN1,2 and CLB5,6 genes, which is dependent on SBF/MBF. The rate for CLB5,6 transcription (k2) was estimated from kinetics of Clb5 production in elutriated cells (this article). The same value was used for CLN1,2 transcription (k1). The rates of translation (k4 and k3, respectively) were set one order of magnitude higher. Rate constants for production of Far1 (k5) and Cln3 (k6) were estimated from kinetics of production of the appropriate protein in elutriated cells growing in glucose and in ethanol medium [15]. Basal level of production of Whi5 (k8) and Sic1 (k9) was set the same as k5. Since the Cdk1 level is never limiting, the rate constant for Cdk1 production (k7) was set much higher than those for Cdk-regulatory proteins. The Cln3-independent production of SBF/MBF (k35), implemented to mimic the contribution of Bck2, was set low. Rate constants for degradation. In the absence of other relevant information, rate constants for degradation of proteins were set to the same value, 0. 01 min−1. Rate constants for degradation of mRNAs were set one order of magnitude higher. The rate constants for degradation of Cln2 and Clb5, k12 and k13, respectively, were derived from previous cell cycle models [19,20]. The rate constant for degradation of cytoplasmic Cdk1 was derived assuming a steady partitioning of Cdk1 between nucleus and cytoplasm, while degradation of nuclear Cdk1 (k21) was considered negligible and set to 0. Rate constants for association and dissociation of protein complexes. Values of association and dissociation constants for proteins involved in Cdk-cyclin-Cki complexes formation were not available. Rate constants derived from BIAcore (http: //www. biacore. com) data regarding interaction between Sic1 and heterologous Cdk-cyclin [40,41] were thus used, assuming that regardless of the actual Cdk, cyclin, or Cki involved, relative scale of interaction affinity should be conserved. The same association value was then used for the association constant between SBF and Whi5 (k34). These values guided us also in choosing the value for dissociation of Cdk-cyclin complexes (k25, k27, k29) and of these complexes containing Cdk inhibitors (k31, k33). Rate constants for phosphorylation reactions. Michaelis–Menten kinetics for phosphorylation reactions were simplified to mass action (i. e. , substrate concentration was considered low compared with Km). Such simplification did not significantly affect simulation results (unpublished data). The resulting constants are thus near to the kcat/Km ratio. Data for kcat and Km for typical Cdk-catalyzed phosphorylation reactions [81] were thus used. Rate constants for transport between nucleus and cytoplasm. In the absence of more specific experimental information regarding “in vivo” constants, rate constants for nuclear import of single proteins (Far1, Cln3, Cdk1, and Whi5) were set equal (k42–k45). The rate constants for nuclear export were then set to favor nuclear localization. Similarly, nuclear localization was favored for the Cdk1-Cln1,2 binary complex. According to experimental data indicating that Sic1 promotes Clb5 nuclear localization [17], nuclear import of the ternary complex Cdk1-Clb5,6-Sic1 (k47) was favored over transport of the cognate binary complex (k48). Rate constants for nuclear export of CLN1,2 and CLB5,6 mRNAs (k50 and k51, respectively) were set larger than those of corresponding proteins. Rate constants of exponential growth. Values for glucose- and ethanol-grown cells were obtained by averaging literature data [15] and unpublished data from our laboratory. Absolute and relative initial concentration of protein and protein complexes. The average number of molecules in glucose-grown cells was taken from Ghaemmaghami et al. [25], which was related to the maximal values of that compound assumed during time course. The Far1/Cln3 ratio in newborn cells was taken from Alberghina et al. [15]. Sensitivity analysis. Sensitivity analysis was performed to test the influence of the parameter choice on the systems dynamics (Figure S1). To this end, we calculated the time-dependent response coefficients [22], defined as These coefficients indicate the direction and amount of change of the time course for the concentration c (t) upon an infinitesimal change of the parameter (or initial concentration) p. Loosely spoken, one can also interpret this as the percentage change of the concentration over time upon a 1% change of the parameter. During model development, the response coefficients were used to indicate appropriate parameter changes, since there are not enough data available to estimate the parameters by a global approach. Probabilistic model for firing of the DNA replication origins. The influence of Cdk1-Clb5,6nuc on the licensing of replication origins and DNA replication is described by a probabilistic three-step model that does not regard molecular details of this highly complex process. Step 1 lumps all events from free origin to pre-replicative complex. The transition time for each of the 440 replication origins is taken from a normal distribution with mean of 15 min and standard deviation of 2 min. Step 2 is Cdk1-Clb5,6nuc dependent. The probability for performing Step 2 at a certain time is determined by the concentration of Cdk1-Clb5,6nuc at that time. The period of this step is necessary for Cdk1-Clb5,6 to exceed a value taken from a normal distribution with a mean of 0. 03 μM and standard deviation of 0. 01 μM. The transition time for Step 3 to reach the “fired” state is again taken from a normal distribution with a mean of 1 min and standard deviation of 0. 01 min. When the origin has fired, then DNA replication proceeds bidirectionally from multiple replication origins, as experimentally reported [82,83]. If replication reaches the neighboring origin before it fires on its own, that origin is set to the state “fired. ” The distance between DNA replication origins is fixed. We considered that in ethanol-growing cells—with a growth rate about 2-fold lower compared with the glucose-growing cells—the fork rate is about one-half (and the time of origins activation is doubling) than the glucose ones. This assumption agrees with the reported data, in which the longer S phase in yeast cells growing in poor nitrogen medium can be accounted for by a reduction in replication fork rate [55]. Probabilistic model for budding. Assuming that the parameters for the individual cells may vary around their mean values as given in the parameter list in Tables 2–5 (standard deviation of ki*0. 287), we can simulate the time courses of Cdk1-Cln1,2cyt for cell populations. Onset of budding is determined by the time tmax/2 when Cdk1-Cln1,2cyt reaches a critical value equal to the half-maximal concentration obtained for mean parameter values. As time for the appearance of the bud, we use tbudding = tmax/2 (Cdk1Cln1,2cyt) + tdelay, with tdelay = 0 min in glucose and tdelay = 120 min in ethanol. Yeast strains, cell growth, and cell size determination. Yeast cells W303-CF (cln3: : KAN1, pCLN3–15Myc, FAR1–15Myc-URA3), FAR1tet (cln3: : KAN1, pCLN3–15Myc, far1: : HIS3, pTet-FAR1–15Myc), and cln3Δ (cln3: : KAN1) were grown in synthetic complete media supplemented with ethanol as a carbon source. Growth conditions, media, and flow cytometry determinations were conducted as described previously [15]. Simulation tool. Numerical calculations have been carried out using Mathematica Version 5. 1 (Wolfram Research, http: //www. wolfram. com).

A major property of living cells is their ability to maintain mass homeostasis throughout cell divisions. It has been proposed that in order to achieve such homeostasis, some critical event (s) in the cell cycle will take place only when the cell has grown beyond a critical cell size. In the budding yeast Saccharomyces cerevisiae, a widely used model for the study of the eukaryotic cell cycle, a large body of evidence indicates that cells have to reach a critical size before they start to replicate their DNA and to form bud, which will give rise to the daughter cell. This critical cell size is modulated by growth rate, hence by nutritional conditions and the multiplicity of genetic material (i. e. , ploidy). The authors present a mathematical model of the regulatory molecular network acting at the G1 to S transition. The major novel features of this model compared with previous models of this process are (1) the accounting for cell growth (i. e. , the increase in cell volume); (2) the explicit consideration of the fact that cells have a nucleus and a cytoplasm, and that key cell cycle regulatory molecules must move between these different compartments and can only react or regulate each other if they are in the same compartment; and (3) the requirement of sequential overcoming of two molecular thresholds given by a cyclin-dependent kinase/cyclin and a cyclin-dependent kinase inhibitor. The model was tested by simulating the processes during G1 to S transition for different growth conditions or for different mutants and by comparing the results with experimental data. A parameter sensitivity analysis (i. e. , testing the model predictions when parameters are varied), newly indicates that the critical cell size is an emergent property of the G1 to S network. The model leads to a unified interpretation of seemingly disparate experimental observations and makes predictions to be experimentally verified.

Abstract Introduction Results Discussion Materials and Methods

biochemistry cell biology yeast and fungi microbiology computational biology eukaryotes genetics and genomics saccharomyces

2007

Cell Size at S Phase Initiation: An Emergent Property of the G1/S Network

Melanization, an important insect defense mechanism, is mediated by clip-domain serine protease (cSP) cascades and is regulated by serpins. Here we show that proteolytic activation of prophenoloxidase (PPO) and PO-catalyzed melanization kill the baculovirus in vitro. Our quantitative proteomics and biochemical experiments revealed that baculovirus infection of the cotton bollworm, Helicoverpa armigera, reduced levels of most cascade members in the host hemolymph and PO activity. By contrast, serpin-9 and serpin-5 were sequentially upregulated after the viral infection. The H. armigera serpin-5 and serpin-9 regulate melanization by directly inhibiting their target proteases cSP4 and cSP6, respectively and cSP6 activates PPO purified from hemolymph. Furthermore, serpin-5/9-depleted insects exhibited high PO activities and showed resistance to baculovirus infection. Together, our results characterize a part of the melanization cascade in H. armigera, and suggest that natural insect virus baculovirus has evolved a distinct strategy to suppress the host immune system. The cotton bollworm, Helicoverpa armigera, is a major pest of cotton and one of the most destructive and polyphagous pest species. It is widespread in central and southern Europe, temperate Asia, Africa, Australia and Oceania, and Brazil [1]. Moreover, the rapid spread of H. armigera through South America into Central America has threatened agriculture in North America and an immense quantity of crops are exposed to H. armigera each year [2]. Due to the broad-spectrum pesticide resistance developed in H. armigera, traditional chemical insecticides do not protect these crops from significant production impact [2] and transgenic crops also lead to resistance development in pests [3], highlighting the urgent need to develop effective biopesticide alternatives to control pests. H. armigera nucleopolyhedrovirus (HearNPV), a natural enemy of H. armigera, has increasingly been used as a biological insecticide for pest control in recent decades [4,5]. Before the baculovirus kills the pest insects, it must invade and overcome the host immune responses. However, molecular interactions between the virus and its host’s immune system are poorly understood. Viruses are a major burden for invertebrate organisms. Besides being used as biological agents against pest insects, these obligate intracellular pathogens have an important impact on economically beneficial insects [6]. However, the immunity to viral infection is still poorly understood in invertebrates. An important defense against viruses in invertebrates is RNA interference (RNAi), which can efficiently degrade virally transcribed RNA [7]. On the other hand, two additional cellular mechanisms, apoptosis and autophagy restrict viral replication and dissemination [6]. Moreover, several studies based on invertebrate larvae, have suggested that melanization is involved in antiviral immunity. Melanization is an important defense mechanism in arthropods and plays an essential role in wound healing and innate immunity [8]. In the mosquito Aedes aegypti, phenoloxidase (PO) activity is required for defense against the Semliki Forest virus [9]. In the shrimp Penaeus monodon, the white spot syndrome virus (WSSV) has evolved a specific strategy to counteract the host’s melanization response [10]. Studies in a few insects revealed that melanization is initiated by the recognition of microbial elicitors such as β-1,3-glucan, lipopolysaccharide, and peptidoglycan by pathogen recognition receptors [8]. Upon binding, a serine protease (SP) cascade is activated to finally cleave the PPO zymogen to active PO. PO catalyzes the formation of quinones and melanin, which kill and immobilize microbes [8,11]. Biochemical studies of Manduca sexta, Tenebrio molitor and Holotrichia diomphalia have indicated that SP cascades are mainly composed of clip-domain serine proteases (cSPs) that harbor regulatory clip domain (s) in their N- termini. In M. sexta, the biochemical model insect, two SP cascades for PPO activation were clarified [11,12]. One consists of HP21 and PAP2/3, while the other HP1, HP6, and PAP1. HP21 activates PAP2 or PAP3; PAP2/PAP3 then converts PPO into active PO. PPO activation can also be catalyzed by the other cascade composed by HP1, HP6, and PAP1. Furthermore, the genes for cSPs and cSP homologues (cSPHs) could be phylogenetically divided into 5 clades: A, B, C, D and E [11,13]. With the exception of CLIPA and CLIPE, most members of the other three clades are expected to have protease activity. Currently, some cSPs in CLIPB, including PPO-activating proteases (PAP1−3) in M. sexta and SPE in T. molitor, are thought to directly cleave PPO. In addition, CLIPCs have been shown to cleave CLIPBs, for instance, M. sexta HP6 activates M. sexta proPAP1 [11] and T. molitor SPE activates T. molitor proSPE [14,15]. However, functions of CLIPDs are poorly characterized, only one member, M. sexta HP1, was recently described to cleave a CLIPC, M. sexta proHP6 [12]. Apart from that, cSPs are strictly regulated by serpins [8] and serpins are a superfamily of proteins, most of which inhibit serine proteases. Containing 350–400 residues typically, serpins have a reactive center loop (RCL) near their C-termini, which acts as the binding site for its target protease. After binding, the serpin is cleaved between the P1 and P1’ residues within the RCL, undergoes a dramatically conformation change, and forms a covalent complex with the protease [16]. In H. armigera, our previous immunotranscriptome study revealed a repertoire of immunity-related genes well conserved in holometabolous insects [17]. Like M. sexta, two PPO transcripts were identified in the H. armigera transcriptome. We also found 11 cSPs and 22 serpins, similar to other lepidopteran insects. These results indicate that H. armigera contains a melanization regulatory system similar to other lepidopteran insects. This established the basis for further studies of serpins regulating the SP cascade in response to viral infection. Baculoviruses are large DNA viruses that infect insects, primarily the lepidopteran insects. Studies on the interaction between baculoviruses and insect host innate immunity are mainly performed in cell lines. Two cellular mechanisms, RNAi [18] and apoptosis [19,20], are implicated in defense against baculoviruses. However, there are a few studies suggesting that melanization plays a role in defense against baculoviruses in larval insects. It has been demonstrated that melanization can inhibit baculovirus infection in the resistant strain of H. zea [21,22]. In addition, hemolymph of Heliothis virescens is virucidal to baculovirus [23]. More interestingly, the melanin precursor 5,6-dihydroxyindole (DHI) shows antiviral activity against baculovirus [24]. However, the mechanism by which baculovirus overcomes the host immune system is unclear. The complete genome of HearNPV was sequenced [25], and multiple viral genes contributing to viral infection have been characterized [26–28]. In our previous study, the HearNPV infection suppressed host defense response [29]. Here, we deciphered molecular mechanisms on how HearNPV modulates host immunity. Quantitative proteomics analyses revealed that the melanization pathway is suppressed in H. armigera infected with HearNPV. In addition, serpin-5 and serpin-9 induced by HearNPV infection downregulate melanization by inhibiting target proteases, cSP4 and cSP6. Our results suggest that the virus has evolved a novel strategy to suppress the host immune system. Our previous study indicated that HearNPV infection dramatically regulated gene expression at 48 and 72 h post ingestion (hpi) in the fat body, the tissue responsible for the synthesis and secretion of most hemolymph polypeptides [29]. Here, we used a proteomic method to examine and quantify hemolymph proteins of H. armigera infected by HearNPV (Fig 1A). The hemolymph samples (48hM, 48hI, 72hM and 72hI) were collected from control and infected larvae at 48 and 72 hpi, separated by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) (S1A Fig), and subjected to liquid chromatography tandem mass spectrometry (LC-MS/MS) analysis. A high quality, non-redundant protein database (HarmgeraNPV_001. fasta) was queried with MS/MS data for protein identification and quantification. We identified a total of 1,028 unique host proteins with the strict criteria that a protein must be identified at least twice, not only in technical repeats but also in biological repeats. We identified more unique proteins than in previous reports, for example, 654 in M. sexta [30], 752 in Bombyx mori [31], and 725 in D. melanogaster [32], indicating a significant increase in coverage of the insect hemolymph proteome. The pairwise Pearson coefficients of the 1,028 proteins demonstrate excellent data consistency within the control (0. 94−1. 00) or infected (0. 94−0. 99) (S1 Table). The control-infected coefficients were between 0. 68 and 0. 84 at 48 hpi and were between 0. 41 and 0. 59 at 72 hpi, suggesting that protein levels dramatically changed between the control and infected hemolymph and the differences in protein level increased over time. Consistent with the information revealed by the correlation analyses, principal component analysis (PCA) revealed a clear difference between control and infected hemolymph proteins, and the magnitude of difference at 72 hpi increased compared to 48 hpi (S1B Fig). This is also consistent with our transcriptomic analysis [29]. A total of 45 HearNPV proteins were identified in infected hemolymph (S2 Table), 18 of which belonged to late proteins, including P10, FP25K, and P26. By contrast, fewer early proteins (9) were identified such as GP37, ME53, and FGF. In addition, two viral proteins (HA91, HA133) have both early and late promoters. Analysis of the levels of these viral proteins revealed that two early proteins, ME53 and HA121, were restricted to infected hemolymph at 48 hpi. Conversely, in infected hemolymph at 72 hpi, 5 late proteins (ODV-E66, P24, FP25K, Calyx/PEP, and HA66) were exclusively identified and 10 late proteins including cathepsin were much more abundant. Notably, three late proteins, P26, EGT and cathepsin, represented three of the top four most abundant viral proteins in infected hemolymph not only at 48 hpi but also at 72 hpi. These results suggest that oral inoculation of HearNPV successfully established the systemic infection of host larvae. Then, we performed pairwise comparisons between control and infected groups to identify the differential proteins (S3 Table). Venn diagram analysis revealed that 230 proteins were exclusively regulated in hemolymph at 48 hpi (78 upregulated and 152 downregulated), while another 366 proteins were specifically regulated in hemolymph at 72 hpi (244 upregulated and 122 downregulated) (Fig 1B). A total of 305 proteins were regulated at both 48 and 72 hpi. Considering the different stages of viral infection, we speculate that time-dependent differential regulation strategies might be adopted by the virus. Importantly, the data showed consistent downregulation of 87 proteins at both 48 and 72 hpi and some of these (11) were immunity-related, including pattern recognition receptors (PPRs), SPs, and PPOs. In addition, it should be noted that although many proteins (244) were exclusively increased at 72 hpi, most (200) had no putative signal peptide, while among the exclusively downregulated proteins (122) at 72 hpi, most (85) did have a putative signal peptide. Meanwhile, 78 proteins were exclusively increased at 48 hpi, 41 of them had no putative signal peptide, while among the exclusively downregulated proteins (152) at 48 hpi, 73 of them have no putative signal peptide. The percentage of proteins without putative signal peptide in increased protein at 72 hpi (82. 0%) is much higher that at 48 hpi (52. 5%). As baculoviruses are liberated from the nuclei of infected cells by lytic release at late stage of viral infection, increased cell lysis after virus challenge might be responsible for the bias. To avoid bias, we only focused on differential proteins identified in both control and infected hemolymph. Functional classification indicated that a large proportion of downregulated proteins were involved in carbohydrate metabolism (48 hpi), glycan biosynthesis/metabolism (72 hpi), and immunity-related processes (72 hpi) (Fig 1C). Other notable functional groups were translation and amino acids (aa) metabolism (48 hpi). Importantly, most immunity-related proteins were downregulated at 72 hpi, as already implied by transcriptomic data, demonstrating the robustness of our approach. Immunity-related proteins were significantly enriched among differential proteins after viral infection. Therefore, we mainly focused on immunity-related proteins identified in hemolymph. Previously, 233 immunity-related transcripts in H. armigera were identified based on homology [17]. Here, 67 immune proteins were further identified in hemolymph of H. armigera, including recognition molecules (21), modulators (24), signaling proteins (3) and effectors (19) (S1C Fig). Forty members of these immune proteins were differently expressed in hemolymph during viral infection (S4 Table). Hierarchical clustering revealed that many more immune proteins exhibited more remarkable downregulation than upregulation at the protein level in hemolymph at 48 hpi (8 upregulated and 15 downregulated) and 72 hpi (6 up and 30 downregulated) (Fig 1D and S1D Fig), of which 11 members were consistently downregulated at both time points. Six of the 11 downregulated proteins were postulated to be involved in the melanization activation pathway, including PPO1 (9. 5-fold at 48 hpi, 2. 5-fold at 72 hpi), PPO2 (14. 1-fold at 48 hpi, 2. 4-fold at 72 hpi), cSP6 (1. 4-fold at 48 hpi, 14. 7-fold at 72 hpi), cSP8 (1. 5-fold at 48 hpi, infinity at 72 hpi) and cSP29 (3. 8-fold at 48 hpi, 22. 7-fold at 72 hpi). Other putative melanization components also exhibited downregulation at either 48 hpi or 72 hpi, such as SP42 (2. 6-fold at 72 hpi), cSP4 (16. 9-fold at 72 hpi), and serpin-3 (2. 2-fold at 72 hpi). Interestingly, two serpins, serpin-5 and serpin-9, exhibited an inverse regulation pattern. Serpin-9 was upregulated at 48 hpi (3. 1-fold) and downregulated at 72 hpi (2. 8-fold), whereas serpin-5 was completely different, being downregulated at 48 hpi (2. 8-fold) and upregulated at 72 hpi (infinity). Serpin-5 is involved in regulating melanization [33,34], and the function of serpin-9 has not been previously reported. These results suggest that baculovirus infection suppresses hemolymph melanization in H. armigera. To test the resultant hypothesis from our quantitative proteomics analysis that baculovirus infection suppresses host melanization, a well-established PO activity assay using H. armigera larvae hemolymph was performed. Time-course analysis showed that PO activity in infected hemolymph did not exhibit any remarkable changes at 0 or 24 hpi. As infection progressed, PO activity decreased sharply at 48 hpi (7. 4-fold) and 72 hpi (4. 2-fold) compared to the controls (Fig 2A). The results demonstrated that at the early stage of baculovirus infection, hemolymph melanization was not affected. However, at the middle and late stages of the virus infection, the hemolymph melanization pathway was persistently downregulated. Our results showed that baculovirus infection reduced not only PPO amount but also host PO activity. A recombinant variant of HearNPV, HearNPV-egfp that expresses green fluorescent protein (GFP) in infected cells, was used in the virus infection assay to evaluate the effects of hemolymph melanization. The HearNPV-egfp was incubated with H. armigera hemolymph and dopa, the substrate of melanization, for 1,3, and 6 h. Hemolymph only and dopa only were set as controls. Virus titer was determined at different time points by end-point dilution assay (Fig 2B). After 1 h incubation, the virus titer of hemolymph only, hemolymph and dopa group both decreased to zero. However, when phenylthiourea (PTU), specific inhibitor of PO, was incubated together with hemolymph, dopa, and virus, the virus titer did not change even after 6 h incubation. Dopa incubated with virus also did not have significant decrease in virus titer. Moreover, when the HearNPV-egfp suspension was incubated with DHI, a reactive intermediate compound generated by PO, the virus titer exhibited 95-fold reduction after 1 h incubation and could hardly be detected after 3 h. Next, the same samples of reaction mixture for 1,3, and 6 h were added to HzAM1 cells, after which the fluorescence microscopy images were taken at 72 hpi. When the HearNPV-egfp suspension was incubated with DHI, fewer infected cells was observed for 1 h incubation and no infection was observed after 3 h. Consistently, no viral infection was observed in hemolymph and dopa treatment group after 1 h incubation (Fig 2C). However the addition of PTU rescues the infection. Infected cells didn’t decrease even after 6 h incubation. These results demonstrate that melanization can block viral infection in vitro. The abovementioned experimental data suggest a critical role of hemolymph melanization in baculovirus infection. We developed eight antibodies against PPO, cSP, and serpins involved in the melanization pathway (S5 Table) and the specificity of the cSP antibodies were confirmed by a knockdown experiment (S2 Fig). We performed immunoblot analyses of hemolymph from control and infected larvae at 48 and 72 hpi using these antibodies. The samples were left for 30 min for PPO to undergo spontaneous activation. The results showed that the amounts of cSP4, cSP6, cSP8, and cSP29 in H. armigera hemolymph decreased over the time course in response to HearNPV infection (Fig 3A and S3A Fig). The protein level of cSP4 in infected hemolymph was almost the same as that in mock hemolymph at 48 hpi, but its level was drastically reduced at 72 hpi. Regarding cSP6, the protein bands in the infected group were much weaker than those in the mock group at both 48 and 72 hpi. Moreover, the protein level of cSP8 significantly decreased at 48 hpi, and was undetectable at 72 hpi. Likewise, cSP29 could barely be detected by immunoblot in infected hemolymph (48 and 72 hpi), suggesting the strong effects of viral infection on its expression. In addition, two key enzymes involved in melanization, PPO1 and 2, were significantly downregulated at the protein level upon viral infection. Apart from the changes in protein levels, the cleavage of PPO was inhibited, with no cleaved forms detected in infected hemolymph (Fig 3B), The same result was observed even when increasing 4–7 times higher loading amount (S3B Fig). Furthermore, serpin-9 and serpin-5 were sequentially increased in protein abundance (Fig 3B and S3A Fig); the former was induced at 48 hpi and the latter was induced at 72 hpi, which is consistent with proteomics data, implying that these two molecules play critical roles in the modulation of host melanization and their targets were likely distinct from each other. Taken together, the immunoblot results indicate systemic inhibition of melanization in H. armigera hemolymph after baculovirus infection. The quantitative proteomic and immunoblot analyses revealed that the protein levels of serpin-9 and serpin-5 were sequentially induced by viral infection, implying their roles in suppression of host melanization. Using quantitative real-time PCR (qPCR), we determined the relative transcriptional levels of serpin-5 and serpin-9 in fat body and hemocytes at three time points (24,48, and 72 hpi), and found that the transcript levels of serpin-5 and serpin-9 were substantially induced by viral infection (Fig 4A). Notably, there was no significant difference in serpin-5 mRNA levels between the control and infected groups at 24 and 48 hpi, but it was significantly induced at 72 hpi in both fat body (>5-fold) and hemocytes (>2-fold). However, serpin-9 mRNA was strongly elevated at both 48 and 72 hpi (>10-fold) in hemocytes, while it was induced at 48hpi (>2-fold) and suppressed at 72 hpi (>2-fold) in fat body. These qPCR results were consistent with the induction of serpin-5 and serpin-9 at protein level in the infected hemolymph. To explore the effects of serpin-5 and serpin-9 on PO activity, we performed a PO activity assay in vitro using recombinant serpins. First, we cloned the cDNA of serpin-5 and serpin-9 from H. armigera. Haserpin-5 cDNA encoded a polypeptide of 400 aa residues with a predicted signal peptide of 21 residues. The calculated molecular weight of mature serpin-5 is 42 kDa. Haserpin-9 cDNA encoded a polypeptide of 406 aa residues, with a predicted signal peptide of 20 residues. The calculated molecular weight of mature serpin-9 is 43 kDa. Sequence analysis indicated that Haserpin-5 is orthologous to M. sexta serpin-5 [33,35] with 78% identity in aa sequence, suggesting that they might share similar regulatory functions. However, the ortholog gene of Haserpin-9 was not found in M. sexta. Interestingly, phylogenetic analysis revealed that Haserpin-5 and Haserpin-9 form a group with Msserpin-4, Msserpin-5 and Dmserpin77Ba [36] (S4A Fig). In addition, comparison of the RCL regions also showed that these five serpins have similar predicated residues at the P1-P1’ positions of the scissile bond (S4B Fig), implying that Haserpin-9 might also participate in melanization. To explore their biochemical functions, serpin-5 and serpin-9 gene were cloned into pET-28a vector and pET-32a vector respectively, and expressed in an Escherichia coli expression system. Soluble serpin-5 and serpin-9 protein were further purified by nickel affinity chromatography. Recombinant serpin-5 and serpin-9 proteins migrated as 42. 8-kDa and 61. 5-kDa single bands on SDS-PAGE under reduced conditions (Fig 4B) and were confirmed by mass spectrometry. As melanization was suppressed by HearNPV infection and serpin-5 and serpin-9 were upregulated in hemolymph from baculovirus infected larvae, it’s presumable that serpin-5 and serpin-9 might inhibit melanization in hemolymph. Therefore, we collected hemolymph from naïve larvae of H. armigera and added serpin-5 and serpin-9 at different concentrations. Before subject to PO activity assay, these samples were exposed to air at 27°C for 5 min to undergo spontaneous melanization. The results showed that PO activities decreased by up to 46% and 83%, respectively (Fig 4C), suggesting that serpin-5 and serpin-9 inhibit melanization by targeting one or more cSPs in the PPO activation cascade. The above results showed that serpin-5 and serpin-9 function as inhibitors of PPO activation. To further identify their target cSPs, we tried to activate the infected larval hemolymph at the two time points by exposure to air, and captured serpin-associating cSPs in the hemolymph samples using a semi-quantitative proteomic approach (Fig 5A). Immunoaffinity chromatography using serpin-5 antibody allowed us to isolate complexes formed by serpin-5 and its associated proteins from the hemolymph at 72 hpi. Similarly, serpin-9 and its associated proteins from the hemolymph at 48 hpi were purified using serpin-9 antibody. Proteins eluted from the column were analyzed and quantified using tryptic peptide nano-LC-MS/MS analysis. The abundances of serpin-5 and serpin-9 increased in infected hemolymph (Fig 5B), as suggested by the quantitative proteomic (Fig 1D) and immunoblot analysis (Fig 3D). Among serpin-5 immunoprecipitated proteins (277), cSP4 and cSP6 were identified. Meanwhile cSP29 and cSP6 were identified among serpin-9 immunoprecipitated proteins (214). cSPs were remarkably increased in the infected hemolymph. Phylogenetic analysis revealed that cSP6 as well as cSP8 belong to the CLIPB group and are orthologous to PAP3 and PAP1 in M. sexta, respectively. Meanwhile cSP4 and its ortholog M. sexta HP6 belong to the CLIPC group (S5 Fig). Interestingly, cSP29 is orthologous to M. sexta HP1, which is reported to activate HP6 in M. sexta and is the only member of the CLIPD group with a known function. Apart from serpins and the target cSPs, serpin-3 and PPOs also tightly associated with the antibody columns. Serpin-3 had similar change patterns to serpin-5 at 72 hpi and serpin-9 at 48 hpi, respectively. The protein abundance of PPO associated with serpin-5 increased at 72 hpi. However, the protein level of PPO bound with serpin-9 substantially decreased. These results indicate that serpin-5 might be a key regulator of cSP4 and cSP6, while serpin-9 might be a critical modulator of cSP6 and cSP29 in the melanization cascade. To further confirm the specific interaction between serpins and associated cSPs, we conducted a yeast two-hybrid assay. We constructed prey plasmids containing serpin-5 or serpin-9, and bait plasmid containing cSP4, cSP6, or cSP29. In selective growth media, we observed that serpin-5 bound to cSP4, whereas serpin-9 interacted with cSP6 and cSP29 (Fig 5C). The results suggest that cSP4 is likely the targeted protease of serpin-5 while cSP6 and cSP29 are the targeted proteases of serpin-9. To further investigate the interaction of serpin-5 with cSP4 and serpin-9 with cSP6, we expressed cSP4 and cSP6 in their zymogen forms using the Drosophila S2 expression system. Predicted activation site of cSP4 and cSP6 was manually changed from LDLH92 to IEGR92 and VGNK156 to IEGR156 respectively. IEGR is the sequence of cleavage, which could be recognized by bovine coagulation factor Xa. They were named as procSP4Xa and procSP6Xa. Expressed recombinant proteins were purified by Ni-NTA agarose column. The amidase activity was detected using the substrate acetyl-Ile-Glu-Ala-Arg-p-nitroanilide (IEAR). Serpin-5 that was incubated with factor Xa-activated cSP4Xa was found to form a higher molecular weight complex by immunoblot (Fig 6A). Once being activated by factor Xa, the cSP4Xa zymogen band disappeared, and the catalytic domain of cSP4Xa band was detected by anti-His antibodies. When serpin-5 was mixed with active cSP4Xa, a novel band with high molecular weight appeared. IEAR activity of cSP4 was inhibited (Fig 6A). Similar results were also observed with serpin-9 and cSP6Xa (Fig 6B), active cSP6Xa was inhibited by serpin-9, as demonstrated by the formation of high molecular complex and decreased IEAR activity. These results confirm that serpin-5 and serpin-9 can inhibit cSP4 and cSP6 enzymatic activity by forming an SDS-stable complex respectively. Because cSP6 is predicted to function as PAP to cleave PPO directly, the inhibition of cSP6 by serpin-9 would lead to suppression of PPO activation. To test the hypothesis, hemolymph PPOs was purified from hemolymph of H. armigera and shown to be a heterodimer formed by PPO1 and 2, which is consistent with results revealed from M. sexta and H. diomphalia (Fig 6C and S7A Fig). It exhibited PO activity after activation by alcohol in native PAGE (S7B Fig). Then, we incubated purified PPO with cSP6Xa with or without factor Xa. The results clearly showed that PPOs were enzymatically cleaved and activated by cSP6Xa (Fig 6C and S7C Fig). Taken together, serpin-5 and serpin-9 suppress PPO activation by the inhibition of cSPs in melanization cascade. To reveal the effect of serpin-5 and serpin-9 in vivo during baculovirus infection, we conducted dsRNA mediated knockdown by injecting double-stranded RNA (dsRNA) of serpin-5 or serpin-9 into 2nd instar larvae of H. armigera. The relative mRNA expression levels were detected by qPCR in whole larvae, and the protein levels were detected by immunoblot in hemolymph. The results showed that the transcript level of serpin-5 and serpin-9 were decrease by 43% or 83% in larvae injected with dsserpin-5 or dsserpin-9 respectively (Fig 7A). Immunoblotting tests also confirmed efficiency of serpin-5 and serpin-9 RNAi (S6A Fig). When injected with dsserpin-5 or dsserpin-9 before HearNPV treatment, the survival of H. armigera increased due to the knockdown of serpin-5 or serpin-9 (Fig 7B). Then we collected hemolymph and measured PO activity. We found that depletion of serpin-5 and serpin-9 significantly increased PO activity compared to the control after HearNPV infection (Fig 7C). In addition, copies of the viral genomic DNA in hemolymph was measured by qPCR. The results showed that the number of viral DNA copies dramatically decreased by 93. 6% or 91. 1% in hemolymph of serpin-5 and serpin-9 depleted larvae compared to the control (iGFP-NPV) (Fig 7D). Consistently, immunoblot indicated that the amount of VP39, the structural protein of HearNPV, dramatically decreased in hemolymph of serpin-5 and serpin-9 depleted larvae in comparison to the control (iGFP-NPV). Conversely, PPO2, cSP6, cSP29 was apparently increased in hemolymph of serpin-5 and serpin-9 depleted larvae, while cSP4 increase in hemolymph of serpin-5 depleted larvae (S6B Fig). Clearly, the decline of serpin-5 or serpin-9 in H. armigera increased PO activity, resulting in elevated resistance against baculovirus infection. After viral infection, it appeared that cathepsin, EGT and P26 represented three of the top four most abundant viral proteins in infected hemolymph not only at 48 hpi but also at 72 hpi. These are secreted proteins, and are not included in the top 20 most abundant viral proteins in infected fat body at 72 hpi [29]. Cathepsin [37], an enzyme involved in insect disintegration, increased over time during infection. EGT was reported to block insect molting and pupation [38], as observed in our previous study [29]. P26 was described as a budded virus (BV) specific structural protein, the high level of P26 suggested that a large number of BV virions were secreted into hemolymph of infected H. armigera at 48 and 72 hpi. There were high abundance of virus proteins in the hemolymph, however, our proteomic study also indicates that most immunity-related proteins were downregulated, including the melanization pathway, an essential component of insect immune response. Baculovirus encoded microRNA may be employed to regulate the host immunity-related genes [39,40], and increased serpin-5 and serpin-9 may also be involved in the host immunity-related gene expression (S6B Fig). Because baculoviruses are natural enemies of insects, specific interactions between the virus and host likely influence virus survival. Our previous report revealed that RNAi and JAK-STAT pathways were not significantly regulated during baculovirus infection [29]. In this study, our proteomic analysis show that levels of most AMPs were not significantly changed. But most of melanization components were downregulated in hemolymph after viral infection. Further results show that melanization can inactivate baculoviruses and block their infection. The fact that hemolymph does not melanize during infection by baculovirus in permissive lepidopteran hosts but melanizes in resistant ones has been reported [21]. The resistance is attributed to the impact of melanization in hemolymph, however, the inhibition of melanization in permissive hosts is still elusive. In this study, our proteomics and immunoblot analyses demonstrated that the protein level of melanization-activating components exhibited a significant decline during the viral infection, while two putative negative regulators of melanization, serpin-5 and serpin-9, increased instead. Consistent with the change at the protein level of melanization genes, PPO activation was inhibited and PO activity in the infected hemolymph dramatically decreased. Hemolymph PO of the larval tobacco budworm, H. virescens, is virucidal to baculovirus [23], and DHI generated by PO has virucidal activity against baculovirus AcMNPV [24]. In this study, similar effects of melanization on HearNPV were observed in H. armigera, hence, we concluded that the permissiveness of baculovirus infection is attributable to the inhibition of melanization by regulation of protein levels in hemolymph. Melanization is an important invertebrate immune mechanism. The major components in the melanization pathway such as SPs and serpins have been characterized in some insects and other arthropod species [41]. However, melanization cascade in H. armigera remains unclear. In our previous work, transcripts for 46 SPs, 22 serpins, 7 SPHs and 2 PPOs were identified in H. armigera [17], providing fruitful ground for deciphering their roles in melanization. In this study, we identified the proteins encoded by 11 SPs, 10 serpins, 2 PPOs, and 3 SPHs in hemolymph using proteomics. Based on phylogenetic analysis, many were found to have orthologs in other insects, strongly suggesting the conservation of melanization-mediated immunity in the cotton bollworm. We deciphered the mechanism of HearNPV suppressing host melanization reaction using integrated approaches (Fig 8). Phylogenetic analysis reveals that H. armigera melanization reactions is homologous to those reported in M. sexta. In H. armigera, cSP29-cSP4-cSP8 was predicted to form a sequential activation cascade. Both cSP6 and cSP8 are PPO activating enzyme. The cleavage and activation of PPO by cSP6 are further demonstrated by in vitro reconstitution (Fig 6C). Moreover, the inhibition of cSP4 activity by serpin-5 was similar to the interaction of HP6 regulated by serpin-5 in M. sexta [33]. However, we did not find orthologue of M. sexta PAP-2 in H. armigera and the inhibition of PPO activation by serpin-9 has not been reported in M. sexta at present. To the best of our knowledge, this is the first time that melanization pathway is clarified in H. armigera, and it is shown to respond to viral infection, which further advances our knowledge of melanization in diverse insects and provides a theoretical basis for better controlling agricultural pests. In our study, serpin-5 and serpin-9 were demonstrated to be involved in modulating the melanization reaction during viral infection. Serpin-5 was previously shown to be induced in hemolymph of M. sexta after bacterial or fungal infection and to inhibit melanization [35]. Similar results were also found in B. mori [34]. In this study, we demonstrated that the inhibition of melanization caused by serpin-5 is indispensable to viral infection. Strikingly, we identify a novel serpin, serpin-9, whose ortholog in the model insect M. sexta has not been found, but that also participates in the suppression of melanization for successful viral infection. The inhibitory mechanisms of these two serpins are different. Serpin-9 acts upstream and downstream to serpin-5, and its induction time is also earlier (48 vs. 72 hpi). In addition, it is also possible that serpin-5 and serpin-9 might inhibit melanization by regulating a signaling pathway (such as Toll pathway) that controls the expression of components of melanization cascade. As the increased expression of components of the melanization cascade in both iserpin-5/NPV and iserpin-9/NPV insects were observed (S6B Fig). Interestingly, a baculovirus encoded serpin has been characterized as an inhibitor of host melanization and its expression increased the virulence of the virus by four-fold in Trichoplusia ni larvae [42]. However, this molecule is not an ortholog of serpin-5 or serpin-9 in H. armigera, and its exact molecular mechanism remains unknown. We have demonstrated that the polyphagous pest H. armigera retains very complex melanization against the invasion of microbes. However, in nature, pathogens have evolved distinct strategies to modulate host immunity by suppressing melanization. An antibiotic produced by the entomopathogenic bacterium Photorhabdus luminescens inhibits melanization by targeting PO activity in M. sexta [43]. Parasitoid wasps associated polydnavirus also expresses two viral proteins, Egf1. 0 and Egf1. 5, to suppress melanization via PAP inhibiton [44–46]. More recently, it was reported that a viral protein encoded by WSSV interacts with PAP to suppress melanization in the shrimp Penaeus monodon [10]. Different from these findings, HearNPVs exploit host serpins to inhibit SPs at different steps of the melanization cascade. The phylogenetic analysis comparing the SPs interacting with serpin-5 or serpin-9 to other insect SPs showed that they belong to three different clades. cSP6 belongs to the CLIPB clade and is an ortholog of M. sexta PAP3. Indeed, cSP6 is inhibited by serpin-9 and acts as a PAP in H. armigera and is found to cleave PPO directly. cSP4, which interacts with serpin-5, is predicted to cleave PAP. cSP4 belongs to the CLIPC clade and is the Helicoverpa ortholog of MsHP6. MsHP6 has been verified experimentally to cleave MsPAP1 (a CLIPB member), while biochemical studies have shown that MsPAP3 cleaved PPO directly in M. sexta [47]. Moreover, cSP29, targeted by serpin-9, is an SP belonging to the CLIPD subfamily, and its hypothesized function is to cleave cSP4. More recently, MsHP1 was identified to be able to cleave MsHP6 and its activation did not require proteolytic cleavage [12]. It is tempting to postulate that HacSP4, cSP6, and cSP29 also function as their orthologs in the melanization cascade. Among them, cSP6 and cSP8 are orthologous to M. sexta PAP3 and PAP1, respectively and may function as PAPs [47,48]. Likewise, cSP4 is postulated to cleave PAPs [49], while cSP29 is postulated to cleave cSP4. Taken together, we envision a mechanism whereby baculovirus overcomes host melanization. The viral infection induces the level of serpin-5 and serpin-9 in hemolymph. Serpin-5 specifically inhibits cSP4, and serpin-9 inhibits cSP6 and cSP29, resulting in a dramatic decline of PO activity to suppress the virucidal capacity of host melanization (Fig 8). This mechanism, which involves baculovirus induced serpin-5 and serpin-9 and then inactivates the host immune system, suggests that these two negative regulators of immune response are important for baculovirus infection in insects. These findings improve our understanding of the interaction between HearNPV and its co-evolutionary host H. armigera. The colony of cotton bollworm H. armigera was raised in the laboratory as previously described [17]. The H. zea cell line HzAM1 was maintained at 27°C in Grace’s insect medium supplemented with 10% fetal bovine serum (FBS). The HearNPV G4 strain was used in this study [25]. Occlusion bodies (OBs) were amplified from infected larvae. HearNPV-egfp was constructed in a previous work [50]. The experiments were performed as described in our previous study [29]. Briefly, 96 day-2 3rd instar H. armigera larvae were divided into two groups. After 16 h of starvation, each larva in the infection group was fed for 10 min on 2 μL 10% sucrose solution containing edible blue dye contaminated with HearNPV at a final concentration of approximately 6. 7×107 OBs/μL (99% lethal concentration, LC99), while each larva in the mock group was fed on 2 μL 10% sucrose solution containing edible blue dye alone. The cell-free hemolymph samples were prepared as previously described [30]. After virus ingestion, the prolegs of larvae were cut at 48 and 72 hpi, and hemolymph was collected into clean tubes containing 1 mM PTU and 1 mM p-aminobenzamidine (pAB). Then the hemolymph was centrifuged at 4,000 ×g at 4°C for 5 min to remove hemocytes. Hemolymph from six infected larvae at 48 hpi was pooled as 48 h infected hemolymph (48hI) and hemolymph from six infected larvae at 72 hpi was used as 72 h infected hemolymph (72hI). Similarly, 48 h mock-infected hemolymph (48hM) and 72 h mock-infected hemolymph (72hM) were prepared from mock-infected larvae. In this study, the sample size of 6 was twice as many as reported in M. sexta [30]. The experiment was repeated twice to obtain biological replicates. The protein concentrations of the twelve samples were determined using the bicinchoninic acid assay (BCA) method. Equal amounts of the samples (25 μg) were separated on 4–15% gradient SDS-polyacrylamide gels (Bio-Rad, Hercules, CA, USA) and stained with Coomassie Brilliant Blue G-250 for visualization. Subsequently, the gel lanes were excised into 12 or 6 slices for the 48 h samples and five slices for the 72 h samples. Gel slices were subjected to reduction, alkylation and in-gel digestion with mass spectrometry grade trypsin (Promega, Madison, WI, USA). After digestion, the resulting peptides were extracted by 1% trifluoroacetic acid in 60% acetonitrile, near-dried in a Speedvac, solubilized in 0. 1% trifluoroacetic acid in 2% acetonitrile 0. 1% trifluoroacetic acid, ultracentrifuged and the supernatant then subjected to LC-MS/MS analysis as previously described [29]. Three technical replicates were performed for each biological sample in the LC-MS/MS analysis. The MS data analyses were performed using Proteome Discoverer software, supported by a protein database (HarmigeraNPV_001). The database was deduced from the putative open reading frames based on the transcriptome unigene dataset [17] and the HearNPV genome [25]. Protein identification and quantification were performed as previously described [29]. The relative abundance of protein was represented as the peak area of the target protein normalized by the average peak area of two internal reference proteins (angiotensin converting enzyme and laminin β-2 chain). The statistical differences across the samples were analyzed by Student’s t-test. A protein with a fold change ≥ 1. 4 and p < 0. 05 was considered differentially regulated. Kyoto Encyclopedia of Genes and Genomes (KEGG) orthology (KO) -based annotation system (KOBAS) and PCA analyses were performed as previously described [29]. Based on the relative abundance of immune proteins, the heatmap package in the R environment was implemented to analyze immune protein abundance in hemolymph. Phylogenetic trees were constructed using MEGA6 with the neighbor-joining method. Cell-free hemolymph samples were collected from mock-infected and infected H. armigera at 0,24,48 and 72 hpi as described above, but without adding PTU or pAB. After incubation at 27°C for 30 min, the PO activity assays were conducted in 96-well plates containing 1 μL hemolymph followed by addition of 200 μL substrate solution (2 mM dopamine in 50 mM sodium phosphate buffer [pH 6. 5]). PO activity was determined at 470 nm using a plate reader. One unit of activity was defined as ΔA470 of 0. 001 in one minute. To measure PO activity of hemolymph in serpin-5 or serpin-9 depleted larvae, hemolymph samples were incubated at 27°C for 5 min to observe the differences between samples. Hemolymph from naïve 5th instar larvae was collected into tubes containing recombinant serpin-5 (10 μL, at final concentrations of 0–320 μg/mL) or serpin-9 (10 μL, at final concentrations of 0–1200 μg/mL). A recombinant baculovirus protein, P33, was used as negative control. After incubation for 5 min at room temperature, the reaction mixtures were subjected to the PO activity assay. procSP4, procSP6, procSP8, procSP29, serpin-5, PPO1, or PPO2 cDNA encoding mature protein was amplified and cloned into pET-28a vectors, and P33 and serpin-9 cDNA encoding mature protein was cloned into a pET-32a vector. The cloning primers are presented in S6 Table. These recombinant proteins were expressed in E. coli BL21 and affinity purified on a Ni-NTA agarose column (Qiagen, Hilden, Germany). Polyclonal antibodies were produced in rabbits against the recombinant proteins (Beijing Protein Innovation, Beijing, China). Cell-free hemolymph samples were collected into clean tubes containing 1 mM PTU and incubated at 27°C for 30 min. Hemolymph samples were resolved on 4–15% gradient SDS-polyacrylamide gels (Bio-Rad) and electrotransferred onto a polyvinylidene difluoride (PVDF) membrane (Invitrogen, Carlsbad, CA, USA). Immunoblot analyses were performed using 1: 5000 diluted polyclonal antibodies as the primary antibodies. H. armigera (heat shock protein 27. 2 kDa (HSP27. 2) was used as a control for loading and transfer. Band intensities of scanned blots were quantified using ImageJ. The integrated intensity of a fixed area was measured, and background levels were subtracted. Total RNA samples were prepared from the fat body, hemocytes, or whole body of H. armigera. qPCR reactions were performed on the MX3000P system (Stratagene, San Diego, CA) using SYBR green PCR master Mix (Tiangen, Beijing, China). Thermal cycling conditions were: 94°C, for 5 s; 59°C, for 20 s; and 72°C, for 20 s. Quantitative measurements were performed in triplicate and normalized to the internal control ribosomal protein S7 for each sample. Primers used for qPCR are listed in S6 Table. The recombinant serpin and P33 proteins were induced in E. coli grown in LB medium (1. 6 L) with 0. 5 mM isopropyl β-D-1-thiogalactopyranoside at 16°C overnight. The E. coli cells were harvested by centrifugation at 4,500 rpm for 10 min, resuspended in 60 mL lysis buffer (50 mM Tris-HCl, pH 7. 5,0. 3 M NaCl, 5 mM β-mercaptoethanol, 1 mM phenylmethane sulfonyl fluoride [PMSF]), and then incubated with lysozyme (1 mg/mL) for 20 min on ice. The suspensions were further incubated with DNase (10 U/mL) for 10 min on ice. After sonication, cell debris was removed by centrifugation at 16,000 rpm at 4°C for 40 min. Suspensions were loaded on a 4 mL Ni-NTA agarose column equilibrated with 50 mM Tris-HCl (pH 7. 5) and 0. 3 M NaCl, and washed with 10 mM imidazole. Concentration and buffer exchanges in the serpin fractions were performed in an Amicon Ultra 10K cartridge (Millipore, Billerica, MA, USA). The purified proteins were stored at –80°C in 20 mM Tris-HCl (pH 7. 5). Rabbit antibody covalently coupled Protein A-agarose was prepared as previously described [51]. Cell-free hemolymph samples (0. 1 mL) were collected from mock-infected or infected larvae. PTU (10 mM) was added to the hemolymph and incubated at room temperature for 30 min, after which PMSF (to a final concentration of 1 mM) and protease inhibitor cocktail (Pierce, Waltham, MA, USA) (1: 100) were added to inhibit protease activity. The samples were incubated with 50 μL of Protein A-agarose beads at 4°C for 5 h to remove non-specific absorption. After centrifugation at 12,000 rpm for 4°C, the supernatants were added to 50 μL of the rabbit antibody coupled Protein A-agarose beads and incubated at 4°C overnight. Haserpin-protease complexes were washed and eluted as described in the Protein A Immunoprecipitation Kit (Sigma, St. Louis, MO, USA). The affinity-purified proteins were subjected to SDS-PAGE and stained with Coomassie Brilliant Blue. Subsequently, the gel lanes were excised into slices for in-gel digestion and LC-MS/MS analysis as described above. At 12 h after dsRNA injection (described below), 24 larvae were starved for 12 h and then challenged with HearNPV (3. 0×105 OBs/μL, LC95) as described above. The larvae were maintained in individual containers and fed continuously. The survival curves were compared using Kaplan-Meier, the p-value threshold was calculated with a log-rank or Mantel Cox test; p < 0. 05 was considered statistically significant. Graphpad 6. 0 software was used in all of the statistical analyses. The experiments were performed as previously described [52], with primers listed in S6 Table. Serpins were cloned into pGBK7 and cSPs were introduced into pGADT7 plasmid. For the validation of serpin-5 and serpin-9 binding proteins, the two types of plasmids were co-transformed into AH109 cells using YeastMaker Yeast Transformation System 2 (Clontech, Mountain View, CA, USA), and the binding was validated in synthetic dropout-Leu-Trp-His medium supplemented with X-gal. Recombinant procSP4Xa and procSP6Xa proteins and high-purity native PPO were prepared as described below. procSP4 (50 ng) activated by bovine clotting factor Xa was mixed with purified serpin-5 at a molar ratio of 1: 5; While in control sample, procSP4 or factor Xa was omitted from the mixture. After incubation at room temperature for 10 min, the mixtures were resolved by SDS-PAGE and subjected to immunoblot analysis as described above, using diluted antibodies against His-tag (1: 8000) or serpin-5 (1: 5000) as primary antibodies. In the immunoblot analysis using antibodies against His-tag, the amount of loading sample in the second lane was six times the expected amount to visualize the band. Amidase activity of the reaction mixtures was also measured using 200 μL, 50 μM acetyl-Ile-Glu-Ala-Arg-p-nitroanilide (IEAR) [53]. The same processing method was used for procSP6 and serpin-9. Activated cSP6 was mixed with native PPO at a molar ratio of 1: 1; in the control samples, procSP6 or factor Xa was omitted from the mixture. After incubation at room temperature for 10 min, the mixtures were resolved by SDS-PAGE and subjected to immunoblot analysis as described above, using diluted antibody against PPO1 as the primary antibody. Amidase activity and PO activity in the reaction mixtures were also assayed as described above. PPO was purified from the hemolymph of H. armigera larvae using the previously described method [54] with modifications. Unless otherwise stated, all of the operations were conducted in a cold room (3–5°C) and centrifugations were performed at 12,000×g for 20 min. Hemolymph (10 mL) from 200 day-3 5th instar larvae was collected (approximately 50 μL/Larva) directly into ice-cold saturated ammonium sulfate (AS) solution to prevent the activation of PPO. The final concentration of AS was adjusted to 35% saturation. After centrifugation, the supernatant was collected and brought to 50% AS saturation. The protein precipitate was collected by centrifugation and dissolved in 2 mL buffer A (10 mM potassium phosphate buffer, pH 6. 8, containing 500 mM NaCl, and 0. 5 mM glutathione). The protein solution was exchanged against buffer A three times and concentrated to 600 μL by filtration over an Amicon Ultra 30K cartridge (Millipore). The 600 μL protein solution was loaded on a 5 mL column prepacked with Ceramic Hydroxyapatite (Bio-Rad) equilibrated with buffer B (10 mM potassium phosphate buffer, pH 6. 8, containing 500 mM NaCl) and washed with 15 mL buffer B. Bound proteins were eluted at 1 mL/min with a linear gradient of 10–100 mM potassium phosphate in buffer B (30 mL). Pooled PPO fractions were adjusted to 1 mM MgCl2 and passed through a 0. 5 mL Concanavalin A (Con A) Sepharose column (Sigma) equilibrated with 20 mM Tris-HCl (pH 7. 4), 0. 5 M NaCl, and 1 mM MgCl2. The flow-through fraction was adjusted with saturated AS to a final concentration of 1 M and applied to a 1 mL Phenyl Sepharose 6 Fast Flow (low sub) column (GE healthcare) equilibrated with 0. 1 M potassium phosphate buffer (pH 7. 1) and 1. 0 M AS and washed with the same buffer (3 mL). Bound proteins were eluted with a descending gradient of 1–0 M AS and 0. 1–0. 01 M potassium phosphate (20 mL). The PPO fractions were combined and concentrated to 0. 5 mL and loaded onto a 25 mL Superdex 200 column (AKTApurifier System, GE Healthcare, Little Chalfont, UK) equilibrated with 10 mM Tris-HCl (pH 7. 5) and 0. 2 M NaCl. Fractions containing PPO were combined and protein concentration was determined by the BCA method. PPO was stored at –80°C. After each chromatographic step, the PO activity in column fractions was assayed by activation of PPO with cetylpyridinium chloride using the 2 mM dopamine solution. The entire procSP4 and procSP6 coding region, excluding the signal peptide, were amplified by RT-PCR from the fat body of 5th instar H. armigera larvae. The primers are listed in S6 Table. The PCR products were gel purified, digested with EcoRI and XbaI and ligated into the same site in pMT-BiP/V5-HisA vector (Invitrogen). After sequence confirmation, the resulting plasmids were used as templates to produce mutants in which the four residues at the predicted activation site were replaced with IEGR, a cleavage site of bovine coagulation factor Xa, using overlap extension PCR [55]. The primers are listed in S6 Table. These constructs were named procSP4Xa and procSP6Xa. After sequence verification, the plasmids were used to transfect Drosophilia S2 cells in combination with hygromycin selection vector pCoHygro (Invitrogen), and stable cell lines for producing recombinant proteins were obtained following the manufacturer’s instructions. The cell cultures were harvested at 6 days after induction of expression with 500 μM CuSO4. Cells were removed by centrifugation at 1,000 g at 4°C for 5 min, and the cell debris was removed by centrifugation at 14,000 g at 4°C for 1 h. The cell-free medium (400 mL) was loaded onto a 5 mL Ni-NTA agarose column equilibrated with phosphate buffered saline (PBS; pH 7. 4) and washed with 10 mM imidazole. Bound proteins were eluted with 100 mM imidazole, exchanged against 20 mM Tris-HCl (pH 8. 0) and 0. 1 M NaCl, and concentrated by filtration over an Amicon Ultra 10K cartridge. The purified proteins were stored at –80°C. A recombinant stock of HearNPV-egfp [50] (10 μL, 5×105 TCID50/mL) was mixed with either 5 μL hemolymph (H. armigera cell-free hemolymph from healthy 5th instar larvae), 20 μL of 10 mM dopa, 5 μL hemolymph and 20 μL of 10 mM dopa, 5 μL hemolymph and 20 μL of 10 mM dopa with 10 μL of 10 mM PTU, or 10 μL of 1. 25 mM DHI in 80μL phosphate buffer (PB; 50 mM sodium phosphate, pH 6. 5), then all of the mixtures were adjusted to a final volume of 100 μL with PB and incubated at room temperature for 0,1, 3, or 6 h. The virus titers of these mixtures were determined by end-point dilution assay using HzAM1 cells. Each virus infection was performed in triplicate. Moreover, the reaction mixtures (3 μL) were added to HzAM1 cells in Grace’s insect medium supplemented with 10% FBS in 24-well plates (1. 25×105 cells/well); after incubation for 2 h, cells were washed with the serum-free medium three times and then medium containing 1% low melting point agarose was added to each well. Each treatment was performed in triplicate. After incubation at 27°C for 72 h, plaque formation in HzAM1 cells infected with variously treated viruses were examined by fluorescence microscopy. dsRNA of target genes was generated in vitro using T7 RiboMAX Express RNAi kits (Promega) according to the manufacturer’s instructions. The GFP gene was used to generate GFP dsRNA. 2 μg dsRNA was injected into the abdomen of late 2nd instar larvae using a microinjector (World Precision Instrument, Sarasota, CA, USA). Primers used to generate dsRNA are listed in S6 Table. The transcript levels of target genes at 3 days after dsRNA injection were confirmed by qPCR. The virus titers in hemolymph of insects where serpins were knocked down were determined by qPCR analyses of viral genomic DNA, using primers of ha39. HearNPV Bacmid DNA was isolated according to Bac-to-Bac manual (In vitrogen) and quantified by spectrophotometry at OD260. A standard curve was generated with ten-fold dilutions of Bacmid DNA. The cell-free hemolymph (1 μL) from serpin-5 or serpin-9 depleted larvae infected with HearNPV, were collected into 19 μL H2O, 80 μL virus disruption buffer (10 mM Tris-HCl pH7. 5,10 mM EDTA, 0. 25% SDS) and 5 μL proteinase K (20 mg/ml). Virions were lysed at 50°C for 1 h, and viral DNA was ethanol precipitated and resuspended in 30 μL H2O. The viral DNA was used as the PCR template. The mass spectrometry data (PXD006126) have been deposited in the PRIDE repository (http: //www. ebi. ac. uk/pride). All sequence data that support the findings of this study are available in GenBank with the following accession numbers: serpin-5 (KY680238), serpin-9 (KY680239), cSP4 (KY680240), cSP6 (KY680241), cSP8 (KY680243), cSP29 (KY680244), PPO1 (KY744277), PPO2 (KY744278).

Melanization is one of important modules in insect defense system. It consists of a cascade of clip-domain serine proteases (cSPs) that converts the zymogen prophenoloxidase (PPO) to active phenoloxidase (PO), which is negatively regulated by serpins. PO then catalyses the formation of melanin that physically encapsulates certain pathogens. Parasites and bacteria have evolved to produce specific proteins or antibiotic to suppress the melanization response of host insects for survival. However, the mechanisms by which virus persists in the face of the insect melanization are poorly understood. In this study, we show that a DNA virus baculovirus infection of the cotton bollworm, Helicoverpa armigera, reduced the levels of most cascade members in the host hemolymph and PO activity. By contrast, serpin-9 and serpin-5 were sequentially upregulated after the viral infection. Our results also reveal that melanization kills baculovirus in vitro. Serpin-5 and serpin-9 regulate melanization by directly inhibiting their target proteases cSP4 and cSP6, respectively and cSP6 activates PPO purified from hemolymph. Moreover, serpin-5/9-depleted insects show resistance to baculovirus infection. Our findings have enriched the understanding of molecular mechanisms by which pathogens suppress the melanization response of host insect for survival.

Abstract Introduction Results Discussion Materials and methods

invertebrates medicine and health sciences viral transmission and infection molecular probe techniques enzymes immunology microbiology enzymology animals developmental biology molecular biology techniques research and analysis methods immune system proteins immunoblot analysis proteins life cycles proteomics recombinant proteins molecular biology insects arthropoda biochemistry eukaryota virology biology and life sciences proteases larvae organisms

2017

Inhibition of melanization by serpin-5 and serpin-9 promotes baculovirus infection in cotton bollworm Helicoverpa armigera

Nuclear-mitochondrial conflict (cytonuclear incompatibility) is a specific form of Dobzhansky-Muller incompatibility previously shown to cause reproductive isolation in two yeast species. Here, we identified two new incompatible genes, MRS1 and AIM22, through a systematic study of F2 hybrid sterility caused by cytonuclear incompatibility in three closely related Saccharomyces species (S. cerevisiae, S. paradoxus, and S. bayanus). Mrs1 is a nuclear gene product required for splicing specific introns in the mitochondrial COX1, and Aim22 is a ligase encoded in the nucleus that is required for mitochondrial protein lipoylation. By comparing different species, our result suggests that the functional changes in MRS1 are a result of coevolution with changes in the COX1 introns. Further molecular analyses demonstrate that three nonsynonymous mutations are responsible for the functional differences of Mrs1 between these species. Functional complementation assays to determine when these incompatible genes altered their functions show a strong correlation between the sequence-based phylogeny and the evolution of cytonuclear incompatibility. Our results suggest that nuclear-mitochondrial incompatibility may represent a general mechanism of reproductive isolation during yeast evolution. Reproductive isolation preventing gene flow between diverging populations is crucial for the process of speciation [1]. One of the general reproductive isolation mechanisms that lead to hybrid inviability or sterility is genetic incompatibility (Dobzhansky-Muller incompatibility), which is caused by improper interactions between genetic loci that have functionally diverged in two different species [2], [3]. Since genetic incompatibility probably plays an important role at the incipient stage of speciation, identifying the incompatible loci and determining the selection forces underlying their functional divergence are vital for our understanding of how speciation occurs. In the past two decades, scientists have discovered several genetic loci causing hybrid sterility or inviability [4], [5]. Genes involved in genetic incompatibility have been cloned from a variety of organisms, including flies, platyfishes, mice, and Arabidopsis [6], [7], [8], [9], [10], [11]. Nonetheless, most of the genes identified were from Drosophila, and in most cases, only one component of the incompatible genetic loci was cloned. Systematic studies that involve more than two species in other organisms are still rare. Baker' s yeast, Saccharomyces cerevisiae, and its close relatives, the Saccharomyces sensu stricto yeasts, represent an interesting system for studying genetic incompatibility. These yeasts can mate with each other freely under laboratory conditions. Diploid hybrids collected from the wild or generated in the laboratory can reproduce asexually without showing any obvious defect. However, the viability of hybrid gametes (spores) is very low (about 0. 5%–1%), suggesting that there is strong postzygotic reproductive isolation between these yeast species [12]. Because yeast differs from flies in cellular complexity, life style, and population structure, studies in yeast will greatly expand our knowledge of speciation processes [13]. Using chromosome replacement lines of hybrids of two yeast species, a previous study identified a strong incompatibility between a S. bayanus nuclear gene, AEP2, and S. cerevisiae mitochondria that leads to interspecific F2 hybrid sterility [14]. It was found that the 5′-UTR regions of a mitochondrion gene, OLI1, have diverged dramatically between these two species. Since interactions between the Aep2 protein and the 5′-UTR region of the OLI1 mRNA are essential for OLI1 translation, the incompatibility is probably caused by the failure of Sb-Aep2 to recognize the divergent 5′-UTR region of Sc-OLI1. The finding raises a few interesting questions: Does the cytonuclear incompatibility play a general role in yeast reproductive isolation, does the AEP2-OLI1 type of interaction (activation of mRNA translation) represent a common mode of cytonuclear incompatibility, and is there a specific selective force driving this type of cytonuclear evolution? The mitochondrion is a critical component of cellular energy production and several metabolic pathways. In many organisms including yeast, proper mitochondrial functions are required for gamete development [15], [16], [17]. The mitochondrion contains its own genome, though one in an advanced state of degeneration. Most genes essential for mitochondrial functions have been transferred from the proto-mitochondrion genome to that of their host [18], [19]. As a consequence of these events, gene products from both mitochondrial and nuclear genomes are required for proper mitochondrial operations [20], [21]. In yeast, for example, the mitochondrial genome encodes only eight proteins, but it is estimated that ∼1,000 proteins function in mitochondria [22]. Although these two genomes are under different mutation and selection pressures, they are constrained to evolve coordinately to maintain optimal functions [23]; any change in mitochondria (adaptive or drifted) may require one or more consecutive changes in the nucleus [23], [24]. This type of interaction provides an ideal background for the evolution of Dobzhansky-Muller incompatibilities; when two populations containing well-adapted cytonuclear mutations mix their genomes, unmatched mitochondria and nuclei cause reduced hybrid fitness [25]. Cytonuclear incompatibility has been observed in a wide range of organisms, including primates, amphibians, flies, wasps, a marine copepod, and a variety of plants [26], [27], [28], [29], [30], [31]. In yeast, cytonuclear incompatibility has been tested directly by transferring mitochondria from one species or strain into cells of another, which clearly showed a species barrier between mitochondrial and nuclear genomes [32], [33]. From these studies, we know that the deleterious effects of cytonuclear incompatibility can lead to reduced fitness, hybrid sterility, or inviability. Nonetheless, molecular descriptions of such an intracellular conflict are rare, and its generality as an engine of speciation remains an open question. Here, we present results of a systematic study aimed at understanding the role of cytonuclear incompatibility in postzygotic reproductive isolation. We screened three sensu stricto yeasts, S. cerevisiae, S. paradoxus, and S. bayanus for incompatible genes causing F2 hybrid sterility and used the information about how these genes diverged in function to reconstitute the evolutionary history of the species. We found that only a few strongly incompatible gene pairs have evolved between these species. Two of them, MRS1 and AIM22, were identified and the molecular mechanisms of incompatibility were characterized. By analyzing the mutations of the MRS1 gene leading to its functional divergence between two species at the nucleotide level, we show that only three mutations make major contributions. Finally, we show that the functional divergence of these incompatible genes is correlated with the phylogeny, suggesting that cytonuclear incompatibility not only represents a general mechanism of reproductive isolation but has also occurred repeatedly during yeast evolution. A previous study has shown that incompatibility between a S. bayanus nuclear gene, AEP2, and S. cerevisiae mitochondria causes interspecific F2 hybrid sterility [14]. To examine whether nuclear-mitochondrial incompatibility represents a general mechanism of reproductive isolation in yeast, a systematic screen for such genes was conducted in three closely related yeast species: S. cerevisiae (Sc), S. paradoxus (Sp), and S. bayanus (Sb). Hybrid diploid strains between S. cerevisiae and S. paradoxus, or between S. cerevisiae and S. bayanus were induced to generate haploid spores containing different combinations of chromosomes from their parental species. These spores were then assayed for cytonuclear incompatibility. In hybrid diploids between species A and B, incompatibility could occur in two directions, between A-nucleus and B-mitochondria or between B-nucleus and A-mitochondria. When generating the yeast hybrids, we deliberately removed one parental type of mitochondria (by using ρ0 mutants, which lack mitochondrial DNA) so that we could unambiguously assign the direction of incompatibility. After sporulation of hybrids, viable spores were measured for their ability to grow on glycerol, a non-fermentable carbon source (Figure 1). The results showed strong cytonuclear incompatibility in most of the interspecific hybrids; we observed 66%±5%, 78%±7%, and 32%±16% of respiration-proficient spores in the interspecific crosses between Sc-ρ0 and Sp, Sb-ρ0 and Sc, and Sc-ρ0 and Sb, respectively, while the intraspecific crosses generated almost 100% of respiration-proficient spores (Figure 2). The percentage of viable spores that could grow on glycerol plates was used to estimate the number of strongly incompatible loci (see Materials and Methods). The results of this growth assay suggest that there are only one or few nuclear genetic loci strongly incompatible with mitochondria in the Sc-nucleus and Sp-mitochondria, the Sb-nucleus and Sc-mitochondria, and the Sc-nucleus and Sb-mitochondria pairs. Although no strong incompatibility was detected for the Sp-nucleus and Sc-mitochondria pair, slow spore growth (on glycerol plates) was commonly observed, suggesting that some incompatibility may exist in this pair as well. Because the incompatibility between the Sb-nucleus and Sc-mitochondria has been described earlier [14], we focus here on the remaining two pairs. In our experimental design, we constructed the hybrid diploids using spo11Δ mutants to prevent meiotic homologous recombination. We expected that hybrid F1 haploids unable to respire (see Figure 1) should carry a specific set of chromosomes containing the incompatible genes. To identify the genes responsible for the observed cytonuclear incompatibility, the respiration-deficient clones were first examined for their chromosome contents using species-specific PCR (see Materials and Methods). In the cross between Sc-ρ0 and Sb, the haploids carrying Sb-mitochondria but unable to respire lacked Sb-Chromosome 6,9, or both (Table S1A). When these hybrid clones were transformed with genomic DNA libraries to screen for those genes capable of rescuing the respiratory defect, Sb-MRS1 encoded on Sb-Chromosome 9 and Sb-AIM22 encoded on Sb-Chromosome 6 were isolated. The results from the cross between Sc-ρ0 and Sp are more complex, as all the respiration-deficient clones did not contain the two Sp-Chromosomes 4 and 9 (Table S1B), but were fully rescued by Sp-MRS1 alone, which is encoded on Sp-Chromosome 9. The reason why all the hybrid clones are missing Sp-Chromosome 4 is unclear. One possibility is that Sp-Chromosome 4 is also incompatible with Sc-Chromosome 9, an issue requiring further investigation. To rule out the possibility that the respiration defects caused by Sc-MRS1 and Sc-AIM22 were simply due to nuclear-nuclear incompatibility in the hybrid haploid clones, we crossed Sp-mrs1Δ, Sb-mrs1Δ, and Sb-aim22Δ with Sc-ρ0 and examined the respiratory ability of the diploid cells. All the diploid cells were still respiration-deficient, even though they contained a complete set of the S. cerevisiae genome (Figure 3). This result demonstrated that Sc-MRS1 is incompatible with Sp-mitochondria and that both Sc-MRS1 and Sc-AIM22 are incompatible with Sb-mitochondria. Yeast cells utilize non-fermentable carbon sources to induce meiosis. Previous studies have shown that respiration-deficient cells were unable to sporulate [34]. To confirm that indeed the cytonuclear incompatibility observed in our experiments contributes to reproductive isolation, the aforementioned diploid cells (Sp-mrs1Δ×Sc-ρ0, Sb-mrs1Δ×Sc-ρ0, and Sb-aim22Δ×Sc-ρ0) were grown on sporulation medium and examined for their sporulation efficiency. No ascus was observed in these cultures, while the control cultures sporulated efficiently. Thus, cytonuclear incompatibility caused by Sc-MRS1 or Sc-AIM22 results in reproductive isolation between these yeast species. Mrs1 is a mitochondrial protein required for excision of the aI5β intron in COX1 and the bI3 intron in COB [35], [36]. A previous study has shown that S. douglasii Mrs1 (S. douglasii is a synonym of S. paradoxus) is required to splice a S. douglasii-specific COX1 intron not existing in the S. cerevisiae COX1 [37]. We observed a similar result in S. bayanus. Sc-MRS1 could not complement the respiratory defect of the Sp-mrs1Δ or Sb-mrs1Δ mutants (Figure S1). When Sb-Mrs1 or Sp-Mrs1 were replaced by Sc-Mrs1, the level of mature COX1 mRNA was drastically reduced but the COB mRNA was not affected (Figure 4A). We further analyzed the translation products from purified mitochondria and confirmed that only the Cox1 protein was missing (Figure 4B). By contrast, Sc-MRS1 transcription and protein transport into mitochondria appear to be normal in both S. paradoxus and S. bayanus (Figure 5). Thus the incompatibility between the mitochondrial and nuclear genomes is most likely due to a change in the splicing specificity of Mrs1 that occurred after S. cerevisiae diverged from the common ancestor of S. cerevisiae and S. paradoxus. In order to understand how the COX1 introns evolved in yeast, we compared COX1 intron patterns between S. cerevisiae, S. paradoxus, S. bayanus, S. servazzii, Candida glabrata, and Kluyveromyces thermotolerans. S. servazzii and C. glabrata are species outside the sensu stricto complex and K. thermotolerans is a pre-WGD (whole-genome duplication) species. The comparison indicates that the intron in the Sp- or Sb-COX1 gene incompatible with Sc-Mrs1 is an ancient intron (Figure 4C). Since this intron was eliminated only in the S. cerevisiae lineage, it is likely that the Sc-MRS1 gene product lost the ability to splice this intron after the intron loss event (by adaptation or drift). The fact that S. cerevisiae and S. paradoxus share a high degree of nucleotide sequence identity allowed us to determine the key mutations underlying the functional change of the MRS1 gene and to reconstruct the process of Mrs1 evolution. To this end, chimeric proteins with regions from Sc-Mrs1 and Sp-Mrs1 were constructed and assayed for their ability to complement the respiratory defect of the Sp-mrs1Δ mutants. We found that the functional difference between Sc-Mrs1 and Sp-Mrs1 is mainly determined by a region comprising 63 amino acids (a. a. sites 179–241; Figure 6A). Nine nonsynonymous changes have accumulated in this region since Sc-MRS1 and Sp-MRS1 diverged (Figure S2A). To determine which of these mutations led to the altered activity of Mrs1, we introduced the S. paradoxus version of each of these sites into Sc-MRS1 and assayed their ability to rescue the Sp-mrs1Δ respiratory defect. Analogous experiments were performed in the reverse direction by introducing Sc-specific amino acids into Sp-MRS1. Only mutations in three amino acids (Sc to Sp: T201A, V211A, and M227I) had obvious contributions (Figure 6B and Figure S2). When all three mutations were combined together in a single mutant clone, it explained most of the effect. The growth rate of cells carrying the mutant plasmid (Sc-MRS1-n123) in a glycerol-containing medium is about 75% of that of wild type cells. Interestingly, these three amino acids are all conserved in S. paradoxus, S. kudriavzevii, and S. bayanus but are changed in S. cerevisiae. Among the other six nonsynonymous changes in this region, only two of them (Sc to Sp: K186E and R223Q) share the same pattern. This observation is consistent with our hypothesis that the functional change of MRS1 occurred only after an ancestral COX1 intron was lost in S. cerevisiae. Our results clearly demonstrate that the observed nuclear-mitochondrial incompatibility results from cumulative effects of multiple mutations. On the other hand, they also suggest that only a small fraction of the nonsynonymous changes between species contributes to the incompatibility. The Mrs1 protein does not contain any specific functional domain. However, a recent study has used computer modeling to predict the Mrs1 protein structure [38]. We examined the relative positions of these residues using the predicted structure. All three residues (a. a. 201,211, and 227) were found to localize on the RNA-binding surface (Figure S3). It is possible that these amino acid changes have altered the substrate specificity of Mrs1 that leads to the incompatibility. S. cerevisiae nuclei can only support S. bayanus mitochondria if the cells contain a S. bayanus AIM22 gene. AIM22 encodes a lipoate-protein ligase homologous to the bacterial lplA protein [39], [40]. In eukaryotic cells, lipoic acid has been shown to be an essential cofactor to a variety of mitochondrial proteins and lipoate-protein ligase (together with other enzymes) is required to lipoylate these mitochondrial targets [39], [41]. We have not investigated which mitochondrial protein or enzyme in Sb-mitochondria is incompatible with Sc-Aim22. However, our data indicate that the incompatibility is not caused by misregulation of the Sc-AIM22 transcription or failure to transport the Sc-Aim22 to Sb-mitochondria (Figure 5). To investigate whether the AIM22 gene in the S. cerevisiae-S. paradoxus branch has been under positive selection during evolution, we measured the ratio of nonsynonymous (Ka) to synonymous (Ks) nucleotide substitution rates between these species. The Ka/Ks values of AIM22 in the S. cerevisiae–S. paradoxus, S. cerevisiae–S. bayanus, and S. paradoxus–S. bayanus pairs are 0. 13,0. 12, and 0. 14, showing no sign of positive selection (Ka/Ks>1). We also ran a PAML' s branch-model analysis on the AIM22 gene [42], [43] but could not detect any signature of significant positive selection (Figure S4). Previous studies of AEP2 have shown that the nuclear-mitochondrial incompatibility is asymmetrical. While Sb-Aep2 is completely incompatible with Sc-mitochondria, Sc-Aep2 retains partial compatibility with Sb-mitochondria [14]. We also found that incompatibility caused by AIM22 and MRS1 only occurred in one direction. Although Sc-AIM22 and Sc-MRS1 are not compatible with Sb-mitochondria, Sb-AIM22 could complement the Sc-aim22Δ mutant and both Sb- and Sp-MRS1 could rescue the Sc-mrs1Δ mutant. Nuclear-mitochondrial incompatibility has been shown to occur commonly between different yeast species [32], [33]. However, it is unclear whether nuclear-mitochondrial incompatibility between different pairs of species has evolved at different periods of time in different species lineages. To address this issue, we tested the compatibility between different orthologues of these incompatible genes and mitochondria from each species. Different orthologous alleles of MRS1, AIM22, or AEP2 were transformed into the S. cerevisiae, S. paradoxus, or S. bayanus mutants in which the wild-type copy had been deleted. The transformants were then tested for their ability to grow on glycerol plates (Figure 7A and Figure S1). Information from these assays was used to deduce the time of occurrence of the functional change leading to incompatibility. A clear correlation between the emergence of cytonuclear incompatibility and the phylogeny is observed (Figure 7B). Sc-MRS1 is incompatible with Sp-mitochondria or Sb-mitochondria, indicating that the functional change of Sc-MRS1 occurred only in the S. cerevisiae lineage. On the other hand, the functional change of AIM22 represents a more ancient event in the common ancestor of S. cerevisiae and S. paradoxus, because Sc-AIM22 and Sp-AIM22 are exchangeable but neither of them is compatible with Sb-mitochondria. Finally, the data suggest that Sb-AEP2 diverged in function only in the S. bayanus lineage since Sb-AEP2 is incompatible with either Sc-mitochondria or Sp-mitochondria. These results provide evidence that nuclear-mitochondrial incompatibility has repeatedly arisen during the history of yeast evolution and probably represents an important reproductive isolation mechanism in yeast species. Previous studies in Saccharomyces yeasts have suggested that the deleterious effect of DNA sequence divergence on meiotic recombination probably contributes in part to reproductive isolation during yeast evolution [12], [44], [45]. Our results, on the other hand, suggest that nuclear-mitochondrial incompatibility is also a promising candidate for causing intrinsic hybrid dysfunction. In fact, these two mechanisms are not mutually exclusive. Genomes from different populations will accumulate enough DNA sequence divergence only after extended periods of allopatric isolation, but the effect of sequence divergence can be applied directly in the diploid F1 hybrid cells. Cytonuclear incompatibility can be achieved by only a few mutations, and its deleterious effect can be carried on to the F1 gamete or F2 progeny. In theory, cytonuclear incompatibility has a stronger impact on blocking gene flow between populations in the early stages of speciation, and reproductive isolation is reinforced later on when populations have accumulated enough DNA sequence divergence. These two mechanisms can be complementary to each other in terms of their effects and evolutionary trajectories. It will be interesting to investigate whether cytonuclear incompatibility exists between different populations of the same species. Reproductive isolation resulting from genetic incompatibility has been discovered in a variety of organisms [7]. Most of the examples characterized so far are caused by interactions between nuclear genes. In yeast, this type of incompatibility has been investigated in a few studies, yet no strongly nuclear-nuclear incompatible genes were identified [14], [46]. On the other hand, cytonuclear incompatibilities were observed in hybrids between different yeast species or populations [47]. Cytonuclear incompatibility probably represents a more general mechanism of reproductive isolation in yeast. By analyzing the functions and the interacting components of the identified incompatible genes, we discovered that cytonuclear incompatibility could be achieved by multiple molecular mechanisms: intron splicing, protein lipoylation, and activation of mRNA translation. This suggests that cytonuclear incompatibility in yeast can occur in various pathways by diverse molecular mechanisms. Scientists usually compare orthologous sequences from different species and use the detected molecular signatures to infer the evolutionary process of a gene. Since our data suggest that Mrs1 changed function only after S. cerevisiae diverged from S. paradoxus, it is reasonable to assume that the altered function originated from amino acid changes that occurred specifically in the S. cerevisiae lineage. By comparing the coding regions of the MRS1 orthologues from S. cerevisiae, S. paradoxus, S. kudriavzevii, and S. bayanus, we observed 22 amino acids that are different in S. cerevisiae but are conserved in the other three species. Interestingly, our functional assays showed that only three of these amino acid changes contribute significantly to the functional differences of Mrs1. The other mutations may have very minor effects unable to be detected by our functional assays or have been fixed simply by genetic drift. In a previous study, Rawson and Burton have also observed that three amino acid changes in a nucleus-encoded cytochrome c (CYC) are responsible for cytonuclear incompatibility between different populations of a marine copepod, Tigriopus californicus [48]. These results illustrate the importance of mapping the critical amino acid changes in order to understand how a gene evolved. What is the major driving force underlying the evolution of mitochondria? It is known that the mitochondrial genome suffers a higher mutation load because it is constantly facing higher levels of oxidative reagents and its DNA protection system is more primitive as compared to the nuclear genome. With a much smaller copy number of mitochondrial DNA in yeast (30–80 molecules of mitochondrial genomes in yeast cells compared to 1,000–5,000 molecules in animal cells), mutations may be fixed frequently in the mitochondrial genome by genetic drift. Since wild yeast often propagate clonally in natural environments [13], a founder cell with mild deleterious mutations in its mitochondrial genome may have a chance to accumulate suppressors in the mitochondrial or nuclear genomes to rescue the fitness before its progeny are outcompeted by cells from another population. Alternatively, mitochondrial evolution may be driven by an “arms race” process between selfish mitochondrial DNA and the “wild-type” mitochondrial or nuclear genomes. It is commonly observed in yeast that by manipulating the host replication or segregation systems, some mitochondrial genomes allow themselves to be inherited more efficiently, even though they may be carrying compromised respiratory functions [49]. In a sexual population, the other mitochondrial or host nuclear genomes will be selected to counteract this selfish behavior or the deleterious effects carried by the selfish DNA. Such an “arms race” may allow mitochondrial genomes to evolve faster than by genetic drift (in a fashion similar to positive selection). Incompatibilities driven by arms races between different genetic components have been suggested in a few recent studies [7]. Among these genetic conflicts, most of them are caused by selfish elements manipulating segregation distortion [11], [50], [51], [52], [53]. The “hybrid necrosis” phenotype observed in Arabidopsis probably results from recurrent conflicts between the host defense system and pathogens [10]. The genetic conflict caused by selfish mitochondrial genomes may represent another type of arms race. It will be interesting to investigate whether the arms race model can explain the cytonuclear incompatibilities observed in other organisms. Finally, ecological adaptation may also contribute to mitochondrial evolution. Evidence suggests that adaptive mutations have occurred in mitochondria in response to different environmental stresses that interfere with cellular energy demands [48], [54], [55]. In yeast, it has been shown that S. bayanus grows much better than S. cerevisiae on media containing only non-fermentable carbon sources, with the opposite observed in fermentable media. It has been speculated that the changes in Sb-AEP2 and Sb-OLI1 are a part of such ecological adaptation [14]. Reciprocal crosses between species often generate asymmetrical hybrid viability or sterility, a general feature of intrinsic postzygotic isolation called Darwin' s corollary [56], [57], [58], [59], [60]. The Dobzhansky-Muller model suggests that alleles causing reproductive isolation act asymmetrically [61]. However, asymmetries in allele action do not necessarily lead to asymmetries in reproductive isolation [62]. Incompatibility between autosomal loci affects both reciprocal crosses identically. Asymmetric reproductive isolation is usually caused by incompatibility between autosomal loci and uniparentally inherited materials, such as cytoplasmic elements and sex chromosomes. It has been suggested that cytonuclear incompatibility caused by different trajectories of mitochondrial evolution in different species may contribute to this phenomenon [59], [63]. Our results provide the molecular basis to support this hypothesis. Since cytonuclear incompatibility can be achieved by multiple molecular mechanisms and evolve at different rates in different lineages, it can serve as a general mechanism of reproductive isolation and also create asymmetrical reproductive isolation between species. Yeast strain genotypes are listed in Table S2. The parental S. cerevisiae strains (JYL1127 and JYL1128) are isogenic with W303 (MATa ura3-1 his3-11,15 leu2-3,112 trp1-1 ade2-1 can1-100). The parental S. paradoxus strains (JYL1137 and JYL1138) are derived from YDG 197 and are a gift from Dr. Duncan Greig (University College London, UK). The parental S. bayanus strains (JYL1030 and JYL1031) were derived from a strain (S. bayanus #180) collected by Dr. Duccio Cavalieri (University of Florence, Italy). The strains JYL1157,917, and 1256 were used for measuring hybrid fertility. Substitutive and integrative transformations were carried out by the lithium acetate procedure [64]. Media, microbial, and genetic techniques were as described [65]. a and α cells of one species (S. cerevisiae, S. paradoxus, or S. bayanus) were crossed to α and a ρ0 cells of another species to generate F1 hybrid cells. In both species, the SPO11 genes were deleted to prevent meiotic recombination between homologous chromosomes. Strains from the first species also had a URA3 marker inserted near the centromere of a chromosome so that haploid spores could be efficiently selected on 5-FOA plates. After the hybrid diploids were induced to sporulate, viable spores were tested for their respiratory ability. Respiration-deficient clones were further crossed with ρ0 mutants of the parental species to confirm that the defect was caused by cytonuclear genomic incompatibility (Figure 1). A genetic analysis was used to estimate the number of the nuclear genes that are incompatible with mitochondrial DNA from another species: in a cross between two spo11Δ mutants, meiotic recombination does not occur and homologous chromosomes segregate randomly [66], [67]. For a specific chromosome A, 25% of the spores will carry two A chromosomes (of both parental types), 50% of them will carry one A chromosome, and 25% of them will have no A chromosome (these cells will not survive so they will not be counted in our later analysis). If only one gene on chromosome A is recessively incompatible with mitochondrial DNA (in our case, all spores from a single cross carry the same parental type of mitochondrial DNA), we expect to see that 66% of the viable spores are Gly+ (1/3+2/3×1/2 = 66%). The spores carrying two A chromosomes should be Gly+ because the incompatibility is recessive. Half of the spores that carry only one A chromosome should be Gly+ if their A chromosome is from the same parent of the mitochondrial DNA. If two genes on different chromosomes are involved, we expect to see that 43% of the viable spores are Gly+ (66%×66% = 43%). Epistatic effects are not taken into account in this analysis because considering such effects would make it impossible to estimate the involved locus number. We examined the chromosome composition of respiration-deficient hybrid lines by PCR using species-specific primers for all 16 chromosomes. Yeast genomic DNA libraries constructed from S. bayanus or S. paradoxus genomes were transformed into the Gly− clones to screen for incompatible genes. Yeast strains were grown in 3 ml YPD liquid cultures at 30°C to stationary phase and total RNA was isolated using Qiagen RNeasy Midi Kits (Qiagen, Valencia, CA). Ten µg of total RNA was separated on a 1. 3% agarose-formaldehyde gel and then transferred to a nylon membrane (Millipore, Billerica, MA). Northern blotting was performed as described [68]. Because both COX1 and COB signals were difficult to be completely washed out and the background was high after the second hybridization, we used the same RNA samples to load three repeats on the same gel, cut the membrane after transferring, and hybridized each repeat with one specific probe. The gene-specific primers for the DIG-labeled probes were described in Rodeheffer et al. [69] except for the probe of COX2. The primers for COX2 were 5′-TTAATGATAGTGGTGAAACTGTTG-3′ and 5′-CCAAAGAATCAAAATAAATGCTCG-3′. The probe generation and hybridization were as described in the Genius System User' s Guide (Roche, Indianapolis, IN). Total RNA was isolated using Qiagen RNeasy Midi Kit (Qiagen). First-strand cDNA was synthesized using High Capacity cDNA Reverse Transcriptase Kit (Applied Biosystems, Foster City, CA) at 37°C for 2 h. A 25-fold dilution of the reaction products was then subjected to real-time quantitative PCR analysis using gene-specific primers, the SYBR Green PCR master mix, and an ABI-7000 sequence detection system (Applied Biosystems). Data were analyzed using the built-in analysis program. Yeast cells were cultured in selection medium at 30°C to the mid-exponential phase. Mitochondria fractions were prepared as described [70]. The post-mitochondrial supernatant (PMS) fractions were collected immediately following centrifugation of mitochondria. The PMS was precipitated with ice-cold 10% trichloroacetic acid (TCA) and centrifuged at 14K rpm for 15 min at 4°C, followed by an ice-cold acetone wash using the same centrifugation conditions. Laemmli sample buffer was added to the mitochondrial pellets resuspended in TE buffer plus protease inhibitors, and to the PMS fractions. For Western blot analyses, mitochondrial protein (40 µg) was separated by SDS-PAGE using the GE Healthcare Life Science system (Mini-Vertical Units SE260) and then transferred to a BioRad Immun-Blot PVDF membrane in transfer buffer (25 mM Tris base, 200 mM glycine, 20% methanol, 0. 01% SDS). The immunopositive bands were visualized by using Western Lightning chemiluminescence reagent (PerkinElmer, Waltham, MA). Anti-G6PDH polyclonal antibody was purchased from Sigma (St. Louis, MO). Anti-c-Myc polyclonal antibody (A-14) was from Santa Cruz Biotechnology (Santa Cruz, CA). Anti-Hsc82 and Anti-Atp2 polyclonal antibodies were obtained from Dr. Chung Wang (Academia Sinica, Taipei). Cells in the early log phase were inoculated in 2 ml standard minimal medium and grown for 30 min. Cycloheximide stock solution (10 ml/mg in dH2O) was added to a final concentration of 100 µg/ml. Cells were incubated for 5 min prior to addition of 0. 1 mCi of [35S]-methionine. The reaction was terminated after 1 h by adding 2 ml of chase solution (1% casamino acid, 2 mg/ml Na2SO4). Mitochondria were prepared as described [65]. The radiolabeled proteins were separated on a 17. 5% polyacrylamide gel. The ancestral AIM22 sequence of S. cerevisiae and S. paradoxus was constructed using a maximum likelihood procedure (free-ratio model) as implemented in the PAML package. The Ka/Ks ratios were calculated using DNAsp 5. 0 [71]. The Ka/Ks ratios were estimated using the free-ratio branch model of PAML. This method allows the Ka/Ks ratios to vary among branches in a given phylogeny and is useful in detecting positive selection acting on particular lineages [42]. The AIM22 sequences from five Saccharomyces sensu stricto species (S. cerevisiae, S. paradoxus, S. mikatae, S. bayanus, and S. kudriavzevii) were used in our analysis. Results of the branch-model PAML analysis suggested that no particular lineage was subjected to positive selection (i. e. , Ka/Ks>1) (Figure S4). Because both Sc-AIM22 and Sp-AIM22 are incompatible to S. bayanus mitochondria, it is possible that there has been an accelerated sequence evolution in either the lineage leading to S. cerevisiae and S. paradoxus, or in the lineage leading to S. bayanus. To test this hypothesis, we performed an analysis as described in Yang and Nielsen [43]. One ω-ratio and two ω-ratio branch models were implemented. The first model assumes that only one ω-ratio leads to whole phylogeny branches and the second model assumes that one ω-ratio leads to the branch that we are interested in and another ω-ratio leads to the rest of the other branches. Twice the difference of their likelihood ratio between any two models (likelihood ratio test; LRT) was then compared against a chi-square distribution. The degree of freedom (d. f.) was obtained based on the difference of parameters used in two different models. From our analysis, no accelerated evolution was observed in any lineage along the phylogeny of AIM22.

Hybrids between species are usually inviable or sterile, possibly due to functional incompatibility between genes from the different species. Incompatible genes are hypothesized to encode interacting components that cannot function properly when paired with alleles from another species. To understand how incompatible gene pairs result in hybrid sterility or inviability, it is important to identify these genes and reconstruct their evolutionary history. A previous study has shown that incompatibility between nuclear and mitochondrial genomes (cytonuclear incompatibility) causes hybrid sterility between two yeast species. To expand on these findings, we screened three yeast species for genes involved in cytonuclear incompatibility, discovering two nuclear genes, MRS1 and AIM22, which encode proteins that are unable to support full mitochondrial function in the hybrids. Of these two genes, Mrs1 is required for removing a specific intron in the mitochondrial COX1 gene. By comparing different yeast species, we find a clear coevolutionary relationship between Mrs1 function and the COX1 intron pattern. We also show that changes in three amino acids in the Mrs1 RNA-binding domain are sufficient to make Mrs1 incompatible in hybrids. Our results suggest that cytonuclear incompatibility may represent a general mechanism of reproductive isolation during yeast evolution.

Abstract Introduction Results Discussion Materials and Methods

evolutionary biology genetics and genomics/microbial evolution and genomics evolutionary biology/microbial evolution and genomics evolutionary biology/evolutionary and comparative genetics

2010

Multiple Molecular Mechanisms Cause Reproductive Isolation between Three Yeast Species

Human Immunodeficiency Virus 1 uses for entry into host cells a receptor (CD4) and one of two co-receptors (CCR5 or CXCR4). Recently, a new class of antiretroviral drugs has entered clinical practice that specifically bind to the co-receptor CCR5, and thus inhibit virus entry. Accurate prediction of the co-receptor used by the virus in the patient is important as it allows for personalized selection of effective drugs and prognosis of disease progression. We have investigated whether it is possible to predict co-receptor usage accurately by analyzing the amino acid sequence of the main determinant of co-receptor usage, i. e. , the third variable loop V3 of the gp120 protein. We developed a two-level machine learning approach that in the first level considers two different properties important for protein-protein binding derived from structural models of V3 and V3 sequences. The second level combines the two predictions of the first level. The two-level method predicts usage of CXCR4 co-receptor for new V3 sequences within seconds, with an area under the ROC curve of 0. 937±0. 004. Moreover, it is relatively robust against insertions and deletions, which frequently occur in V3. The approach could help clinicians to find optimal personalized treatments, and it offers new insights into the molecular basis of co-receptor usage. For instance, it quantifies the importance for co-receptor usage of a pocket that probably is responsible for binding sulfated tyrosine. Specific protein interactions are central to biological processes, and the infection of cells with viruses is no exception there. In the case of pathogenic viruses, such protein interactions are potential targets for medical intervention. An example of particularly high relevance is Human Immunodeficiency Virus 1 (HIV-1). HIV-1 enters human cells in a process that comprises several steps, including the binding of the viral gp120 protein to the cellular receptor protein CD4 and a co-receptor protein, usually one of the two chemokine receptors CCR5 and CXCR4 [1]. The type of co-receptor used by the virus, the so-called co-receptor tropism, has a prognostic value, since patients with a CXCR4-tropic virus (“X4 virus”) progress faster to Acquired Immunodeficiency Syndrome (AIDS) compared to patients with a CCR5-tropic virus (“R5 virus”) [2]. In addition to the purely X4- and R5-tropic viruses, there are also “dual-tropic” strains, able to use both co-receptors (“R5X4 virus”). Recently, the first drug (Maraviroc [3]) that binds to CCR5, and thus inhibits productive binding of gp120, has been approved by regulatory authorities in several countries. This has made the determination of co-receptor tropism directly relevant to anti-retroviral treatment, as CCR5-inhibitors are of course inactive against X4 virus. The standard way of determining co-receptor tropism is by cell-based assays [4], [5]. The main drawbacks of these assays are that they are currently only carried out by a handful of specialized laboratories worldwide, and that the overall procedure typically takes several weeks. These impediments to the wide application of entry inhibitors could be overcome by an approach similar to genotypic drug resistance testing [6], where drug resistance of a viral strain is inferred from comparison of mutational patterns obtained from sequencing parts of the genome of that strain with patterns of validated resistance mutations. This is a relatively fast and cheap standard procedure established in many clinics. At first glance, genotypic testing for co-receptor tropism seems to be possible since the main molecular determinant of tropism is known to be the third variable loop (V3) of the viral glycoprotein gp120 [7], a peptide stretch of about 35 amino-acids with a disulfide bridge connecting the terminal cysteins. Unfortunately, as suggested by its name, V3 is notorious for its high sequence variability [8] including also some variability in length, and this has made it difficult to use it as a basis for genotypic co-receptor tropism testing. Nevertheless, the relevance of the quest has prompted many groups to develop models that link properties of V3 to co-receptor tropism. The importance of electrostatics for co-receptor tropism has been recognized early on, and the best-known model, the so-called 11/25-rule, refers to charges of V3-residues 11 and 25: if one of these is positive, then the virus is CXCR4-tropic [9], [10]. This rule has a specificity of more than 0. 9 (few false positives), but only a low to moderate sensitivity (many false negatives) of about 0. 4–0. 6, depending on the test data, which is not satisfactory for routine clinical application. To improve predictions from sequence, several groups have applied machine learning methods, such as artificial neural networks [11], position specific scoring matrices [12], decision trees, or support vector machines [13]. Still, prediction accuracies fall short of what seems reasonable for regular clinical use [14]. It is unclear whether the limited accuracies are the footprint of tropism-determinants outside V3, or the consequence of model imperfections. A milestone for the understanding of co-receptor tropism was the X-ray structure of gp120 with the V3 loop in a biological context [15]. This paved the way for the development of prediction methods that use, in addition to V3 sequence, structural information. To our knowledge, the first of these methods has been that of Sander et al. [16], which was mainly based on geometric distances of amino-acid pairs within the structure of V3. Although our method, detailed in the following, relies on the same experimental structure by Huang et al. [15], it differs from that of Sander et al. in several respects, e. g. it deals with indels, and, perhaps most crucially, it uses as descriptors properties that directly determine interaction of V3 with the co-receptors. By the latter we consider a seemingly trivial but fundamental fact that so far has not been thoroughly exploited: although V3 is highly variable, all X4-tropic V3 loops share one property, namely, they preferentially have a physical binding interaction with CXCR4, while R5-tropic V3 loops preferably interacts with CCR5. The accuracy of the method makes it attractive as clinical tool for patient tailored decisions on treatment with entry inhibitors, and it suggests that co-receptor tropism can be explained almost exclusively based on V3. We aim at a computational method that for a given amino acid sequence of V3 predicts the tropism class “X4” (including dual-tropics), or “R5”. Predictions by the method should have an accuracy close to 100%, and be robust against the high diversity of V3, both in terms of sequence and length. In agreement with experimental data, we based the method on the assumption that the co-receptor tropism of HIV-1 is determined by a preferential physical interaction between a V3 loop and one of the co-receptors. We further assumed that both molecules interact while taking specific conformations. While little is known about the conformations of the extracellular parts of the co-receptors, there is a crystal structure available for a CCR5-tropic V3 loop [15]. In the first step of our approach we therefore modeled the conformations of V3 sequences of known tropism using this crystal structure as a template (see Materials and Methods). The modeled conformations enable the estimation of spatially distributed physical quantities that contribute to differential interactions of the V3 loops with the respective co-receptor, namely the values of the electrostatic potential around each V3 loop (“electrostatics hull”). Using these sets of and the corresponding tropism “X4” and “R5”, respectively, we trained a first random forest [17] classifier. Tropism classification of unseen V3 sequences is performed by automated modeling of the new V3 conformation, computation of, and application of the previously trained random forest. The output is a probability for the given V3 sequence to belong to the X4 class (and not to the R5 class). Although the first step explicitly takes into account conformation dependent physical properties that are of direct relevance to the differential interaction with the two co-receptors, we do not expect a perfect classifier from this first step for a number of reasons. For example, it is unclear whether the crystal structure is an appropriate template for all V3 sequences. In fact, V3 is known to be flexible [18], and there may even be a conformational switch between X4- and R5-tropic V3 loops [19]. Hence, we trained in a second step another random forest classifier solely on V3 sequences and with the hydrophobicity scale of Kyte and Doolittle [20] as descriptor. This descriptor has been derived by amalgamating several properties of amino-acids into a single scale, notably experimental results on solubility; it happens also to map amino-acids of opposite electrical charges to different scale values. Thus, this second classifier probably captures aspects of the relation between sequence and tropism that are at least partially complementary to those considered by the first classifier. In the final step of our approach, we trained a third random forest classifier with the two tropism class probabilities obtained from the previous two steps as input. Thus, application of the whole approach to an unseen V3 sequence includes application of a first level set of two random forests considering conformational and sequence properties, and a second level random forest using the outcomes of the first level for the final classification. Application of the classifier to a new V3 sequence to predict its co-receptor tropism takes a few seconds on a state-of-the-art CPU core. In the cross-validation, X4 sequences were detected with a sensitivity of (at a specificity of 0. 97), and the area under the ROC curve (AUC) was (full set of sequence and tropism data used for training and cross-validation is provided as Supporting Information). The method is described in greater detail in the following sections. The findings outlined in the introduction are compatible with a direct physical interaction between V3 and the respective co-receptor. Specifically, the 11/25 rule and the association of V3 net charge with tropism [9], [10] point to the impact of electrostatics on co-receptor tropism. In previous work, electrostatics has been considered in several ways, including the mentioned 11/25 rule, both alone and in combination with overall net charge [11], and also more complex relations such as an 11/24/25 rule [21]. Although these phenomenological rules have been helpful in guiding research, they are too simple to accurately capture the underlying molecular process, which limits their predictive power. To develop a more accurate model, we therefore first considered the one conserved feature that defines each of the tropism classes, namely the preferential interaction of V3 with one of the co-receptors, in particular their electrostatic interaction. Unfortunately, it is currently not possible to compute electrostatic energies of complexes of V3 and co-receptors since this necessitates availability of the structures of these complexes, which are unknown as yet. Thus, we resorted to the electrostatic potential around the V3 loops as alternative descriptor. Fulfillment of the following three assumptions is sufficient, though not necessary, to justify the choice of as descriptor: first, electrostatics is crucial for preferential interaction; second, the X-ray structure of the V3 loop from Huang et al. [15] represents the typical conformation of V3 loops, and conformations of all V3 loops can be derived as homology models from this X-ray structure; third, V3 loops bind to the co-receptors in the same binding mode. If these conditions are satisfied, preferential interactions of V3 loops with co-receptors can be mapped on differences in, essentially because different will in general lead to different interaction energies with unknown but constant co-receptor charge densities. Technically, we restricted computation of to an “electrostatics hull”, a discretized surface of points in space around the template V3 structure of Huang et al. [15]. The hull should be, on one hand, wide enough to enclose all superimposed V3 loops with a certain safety margin, and, on the other hand, tight enough to reflect the differences of from different V3 loops. We obtained good results with a hull in a distance of 0. 6 nm to the solvent accessible surface of the template V3 structure. For each V3 sequence of known co-receptor tropism in the training set, a homology model was generated based on the template X-ray structure. Then the electrostatic potential at the points of the electrostatics hull was computed by solving the Poisson-Boltzmann Equation [22]. A random forest [17] was trained using vectors of length as input, and as responses the corresponding measured tropisms, with. Using the leave-one-patient-out scheme for cross-validation (see “Materials and Methods”) we arrived for this classifier at an AUC of (“ESP” in Fig. 1). The analysis based on the electrostatics hull opens the possibility of deriving a co-receptor specific pharmacophore pattern of V3 loops. Fig. 2 shows points of the electrostatics hull that are of highest importance for the classification by the random forest, with importance here defined as percentage decrease in accuracy in classification if, for the respective point of the electrostatics hull, descriptor values are randomly permuted [17]. As could be expected, some important points cluster in the region around residues 11,24, and 25, though their dispersion makes it difficult to associate them with single residues. The majority of these points are located on the side to which most of the amino-acid side-chains point in the crystal structure (see also Supporting Information file Text S1). Interestingly, there is another important region on the opposite side of the loop between residues 6 and 30 that may be involved in the binding of sulfated tyrosines in the N-terminal region of CCR5 [23]. In Fig. 2 important positions are colored according to average electrostatic potential in the R5, R5X4, and X4 classes. The potential around R5-tropic V3 is generally lower as compared to X4-tropic V3, in particular around residues 24 (in agreement with the 11/24/25 rule) and 30. The coloring shows that R5X4-tropic V3 usually have values between those of R5 and X4, while at a few patches they are chimeras of the mono-tropic classes. The latter is true between residues 6 and 30 and close to residue 25 where R5X4 on average resembles R5, and around residue 11 and close to residue 24 where R5X4 is more similar to X4. The classification based on the values of on the electrostatics hull may fail in some cases, e. g. because some V3 sequences could prefer conformations not adequately represented by the X-ray structure of Huang et al. [15] that forms the basis of the electrostatics hull computation. We have therefore trained a second random forest, basically using as input the Kyte-Doolittle hydrophobicity values [20] of the residues along the V3 sequences, and as response again the measured tropisms. The hydrophobicity scale seemed suitable as it also captures physically motivated properties that are relevant for binding. An obstacle to sequence based learning was the high sequence diversity in our dataset so that standard multiple sequence alignment methods did not return clear profiles. This may have been the reason why other groups used for preparation of sequence data e. g. pairwise alignments to a reference sequence [11], manual alignments [24], or combinations of computational and manual multiple sequence alignments [12]; in these methods insertions and deletions were usually treated ad hoc, e. g. by removing insertions beyond a sequence length of 35. We have sought a simple algorithm that considers all sequences in a systematic and automated way irrespective of sequence length. This algorithm essentially leads to “normalized sequences” of uniform length with interpolated hydrophobicity values as descriptors. In detail, we normalized all sequences to the maximum length of occurring in the dataset. In the normalization procedure each sequence of residues is first arranged along a continuous pseudo-sequence axis with equal distances of between all neighbor residues. If the first residue is placed at pseudo-sequence position 1, this equidistant arrangement brings the th residue to pseudo-sequence position, while the residues in-between are in general at non-integer positions. In the second step of the normalization procedure, hydrophobicity values at the integer positions of the normalized sequence are linearly interpolated from the neighboring positions of the previously determined pseudo-sequence and their respective Kyte-Doolittle values, i. e. if the normalized sequence position has two neighbors in the pseudo-sequence at and with Kyte-Doolittle values and, respectively, then the hydrophobicity descriptor value at normalized sequence position is. This normalization leads to uniform sequence lengths with a consistent and automated treatment of insertions and deletions. Random forests trained on normalized sequences with interpolated Kyte-Doolittle descriptors had an AUC of, and thus about the same prediction performance in cross-validation as that trained on the electrostatics hull (see Fig. 1). Fig. 3 shows the distribution of the importance for the random forest error of the normalized sequence positions 1 to 38, with importance here defined as percentage decrease in accuracy in classification if, at the respective normalized sequence position, descriptor values are randomly permuted [17]. The highest peak is in the vicinity of position 11, in agreement with the 11/25 rule (note that in the sequence normalization procedure described above most sequences are stretched towards the maximum length of 38, and this stretch shifts position 11 of the amino-acid sequence towards position 12 of the normalized sequence). Position 25 does not stick out prominently; in fact, at position 25 of the normalized sequence there is a dip in a broad hill. However, positions 22,23,24 and 27 in the normalized sequence have sizable importance values. The next highest peaks are around positions 8 and 29. These two positions are close in space but on opposite sides of the V3 loop in the so-called stem region (the central bulge of the V3 structure). As mentioned above, there is evidence [23] that in R5 tropic virus this region is involved in the binding of sulfated tyrosins near the N-terminus of CCR5, and that X4 and R5 tropic viruses interact differentially with these sulfated tyrosins [25]. In Fig. 4 the class probabilities according to the two previously described random forests are plotted for all V3 sequences in the dataset. The figure suggests that the two computational models are in part complementary, as the distribution of both tropism classes extends into the upper left and lower right quarters. More importantly for classification, the two sets of R5 and X4/R5X4 seem to be rather well separable in Fig. 4. Hence, in the spirit of “stacking” [26], we have trained another random forest for classification using the output probabilities of the electrostatics and hydrophobicity random forests as inputs and again the measured tropism classes as response. This second-level classifier performed well (“Combined” in Fig. 1), with an AUC of in leave-one-patient-out cross-validation. The ROC curves in Fig. 4 have several remarkable features. First, there is a striking difference between the ROC curve from sequence-wise cross-validation (dashed) and leave-one-patient-out cross-validation, with the first procedure having a clearly higher performance (). This suggests that the algorithm perceptibly takes advantage of similarities of sequences originating from the same patient. Focusing therefore on the more conservatively estimated ROC curve from patient-wise cross-validation, and on the region of low false positive rates of, say, 0. 1 and less, we find that both first-level classifiers perform similarly well, and that in this region we also have the strongest added value of the second-level classification of the order of 10% in sensitivity. The dataset used for training and cross-validation is composed of sequences from several subtypes, and we could therefore study the dependence of prediction performance of subtype. To this end we set up a contingency table of subtypes (B, C, D, other) as rows, and correct (T) and false (F) predictions in the cross-validation as columns (for example see Supporting Information file Text S1). We then carried out a -test with the null hypothesis of subtype-independence of performance, as given by the Ts and Fs. This was done for probability cutoffs between 0 and 1 in steps of 0. 01 for the assignment of a sequence to the tropism class X4. It turned out that the p-value in all cases remained below, so that we should accept at this significance level the alternative hypothesis: performance depends on subtype. Specifically, the two-level random forest performs somewhat better on subtypes C and D than on subtype B (see also Supporting Information file Text S1). Finally, one may ask whether classification with a single joint descriptor set, encompassing both electrostatics and hydrophobicity variables, could perform better than the two-level classification. Theoretical and empirical results from other groups [27], [28] suggest that second-level learning on ensembles of classifiers trained on different descriptor sets improves accuracy compared to single-level learning. A possible advantage of single-level learning with a joint descriptor set could be a consistent importance analysis across all descriptors. However, it has recently been shown that such an importance analysis in such a joint feature space is biased, and thus may be difficult to interpret [29]. Despite these caveats we tested classification with a single-level random forest with joint descriptor set, and found a performance that was good, but lower than that of the two-level approach; e. g. using the sequence-wise cross-validation the two-level approach had an AUC of, while the single-level random forest achieved an AUC of, which is significantly lower (p-value of according to Wilcoxon test). For comparison with other methods we compiled an independent test set of recently published data comprising 74 sequences of various subtypes as described in the last section of “Materials and Methods”. These data are disjunct to the training set of the two-level classifier. As the data are recent, it is plausible that they were also not included in the training sets of the other machine-learning methods, though we cannot rule out this inclusion. Apart from our two-level method, we selected for comparison the following methods: 11/25 [9], [10], 11/24/25 [21], geno2pheno [30], and wetcat [13], i. e. two simple rule-based and two machine-learning methods, the latter two via their respective web-interfaces. Since wetcat did not allow for cutoff changes, we took the specificity of 0. 98, resulting from the application of wetcat to the independent test set, as reference specificity. We computed the sensitivity of the two-level approach at this specificity. Choosing a false positive rate of 1% as input parameter for geno2pheno fortunately resulted also in a specificity of 0. 98, so that the sensitivities of all three machine learning methods could be compared at the same specificity. Tab. 1 shows that the two-level approach gives a higher sensitivity at this specificity than geno2pheno and wetcat. For comparison with the two rule-based methods we computed the sensitivities of the two-level approach at the specificities of these methods. Tab. 1 shows that the two-level approach has a higher sensitivity than the rule-based methods at their respective specificity, though the two simple rules do surprisingly well. The numbers in Tab. 1 should not be over-interpreted as the size of the test dataset is rather limited. For instance, on the much larger dataset used for patient-wise cross-validation of the two-level approach (provided as Supporting Information), the sensitivity of that approach was 0. 762 at a specificity of 0. 98, compared to the sensitivity of 0. 68 on the test dataset reported in Tab. 1. Conversely, the sensitivity of the 11/25 rule decreases as we go from the smaller test set to the larger set from 0. 71 (specificity 0. 95) to 0. 53 (specificity 0. 97). The high prediction performance suggests that tropism of the investigated sequences can be attributed almost exclusively to properties of V3, and that determinants outside V3 [31], [32] may be rare. Still, there remain a few instances of V3 sequences that were misclassified after second-level learning. For instance, there are X4 and R5X4 tropic sequences in the lower left corner of Fig. 4, i. e. both first-level classifiers, and therefore also the second-level classifier, are almost certain to see a R5 sequence, while the experimentally determined class is X4/R5X4. It is unclear whether these remaining discrepancies are due to deficiencies of our approach, tropism determinants outside the V3 loop, or experimental errors. Further points have to be considered in view of a clinical application. First, most data for genotypic testing currently comes from bulk sequencing of blood samples that in general can contain mixtures of X4 and R5 virus. Since our method in its described form is intended for clonal sequences, the predictive performance on bulk sequencing data will be lower. Fortunately, due to the current development of “deep” sequencing [33], more and more clonal data will become available. Second, although sequences from several subtypes were present in the training set, and a first analysis of the influence of subtype was encouraging, it cannot be excluded that the performance will drop if the method is applied to subtypes that were not present in the learning set. In such cases, the method should possibly be re-trained and tested anew. Third, we have already mentioned above the occurrence of non-V3 determinants of co-receptor tropism as a possible source of errors. The frequency of such non-V3 determinants is not known, and it could be even imagined that this frequency may change over time due to a wider administration of entry inhibitors. The presented method is based on training data comprising sequences, protein structures, and outcomes of assays. This mixture of data is available also for other cases of biological or medical interest. For instance, it would be interesting to apply the method to influenza, where structures and sequences of hemagglutinin and neuraminidase proteins responsible for contacts with host cells are available, as well as many data from immunological assays. An interesting question corresponding to prediction of co-receptor tropism in HIV could be: Is an influenza virus with a given set of protein sequences likely to infect bird, swine, or human? For training and cross-validation all V3 sequences with tropism information available from the Los Alamos HIV sequence database (http: //www. hiv. lanl. gov/) were retrieved. Sequences were excluded from the analysis if they occurred with contradictory tropism annotation in the database, if they contained non-canonical amino-acid symbols, or if sequences were shorter than 30 residues. Duplicated sequences were included only once. R5X4 tropic viruses of non-clonal nature were excluded to avoid possible discordance between genotyped sequence and sequence effectively used in the phenotypical assay. These criteria led to 1151 R5 sequences, 166 X4 sequences, and 34 R5X4 sequences. 284 of these sequences contained indels (17% of R5,51% of X4,10% of R5X4). Most of the sequences came from subtypes B (619 R5,81 X4/R5X4), C (218 R5,14 X4/R5X4), and D (75 R5,51 X4/R5X4), with the rest (239 R5,54 X4/R5X4) spread over many different subtypes. In training and cross-validation, R5X4 sequences were assigned to the X4 class. All sequences used for training and cross-validation are provided as Supporting Information files Dataset S1 (R5), Dataset S2 (X4), and Dataset S3 (R5X4). For comparison with other methods an independent test set was collected (see below “Comparison with other methods”). V3 structures were modeled using Modeller [34], version 9. 6. First, V3 sequences were subjected to pairwise alignment with the V3 sequence in the X-ray structure by Huang et al. [15]. Based on this alignment the structures of the V3 loops were modeled with refinement limited to optimization of side-chain positions and accommodation of insertions and deletions, if present. The root mean square deviation of the modeled structures to the template on average was 0. 085 nm with a standard deviation of 0. 018 nm. The electrostatic potential around the modeled structures was computed by solving the Poisson-Boltzmann equation with APBS [22] on a cubic grid with a spacing of 0. 3 nm. PDB2PQR [35] was used to determine charges and radii. Values for the dielectric constant inside and outside V3 were scanned. Best results in the tropism prediction were achieved with a value of both inside and outside V3. Ionic strength was set to zero. For the training of the machine learning model below, the values of the electrostatic potential on a hull around the modeled V3 structures was taken as input. The hull was defined as the set of grid points with minimum distance to the solvent accessible surface of the template V3 loop (solvent radius 0. 14 nm) of times the grid spacing distance, i. e. we tested hulls with distances of 0. 3 nm, 0. 6 nm, 0. 9 nm, etc. to the solvent accessible surface. Best results were obtained with a distance of 0. 6 nm. Random forest analyses were carried out with the package randomForest [17] of R [36]. ROC curves were analyzed with package ROCr [37]. Cross-validation was performed in two ways. Firstly, the out-of-bag error, as provided by the random forest package was computed for the training and cross-validation set of sequences described above (alternatively, we have employed ten-fold external cross-validation but with essentially the same results). The out-of-bag error is estimated by repeatedly bootstrapping datasets, generating training sets comprising two thirds of these datasets, and predicting the remaining third [17]. Secondly, we have assessed the influence of sequence clusters originating from the same patient by a leave-one-patient-out procedure, where the random forest was trained on sequences of all patients except one, and the tropisms of the sequence or sequences of this patient were predicted; this was repeated with sequences of each patient being used as test set once. If not mentioned otherwise performance results reported in “Results/Discussion” refer to the leave-one-patient-out procedure. AUC values of the form given in the text are averages over ten random forest trainings with marking a 95% confidence interval estimated with a -distribution. For comparative testing with other methods we collected from recent publications [38]–[41] an independent test set. All sequences from these publications were considered that did comply with the criteria applied to the training dataset described above, and, additionally were not already contained in that training set. In this way we obtained a test set of 74 sequences (43 R5,31 X4). The test set contained sequences of subtypes B [39], AE [40], D [41], and possibly A, C, D, F, G, H, J, AE, AG, CRF11, CRF12_BF, CRF14_BG, URF from Ref. [38]. Since the last reference did not contain assignments of sequences to subtypes, and as we had to exclude some of the sequences from that reference because they were already contained in our training set, the subtypes contributed by Ref. [38] are not clear. For the application of the 11/25 and 11/24/25 rule, sequences were pairwisely aligned with the reference V3 sequence of the HXB2 strain using Modeller [34]. HXB2 was taken from the Los Alamos sequence database. geno2pheno and wetcat (SVM) were used via their web-interfaces at http: //coreceptor. bioinf. mpi-inf. mpg. de/ and http: //genomiac2. ucsd. edu: 8080/wetcat/, respectively.

Human Immunodeficiency Virus is the pathogen causing the disease AIDS. A precondition for virus entry into human cells is the contact of its glycoprotein gp120 with two cellular proteins, a receptor and a co-receptor. Depending on the viral strain, one specific co-receptor is used. The type of co-receptor used is crucial for the aggressiveness of the viral strain and the available treatment options. Hence, it is important to identify which co-receptor is used by the virus in an individual patient. Since the genome of the virus in the patient can be readily sequenced, and thus the composition of the viral proteins be determined, it could be possible to predict co-receptor usage from the viral genome sequences. To this end, we developed a method that is motivated by the insight that physical properties of gp120 will determine its specificity for a co-receptor. The method learns a computational model from structures and sequences of a crucial part of gp120, and the corresponding experimentally measured co-receptor usage. It then employs the model to predict co-receptor usage for new sequences. The high accuracy of the method could make it helpful for diagnosis and suggests that the model captures the determinants of co-receptor usage.

Abstract Introduction Results/Discussion Materials and Methods

computational biology/macromolecular structure analysis computer science/applications virology/diagnosis virology/immunodeficiency viruses molecular biology/bioinformatics computational biology/macromolecular sequence analysis biophysics virology/host invasion and cell entry

2010

Prediction of Co-Receptor Usage of HIV-1 from Genotype

Methotrexate (MTX) is widely used for the treatment of childhood acute lymphoblastic leukemia (ALL). The accumulation of MTX and its active metabolites, methotrexate polyglutamates (MTXPG), in ALL cells is an important determinant of its antileukemic effects. We studied 194 of 356 patients enrolled on St. Jude Total XV protocol for newly diagnosed ALL with the goal of characterizing the intracellular pharmacokinetics of MTXPG in leukemia cells; relating these pharmacokinetics to ALL lineage, ploidy and molecular subtype; and using a folate pathway model to simulate optimal treatment strategies. Serial MTX concentrations were measured in plasma and intracellular MTXPG concentrations were measured in circulating leukemia cells. A pharmacokinetic model was developed which accounted for the plasma disposition of MTX along with the transport and metabolism of MTXPG. In addition, a folate pathway model was adapted to simulate the effects of treatment strategies on the inhibition of de novo purine synthesis (DNPS). The intracellular MTXPG pharmacokinetic model parameters differed significantly by lineage, ploidy, and molecular subtypes of ALL. Folylpolyglutamate synthetase (FPGS) activity was higher in B vs T lineage ALL (p<0. 005), MTX influx and FPGS activity were higher in hyperdiploid vs non-hyperdiploid ALL (p<0. 03), MTX influx and FPGS activity were lower in the t (12; 21) (ETV6-RUNX1) subtype (p<0. 05), and the ratio of FPGS to γ-glutamyl hydrolase (GGH) activity was lower in the t (1; 19) (TCF3-PBX1) subtype (p<0. 03) than other genetic subtypes. In addition, the folate pathway model showed differential inhibition of DNPS relative to MTXPG accumulation, MTX dose, and schedule. This study has provided new insights into the intracellular disposition of MTX in leukemia cells and how it affects treatment efficacy. Methotrexate (MTX) is one of the primary anticancer agents used for the treatment of acute lymphoblastic leukemia (ALL) [1]–[3]. The ability of cells to accumulate intracellular polyglutamate metabolites of MTX (MTXPG) is an important factor in its antileukemic effects [4]. Specifically, MTXPG inhibits the folate pathway by competitively inhibiting several important enzymes including: dihydrofolate reductase (DHFR), thymidylate synthase (TS), glycinamide ribonucleotide transformylase (GART), and aminoimidazole carboxamide ribonucleotide transformylase (AICART). This inhibition leads to reduced or blocked TS and de novo purine synthesis (DNPS), which are needed for DNA synthesis. There is large variability in MTXPG accumulation and a variety of studies have related differences in its accumulation to ALL lineage, ploidy, molecular subtype, and folate pathway gene expression [5]–[8]. Thus, developing a better understanding of the underlying mechanisms responsible for these differences in cellular disposition of MTX is important for understanding the basis for inter-patient differences in MTX' s antileukemic effects and to identify strategies to circumvent mechanisms of resistance. Pharmacokinetic and pharmacodynamic modeling is a useful approach to quantify the intracellular kinetics of MTX and to aid in understanding the underlying mechanisms related to differences in MTXPG accumulation [9]. For example, modeling can be helpful in addressing whether higher accumulation of intracellular MTXPG is related to higher formation of polyglutamates via higher folylpolyglutamate synthetase (FPGS) activity, lower degradation of polyglutamates via γ-glutamyl hydrolase (GGH), or differences in MTX influx or efflux from leukemic cells, In addition, there are numerous models describing MTX inhibition of target enzymes in the folate pathway [10]–[18], which can be exploited to advance our understanding of how folate inhibitors such as MTX alter folate homeostasis leading to its antileukemic effects. In an effort to better understand the underlying dynamics of the observed differences in MTXPG accumulation along with their differential effects on folate kinetics, we used a pharmacokinetic model to characterize the disposition of plasma MTX and intracellular MTXPG [9] along with a pharmacodynamic model to describe the dynamics of perturbations in the folate pathway [12]. These models allowed us to relate differential disposition of intracellular MTXPG to changes in transport of MTX into and out of leukemic blasts along with metabolism of intracellular MTXPG. In addition, the folate pathway model allowed us to investigate how this differential disposition of intracellular MTXPG alters folate homeostasis and its downstream consequences. Therefore, the objectives of this study were to determine the intracellular pharmacokinetics of MTXPG in circulating leukemic blasts, and to assess the relationship between these pharmacokinetic parameters and covariates including ALL lineage, ploidy, molecular subtype, and gene expression of and polymorphisms in or flanking genes related to MTX transport and metabolism. In addition, we analyzed the effects of intracellular MTXPG disposition, MTX dose, and MTX infusion schedule on the folate pathway. A total of 791 plasma samples in 194 patients were assayed to determine the plasma MTX disposition. Figure 1A shows the concentration vs time plot of these data along with the population average model fit of the data sub-grouped by infusion length. The median clearance of MTX was higher in the 24 hr infusion group compared to the 4 hr infusion group (122. 6 ml/min/m2 vs 108. 6 ml/min/m2; p<0. 001). A total of 732 peripheral blood leukemia cell samples in 194 patients were assayed for intracellular MTXPG disposition. Fixing the plasma MTX pharmacokinetic parameters to each individual' s estimates, the intracellular population pharmacokinetic parameters for MTXPG were determined and the descriptive statistics of the individual estimates (conditional means) are shown in Table 2. In addition, the population estimates, relative standard error estimates of the population estimates, inter-individual variability estimates, and sensitivity analysis of the individual estimates are summarized in Table S2. Figure 1 shows the concentration vs time plot of intracellular MTX (or MTXPG1) (Figure 1B) and total intracellular MTXPG2-7 (Figure 1C) along with the population average model fit (for non-hyperdiploid B-lineage ALL) of the data. In addition, several representative plots of individual model fits to the data are shown in Figure S1. It has been previously reported that there are significant differences in intracellular MTXPG accumulation by ALL lineage, ploidy, and molecular subtype [4], [5], [19]. Using the pharmacokinetic model of the intracellular disposition in peripheral blasts of MTXPG, we quantified how differences in MTXPG disposition related to the model estimated parameters describing MTXPG influx, efflux, FPGS, and GGH activity. ALL chromosomal ploidy exhibited differences in influx and efflux parameters for MTX. Specifically, NET-influx was 2 times higher (p<0. 0009) in hyperdiploid ALL compared to non-hyperdiploid ALL (Figure 2). In addition, we observed higher efflux (1. 8 times higher; p<0. 003) and lower NET-influx (1. 5 times lower; p<0. 02) in patients randomized to the 24 hr infusion compared to the 4 hr infusion. The model parameters describing FPGS activity differed significantly by ALL lineage and molecular subtype. Specifically, the maximum FPGS activity was 2. 1 times higher (p<0. 0002) in B-lineage ALL compared to T-lineage ALL. In addition, there was a significant difference (p<0. 0002) in the maximum FPGS activity among the different molecular subtypes of ALL with the highest activity in B-lineage hyperdiploid ALL followed in decreasing order by B-lineage non-hyperdiploid, t (12; 21) [ETV6-RUNX1], t (1; 19) [TCF3-PBX1], and T-lineage ALL (Figure 3A). These differences translated to differential net accumulation (p<0. 003) of MTXPG (NET-PG) with highest accumulation in B-lineage hyperdiploid and B-lineage non-hyperdiploid ALL, followed by t (12; 21) [ETV6-RUNX1], T-lineage, and t (1; 19) [TCF3-PBX1] ALL (Figure 3B). We also investigated how the MTXPG model parameters related to gene expression (mRNA) in ALL cells and germline polymorphisms in or flanking genes related to MTX transport and metabolism. These data were available for 168 and 190 of the patients, respectively. First, we assessed how MTX transporter gene expression and polymorphisms related to the model estimated parameters for MTX influx and efflux. Specifically, MTX influx (Vmax-in/Km-in) increased as the expression of SLC19A1 (probe set ID: 209775_x_at) increased (p<0. 0005) and NET-influx increased as the expression of SLC19A1 (probe set ID: 211576_s_at) increased (p<0. 005) (Figure 4A–B). None of the polymorphisms in or flanking transporter genes that we evaluated were significantly related to the MTX influx or efflux parameters. Next, we studied how FPGS and GGH gene expression and polymorphisms related to the model estimated parameters for FPGS and GGH activity. Specifically, net accumulation of MTXPG (NET-PG) increased as the expression of FPGS (probe set ID: 202945_at) increased (p<0. 005) (Figure 4C). In addition, two SNPs upstream of FPGS (DB SNP ID: rs1544105,2782 base pairs (bp) upstream and DB SNP ID: rs7033913,4440 bp upstream) showed a significant relation to maximum FPGS activity (CC 2. 6 times higher activity compared to TT: p<0. 005; CC 2. 4 times higher activity compared to TT: p<0. 01, respectively) (Figure 5A–B). We simulated the effects of differential MTXPG accumulation on the MTX targets in the folate pathway to assess the effects of varying dose and schedule on these targets. We used the previously described enzyme kinetic parameters [12], the MTX and MTXPG inhibition parameters [11], along with the MTX plasma and MTXPG intracellular PK parameters defined in this study. Figure 6 depicts an individual simulation of the dynamics of the various folate components after infusion of 1 g/m2 of MTX over 24 hours. This predicted a two-fold increase in DHF, a one-fold decrease in 5mTHF, and only small changes in the remaining folate components relative to the untreated steady-state levels. Because MTXPG accumulation was significantly lower in T-lineage ALL compared to B-lineage hyperdiploid ALL, we investigated how this differential accumulation affected the inhibition of DNPS by comparing the simulated baseline DNPS activity to its activity over a 72 hr post MTX treatment interval. The simulations showed that there was both greater and longer inhibition of DNPS in the B-lineage hyperdiploid group (Figure 7A). Next we used simulations to compare the 44 hr post MTX treatment inhibition of DNPS between different doses (100 mg/m2 to 5 g/m2) and schedules (4 vs 24 hour infusions). As expected, we observed that as we increased dose the simulations predict greater inhibition of DNPS. We also observed greater DNPS inhibition for doses infused over 24 hours compared to 4 hours. Specifically, while a 1 g/m2 dose infused over 24 hours was predicted to inhibit about three quarters of the patients' DNPS more than 90%, it was predicted to take approximately a 2. 5 g/m2 dose infused over 4 hours to produce the same antifolate effects (Figure 7B). MTX is one of the primary anticancer agents used to treat children with ALL and its intracellular accumulation has been shown to relate to its antileukemic effects [4]. The current study allowed us to better understand the basis of differential MTXPG accumulation and how it relates to ALL lineage, ploidy, molecular subtype, gene expression, and genetic polymorphisms. We accomplished this by developing innovative mechanistic pharmacokinetic and pharmacodynamic models of intracellular MTXPG and its interaction with the folate pathway. This gave us a new approach to describing the intracellular disposition of MTXPG (e. g. influx, efflux, FPGS, and GGH activity) along with the effects of MTXPG on the folate pathway. In addition, the model allowed us to easily test hypotheses about which factors have the strongest effects on MTXPG accumulation along with which MTX doses and schedules more effectively inhibits the folate pathway. Specifically, using the pharmacokinetic and pharmacodynamic models, we were able to evaluate a) the mechanisms of intracellular MTXPG accumulation, b) the causes of differential accumulation by lineage, ploidy, and molecular subtype, c) the difference between 4 vs 24 hour MTX infusion (validating simulations in the previous study [9] which showed that longer infusions of MTX at equivalent doses related to higher accumulation of MTXPG), d) the relations between the pharmacokinetic and pharmacodynamic model parameters and mRNA expression of and polymorphisms in and flanking related genes, and e) how MTXPG accumulation affected target enzymes in the folate pathway. We observed that net influx of MTX was highest in B-lineage hyperdiploid ALL cells which also corresponded to higher RFC expression (SLC19A1). This relation has also been observed in our previous modeling [9] and experimental studies [20]. In addition, we observed differences in the influx and efflux parameters relative to the infusion length of MTX. These differences are most likely attributed to significantly different intracellular disposition of MTXPG1 in the 4 hr infusion group compared to the 24 hr infusion group. Specifically, while the population average intracellular concentration of MTXPG1 is higher during the first 6 to 8 hours after the start of infusion in the 4 hour group compared to the 24 hr group, the concentrations fall below that of the 24 hour group for the remaining time (Figure 1B). These differences could cause an overall increase in the efflux activity for the individuals with higher intracellular concentrations over much of the treatment interval of those in the 24 hr infusion group. Next, we observed differential FPGS activity and net MTXPG accumulation with respect to ALL lineage, ploidy, and molecular subtype. This is concordant with observed differences in FPGS mRNA expression and SNPs in the gene encoding FPGS in both the current study and others [21]–[23]. The folate pathway simulations allowed us to assess the effects of differential MTXPG accumulation on the inhibition of important biosynthetic pathways that are known targets of MTX. One advantage of the modeling and simulation approach was that we could efficiently evaluate multiple situations that would otherwise be difficult, time consuming, and in many cases not practical from a clinical trials perspective. In fact, this is the first time a system of models combining the intracellular disposition of MTXPG and its inhibition of the folate pathway have been used to aid in the understanding of effective MTX therapy. There are two important issues to consider when performing modeling and simulations: the availability of and the sensitivity to the model parameters. Due to the available studies of the folate cycle, there are numerous published estimates of all the primary enzyme kinetic parameters involved (see [12] for a summary). In addition, previous studies have addressed the parameter sensitivity of these folate cycle models by showing that in most cases the effects of changes in the model parameters were local [12]. For example, it was shown that changes in the enzyme kinetic parameters for DHFR had a proportional effect on THF and a much smaller effect on other folates. Figure S2 shows plots of the effects of changes in VMAX DHFR and VMAX ACAIRT on their respective activities. For these two parameters, only VmaxACAIRT had a proportional effect on ACAIRT activity and the remaining effects were all minimal. Thus, this effect is considered a local effect. Therefore, the folate model is not sensitive to the parameter choice for VMAXDHFR and only locally sensitive to the parameter choice for VmaxACAIRT. These simulations helped increase our understanding of how MTXPG accumulation, MTX dose, and MTX schedule affect antileukemic effects. In addition, our simulation results compared qualitatively to previously published studies on MTX inhibition of target enzymes, further validating them. Specifically, the simulation which compared differential accumulation of intracellular MTXPG by ALL lineage showed that in the T-lineage group only about half of the individuals had DNPS inhibition greater than 90% at 44 hrs compared to more than three quarters of the individuals in the B-lineage hyperdiploid group. Also, about half the B-lineage hyperdiploid individuals' DNPS was still inhibited greater than 90% by 72 hours post treatment. These two results are in line with our previously published results [24] which showed that individuals with higher MTXPG accumulation were more likely to achieve full inhibition of DNPS (defined as inhibition greater than 90%). In addition, the simulations describing the effects of MTX dose and schedule showed that there was increased inhibition of DNPS with larger doses and longer infusion schedules. These results are in line with our current clinically measured changes in DNPS in the subset of our patients in which DNPS was directly measured (unpublished data). These results suggest that higher doses of MTX are needed to obtain similar inhibition patterns with shorter (4H) compared to longer (24H) infusions. A recent COG study randomized patients with ALL to receive either a 2 g/m2 dose infused for 4 hours or a 1 g/m2 dose infused for 24 hours [25]. The results of this study have yet to be reported, but they will provide treatment outcome data that will complement the current study. In summary, our pharmacokinetic and pharmacodynamic model of plasma MTX, intracellular MTXPG, and the folate cycle provides an important new tool for elucidating mechanisms underlying inter-individual differences in MTXPG intracellular disposition and inhibition of target enzymes. Furthermore, this model permits assessment of how the dosage or schedule of MTX administration alters the delivery of active drug to leukemia cells of different lineage and molecular subtypes. This will facilitate the design of more effective therapy for pediatric ALL. A total of 356 patients were enrolled on St. Jude Total XV protocol for newly diagnosed ALL between 2000 and 2007 which stratified and randomized patients to receive MTX during the first day of therapy [26]. This study included the 194 patients who had adequate circulating leukemia cells for intracellular MTXPG quantification at 3 to 4 serial time points during the initial 42 hours of therapy. The institutional review board approved the study, and informed consent was obtained from parents/guardians or patients. This study was compliant with the regulations of the Health Insurance Portability and Accountability Act of 1996 (HIPAA). Patients were randomized during the first day of therapy to receive either 1 g/m2 MTX infused intravenously over 24 or 4 hours. Serial plasma samples were obtained at 1,4, 24, and 42 hours after the start of the MTX infusion and MTX concentrations were assayed by the Abbottbase TDx-FPIA II assay (Abbott Diagnostics, Irving, TX). In addition, circulating leukemia cells were obtained at 1,4, 24, and 42 hours after the start of the MTX infusion. Intracellular concentrations of MTXPG were assayed by HPLC as previously described [6], [19]. The pharmacokinetic model used to describe the plasma MTX was a first-order two-compartment model (see first two equations in (1) ). The pharmacokinetic model to characterize the intracellular disposition of MTXPG was previously described [9]. Briefly, it involves two compartments, one for the intracellular concentration of MTXPG1, or intracellular MTX (the third equation in (1) ), and the second for the intracellular concentration of MTXPG2-7, the sum of MTXPG2 through MTXPG7 (the fourth equation in (1) ), where the subscripts denote the number of glutamates attached to each MTX molecule. A diagram of the model is shown in (Figure 8A) and the model is described by the following system of ordinary differential equations: (1) The parameters are defined as follows: ke, k12, and k21 (1/hr) are the first-order parameters describing the elimination of plasma MTX and the transition between the central (MTX) and peripheral (MTXp) compartments respectively; V (L/m2) is the systemic volume; Vmax-in (pmol/109 cells/hr) and Km-in (µM) are the Michaelis-Menten parameters describing the active influx of MTX into the leukemic blasts via the reduced folate carrier and various ABC transporters; kp (1/hr) is the first-order passive influx parameter; keff (1/hr) is the first-order efflux parameter; Vmax-FPGS (pmol/109 cells/hr) and Km-FPGS (pmol/109 cells) are the Michaelis-Menten parameters describing the FPGS activity; and kGGH (1/hr) is the first-order parameter describing the GGH activity. In addition, we defined several secondary parameters which were combinations of the above parameters. These included: Influx (Vmax-in/Km-in); NET-influx (Vmax-in/keff), the ratio of maximum influx activity to efflux—the net influx of MTX; FPGS (Vmax-FPGS/Km-FPGS); and NET-PG (Vmax-FPGS/kGGH), the ratio of maximum FPGS to GGH activity—the net accumulation of MTXPG. We assumed that the amount of drug in the plasma significantly exceeded the intracellular amount. Thus, we did not account for the intracellular drug efflux into the plasma. This allowed us to uncouple the system of four differential equations to two independent systems—one for the plasma pharmacokinetics and the other for the intracellular pharmacokinetics. First we estimated the plasma MTX pharmacokinetics using the maximum a posteriori probability (MAP) parameter estimation method implemented in ADAPT 5 [27] along with the prior parameter distribution obtained from previous St. Jude Total protocols [4]. Then, fixing, per individual, these plasma pharmacokinetic parameters, the intracellular MTXPG model parameters (both population estimates and individual conditional means) were determined using the Monte Carlo Parameter Expectation Maximization (MCPEM) [28] with importance sampling population estimation algorithm in ADAPT 5 [27]. This approach was used since, unlike the plasma pharmacokinetics where we had abundant prior parameter information from previous studies, minimal prior information on the distribution of the intracellular MTXPG model parameters was available. Due to the lack of identifiability of the passive influx parameter kp we fixed it to 0. 4 (1/hr) —its previously reported value [29]. Finally, due to the known significant differences in the intracellular disposition between B and T-lineage ALL, we fit each lineage group separately in the population model. The individual conditional means were used for comparison to covariates and for the below described folate pathway simulations. The percent relative standard error of the population estimated parameters, as determined in ADAPT 5, was used to assess their sensitivity. In addition, the individual conditional means were estimated ten times using randomly chosen initial parameter values for each run. From these runs the sensitivity of the individual conditional means to changes in initial parameter values was determined by calculating their average relative absolute error. The model used to characterize the folate pathway was taken from Nijhout et al. [12] and modified to include the inhibitory effects of MTXPG on target enzymes (Figure 8B; equations in Figure S3). Specifically, MTXPG was modeled to stoichiometrically inhibit DHFR, TS, and AICART/GART via competitive binding. We simulated the effects of MTXPG on the folate pathway in each patient in the current study by using their respective MTX plasma and intracellular MTXPG model parameters along with published folate pathway enzyme kinetic parameters [11], [12]. We considered simulations over the dose range from 100 mg/m2 to 5 g/m2 and with a 4 or 24 hr infusion. Gene expression in ALL cells at diagnosis and germline SNPs in or flanking (within 10,000 bp of the gene) folate transporter (SLCO1B1, SLC19A1, ABCC1, ABCG2) and polyglutamation (FPGS, GGH) genes were determined by Affymetrix HgU133A Human GeneChip arrays and by Affymetrix 500K mapping array genotyping as previously described [5], [30], [31]. Differences in the individual pharmacokinetic model parameters (e. g. the conditional means determined by the above described methods) due to lineage, ploidy, molecular subtype, gene expression, and SNPs were determined by either the Kruskal-Wallis ANOVA or the Mann-Whitney U-test.

One of the primary agents used in the treatment of childhood acute lymphoblastic leukemia (ALL) is methotrexate (MTX). By better understanding its intracellular disposition, we are able to better design treatments that circumvent drug resistance and thus help improve ALL cure rates. In this study, we develop a system of mathematical models that describe the intracellular disposition of MTX along with its inhibition of important biosynthetic pathways necessary for cell division. First, we used the models to describe the disposition of intracellular MTX in a cohort of 194 patients enrolled on St. Jude Total XV protocol for newly diagnosed ALL. The results of this modeling allowed us to determine mechanisms of in vivo variability in MTX accumulation. These mechanisms related to both the influx and efflux of the drug along with the enzymes related to its metabolism. Next, we used model simulations to show the effects of changes in MTX dose and schedule on its efficacy. The results of these simulations show that longer infusions yield better efficacy and that higher MTX doses can circumvent resistance observed in ALL subtypes with lower intracellular MTX accumulate. The results from this study provide new insights into the design of more effective therapy for pediatric ALL.

Abstract Introduction Results Discussion Methods

mathematics/statistics oncology/pediatric oncology computational biology/systems biology hematology/acute lymphoblastic leukemia

2010

Modeling Mechanisms of In Vivo Variability in Methotrexate Accumulation and Folate Pathway Inhibition in Acute Lymphoblastic Leukemia Cells

The role of host genetic variation in the development of complicated Staphylococcus aureus bacteremia (SAB) is poorly understood. We used whole exome sequencing (WES) to examine the cumulative effect of coding variants in each gene on risk of complicated SAB in a discovery sample of 168 SAB cases (84 complicated and 84 uncomplicated, frequency matched by age, sex, and bacterial clonal complex [CC]), and then evaluated the most significantly associated genes in a replication sample of 240 SAB cases (122 complicated and 118 uncomplicated, frequency matched for age, sex, and CC) using targeted sequence capture. In the discovery sample, gene-based analysis using the SKAT-O program identified 334 genes associated with complicated SAB at p<3. 5 x 10−3. These, along with eight biologically relevant candidate genes were examined in the replication sample. Gene-based analysis of the 342 genes in the replication sample using SKAT-O identified one gene, GLS2, significantly associated with complicated SAB (p = 1. 2 x 10−4) after Bonferroni correction. In Firth-bias corrected logistic regression analysis of individual variants, the strongest association across all 10,931 variants in the replication sample was with rs2657878 in GLS2 (p = 5 x 10−4). This variant is strongly correlated with a missense variant (rs2657879, p = 4. 4 x 10−3) in which the minor allele (associated here with complicated SAB) has been previously associated with lower plasma concentration of glutamine. In a microarray-based gene-expression analysis, individuals with SAB exhibited significantly lower expression levels of GLS2 than healthy controls. Similarly, Gls2 expression is lower in response to S. aureus exposure in mouse RAW 264. 7 macrophage cells. Compared to wild-type cells, RAW 264. 7 cells with Gls2 silenced by CRISPR-Cas9 genome editing have decreased IL1-β transcription and increased nitric oxide production after S. aureus exposure. GLS2 is an interesting candidate gene for complicated SAB due to its role in regulating glutamine metabolism, a key factor in leukocyte activation. Staphylococcus aureus is a significant human pathogen and leading cause of skin and soft tissue infection (SSTI) and bacteremia (SAB) in community and healthcare settings. Incidence of SAB ranges from 10 to 30 per 100,000 person-years in developed countries and may present as an “uncomplicated” bloodstream infection or as a “complicated” infection involving a device implant, infective endocarditis, or bone and joint infection [1]. The etiology of SAB is complex, involving host susceptibility, microbial virulence, and healthcare-associated factors [1]. Efforts to identify common host genetic factors underlying SAB initially examined biologically plausible candidate genes involved in the innate immune response in animal models and human samples (reviewed by [2]). Several genes have been implicated in mouse models of infection [3–5], but variation in these genes has not yet been associated with SAB in humans. In contrast, genome-wide screens of S. aureus infections in individuals of European ancestry [6] and SAB in individuals of African-American ancestry [7] have reproducibly associated S. aureus infections with common genetic variants in the class II region of the major histocompatibility complex (MHC). In addition to influencing risk of developing SAB, host genetic factors may also contribute to development of “complicated” infections such as infective endocarditis (IE). These more severe infections have been associated with specific bacterial strains (clonal complexes (CC), defined by patterns on multi-locus sequence typing and spa typing). For example, the CC5 and CC30 clonal complexes are associated with increased risk of IE; however, host response to SAB due to these strains is variable, and not all individuals with CC5 or CC30-related SAB develop IE [8,9]. Candidate gene studies have associated IE with variation in IL6, IL1B [10] and TLR6 [11], although these findings considered multiple bacterial infections underlying IE and have yet to be replicated. Taken together, these observations suggest that host genetic susceptibility and microbial strain variation influence development of IE. Hypothesizing that such host genetic susceptibility is due in part to variants in coding sequences of genes (e. g. variants leading to protein-coding changes that might disrupt gene function or host-microbe interaction), we conducted a two-stage study to identify variants associated with complicated SAB, selecting candidate genes in a whole-exome sequencing discovery stage followed by a custom-sequencing replication stage. Such an approach captures both common and rare coding sequence variants, including very rare variants not included on standard genotyping arrays. Genetic variants then can be analyzed for association with complicated SAB individually or in gene-based tests defined by function, location, and allele frequency. Subsequent gene expression studies in human whole blood samples and mouse cell lines examined changes in GLS2 expression in the context of S. aureus infection. The results of this study implicate variants in the GLS2 gene, which regulates plasma glutamine levels important for modulating the adaptive immune response as risk factors underlying development of complicated SAB. The discovery sample of 168 individuals (84 complicated SAB, 84 uncomplicated, frequency matched by age (in deciles), sex, and bacterial clonal complex) is described in Table 1. The majority of the sample was male (65%), and average age was 59. 1 years. All participants were white and non-Hispanic. By design, the majority of the sample was infected with strains of S. aureus previously associated with complicated SAB (CC5 or CC30,72%). All individuals were white, non-Hispanic ethnicity, and little population stratification was detected by EIGENSTRAT analysis. None of the ten principal components extracted by EIGENSTRAT was significantly associated with complicated SAB in the discovery sample (p>0. 05), and therefore these variables were not included in subsequent analyses to adjust for potential confounding by population stratification. After sequence alignment, base calling, and quality control steps, 404,808 autosomal single nucleotide variants (SNV) were analyzed for association with complicated SAB, adjusting for age (in deciles), sex, bacterial clonal complex (CC5 and CC30 vs. CC8) and sequencing batch. No SNV was significantly associated with complicated SAB at a genome-wide corrected threshold (p<5 x 10−8) in the overall sample (S1 Fig) or when stratified by bacterial clonal complex (CC5 and CC30 separate from CC8 (S2 and S3 Figs) ), and no inflation of SNV test statistics was observed on quantile-quantile plots and estimates of the genomic inflation factor (λ = 0. 75 (overall; S4 Fig), 0. 77 (CC5 and CC30; S5 Fig) and 0. 83 (CC8; S6 Fig). Gene-based analysis using SKAT-O (allowing for cumulative independent effects of SNVs annotated as being in a gene by SeattleSeq) did not detect any significant associations after Bonferroni correction (p<2. 5 x 10−6) for testing all variants in 20,000 genes or when restricting analysis to SNV annotated by SeattleSeq as missense, nonsense and splice-site variants. The top gene-based results (p<1 x 10−4) overall and in the subsets (CC5 and CC30, CC8) are presented in Table 2, and results for all genes are presented in S1 Table (overall), S2 Table (CC5 and CC30) and S3 Table (CC8). Slight inflation of test statistics was observed on quantile-quantile plots (overall; S7 Fig, CC5 and CC30; S8 Fig, and CC8; S9 Fig). Analysis restricted to functional variants weakened evidence of association at all top genes other than LCMT2 in the CC5 and CC30 subset, and no genes were included among the top results solely by analysis of functional variants. Complete SKAT-O results using only functional variants and corresponding quantile-quantile plots are provided in S4 Table and S10 Fig (overall), S5 Table and S11 Fig (CC5 and CC30), and S6 Table and S12 Fig (CC8). Little inflation of test statistics was observed on quantile-quantile plots. The SKAT-O results overall and in the two subsets were used to identify the most significant genes for consideration in the replication phase. Starting with genes with p < 1 x 10−4 in at least one analysis, the list was expanded by including the next-most-significant genes from each subset until a 2 Mb capture set was generated. This occurred at p<3. 5 x 10−3 and yielded a set of 334 genes. Eight additional biologically interesting candidate genes suggested by prior studies in humans and mice (DUSP3, FGA, FGB, FGG, FN1, PSME3, SPINK5, TNFAIP8) were added to this set for a final total of 342 genes that were captured and analyzed in the replication set (S7 Table). These 342 genes contained 8,915 variants detected in the discovery dataset. The replication set of 240 individuals (122 with complicated SAB and 118 with uncomplicated SAB), frequency matched by the same covariates as the discovery sample, is described in Table 1. The sample was 66% male, and average age was 65. 4 years. All participants were white and non-Hispanic, and 80% of the sample was infected by bacterial clonal complexes previously associated with IE (CC5 or CC30). The replication sample was thus comparable to the discovery sample in these respects, but slightly older and more likely to carry CC5 or CC30. Like the discovery sample, little population stratification was detected by EIGENSTRAT analysis. None of the ten principal components extracted by EIGENSTRAT was significantly associated with complicated SAB in the replication sample (p>0. 05), and therefore these variables were not included in subsequent analyses to adjust for potential confounding by population stratification. After sequence alignment, base calling, and quality control steps, 10,931 single nucleotide variants (SNV) were analyzed for association with complicated SAB, adjusting for age (in deciles), sex, and bacterial clonal complex (CC5 and CC30 vs. CC8) (S13 Fig). No SNV was significantly associated with complicated SAB at a Bonferroni corrected threshold (p<4. 5 x 10−6) in the overall sample or when stratified by bacterial clonal complex (CC5 and CC30, S14 Fig. ; too few CC8 cases were included to analyze separately). No inflation of SNV test statistics was observed on quantile-quantile plots and estimates of the genomic inflation factor (λ = 0. 70 (overall; S15 Fig), 0. 71 (CC5 and CC30; S16 Fig). Gene-based analysis using SKAT-O in the overall replication sample detected significant association (p = 1. 2 x 10−4) at one gene (GLS2) after Bonferroni correction for 342 genes tested (p<1. 5 x 10−4). The top gene-based results (p<1 x 10−2) overall and in the CC5 and CC30 subset are presented in Table 3, and complete gene-based test results in the replication sample are presented in S8–S11 Tables, with corresponding quantile-quantile plots in S17–S20 Figs. Complete association results for individual SNV in the GLS2 gene for both subsets and the meta-analysis are presented in S12 Table. While not significant after multiple testing correction, it is notable that the strongest overall association at an individual SNV in the replication sample is with intronic variant rs2657878 (p = 5 x 10−4) in GLS2, which is in strong linkage disequilibrium (r2 = 0. 85) with missense variant rs2657879 (p = 4. 4 x 10−3) (Fig 1, S21 Fig). While rs2657878 was not significantly associated with complicated SAB in the discovery dataset (p = 0. 47; rank 262,339 of 404,809), the meta-analysis across subsets remained nominally significant (p = 2. 4 x 10−3). The most significant GLS2 result from the meta-analysis (which considered only SNV found in both subsets) was at less common (minor allele frequency 1. 6%) intronic variant rs937115 (OR 8. 01, p = 2. 0 x 10−4), which was also not significant after multiple testing correction. When meta-analyzing gene-based test results across the discovery and replication datasets, GLS2 remains the top-ranked gene (p = 7. 9 x 10−4), with the sum test across 11 SNV more significant than the burden test (rho = 0). The gene-based and individual SNV tests were not significant when considering CC30 alone in the discovery and replication datsets, suggesting that the more significant overall results in the replication set are not attributable to the greater proportion of the sample carrying CC30. To evaluate the clinical relevance of GLS2 in human bloodstream infections, we compared microarray expression data from patients with S. aureus (n = 32) or Escherichia coli (n = 19) bloodstream infections (BSIs) against healthy controls (n = 44). We found that GLS2 expression was significantly suppressed in S. aureus and E. coli BSI patients relative to healthy controls (Fig 2A), and that the significant difference in was present in both white and African-American subsets. Notably, no difference in expression in other genes adjacent to GLS2 (SPRYD4, MIP, RBMS2) was observed, supporting a focus on that gene. To validate our microarray expression data, we next challenged RAW 264. 7 macrophages with S aureus and measured Gls2 expression by qRT-PCR. We observed the same pattern of Gls2 suppression in macrophages challenged with S aureus (Fig 2B). Having shown that GLS2 expression is suppressed in patients with S. aureus or E. coli BSI and in RAW 264. 7 macrophages challenged with S. aureus, we next sought to understand the significance of the observed GLS2 expression pattern. To this end, we used CRISPR-Cas9 technology to silence Gls2 in RAW 264. 7 macrophages, then challenged them with S. aureus and evaluated transcription of IL-1β, an important pro-inflammatory cytokine responsible for macrophage and neutrophil activation in response to S. aureus [12]. Silencing Gls2 in RAW 264. 7 macrophages significantly decreased IL-1β transcription compared to wild type (WT) cells (Fig 3A). Furthermore, the concentration of nitric oxide (NO), a prominent macrophage signaling molecule generated by inducible NO synthase, was significantly increased in S. aureus Gls2-silenced macrophage as compared to WT (Fig 3B). These data indicate that GLS2 modulates innate immune responses to S. aureus stimulation both in vitro and in vivo. This two-stage study utilized whole-exome sequencing to identify genes with differences in a discovery stage of patients with complicated and uncomplicated SAB, followed by a replication stage utilizing targeted capture and sequencing of 342 such genes. This strategy revealed a novel candidate gene for SAB, GLS2, in which multiple variants are more frequent in complicated SAB compared to uncomplicated SAB. The initial gene-based test results in GLS2 in the discovery dataset were nominally significant in multiple subsets, resulting in its inclusion in the replication study. However, no individual variant was significantly associated with complicated SAB in the discovery sample, indicating that the gene-based test result was due to the cumulative effect of multiple variants that did not have significant individual effects. The gene-based test of GLS2 was the strongest result in the replication dataset, and several individual variants were nominally significantly associated with complicated SAB. The strongest single-variant association in the replication dataset was with a common variant (rs2657878) in intron 14 of GLS2. While this variant does not have a known functional consequence, it is in strong linkage disequilibrium with a coding-sequence variant in exon 18, rs2657879, which encodes a leucine to proline change at amino acid 581. The less common G allele at this variant has been reproducibly associated with lower plasma glutamine concentration in several genome-wide association studies of plasma metabolic markers [13–16] and was twice as frequent in complicated SAB cases in the replication sample (OR = 2. 0, p = 0. 004). These associations are biologically plausible, given the role of GLS2 (with GLS) in metabolizing glutamine. While some GLS2 variants have been suggested to be eQTLs for the adjacent SPRYD4 gene in GTEx data (www. gtexportal. org, accessed April 10,2018), only GLS2 was significantly differentially expressed in human BSI and was further shown to modulate response to S. aureus in mouse macrophages, suggesting that it is the most likely functional candidate gene. The observed pattern of association and inconsistent individual-variant results between discovery and replication samples suggest that the gene-based test might be an indirect association, whereby the coding-region variants are not themselves the biologically relevant variants, but are in linkage disequilibrium with as yet unidentified non-coding variants that might influence gene regulation and function. Therefore, the next steps in evaluating the role of GLS2 in SAB involved further examining gene expression and cellular response (cytokine production, nitric oxide production) in the context of exposure to S. aureus, and later examining regulatory sequence variants for association with these responses. Existing gene expression studies in individuals with bacteremia showed that GLS2 expression was lower in SAB cases compared to controls, and this pattern was reproduced in mouse RAW 264. 7 cells. However, we were not able to examine the association between GLS2 expression and complicated SAB, as sufficient RNA samples were not available in this dataset. Therefore, while these initial functional data demonstrate changes in GLS2 expression in response to SAB, they do not fully explain the association of genetic variants with complicated SAB in particular. Studies designed to demonstrate differences in GLS2 expression in complicated vs. uncomplicated SAB cases are needed, but sufficient numbers of RNA samples do not yet exist in the SABG and DANSAB datasets used for this study. GLS2 is an intriguing candidate gene for complicated SAB due to its role in glutamine metabolism, which is an important process for proliferation and activation of white blood cells such as neutrophils, macrophages and T-cells in response to S. aureus infection. Inhibition of GLS has been shown to reduce Th17 response [17], which is essential for effective neutrophil recruitment in response to S. aureus infection [18–20]. In mice, IL-17 is essential for host defense against S. aureus infections of the skin [21] and inhibition of IL-17 is associated with development of acute colitis after exposure to dextran sulfate sodium [22]. In humans, reduced levels of Th17 cells lead to increased susceptibility to S. aureus infections in patients with hyper-IgE syndrome [23,24], atopic dermatitis [25], and mucocutaneous candidiasis [26]. Taken together, these findings suggest that poorer Th17 responses elicited by reduced GLS2 production may influence the development of complicated SAB through a less robust T-cell response to infection. Further, nitric oxide, an endogenous signaling molecule produced by macrophages, is well known for its role in host defense mechanisms against various pathogenic bacteria [27]. When macrophages are challenged with tuberculosis for example, they produce NO which is converted into reactive nitrogen species (RNS) within infected macrophages resulting in bacterial death. The cytotoxic effects are thought to be indirect, and the role of NO is more complicated than a simple binary (on/off) response. In fact, when NO reacts with oxygen radicals, it generates peroxynitrite, nitrogen dioxide and dinitrogen trioxide that are highly toxic to the cells [28]. Consequently, NO levels are tightly regulated and the exact amount of NO produced determines whether an overall pro- or anti-inflammatory response predominates. In sepsis for example, increased production of NO triggers vasodilatation and consequent hypotension [29]. There is also prior evidence from S. aureus sepsis models implicating the level of NO generated in regulating neutrophil migration to the site of infection [30]. Here, we found that silencing Gls2 not only results in overproduction of NO in macrophages challenged with S. aureus bioparticles but it also reduced expression of IL-1β relative to WT controls, suggesting that Gls2 silencing alters macrophage responses to S. aureus. Although GLS2 has been suggested to act as a tumor suppressor gene, nothing is known about its role in S. aureus infection. Our results indicate that Gls2 acts to regulate the amount of NO in S. aureus-challenged cells and thus protect those cells from damage induced by reactive NO byproducts. Consistent with our findings, it was reported that one of the functions of Gls2 is to limit reactive oxygen species levels in cells and thus protect cells from oxidative stress-induced cell death [31,32]. The fact that GLS2 mutation is associated with complicated bacteremia could thus be the consequence of increased NO levels in these patients, resulting in the production of excessive cytotoxic oxygen radicals. Ochoa et al [33] similarly showed that high levels of circulating NO metabolites in the blood of general surgery patients with clinical sepsis correlated with severity of disease. Another possible explanation could be reduced neutrophil migration to the site of infection which in turn hinders bacterial clearance [34]. Consistent with this, it was reported that high levels of NO inhibited neutrophil migration to the site of infection in a S. aureus sepsis model of infection [30,35]. We also found that silencing of Gls2 significantly reduced transcription of IL-1β, an important part of the innate immune response to S. aureus. IL-1β is required for both neutrophil recruitment [12] and regulation of pro-inflammatory Th17 response to S. aureus [36], and is sufficient for abscess formation in immunity against S. aureus in mice [37]. Taken together, these data strongly suggest a potential role of GLS2 in host susceptibility to S. aureus infection, whereby (as of yet not identified) variation in regulatory regions of the GLS2 gene may alter gene expression in response to S. aureus infection, increasing NO and decreasing IL-1β, allowing complicated infection to develop. While the association of GLS2 variation with complicated SAB provides a novel target for additional study and potential intervention, there are caveats to the interpretation of these findings. The study was conducted in two samples of non-Hispanic whites of European descent, and therefore the results may not generalize to other populations. Also, while there is previous association of the L581P variant (rs2657879) with plasma glutamine concentration, a biological mechanism for this association has yet to be elucidated. Finally, this study did not detect association with other loci previously associated with endocarditis (SLC7A14 [38], in a study that included a subset of the Danish sample used here in the non-significant replication), with risk of SAB (HLA class II region [6]), or with biologically plausible candidate genes identified from prior human and mouse studies (DUSP3, FGA, FGB, FGG, FN1, PSME3, SPINK5, TNFAIP8). This is not surprising, as most of these genes (SLC7A14 being the exception) were associated with SAB overall, rather than complicated SAB in particular. Indeed, sub-analyses of prior studies did not find association between complicated SAB and HLA or other candidate genes. The lack of such associations might reflect lower power due to smaller sample sizes, or alternatively may indicate that genetic factors influencing initial development of SAB are distinct from those governing development of complicated infections such as endocarditis and/or bone and joint infection. The association of GLS2 variants with complicated SAB reinforces the conclusion that the strongest genetic susceptibility factors for S. aureus infection involved the adaptive immune response. Genome-wide association approaches in white [6] and African-American [7] samples reproducibly implicate the HLA class II region, which encodes cell surface molecules involved in antigen presentation and stimulation of the immune response to pathogens. Taken together, these results suggest that genetic susceptibility to SAB is influenced by several genetic variants that potentially modulate the macrophage and T-cell response to infection. The study was approved by the Duke University Institutional Review Board and participants recruited at Duke University provided written informed consent according to institutional policy. Patients dying from SAB prior to consent were included in the study in accordance with IRB-approved policies for decedent research. Danish samples were collected as a “treatment biobank” under protocols approved by the Danish Data Protection Agency (GEH-2014-053 // I-suite no 0337203372 and journal no. 2007-58-0015). Patient consent was not obtained for this study. As retrospective consent of treatment biobank participants for this specific study was not feasible, the Danish Regional Ethics Committee (journal no. H-4-2014-132) approved a waiver of consent for this study. As a condition of this approval, all samples were permanently anonymized prior to genetic analysis. The discovery dataset consisted of 168 individuals with monomicrobial SAB, selected from the S. aureus bacteremia group (SABG) repository [39], a prospective biobank of DNA, bloodstream microbial isolates, and clinical data from all individuals diagnosed with SAB and enrolled in the SABG repository since 1994 at Duke University Medical Center. As previously described in detail [9], individuals were classified as having complicated SAB if they had infective endocarditis ( (IE) (native or device-associated) ) or hematogenous bone and joint infection (vertebral osteomyelitis, septic arthritis) and were classified as uncomplicated SAB if no other types of complicated infection (meningitis, abscess, etc) were present. Exclusion criteria included outpatient status, age younger than 18 years, polymicrobial infection, and neutropenia. Equal numbers of complicated and uncomplicated SAB individuals (n = 84) were selected for study, frequency matched on age (in deciles), sex, and the clonal complex of the bloodstream S. aureus isolate. All individuals selected for study were white, non-Hispanic individuals of European descent. A replication data set was created following the approach described for the discovery sample. Complicated cases (native or device-associated IE or hematogenous bone and joint infection) were matched to uncomplicated cases by age (in deciles), sex, and CC of the bloodstream isolate (CC 5,8, or 30). In this way, a replication dataset of 240 patients was created from two sources. A total of 196 individuals were selected from the Danish Staphylococcal Bacteremia study group (DANSAB) biobank, a national resource of blood samples with over 2500 SAB cases maintained by the Statens Serum Institut and Herlev-Gentofte University Hospital which had been combined with clinical information from the Danish Staphylococcus aureus bacteremia registry [40]. An additional 44 SAB patients were identified by combining information from the national Danish Bacteremia Registry, patient journals and the Copenhagen Hospital Biobank [41]. Individuals were white, non-Hispanic (by definition) and of Northern European descent. For the discovery sample, spa typing was used to infer clonal complex (CC) as previously described [8,42]. Briefly, bacterial DNA was amplified using established PCR primers and sequences were determined via capillary electrophoresis. Sequences for spa were evaluated against eGenomics software (eGenomics, Inc, New York) and were used to determine CCs using a validated database [43]. For the replication sample, spa typing was used to classify isolates into clonal complexes, using similar laboratory methods and comparison to the MLST database (www. mlst. net) as previously described [40]. In both discovery and replication samples, only patients whose bloodstream isolate was unambiguously mapped to CC5, CC30, or CC8 were selected for this study. Whole-exome capture was performed in four batches on genomic DNA isolated from peripheral blood leukocytes using the Agilent SureSelect 50Mb AllExon v5, including untranslated regions (UTRs), capture kit (Agilent, Santa Clara, CA). Samples were ‘barcoded’ for multiplex analysis and sequencing was performed with three samples pooled per lane on an Illumina HiSeq2000 instrument in the Center for Genome Technology, John P. Hussman Institute for Human Genomics, University of Miami. Sequence reads were assessed for quality and bases were called using the Illumina CASAVA 1. 8 pipeline. Calls were then exported for alignment against the human reference genome (hg19) using the Burrows-Wheeler Alignment (BWA) software [44]. Variants were called using the GATK UnifiedGenotyper with VQSR recalibration [45]. Genotype calls with genotype quality <30, read depth <8, or Phred-scaled likelihood of reference genotype <99 were removed from analysis. Genotype variants with VQSR recalibration scores < = -2 were excluded. No allele frequency threshold was applied. After quality control, 404,808 variants were retained for analysis. SeattleSeq [46] version 138 was used to annotate variants to genes, and evaluate functional consequence (missense, nonsense, splice site variation). These variants were then analyzed for association with complicated SAB individually using Firth-bias corrected logistic regression [47], controlling for age, sex, clonal complex, and sequencing batch, as implemented in EPACTS (Efficient and Parallelizable Association Container Toolbox http: //genome. sph. umich. edu/wiki/EPACTS). Cumulative effects of variants across a gene, controlling for the same covariates, were evaluated using SKAT-O specifying the optimal adjust option and small sample size adjustment [48]. With these parameters, SKAT-O either conducts a variant burden test, assuming all variants influence the trait in the same direction, or a weighted SKAT test, that gives variants with frequency less than 1% in the sample greater weight and allows for the presence of both rare risk and protective alleles. Analyses were conducted for the entire dataset (including all variants identified in the sequence captured by the whole exome capture kit, including some flanking intronic sequences and untranslated regions), as well as stratified by bacterial clonal complex (to allow for possible host-microbe interactions) and functional effect of variant (considering missense/nonsense/splice variants separately). A traditional genome-wide association significance level of p < 5 x 10−8 was used to evaluate statistical significance of individual variant tests and a Bonferroni-corrected threshold of p<2. 5 x 10−6 (0. 05/20,000 genes) was used for gene-based tests. Because no results were significant after multiple-testing correction, the top gene-based test results (with p-values < 10−4) are presented. Results of the gene-based SKAT-O analysis in the discovery sample were used to select targets for analysis in the replication sample. First, 334 genes with nominally significant gene-based results (p<3. 5 x 10−3) overall or in one of the subset analyses (by clonal complex or functional status) were selected for analysis, and eight biological candidate genes were added based on results from mouse model studies. A targeted capture array for 342 genes (S7 Table) was designed using the Agilent SureSelect website (Agilent, Santa Clara, CA). Capture array probes were selected from the probes used in the whole exome capture (Agilent SureSelect AllExon 50 Mb + UTR v5), so that the sequences for these genes captured in the replication dataset were the same as those captured for the discovery dataset. Targeted capture was performed in a single batch on genomic DNA isolated from peripheral blood leukocytes. Samples were ‘barcoded’ for multiplex analysis and sequencing was performed with samples pooled 48 per lane on an lllumina HiSeq2500 instrument in the Center for Genome Technology, John P. Hussman Institute for Human Genomics, University of Miami. Sequence reads were assessed for quality and bases were called using the Illumina CASAVA 1. 8 pipeline. Calls were then exported for alignment against the human reference genome (hg19) using the Burrows-Wheeler Alignment (BWA) software [44]. Variants were called using the GATK UnifiedGenotyper [45]. Genotype calls with genotype quality <30, read depth <8, or Phred-scaled likelihood of reference genotype <99 were removed from analysis. All variants, regardless of frequency, that passed these QC steps were retained for analysis. After quality control, 10,931 variants were retained for analysis. SeattleSeq [46] version 138 was used to annotate variants to genes, and evaluate functional consequence (missense, nonsense, splice site variation, scaled combined annotation dependent deletion (CADD) score [49]. As in the discovery sample, individual variants were analyzed for association with complicated SAB using Firth-bias corrected logistic regression [47], controlling for age, sex, and clonal complex, as implemented in EPACTS. Cumulative effects of variants across a gene were evaluated using SKAT-O specifying the optimal adjust option and small sample size adjustment [48]. Analyses were conducted for the entire dataset, as well as stratified by bacterial clonal complex (to allow for possible host-microbe interactions) and functional effect of variant (defined the same as for the discovery data set). Bonferroni multiple test corrections were applied to the single variant tests (p<4. 5 x 10−6,0. 05/10,931 variants) and gene-based tests (p < 1. 5 x 10−4,0. 05/342 genes). To evaluate consistency of results across the two samples, SKAT-O results were meta-analyzed across the two samples using the seqMeta package (https: //github. com/DavisBrian/seqMeta). To summarize single-variant results across the most significant locus (GLS2) in both the replication and discovery datasets, we used LocusZoom[50] to create regional association plots of–log p-values for each variant in a 50kb window centered on GLS2, evaluating pairwise linkage disequilibrium and recombination rate using the 1000Genomes November 2014 EUR sample as a reference sample. Existing data on gene expression profiles in individuals with bloodstream infections were used to examine differences in GLS2 and adjacent gene expression in SAB cases, E. coli bacteremia cases, and unaffected controls. Subjects were enrolled at Duke University Medical Center (DUMC; Durham, NC), Durham VAMC (Durham, NC), and Henry Ford Hospital (Detroit, Michigan) as part of a prospective, NIH-sponsored study to develop novel diagnostic tests for severe sepsis and community-acquired pneumonia. All participants were adults. All detail regarding clinical information of these patients, including age, gender as well as the microarray analysis has been previously published [51], and these data are publicly available (GSE33341). Gene expression results from that study for GLS2, SPRYD4, MIP, and RBMS2 were examined for significant differences. The RAW 264. 7 mouse macrophage cell line was maintained at 37°C and 5% CO2 in Dulbecco’s Modified Eagle’s Medium (DMEM) supplemented with 10% Fetal Bovine Serum (FBS). A total of 5x105 cells were pre-seeded in a 24-well plate for 24 h. S. aureus clinical strain, Sanger 476 was used for infection studies. S. aureus for infection was prepared exactly as described previously [5]. The cells were incubated with 5x106 bacteria for 1 hour at 37°C. The non-phagocytized bacteria were removed by washing, and fresh medium was added. RNA was extracted at 5 hours post-infection using a Direct-zoL RNA MiniPrep kit (Zymo Research) according to the manufacturer’s instructions. The RNA was quantified using a Nanodrop 2000 instrument (Thermo Fisher Scientific). After quantification the RNA was reversed transcribed using High Capacity cDNA Reverse transcription kit (Thermo Fisher Scientific). Quantitative real-time PCR (qRT-PCR) was performed using SYBR Select Master Mix (Thermo Fisher Scientifc) and an ABI Prism 7500 Fast real-time PCR system (Life Technologies). The mRNA of Gls2 was normalized to Actin rRNA. The Gls2, IL-1β and Actin primers used here are as follow: Gls2 (5’-AAACGCCCCATCAGTTCAGT-3’/5’-AGGCTCTCCAAGGAAGTTGC-3’); Actin (5’-AGGTGTGATGGTGGGAATGG-3’/5’-GCCTCGTCACCCACATAGGA-3’). Il-1β (5’-GAGAACCAAGCAACGA-3’/5’-CAAACCGTTTTTCCATCTTCT-3’). Statistical analysis was performed with GraphPad Prism, version 5, software using a Mann-Whitney U nonparametric test. To generate CRISPR/Cas9-mediated Gls2 knockdown RAW cells, we cloned sgRNAs targeting exon 2 or exon 6 of Gls2 into LentiCRISPR. v2 (Addgene #52961), for coexpression of sgRNAs with S. pyogenes Cas9. Oligonucleotide primers sgRNA-1 (5’- ACCGTGGTGAACTTGTGGAT-3’), sgRNA-2 (5’- AGCGGCATGCTGCCTCGACT-3’) and sgRNA-3 (5’- GGCAGAAGGGGATCTTCGTG-3’) were ordered from Integrated DNA Technologies and cloned into LentiCRISPR as described (http: //genome-engineering. org/gecko/). We prepared lentiviral particles for each sgRNA vector by cotransfecting HEK293T cells with the LentiCRISPR vector, psPAX2 and pMD2. g using TransIT-LT1 (Mirus) and harvesting virus-containing supernatant at 48 hours post transfection. RAW cells were transduced with virus at a multiplicity of infection (MOI) of <1 by spinfection in the presence of 8 ug/ml polybrene. Twenty-four hours post infection, cells were selected with 5 μg/ml puromycin for 72 hours and then expanded. Cells were harvested one-week post infection and genomic DNA was prepared (Qiagen QIAamp DNA Blood Mini kit). The Gls2 locus was PCR amplified and assessed for editing using Surveyor assays (Integrated DNA Technologies) to confirm introduction of mutations in the gene. A total of 5x105 cells pre-seeded in a 24-well plate for 24 h were treated with S. aureus bioparticles (Invitrogen) to a final concentration of 10 μg/ml. At 24 hours post-infection, supernatants were collected, and nitric oxide production was determined using Nitrate/Nitrite fluorimetric assay kit (Cayman) according to the manufacturer’s protocol.

Complications from bloodstream infection with Staphylococcus aureus (S. aureus) are important causes of hospitalization, significant illness, and death. The causes of these complications are not well understood, but likely involve genetic factors rendering people more susceptible to such infections, differences in the bacteria that cause the infection, and the interactions between them. We examined the parts of the human genome that code for proteins to find variations that were more common in people with complicated S. aureus bacteremia (SAB), and identified one gene, called GLS2, in which variation is more common in complicated SAB cases than uncomplicated cases. Expression of GLS2 is lower in people with SAB than controls and in mouse white blood cells exposed to S. aureus. GLS2 encodes a protein that regulates the metabolism of glutamine, a regulatory process that activates white blood cells. These cells are very important in the immune response to S. aureus infection, and therefore genetic variants that might influence their growth are important potential genetic risk factors for complicated SAB.

Abstract Introduction Results Discussion Methods

blood cells medicine and health sciences neurochemistry immune cells pathology and laboratory medicine statistics pathogens immunology microbiology neuroscience staphylococcus aureus bacterial diseases mathematics test statistics bacteria bacterial pathogens research and analysis methods statistical distributions infectious diseases white blood cells animal cells staphylococcus medical microbiology neurochemicals mathematical and statistical techniques nitric oxide microbial pathogens gene expression hematology probability theory biochemistry bacteremia bloodstream infections cell biology genetics biology and life sciences cellular types physical sciences macrophages statistical methods organisms

2018

Human genetic variation in GLS2 is associated with development of complicated Staphylococcus aureus bacteremia

Borna disease virus (BDV) is a nonsegmented, negative-strand RNA virus that employs several unique strategies for gene expression. The shortest transcript of BDV, X/P mRNA, encodes at least three open reading frames (ORFs): upstream ORF (uORF), X, and P in the 5′ to 3′ direction. The X is a negative regulator of viral polymerase activity, while the P phosphoprotein is a necessary cofactor of the polymerase complex, suggesting that the translation of X is controlled rigorously, depending on viral replication. However, the translation mechanism used by the X/P polycistronic mRNA has not been determined in detail. Here we demonstrate that the X/P mRNA autogenously regulates the translation of X via interaction with host factors. Transient transfection of cDNA clones corresponding to the X/P mRNA revealed that the X ORF is translated predominantly by uORF-termination-coupled reinitiation, the efficiency of which is upregulated by expression of P. We found that P may enhance ribosomal reinitiation at the X ORF by inhibition of the interaction of the DEAD-box RNA helicase DDX21 with the 5′ untranslated region of X/P mRNA, via interference with its phosphorylation. Our results not only demonstrate a unique translational control of viral regulatory protein, but also elucidate a previously unknown mechanism of regulation of polycistronic mRNA translation using RNA helicases. The control of translation initiation on mRNA is one of the most fundamental processes in the regulation of gene expression. Most eukaryotic mRNAs initiate translation via the so-called “scanning mechanism”, in which the 40S ribosomal subunit binds to the cap structure at the 5′-terminus of mRNA and slides to the proximal AUG codon [1]. In this mechanism, translation initiation from the downstream AUGs is generally inefficient. Thus, the eukaryotic cellular genes are transcribed individually, generating monocistronic mRNAs. On the other hand, many animal viruses produce polycistronic mRNAs and express efficiently functionally different proteins from a single mRNA molecule [2]–[5], suggesting that eukaryotic ribosomes have the potential to initiate the translation of downstream ORFs, under the control of sequence- and/or structure-dependent features of the mRNAs. Polycistronic coding by mRNAs is a means of coordinating the expression of more than two proteins, which are arranged in tandem or overlapping in a single mRNA molecule [6], [7]. Analysis of polycistronic mRNAs therefore provides a better understanding of the regulatory mechanisms of ribosomal scanning during mRNA translation. In the leaky scanning mechanism, ribosomes bypass the first start codon when the context is poor and thus reach a start codon further downstream. Some viruses, such as Sendai virus and papillomaviruses, use such mechanisms to enable a multifunctional mRNA to express several proteins with different functions in viral replication [8]–[10]. Another strategy for translation of downstream cistrons from an mRNA is termination/reinitiation, is the major method of translation of prokaryotic and some viral mRNAs [11]–[13]. In this case, ribosomes resume the scanning of the mRNA and reinitiate translation efficiently at a downstream AUG codon, following the termination of an upstream cistron. Although eukaryotic ribosomes are in general unable to reinitiate downstream cistrons on an mRNA, it is also true that about 10 to 30% of eukaryotic mRNAs contain upstream AUG codons (uAUG), which have the capacity to initiate translation of a short upstream ORF (uORF), usually consisting of fewer than 30 codons [14]–[16]. The uORF-mediated reinitiation of downstream ORFs also has been demonstrated in eukaryotic mRNAs [17]–[20], suggesting that ribosomal termination/reinitiation may be a key mechanism for the regulation of complex gene expression in eukaryotic cells. However, we know little about the molecular mechanisms underlying the regulation of ribosomal initiation in the translation of polycistronic mRNA, especially how eukaryotic viruses use translational regulation in the expression of viral proteins. Borna disease virus (BDV) is a non-segmented, negative-sense RNA virus that belongs to the Mononegavirales and which is characterized by highly neurotropic and persistent infection. BDV replicates and is transcribed in the cell nucleus and employs several unique strategies for gene expression [21], [22]. One of the most striking characteristics of this virus is that all of the BDV transcripts have polycistronic coding capacity. The shortest, 0. 8 kb transcript of BDV, X/P mRNA, encodes at least three ORFs: uORF, X, and P in the 5′ to 3′ direction (Figure 1A). The X and P ORFs produce major viral proteins, which overlap by 215 nucleotides (nt) [23], [24]. In contrast, the uORF, whose stop codon overlaps the X translation start codon (X-AUG) by one nt, UGAUG, (Figure 2A), encodes an 8 amino acid peptide, the expression of which has not yet been shown in infected cells. BDV X is a negative regulator of viral polymerase activity, while the P phosphoprotein is a necessary cofactor of the polymerase complex [25], [26]. Thus, the expression ratio between X and P is critical for viral polymerase activity [25]–[28]. Previous studies revealed that, despite an optimal sequence context for initiation of X compared to P, translation of X seems to be suppressed at an early stage of viral infection and gradually increases along with the establishment of persistent infection (Figure S1) [29], [30]. This finding allows us to hypothesize that translational regulation of such a short polycistronic mRNA may have evolved to control rigorously the ratio between X and P in the infected cell nucleus and is essential for the maintenance of the persistent infection. Recent studies have suggested that the 5′ untranslated region (5′ UTR) of X/P mRNA plays a critical role in the translational regulation of X from the polycistronic mRNA [28], [31]. However, the translation mechanism used by the X/P polycistronic mRNA has not been determined in detail. In this study, we demonstrate the autogenous translational regulation of the X/P polycistronic mRNA mediated by host RNA helicases. We show that DDX21, also known as RNA helicase II/Gu, is a regulator of ribosomal reinitiation of X via interaction with the 5′ UTR of X/P mRNA (X/P UTR) and that expression of the downstream P protein may regulate the translation of X by interfering with the binding of DDX21 to the 5′ UTR. Our results provide not only a unique insight into translational control of a viral polycistronic mRNA but also a novel role for RNA helicase in the regulation of ribosomal reinitiation during eukaryotic mRNA translation. To investigate translational regulation of the X/P polycistronic mRNA, we first used a plasmid, pX/Pwt [30], which encodes a cDNA clone corresponding to the X/P mRNA, and assessed whether this plasmid is able to reproduce the translational regulation of X/P mRNA independently of BDV infection. Upon transfection into COS-7 and OL cells, the plasmid produced efficiently both X and P, and expression of X appeared to increase following the course of time after transfection (Figure 1B), similar to the expression dynamics of X in BDV-infected cells (Figure S1) [30], [32]. Previous studies revealed that P translocates to the cytoplasm from the nucleus via interaction with X [28], [30], [33]. As shown in Figure 1C, although the cells transfected with pX/Pwt exhibited the nuclear distribution of P at an early time after transfection, P was shown to move to the cytoplasm of the cells expressing X at 72 h post-transfection (arrows). These observations suggested that the X/P mRNA by itself regulates the translation of X independently of BDV infection and, therefore, that pX/Pwt provides a useful tool to investigate the translational regulation of the polycistronic mRNA. To understand the role of the uORF in the translation of the X and P ORFs, we generated a series of mutant plasmids (Figure 2A) and examined the expression of X and P by Western blotting at 12 h post-transfection, at which point the expression of X had not yet been upregulated. Mutants with the uAUG replaced by TAG or TTG exhibited markedly increased expression of X compared to the wt plasmid (Figure 2B and 2C, lanes 1 and 2). In addition, the involvement of uORF in the translation of X was demonstrated by using a series of deletion mutants of the 5′ UTR (Figure S2). In contrast, the translation of P was reduced to approximately 50% of the wt plasmid (lanes 1 and 2). In addition, changing the initiation codon of X (X-AUG) to TTG in the above uAUG mutants recovered the expression level of P to the equivalent of the wt plasmid (lanes 3 and 4). Furthermore, single mutants, which lack only the X ORF, with substitution of the X-AUG by AGC or ACG produced P at levels comparable to the wt plasmid with complete abolition of the expression of X (lanes 5 and 6). These observations suggested that the uAUG is recognized efficiently by scanning ribosomes and that the presence of the uORF seems to downregulate the basal expression level of X, while enhancing the translation efficiency of the P ORF to a level equivalent to an mRNA lacking both the uORF and X ORF. To determine how translation of the X and P ORFs is initiated from the X/P mRNA, we next introduced mutations into the termination codon of the uORF. A mutant in which the uORF stop codon had been changed to TTA expressed both X and P at equivalent levels to the wt plasmid (lane 7). Furthermore, a 3-nt downstream extension of the uORF termination codon also appeared not to influence the translation of either protein (lane 8). Meanwhile, downstream extension of the uORF to 29 nt reduced the expression of X, but not P, to 70% of the wt level (lane 9). Interestingly, a 164 nt extension of the uORF, so that a stop codon is introduced within the P ORF, showed significant decreases (∼30%) in the expression of both X and P (lane 10). The introduction of premature stop codons within the uORF also reduced the expression level of X (Figure S3). These results indicated that the presence of the uORF termination codon in close proximity to the X-AUG seems to be important for efficient translation of the X ORF. Furthermore, termination of the uORF before initiation of the P ORF is required for the expression of P. We also demonstrated that P is unlikely to be expressed by ribosomal shunting or internal ribosome entry site-mediated mechanism (Figure S4). Taken together, both the X and P ORFs were shown to be translated predominantly by ribosomal reinitiation, dependent on uORF termination, and the remainder of the translation (approximately 30%) might be initiated by leaky scanning of upstream start codons. The results shown above suggest that uORF-termination/reinitiation may play a key role in the regulation of translation of X. Therefore, we sought next to investigate the factors that influence ribosomal reinitiation at the X ORF. At first, we examined the role of the peptide predicted to be produced by the uORF. However, we could not detect any effects of this predicted peptide on the translation of the downstream ORFs (Figure S5). We next examined the effect of the protein encoded downstream, P, because this may accumulate in the nucleus in association with the expression of the X/P mRNA [34]. We generated mutant plasmids, pX/PΔP, in which the initiation codon of P, P-AUG, was substituted by TTG, and puORF-X/PΔP, in which the uORF was fused in-frame to the X ORF in pX/PΔP, and assessed the production of X at 48 h post-transfection, at which point P should have accumulated sufficiently in wt plasmid-transfected cells. As shown in Figure 3A, translation of X was significantly higher in cells transfected with pX/Pwt than with pX/PΔP. In contrast, initiation of translation of the uORF-X fusion protein appeared not to be affected by deletion of the P-AUG. These results indicated that expression of P upregulates the translation of the X ORF without affecting the ribosomal initiation from the uAUG. To verify this, a plasmid expressing P, pcP [33], was cotransfected with pX/PΔP or puORF-X/PΔP. Interestingly, expression of X, but not the uORF-X fusion protein, was enhanced markedly in a dose-dependent manner by co-expression of P, while the nucleoprotein (N) of BDV failed to enhance the expression of X in transfected cells (Figure 3B). A ΔP mutant based on a 164-bp uORF-extension plasmid (Figure 2A, lane 10), puORF164ΔP, also upregulated the expression of X in the presence of P (Figure 3C), although the basal expression level of X in the puORF164ΔP is significantly lower than that in pX/PΔP both with or without P. These data suggested that P influences the ribosomal initiation at the X-AUG predominantly by the uORF-termination-coupled reinitiation mechanism, while a leaky scanning mechanism may be involved to some extent in the upregulation of the translation of X by P. We set out to address next the question of how the expression of P can enhance the reinitiation of translation of X. First, we investigated the possibility that P may interact directly with the X/P mRNA and then influence ribosomal reinitiation at the X ORF. We could not demonstrate, however, a direct interaction between P and the X/P mRNA by immunoprecipitation (IP) -RT-PCR analysis (data not shown). Second, it might be possible that P affects the functions of eukaryotic initiation factors (eIFs), such as eIF2α and eIF3. However, an interaction between eIFs and BDV P was not demonstrable in cells transfected with BDV P (Figure S6A). Furthermore, expression of eIF2α, 2Bε and 3A, as well as the serine phosphorylation level of eIF2α, appeared not to be changed in cells expressing P (Figure S6B), indicating that the expression of P is unlikely to affect the quantitative and qualitative properties of eIFs. We therefore considered the possibility that P enhances the translation of X indirectly by interaction with cellular factor (s), which may affect ribosomal reinitiation at the X ORF. To investigate this, we conducted first an in vitro translation assay using in vitro transcribed X/P mRNAs encoding firefly luciferase fused with the X ORF and cellular extracts of OL cells. As shown in Figure 4A, luciferase activity was markedly reduced in the presence of the nuclear extract, but not the cytoplasmic extract, demonstrating that the nucleus may contain factor (s) that suppress translation initiation from the X-AUG. Interestingly, a mutant X/P RNA, which lacks a 38-mer from the 5′ end of the UTR, retained luciferase activity even in the presence of the nuclear extract (Figure 4A), suggesting that nuclear factors may influence the translation of X via interaction with the 5′ UTR. To verify this, we used 48-mer decoy RNAs, which represent the X/P UTR and, as a control, the 5′ UTR of another BDV polycistronic mRNA (M/G UTR). As shown in Figure 4B, incubation with the X/P UTR decoy RNA, but not the M/G UTR control, interfered with the inhibitory effect of the nuclear extract in a dose-dependent manner. We also performed an RNA-electrophoretic mobility shift assay (RNA-EMSA) using 32P-labeled riboprobes corresponding to the X/P UTR and M/G UTR to determine whether the nuclear extract interacts specifically with the X/P UTR. As shown in Figure 4C, we found that the X/P UTR riboprobe forms complexes with the nuclear extract (arrows) and that an excess of cold riboprobe efficiently interferes with complex formation. On the other hand, the M/G UTR probe failed not only to generate clear complexes with the extract (Figure 4C) but also to interfere with the complex formation of X/P UTR probe (Figure 4D). All these observations suggested the presence of nuclear factors that inhibit the translation of the X ORF via interaction with the 5′ UTR of X/P mRNA. To investigate in more detail the involvement of nuclear factors in the translational regulation of X, we tried to identify the nuclear factors using a stepwise purification assay and RNA-affinity columns coupled with short (20 mer) and full-length (48 mer) X/P UTR RNAs (see Materials and Methods). The nonspecific RNA binding was visualized using an RNA-affinity column coupled with M/G UTR control RNA. As shown in Figure 5A, seven specific bands were detected by SDS-PAGE. The bands were excised and digested with trypsin and analyzed further by LC-MS/MS. We found that these bands represent DDX21, DDX50, DNA topoisomerase 1 (TOP1), hnRNPQ1/2 and nucleolin (Figure 5A). The accuracy of this analysis was confirmed by Western blotting with antibodies specific for each protein (Figure 5B). Among the X/P UTR-binding proteins (UBPs), direct interaction has been demonstrated only between nucleolin and TOP1 [35]. Coimmunoprecipitation experiments using Flag-tagged UBPs revealed interactions among DDX21, nucleolin, and TOP1 (Figure 6A and 6B). In addition, interactions of DDX50 and hnRNPQ1 with nucleolin and DDX21, respectively, were demonstrated when hemagglutinin (HA) -tagged proteins were expressed as targets (Figure 6C and 6D). Furthermore, we demonstrated that nucleolin interacts with DDX21 through its C-terminal region using a pull-down analysis with GST- and His-fused recombinant proteins (Figure S7). These observations indicated that the UBPs interact with each other and may have been isolated from the affinity columns as complexes. To determine which UBPs contribute dominantly in binding to the X/P UTR, we performed IP-RT-PCR analysis using BDV-infected OL cells and X/P mRNA-specific primers. As shown in Figure 7A, X/P mRNA was amplified clearly from the cells transfected with Flag-tagged DDX21 and nucleolin. Furthermore, RNA EMSA using GST-fused DDX21 and nucleolin revealed that DDX21 interacts directly only with the X/P UTR (Figure 7B), while nucleolin binds to both the X/P UTR and the control M/G UTR (arrowhead). We also found by competitive EMSA that DDX21 binds more strongly to the X/P UTR than nucleolin (data not shown). Nucleolin binds a wide variety of DNA/RNA molecules and is known usually to work in concert with other proteins, which may provide the functional specificity [36], [37]. Along with these properties of nucleolin, our results strongly suggested that DDX21 is a core protein that interacts with the X/P UTR. DDX21 is known to have RNA helicase activity, which may link to RNA-folding and/or -unwinding through its binding directly to target RNA elements [38]. These observations led us to hypothesize that the interaction of DDX21, along with other UBPs, causes a structural change of the X/P UTR, impacting on ribosomal reinitiation at the X ORF. To determine whether DDX21 alters the structure of the 5′ UTR, therefore, we performed an in vitro RNA folding assay using a 32P-labeled X/P UTR probe. At first, we monitored the mobility of the X/P UTR riboprobe in a 12% native PAGE with or without boiling. As shown in Figure 8A and 8B, the probe produced low mobility bands after boiling and quick-cooling (lane 2, arrows), in addition to major bands (arrowheads), which are also seen in the gel without boiling (lane 1). The intensities of these low mobility bands were relatively stable during the equilibration period after the quick-cooling step (data not shown), indicating that the low mobility bands represent the extended form of X/P UTR RNA. To determine the effect of DDX21 on X/P UTR structure, we added GST-fused DDX21 to the RNA probe during the equilibration period. When the X/P UTR probe was reacted with the active DDX21, high mobility bands were observed in the gels (Figure 8A and 8B, asterisks). On the other hand, the heat-inactivated DDX21 failed to produce such bands (Figure 8A). The major bands were restored even in the presence of proteins, when the re-boiling process was conducted after the equilibration step (Figure 8B, double-asterisks). These results suggested that DDX21 caused the folding of the UTR probe and produce the high mobility bands in the gels. Therefore, DDX21 is likely to cause structural alteration of the 5′ UTR of the X/P mRNA. To examine the effect of DDX21 on the translation of the X ORF, we performed a coupled assay of in vitro RNA binding and in vitro translation using in vitro transcribed X/P mRNA and recombinant DDX21. As shown in Figure 9A, incubation with DDX21 reduced the translation not only of X, but also of P, from the X/P mRNA. This result suggested that DDX21 inhibits ribosomal scanning through its binding to the X/P mRNA, resulting in suppression of translation of both X and P. However, the difficulty of the reaction conditions, which must be suitable for in vitro RNA binding and in vitro translation reactions in the same tube, as well as the possibility that the effect of DDX21 on ribosomal scanning may require its interaction with other UBPs, suggested that this in vitro assay is insufficient to determine completely the role of DDX21 in translation. Furthermore, although we generated short-interfering RNAs for UBPs, including DDX21, expression of the siRNAs appeared to induce nonspecific inhibition of the translation of other mRNAs (data not shown). Therefore, we sought to investigate further the effect of DDX21 on the translation of the X ORF by focusing on its interaction with P. P is phosphorylated and acts as a protein kinase substrate, inhibiting the phosphorylation of host proteins to modify their functions [39], [40]. A recent study demonstrated the phosphorylation of DDX21 [41]. Furthermore, the phosphorylation of RNA helicases, such as nucleolin, is known to be critical for RNA-binding activity [42]. Therefore, it is tempting to speculate that interference with phosphorylation by P affects the ability of DDX21 to bind to the X/P UTR. To address this, we examined whether the phosphorylation of DDX21 is affected by the expression of P. OL cells were transfected with wt or mutant P, PS26/28A, in which two major phosphorylation sites (Ser26, Ser28) were substituted by alanine [39], [43], and the phosphorylation of DDX21, as well as nucleolin, was monitored. Although the expression levels of the UBPs were unchanged by the expression of P (Figure 9B), the phosphorylation levels of both DDX21 and nucleolin decreased clearly in the cells transfected with wt P, but not with PS26/28A (Figure 9C). To investigate whether the hypophosphorylation of DDX21 in the presence of P modulates its RNA binding activity, we extracted Flag-tagged DDX21 from the cells transfected with wt P or PS26/28A and then estimated its binding ability to the 32P-labelled X/P UTR probe using an in vitro RNA binding assay. As shown in Figure 9D, Flag-tagged DDX21, as well as nucleolin, from wt P-transfected cells exhibited significant reduction of binding to X/P UTR. The binding activities of the tagged proteins from the cells transfected with PS26/28A were significantly higher than those with wt P, suggesting that interference with phosphorylation by P decreases the RNA binding activity of DDX21. Therefore, we finally examined whether phosphorylation of P directly affects translation of the X ORF. Consistent with Figure 3B, the expression of X was significantly upregulated in the pX/PΔP-transfected cells in the presence of wt P in a dose-dependent fashion, whereas the PS26/28A mutant was not able to upregulate the translation of X (Figure 9E). Altogether, these results suggested that BDV P may inhibit the binding of DDX21 to the 5′ UTR by interfering with its phosphorylation, resulting in the upregulation of the ribosomal reinitiation from the X-AUG. In this study, we demonstrated translational regulation of polycistronic mRNA in a unique animal RNA virus. The BDV X/P polycistronic mRNA encodes three overlapping ORFs within a short, 0. 8 kb sequence. We showed that the X and P ORFs are translated predominantly by a reinitiation strategy, following the termination of translation of the uORF, although a leaky scanning mechanism is implicated to some extent in the translational processes. In this study, we employed an RNA polymerase II-controlled vector for expression of the X/P mRNA in transfected cells. We have carefully investigated the expression, as well as the structure, of the transcripts from pX/P plasmid DNAs in each experiment and then verified that our system could recreate the translational regulation of X/P mRNA in BDV-infected cells (data not shown). Currently, two types of reinitiation mechanism have been identified in eukaryotic and viral mRNAs [2], [3], [6], [11], [17], [18]. The first type of mRNAs contain short uORFs (<30 codons) upstream of the main ORFs. In this mechanism, the efficiency of reinitiation is controlled by the length of the uORF and by the intercistronic region, an appropriate distance being necessary for the recharging of reinitiation factors, including eIF2 and Met-tRNAiMet, to the ribosomes. Cellular mRNAs such as C/EBP and AdoMetDC, are representative examples of this type of regulation [17], [44]. In the X/P mRNA, initiation of translation of the P ORF may be mediated by this type of reinitiation mechanism. The scanning ribosomes, which travel continuously on the mRNA after termination of translation of the uORF, must recharge the initiation factors on the intercistronic region between the uORF and P ORF and efficiently initiate translation from the P-AUG. Note that the expression level of P is quite invariant, with or without translation of X, if the uORF is present (Figure 2), indicating that the number of ribosomes, which move continuously along the mRNA after uORF termination, is relatively constant on the X/P mRNA. This may be the mechanism underlying the stable and persistent expression of P in infected cells. The second type of reinitiation mechanism involves mRNAs containing long 5′ ORFs, which usually encode functional proteins. These mRNAs display only short intercistronic distances between the upstream and downstream ORFs, or even may overlap. It has been shown that efficient reinitiation in this mechanism is determined by the stability/mobility of ribosomal complexes to allow reinitiation at the downstream initiation codon [17], [18]. Among viral mRNAs, segment 7 of influenza B virus [45], the ORF-2 of the M2 gene of respiratory syncytial virus (RSV) [46], and the 3′ terminal ORF (VP2) of feline calicivirus (FCV) [47], [48] represent are examples of this type of reinitiation process. Our experiments revealed that reinitiation of the X ORF may be regulated by this type of mechanism, although the uORF encodes only a short and, probably, non-functional peptide. Interestingly, the uORF and X ORF feature an overlapping stop-start codon, UGAUG, as found in other viral polycistronic mRNAs [47], [49], [50]. This feature indicates that the overlapping stop-start codon of the X/P mRNA may play a key role in the regulation of translation of the X ORF. Previous studies revealed that genes divided by such an overlapping stop-start codon are expressed predominantly by termination-coupled translation, in which translation of the downstream ORF is initiated by ribosomes which have read the uORF and stalled at the overlapping stop-start codon [48], [51]. The downstream extension of the termination signal of the uORF in the X/P mRNA significantly reduced the expression of X, suggesting that ribosomal reinitiation from the X-AUG is also carried out by the coupled translation mechanism associated with uORF termination. The regulation of ribosomal movement/stability around the overlapping stop-start codon must be crucial for controlling the translation of the downstream ORF. The scanning ribosomes, which have not dissociated from the mRNA after stalling at the uORF termination codon, may be reutilized efficiently for the reinitiation of translation of the X ORF. In favor of this hypothesis, we found that host nuclear factors influence ribosomal initiation of the X ORF through interaction with the 5′ UTR and identified RNA helicase complexes, mainly involving DDX21. DDX21 is a DEAD-box RNA helicase that localizes to the nucleoli and is involved in ribosomal RNA synthesis or processing [38], [52], [53]. Although detailed functions of DDX21 have not been elucidated yet, this helicase appears to fold or unwind RNA or ribonucleoprotein structures through regulation of RNA-RNA or RNA-protein interaction [38], [52], [53]. We found that DDX21 may be a scaffold protein that interacts with the X/P UTR, among the UBPs, and causes structural alteration of the 5′ UTR. Numerous reports have demonstrated that RNA secondary structure contributes to translational control by affecting the constancy of ribosomal scanning on mRNAs or the recognition of initiation signals [6], [54]. The ribosomes may stack or pass through the initiation codons if secondary structures are formed around the initiation site, leading to enhancement or reduction of the translation efficiency of the ORFs. Therefore, it is conceivable that structural modification of the X/P UTR by DDX21 and the UBPs decreases the ribosomal reinitiation at the X ORF or increases ribosomal dissociation from the mRNA after termination of translation of the uORF, both resulting in the suppression of the translation of the X ORF (Figure 10, left arrow). We found that the structural alterations induced by the base-pair changes in a short stem-loop structure within the X/P UTR influence the translation of the X ORF (Figure S8), supporting this conclusion. On the other hand, in this model the structural change of the X/P UTR should occur in the cytoplasm. Considering DDX21 is mostly a nuclear protein [53], it is possible that the transient interaction of DDX21 with X/P mRNA in the nucleus is enough to maintain the structure of X/P UTR in the cytoplasm by introducing the UBPs (Figure 10, left arrow). Alternatively, DDX21 may be transported to the cytoplasm along with the mRNA in this case. We revealed that phosphorylation of DDX21, as well as nucleolin, is inhibited by expression of P. Previous studies demonstrated that hyperphosphorylation of nucleolin increases its RNA binding affinity, whereas dephosphorylation reduces the affinity [42]. In this study, the RNA-binding activity of DDX21 was shown to be reduced significantly in cells expressing P. These data suggested that accumulation of P in infected cells blocks interaction of DDX21 with the X/P UTR, resulting in upregulation of translation of the X ORF by promotion of ribosomal reinitiation (Figure 10, right arrow). Note that in Figure 9D the PS26/28A did not fully recover the binding activity of DDX21 to the X/P UTR. This suggests that hypophosphorylation of DDX21 may be not exclusively involved in the promotion of the translation of X, although the in vitro binding assay based on the transfection may be insensitive for the detection of the binding activity of DDX21 to the 5′ UTR. Previous studies showed that the intranuclear stoichiometry of N and P is important for BDV polymerase activity and that accumulation of P in the nucleus markedly disturbs both viral replication and persistent infection [21], [55], [56]. Interestingly, it has been demonstrated that X binds directly to P and promotes translocation of P to the cytoplasm from the nucleus [30], [33]. Therefore, P-dependent translational regulation of X must be a convenient and effective mechanism for ensuring an optimal level of P in the nucleus. The nuclear accumulation of P above the threshold level upregulates the translation of X, thereby leading to the translocation of P to the cytoplasm. This could keep the amount of P at the optimal level in the nucleus, which is unequivocally necessary for productive replication and/or persistent infection of BDV in the nucleus. A previous study, which demonstrated that the mutations in Ser26 and Ser28 of P aberrantly upregulate the viral polymerase complex activity, and that recombinant BDVs containing the phosphorylation mutations (rBDV-PS26/28A) reduce the expression of X in infected cells [43], may be consistent with our findings, although the possibility that two amino acid changes inevitably introduced in the X ORF of rBDV-PS26/28A affect the expression level of X has remained. In addition, a recent work using a mutant rBDV, which ectopically expresses X under the different transcriptional unit, demonstrated that the expression of X from the mutant virus is not as tightly linked to expression of P as in the wild type BDV, resulting in strong attenuation of the replication of the rBDV in rat brains [57]. This observation may also support our conclusion that the X/P UTR is not only specifically involved in the regulational expression of X but also essentially controls the balanced expression between X and P in infected cells. Furthermore, a recent work by Poenisch et al. [31] showed that recombinant BDVs containing either a premature stop codon in the uORF or mutations ablating the stop codons of the uORF express wild-type like X and P in cultured cells and efficiently replicate in the brains of adult rats. Although this observation may seem to conflict with our findings that the overlapped termination of uORF is critical for the translation reinitiation of X, the recombinant viruses may be able to recover the translation level of X by the expression of other transcription unit, such as a 1. 9-kb mRNA, resulting in the efficient replication in infected cells. In fact, Poenisch et al. [31] have demonstrated that the 1. 9-kb mRNA not only serves as a template for the synthesis of N but also might be used for the translation of the viral P protein and possibly X, suggesting that the 1. 9-kb transcript is a multicistronic mRNA of BDV. This is the first example, to our knowledge, of autogenous translational regulation of polycistronic mRNA mediated by its own encoding protein and host RNA helicases. The detailed description of the mechanism should provide novel insights into not only an ingenious strategy of virus replication but also the roles of RNA helicases in the translation of eukaryotic mRNAs. Further study remains to be done to discover cellular mRNAs using a similar translation strategy. The COS-7 cell line was grown in Dulbecco' s modified Eagle' s medium (DMEM) supplemented with 5% heat-inactivated fetal calf serum (FCS) at 37°C in a humidified atmosphere of 95% air and 5% CO2. The OL cell line, derived from a human oligodendroglioma, was grown in high-glucose (4. 5%) DMEM supplemented with 5% FCS. Cells were passaged every 3 days. The BDV strain huP2br [32], [58] was used for analyses in this study. Construction of the expression plasmids for BDV X/P mRNA, N, P and P phosphorylation mutants has been described elsewhere [29], [30], [32], [33], [39], [43]. The mutant forms of the plasmids were generated using PCR-based site-directed mutagenesis. To generate X/P-Luciferase hybrid mRNAs, a luciferase gene was fused in frame with the X and P ORFs at the 148 nt and 149 nt positions of the coding sequences, respectively, and introduced into the pcDNA3 vector (Invitrogen) at the Kpn Ι-Not Ι sites. The first AUG codon of Luc was replaced by AAG. For expression of DDX21, DDX50, nucleolin, TOP1 and hnRNPQ1, corresponding cDNAs were amplified by RT-PCR from OL cells and inserted into pcXN2, pET32a or pET42a vectors (Novagen). Cells were transfected with equimolar ratios of plasmid DNAs using Lipofectamine™ 2000 (Invitrogen) or FuGENE6 (Roche Applied Science), according to the manufacturer' s instructions, and cellular samples were collected at the desired times. The introduction of the correct sequences for the wild type and its mutant were confirmed by DNA sequencing and Western blotting analysis of protein production. To generate glutathione S-transferase (GST) -tagged DDX21 and nucleolin recombinant proteins used in the Escherichia coli system, we cloned the amplified cDNAs into the pET42a vector (Novagen). The vectors were transformed into BL21 (DE3) (Novagen), and the expression of the GST-tagged proteins was induced by the addition of 0. 3 mM IPTG. The cell pellets were resuspended in PBS (-) and then lysed by sonication. After centrifugation, the supernatants were loaded on glutathione sepharose 4B (Amersham Biosciences). Eluted proteins were concentrated using Centricon spin columns (Millipore Corporation) and dialyzed against a 20 mM HEPES (pH 7. 5) -100 mM KCl buffer. The His-tagged DDX21 was generated by the insertion of the PCR-based DDX21 cDNA into PET32a vector (Novagen) and the resultant plasmid was transformed into Rosetta-gami B (De3) pLysS competent cells (Novagen). Purification of the recombinant DDX21 using Ni-NTA agarose (QIAGEN) was performed according to the manufacturer' s recommendations. COS-7 and OL cells cultured in 12-well plates were transfected with plasmids expressing X/P-Luc hybrid mRNA. At 6 h post-transfection, cells were lysed and subjected to luciferase assay system (Promega Corporation), according to the manufacturer' s recommendations. The relative levels of luciferase activity were calculated for each fusion plasmid. For Western blotting, equal amounts of total lysate proteins of COS-7 or OL cells transfected with expression plasmids were subjected to SDS-PAGE and transferred onto polyvinylidene difluoride membrane (Millipore Corporation). Antibodies used in this study were as follows: anti-BDV P mouse monoclonal, anti-BDV X rabbit polyclonal antibodies [30], [33], mouse anti-Flag M2 (Sigma-Aldrich), mouse anti-HA 12CA5 (Roche Applied Science), rabbit DDX21 (Bethyl Laboratories), rabbit Nucleolin (Novus Biologicals), rabbit anti-Topo1 (TopGEN, Inc), mouse hnRNP-Q (ImmunoQuest Ltd), rabbit anti-phosphoserine (ZYMED Laboratories). For immunoprecipitation (IP) assay, OL cells transfected with Flag-tagged plasmids were lysed with RIPA buffer [20 mM Tris-HCl (pH 7. 4), 150 mM NaCl, 2 mM EDTA, 1% Nonidet P-40 (NP-40), 1% Na-deoxycholate with protease inhibitors]. After centrifugation at 15,000 rpm for 30 min, the supernatants were incubated with 40 µl of pre-equilibrated anti-Flag M2 agarose (Sigma-Aldrich) overnight at 4°C with gentle rotation. After incubation, beads were collected by centrifugation at 6,000 rpm for 40 s and washed four times with 1 ml of RIPA. The proteins immunoprecipitated with anti-Flag agarose were eluted with 3×Flag peptide (Sigma-Aldrich) in RIPA buffer and detected by Western blotting as described above. In IP for detection of phosphoserine, NaF and Na3VO4 were added in RIPA buffer, and the serine-phosphorylated proteins were detected by anti-phosphoserine antibody. To detect the interaction of host factors with BDV X/P mRNA in vivo, BDV-infected OL cells were transfected with Flag-tagged targeted proteins and lysed with RIPA buffer with RNasin (Promega). After IP with anti-Flag M2, the co-immunoprecipitants were boiled in TE buffer and then treated with RNase-free DNase Ι for 20 min. Total RNAs were isolated from the aqueous solution and used as templates for RT-PCR using specific primers of X/P mRNA. In vitro transcribed X/P-Luc mRNAs were prepared with Maxiscript Kits (Ambion). About 1. 0 pmol of X/P-Luc mRNAs were pre-incubated with nuclear extracts of OL cells (total protein 1 to 4 µg) in a total 20 µl of binding mixture [10 mM HEPES (pH 7. 6), 67 mM NaCl, 2 mM MgCl2,1 mM DTT, 1 mM EDTA, 5% glycerol, 10U RNasin] for 30 min at room temperature. For competition, a serial dilutions of decoy RNAs were pre-incubated with the extracts prior to the reaction. Binding mixtures were then subjected to the in vitro translation system using 50 µl of rabbit reticulocyte lysate (Promega), according to the manufacturer' s recommendations. After incubation period of 2 h at 30°C, 10 µl of mixture was subjected to luciferase assay as described above. The 32P labeled-transcripts corresponding to the X/P and M/G UTRs were prepared with a mirVana miRNA Probe construction kit (Ambion), using PCR products or synthetic oligonucleotides as dsDNA templates. Transcription of the X/P and M/G UTRs was confirmed by their mobility in native PAGE. Unlabeled transcripts were prepared with MEGAshortscript™ T7 Kit (Ambion). The cell extracts were obtained from exponentially growing OL cells. The cells were lysed with buffer A [20 mM HEPES (pH 7. 6), 10 mM NaCl, 1. 5 mM MgCl2,0. 2 mM EDTA, 1 mM DTT, 0. 1% NP-40,20% glycerol and protease inhibitor cocktail] and then incubated on ice for 5 min. After collection of the cells, the lysate was incubated for a further 10 min. After centrifugation at 2,000 rpm for 5 min, the supernatant was collected as the cytoplasmic extract. The pellet was lysed with buffer B [20 mM HEPES (pH 7. 6), 500 mM NaCl, 1. 5 mM MgCl2,0. 2 mM EDTA, 1 mM DTT, 0. 1% NP-40,20% glycerol and protease inhibitor cocktail], incubated on ice for 30 min and separated by centrifugation at 15,000 rpm for 15 min. The soluble nuclear fraction was diluted in binding buffer [10 mM HEPES (pH 7. 6), 100 mM NaCl, 1. 5 mM MgCl2,1 mM EDTA, 1 mM DTT, 0. 1% NP-40,10% glycerol]. About 1. 0 pmol of 32P-labeled gel-purified probes was incubated with the nuclear extracts (4 µg) in a total of 30 µl of binding mixture [10 mM HEPES (pH 7. 6), 67 mM NaCl, 2 mM MgCl2,1 mM DTT, 1 mM EDTA, 5% glycerol, 20 µg tRNA, 10 U RNasin] for 20 min at room temperature. For competition, non-labeled probes were incubated with the nuclear extract for 20 min at room temperature prior to incubation with the labeled probes. For the assays using recombinant proteins, the probes were incubated with 5 pmol of GST-tagged DDX21 or nucleolin in a total of 20 µl of binding mixture [20 mM HEPES (pH7. 5), 70 mM KCl, 2 mM MgCl2,2 mM DTT, 0. 2 mg/ml BSA, 20 U RNasin] for 10 min at 30°C and for 10 min at room temperature. The reaction mixtures were applied to 4% native polyacrylamide gels (40∶1 acrylamide-bisacrylamide) in TBE buffer. After electrophoresis, the gels were exposed to X-ray film overnight at −80°C. Nuclear extracts of OL cells were prepared as described above. The nuclear extracts were passed through the RNA-negative coupled column and then loaded onto a consecutive RNA-positive column to remove nonspecific binding proteins. The extracts (total 2. 5 mg of protein) were loaded on HiTrap Streptavidin HP column (1. 0 ml bed volume; GE Healthcare) equilibrated with binding buffer three times (0. 2 ml/min). The flow-through was incubated with 0. 02 µmol of a 5′-biotinylated short (20 mer) RNA probe in binding buffer on ice for 30 min, and passed through a HiTrap column three times (0. 2 ml/min). The column was washed with 30 ml binding buffer and then the proteins were eluted from the columns by the addition of binding buffer containing 600 mM NaCl. After dialysis with binding buffer, the sample was subjected to an X/P UTR- or M/G UTR-coupled column as a second step of RNA-affinity purification. After washing, the binding proteins were eluted from the column with the same as for the short RNA probe-coupled column. Samples eluted from the RNA affinity columns were separated on 10% SDS-PAGE and visualized by silver-staining (Wako). The protein bands of interest were excised, digested in-gel with trypsin, and analyzed by nanocapillary reversed-phase LC-MS/MS using a C18 column (φ 75 µm) on a nanoLC system (Ultimate, LC Packing) coupled to a quadrupole time-of-flight mass spectrometer (QTOF Ultima, Waters). Direct injection data-dependent acquisition was performed using one MS channel for every three MS/MS channels and dynamic exclusion for selected ions. Proteins were identified by database searching using Mascot Server (Matrix Science). For the protein pull-down assay, 200 pmol of recombinant His-DDX21 and approximately 100 pmol of truncated forms of GST-nucleolin were incubated with RIPA buffer for 1 h at 4°C. After the incubation, reaction mixtures were bound to glutathione-Sepharose 4B (Amersham Biosciences) in RIPA buffer overnight at 4°C. After washing with the same buffer three times, bound proteins were analyzed by immunoblotting with anti-DDX21 antibodies. About 1. 0 pmol of 32P-labeled probes were heated at 85°C for 5 min, quickly cooled on ice and equilibrated at 23°C for 20 min prior to the reaction, unless manipulated further. These RNAs were incubated with 5 pmol of GST-tagged DDX21 and GST-tagged truncated nucleolin, Nuc (1234R) in total 15 µl of binding mixture [20mM HEPES (pH 7. 5), 70 mM KCl, 3 mM ATP, 0. 2 mg/ml BSA, 20 U RNasin] at 23°C for 20 min. After the incubation, the reaction was terminated by the addition of 5×loading buffer [20 mM HEPES (pH 7. 5), 70 mM KCl, 50% glycerol, 0. 5% SDS, 0. 2 mg/ml proteinase K, 0. 01% BPB, 0. 01% XC], which also inactivated the enzyme. A part of the reaction mixtures was then applied to 12% native polyacrylamide gel (40∶1 acrylamide-bisacrylamide) in TBE buffer. After electrophoresis, the gels were exposed to X-ray film overnight at −80°C. The OL cells expressing Flag-tagged recombinant proteins were lysed with RIPA buffer including protease inhibitors and 40 µg/ml of RNase A, and IP were performed using anti-Flag M2 as described above. The precipitants were washed twice with washing buffer [20 mM Tris-HCl (pH 7. 5), 70 mM NaCl, 70 mM KCl, 0. 1% NP-40] and once with binding buffer [20 mM Tris-HCl (pH 7. 5), 70 mM KCl, 0. 1% NP-40] and subjected to in vitro binding assay. 10 pmol of 32P-labeled X/P UTR probe was added to 20 µl of 50% suspension of the protein-loaded beads. After adjusting the total volume to 250 µl with binding buffer, the reaction mixture was incubated at 4°C for 10 min with gentle agitation. Unbound probe was removed by washing three times with 1 ml of binding buffer. The amount of bound radio-activity was measured by scintillation counting and the specificity was achieved by eliminating background activity obtained from the bead with the mock-transfected cell extract.

All viruses rely on host cell factors to complete their life cycles. Therefore, the replication strategies of viruses may provide not only the understanding of virus pathogenesis but also useful models to disentangle the complex machinery of host cells. Translation regulation of viral mRNA is a good example of this. Borna disease virus (BDV) is a highly neurotropic RNA virus which is characterized by persistent infection. BDV expresses mRNAs as polycistronic coding transcripts. Among them, the 0. 8 kb X/P mRNA encodes at least three open reading frames (ORFs), upstream ORF, X, and P. Although BDV X and P have opposing effects in terms of viral polymerase activity, the translational regulation of X/P polycistronic mRNA has not been elucidated. In this study, we show an ingenious strategy of translational control of viral regulatory protein using host factors. We demonstrate that host RNA helicases, mainly DDX21, can affect ribosomal reinitiation of X via interaction with the 5′ untranslated region (UTR) of X/P mRNA and that the downstream P protein autogenously controls the translation of X by interfering with the binding of DDX21 to the 5′ UTR. Our findings uncover not only a unique translational control of viral regulatory protein but also a previously unknown mechanism of translational regulation of polycistronic mRNA using RNA helicases.

Abstract Introduction Results Discussion Materials and Methods

infectious diseases/infectious diseases of the nervous system virology/persistence and latency molecular biology/translational regulation virology/viral replication and gene regulation

2009

Autogenous Translational Regulation of the Borna Disease Virus Negative Control Factor X from Polycistronic mRNA Using Host RNA Helicases

We describe methods for rapid sequencing of the entire human mitochondrial genome (mtgenome), which involve long-range PCR for specific amplification of the mtgenome, pyrosequencing, quantitative mapping of sequence reads to identify sequence variants and heteroplasmy, as well as de novo sequence assembly. These methods have been used to study 40 publicly available HapMap samples of European (CEU) and African (YRI) ancestry to demonstrate a sequencing error rate <5. 63×10−4, nucleotide diversity of 1. 6×10−3 for CEU and 3. 7×10−3 for YRI, patterns of sequence variation consistent with earlier studies, but a higher rate of heteroplasmy varying between 10% and 50%. These results demonstrate that next-generation sequencing technologies allow interrogation of the mitochondrial genome in greater depth than previously possible which may be of value in biology and medicine. The first complete human ‘genome’ sequenced was that of the mitochondrion in 1981 [1]. Since then, over 8,250 complete human and 3,220 complete non-human vertebrate mitochondrial genomes have been sequenced (http: //www. ncbi. nlm. nih. gov). These contributions have come from numerous laboratories, where obtaining the complete sequence of even the ∼16. 5 kb circular mitochondrial genome has been labor intensive and expensive. As an exemplar, it would be desirable to obtain this sequence on tens of thousands of samples in a simple, inexpensive, yet accurate manner. Beyond enriching many aspects of human biology, this development could be considered as a prelude, or even as a prerequisite, to sequence-based individualized medicine. Indeed, the mitochondrial genome, despite its unique structure and function, is an excellent ‘model system’ to identify and solve the technical, biological and medical problems that genomic medicine will encounter. The mitochondrial genome (mtgenome) has multiple attractive structural and functional features. First, it is small at 16,569 bp (revised Cambridge Reference Sequence, rCRS) [2]. Second, it is divided into a small (6. 8%) non-coding displacement loop (D-loop) or control region which provides the origin for mtDNA replication, and a large (93. 2%) coding region compactly housing 37 genes (22 tRNAs, 13 proteins and 2 rRNAs) that encode proteins critical to the electron transport chain [1]. The unique biochemical functions of the mitochondria and its high functional content suggest that a higher fraction of mitochondrial, as compared to nuclear, mutations is likely to be functionally deleterious and have distinct phenotypes. Consequently, we have an enhanced possibility of understanding the logic of how sequence variation affects biochemical functions and organismal phenotypes. Third, depending on cell type, each cell contains hundreds or more of mitochondria, each mitochondrion harboring 2–10 genomes. Thus, the functional consequences of mtgenome variation acutely depend on the tissue, and are thus a model for all genes. Genetic variation in the mtgenome has been critical to demonstrating its unique features of matrilineal inheritance [3], [4], lack of recombination [5], higher variability than the nuclear genome [6], [7] and hypervariability within the D-loop as compared to the rest of the mtgenome [8], [9]. These features have allowed delineation of mitochondrial haplotypes and haplogroups along maternal lines of descent in different human populations, and greatly contributed to our current understanding of human population structure and evolution. In turn, mitochondrial haplogroups have become a marker of an individual' s ancestry. A surprising aspect of the mitochondrial genome has been its unusually large impact on human disease given its small size, owing to its high coding ratio and high mutation rate. The impact of mutations in the mtgenome on tissues with high-energy needs, such as muscle, has long been recognized in genetic disorders such as myoclonus epilepsy with ragged red fibers (MERRF) and Leber' s hereditary optic neuropathy (LHON) [10], [11]. More broadly, mutations in the mtgenome have been identified in, or associated with, many complex disorders such as cancer, cardiovascular disease, neurodegeneration, diabetes and hearing loss [10], [12]–[16]; accumulation of mutations in the mitochondrial genome is a natural part of aging [17], [18] and the development of tumors as well [12]. Therefore, improved methods to sequence the mtgenome are of value to both biology and medicine. The 100–1,000-fold higher mutation rate in mitochondria, as compared to the nuclear genome, is owing to the lack of a DNA repair system within the organelle [19]. Thus, alterations in the mtgenome sequence occur frequently, visualized as two or more mitochondrial genomes of different sequence within a single human. Such ‘heteroplasmy’ has long been considered rare but it is one major explanation for the variation in phenotypes between maternally related individuals with a deleterious mitochondrial mutation since different individuals within the same maternal lineage may harbor different proportions of wildtype to mutant mitochondria. However, strictly on theoretical grounds, heteroplasmy must be common since each oocyte has multiple mitochondria, as compared to the single nuclear genome. Therefore, any new mutation has a significant probability of being lost through mitochondrial segregation in the daughter cells after fertilization (mitochondrial “drift”) and needs to be balanced by additional mutations to allow variation. This may be a second reason for the higher mitochondrial mutation rate observed through heteroplasmy in all tissues. Indeed, some have proposed that, under the “mutation-drift-selection” scenario, heteroplasmy should be the default state for mtDNA in all tissues of the body from mitochondrial segregation of inherited variation or from somatic mutation [20]. Indeed, all extant mitochondrial polymorphisms must have gone through a heteroplasmic state after their origin by mutation. A number of studies have demonstrated heteroplasmy, but its mechanism and incidence in the general population remains unknown since the detection of heteroplasmy has been hindered by the resolution of available sequencing technologies. While Sanger sequencing allows for complete coverage of the mtgenome, it is limited by the lack of deep coverage and low sensitivity for heteroplasmy detection when it is much less than 50% [21]. The Affymetrix Mitochip Array 2. 0 containing the full mtDNA sense and antisense sequences tiled on an array has been successfully used in our laboratory for full mtgenome sequencing with slightly improved heteroplasmy detection [22], [23]. However, neither of these technologies allows the assessment of individual mitochondrial molecules. In contrast, next generation sequencing technology is an excellent tool for obtaining the mtgenome sequence and its heteroplasmic sites rapidly and accurately since it allows deep coverage of the genome through multiple independent sequence reads. In fact, two recent studies demonstrate that the degree of heteroplasmy can vary across an order of magnitude (typically <5% but occasionally >50%) [24] and multiple sites with the mtgenome have heteroplasmy rates >10% [25]. In this study, we present the complete mitochondrial genomic sequence and heteroplasmic status of 40 samples from the International HapMap Project [26] using the next-generation 454 GS FLX pyrosequencing platform. The samples include 20 individuals from the CEU (European ancestry) and 20 individuals from the YRI (African ancestry) reference panels; these are mtgenome sequences isolated without any contamination from nuclear embedded numts (see results) and from publicly available reference samples. The availability of such reference samples is critical as the samples could serve as a basis for reproducing and benchmarking new sequencing technologies. To enable analyses, we developed novel sequence processing and analysis algorithms, both for mapping against the reference sequence and for de novo assembly, for confident determination of the mitochondrial sequence. Our analyses demonstrate sequence accuracy of near 100%, nucleotide diversity of 1. 6×10−3 for CEU and 3. 7×10−3 for YRI, patterns of sequence variation consistent with earlier studies, but a high rate of heteroplasmy varying between 10% and 50%. Twenty-two unique CEU (European ancestry) and twenty-two unique YRI (African ancestry) samples from the International HapMap Project [26], including two sets of duplicates for each population (CEU: NA10851, NA10856; YRI: NA18500 & NA18503), were sequenced. The DNA used was enriched for mitochondrial sequences by long range PCR (LPCR) of three ∼5–6 kb segments using mtgenome-specific primers. Although mitochondrial sequencing using total cellular DNA is possible and easy, and is being routinely performed with heteroplasmy detection [27][28], we avoided this approach because the human nuclear genome has >1,200 non-functional mtgenome fragments (numts) [29] and mitochondrial pseudogenes that complicate mtgenome sequence assembly and introduces numerous polymorphism and heteroplasmic artifacts. Thus, despite its simplicity it is quite erroneous, as we will demonstrate. LPCR reduced this possibility greatly since <5% of insertion sites are >5 kb. Additionally, our primers are designed to avoid nuclear genome amplification; each primer set is specific for the mtgenome as verified by BLAST (refer to methods). We completed sequencing using the 454 GS FLX system by pooling 12 individually tagged samples into each lane of a 4-region gasket PicoTiterPlate (PTP). Two YRI samples (NA19209 and NA19116) were discarded from analysis as both samples showed only one of three amplicons with an unusually high number of sites containing two different nucleotides at high frequencies; this could have arisen from a sample mixture. In addition, two CEU samples (NA12750 and NA12872) were removed due to suspected mislabeling. The results presented are from the remaining 40 samples. On average, each sample had 10,554 reads with a standard deviation of 2,652 reads. The read length distributions were similar and consistent across all samples; the distribution across all 44 samples (including duplicates) show read lengths across a wide range but 93. 7% of them are between 200–300 bp. The average read length was 250 bp with a standard deviation of 36 bp (Supporting Figure S1) so that the yield per sequencing run was ∼2. 6 megabases (mb). Our approach for obtaining the mtgenome sequence was to map quality filtered reads against the reference sequence (rCRS) to identify homoplasmic and heteroplasmic variant sites. We also introduce a novel method for de novo assembly of the reads into a circular genome. An important consideration in our study was to obtain high accuracy of the resulting called bases. We accomplished this by quantitative filtering of reads that were error prone. We finally estimated an accuracy of the resulting sequence and an analysis of its genetic features. The overall quality of the data is summarized in Figure 1. It portrays normalized coverage and the 0-centered ratio of forward/reverse reads at each position of the mtgenome. The average coverage across all 40 samples in YRI and CEU was ∼120-fold. However, the total number of reads varied per sample so that we normalized coverage by a sample' s total number of reads. Second, we assessed the directionality bias in the reads by computing ρ = (r−1) / (r+1) where r is the ratio of forward to reverse reads at a position. We present data on normalized coverage and read ratio as an average across the 20 samples for each population, YRI and CEU respectively. This is displayed along the mitochondrial genome (Figure 1) as a function of local GC-content, calculated using a sliding window of length 51 bp (25 bp before and after each position) across the circular genome. The figure also illustrates where the D-Loop and amplicons lie along the mitochondrial genome. As can be seen, the average coverage falls and the read ratio spikes prior to the PCR amplicon overlap regions in both populations. However, ρ fluctuations are not due to variations in GC content. The human mitochondrial genome can be sequenced at very high accuracy and rapidly using next generation sequencing technology as we, in this study, and other recent studies [24], [25], have shown. All of these studies have in common that they have uncovered patterns of sequence variation as has been described before but quantified the novel finding of a high rate of heteroplasmy in multiple individuals and across the mtgenome. Our study, however, has made three additional and important contributions. First, we have sequenced widely and publicly available biological samples so that our experiments can be replicated and provide a basis for future benchmarking and technology comparisons. Second, our methodology for variant and heteroplasmy detection is quantitative and parametric so that the method can be further optimized with additional experiments and new data. Third, we have developed a method for de novo sequence assembly of the mitochondrial circular genome with an internal test of sequence accuracy (identity of antegrade and retrograde assembly along a circular genome). Each of the above developments is significant for understanding mitochondrial biology and medicine. First, DNA sequencing technology is advancing and new platforms that include single-molecule sequencing are on the horizon [35]. The availability of multiple sequencing methods on publicly available biological samples, such as those we have used, is the only certain way for comparing different technologies and their relative advantages and disadvantages. Second, we believe that the parameters we have used for identifying variants and heteroplasmy will need to be varied depending on the specific technology used and its features such as directional bias, read accuracy, difficulty in reading through homopolymeric tracts and coverage. Consequently, our approach is general and generalizable. Third, mapping reads against a reference suffers from the disadvantage of not being able to confidently identify insertions or inversions. The de novo methods we have introduced can rectify this deficiency particularly since our preliminary exploration of 40 sequences suggests that it produces high-quality assemblies. The problems associated with recovery of target mitochondrial DNA from a biological sample, its DNA sequencing using short reads, the assembly of these reads into an mtgenome and its interpretation of variation and heteroplasmy are invariably confounded. We chose to recover the mtgenome in each individual by three distinct long-range PCR segments, analogous to Li et al. (2010) and in contrast to He et al. (2010). Our primers are designed to specifically target mtDNA and avoid introducing any artifacts from the numerous mitochondrial fragments (numts) in the nuclear human genome. Even if there is indeed some contamination from numts, this effect is expected to be small since it is assumed that there are many more copies of the entire mtgenome than two numts copies per the >1,200 autosomal insertion sites. However, specific fragments are present in >100 copies and can, and do, get amplified [29]. We expect that single molecule sequencing will reduce or eliminate this potential technical artifact. It is currently popular to extract and assemble the mitochondrial genome from whole genome sequencing of total cellular DNA [27]; Picardi and Pesole (2012) have recently done so from off-target exome sequencing data. But, these latter authors also show that ∼1% of all reads map to the mtgenome and not to known numts! Consequently, extensive filtering may be necessary to derive the mtgenome but this might also lose the genome-specific features including heteroplasmic sites. In other words, comparison of our data with those of others needs to consider how the mt DNA was isolated in the first place. In this study, we have made no attempt to estimate the cost of sequencing a single mtgenome in any accurate way. In any case, we have demonstrated that we can obtain such sequence rapidly and with an error rate <5. 63×10−4. Our crude estimate is that each sequence can be obtained for ∼$50 at high throughput much of this cost being the cost of mt DNA recovery. If so, studies of an entire cohort of individuals who have been measured for numerous medically relevant traits and are being followed for disease outcomes would be an ideal pilot experiment for individualized medicine. Forty-four reference DNA samples of unrelated individuals from the International HapMap project were studied using 454 pyrosequencing technology. The samples included 22 Yoruba samples from Nigeria (YRI: NA18500, NA18503, NA18506, NA18516, NA18523, NA18852, NA18855, NA18858, NA18861, NA18870, NA18912, NA19092, NA19101, NA19116, NA19137, NA19140, NA19152, NA19159, NA19171, NA19200, NA19203 & NA19209) and 22 Utah residents of European ancestry (CEU: NA06993, NA06994, NA07019, NA10851, NA10854, NA10856, NA10863, NA11831, NA11881, NA11882, NA11995, NA12004, NA12005, NA12144, NA12145, NA12146, NA12156, NA12248, NA12750, NA12760, NA12872 & NA12891), four of which were studied in duplicate (NA18500 and NA18503 from YRI; NA10851 and NA10856 from CEU). Additionally, four of these samples were sequenced using Sanger sequencing and the Affymetrix Mitochip Array 2. 0 (NA06994, NA12146, NA18516, and NA18523) for comparison. We also evaluated the Standard Reference Material (SRM) 2394 developed by the National Institute of Standards and Technology (NIST). These are a set of eight mixtures (mass percentages of 1%, 2. 5%, 5%, 10%, 20%, 30%, 40%, and 50%) of two 285 bp mitochondrial amplicons that differ in sequence by only one nucleotide and is obtained from two different human cell lines. After QC checks that detected sample contamination, data from NA19209, NA19116, NA12750 and NA12872 were dropped from further analysis. For pyrosequencing, we enriched for the mitochondrial genomic DNA by long range PCR (∼5–6 Kb) for three overlapping amplicons using high-fidelity TaKaRa LA Taq (TaKaRa Biomedicals) in 50 µl reactions (50 ng gDNA, 1× LA PCR buffer, 0. 3 µM of each primer, 400 µM dNTPs, 2. 5 U LA Taq). The primer sequences used were those described in Maitra et al (2004). Each primer set was blasted against the entire human genome to verify that there was no nuclear genome amplification. In silico PCR also confirmed no nuclear genome targets amplification by any of the three distinct primer sets. The success of the amplification reaction was checked by gel electrophoresis. The PCR products were then cleaned using the QIAquick PCR purification kit (QIAGEN) following the column purification protocol and the DNA was eluted in 30 µl of Elution Buffer to obtain a higher concentration. The actual concentration was determined using the Quant-iT PicoGreen dsDNA kit (Invitrogen). To obtain a uniform representation of the entire mtgenome, the amplicons were pooled in equimolar amounts (amount per amplicon [ng] = fraction of total x total amount needed). Since the pyrosequencing protocol required more than 5 µg of total DNA at a concentration of 300 ng/µl we performed at least two PCR reactions per amplicon. After pooling the three amplicons per reference sample in equimolar amounts, the samples were run through a QIAquick purification column to concentrate the pool to the desired 300 ng/µl concentration. For Sanger sequencing, the mtgenome was amplified in 24 overlapping PCR fragments (800–900 bp) as described in Rieder et al 1998. For easy detection during sequencing, M13 tags were added to all forward and reverse primer sets. PCR reactions and cycling conditions were optimized across all primer sets and used 1× PCR Buffer, 200 µM dNTP, 0. 5 U Taq2000,10 ng DNA, and 0. 5 µM of each primer. Confirmation of the reactions' specificity was assessed by 2% agarose gel electrophoresis. The final concentration of each amplicon was determined using the Quant-iT PicoGreen dsDNA kit (Invitrogen). All sequencing using Sanger chemistry were performed by a commercial entity (Agencourt) for each individual PCR product on an automated ABI3730xl platform using a concentration of 15–25 ng/µl in 30 µl of TE buffer; individual sequence traces were provided. The Sanger sequence for each sample was assembled and analyzed in the SeqManII program from the DNASTAR Lasergene® v. 7. 0 analysis software suite. All sequencing reads for an individual sample were imported and assembled into one contiguous consensus sequence by aligning them to the revised Cambridge Reference Mitochondrial Sequence (rCRS). The variant bases for each sample were determined and used as the genotype for that sample for further analysis. Peak intensities for each sequence variant identified by the program were manually reviewed. For pyrosequencing of the 48 samples, including duplicates, we pooled the pooled long range PCR products per sample in four batches of 12 each using 454' s Multiplex Identifiers (MID) that are molecular barcodes that serve as unique tags to identify each sample post-sequencing. These mitochondrial DNA pools were sequenced on a 4-gasket PicoTiterPlate (PTP) using the GS FLX sequencing system. Standard emPCR and sample preparation were followed as recommended by the manufacturer (Roche Inc.) As an additional precaution against misalignments, we developed an improved version of the BLAST algorithm. BLASTN uses an affine gap costs model and allows control of gap opening, gap extension and mismatch penalties and are particularly problematic for homopolymer stretches due to undercalls and overcalls. To accurately align these reads against a reference sequence, we needed an aligner that adjusts the gap penalties depending on the presence and length of the homopolymer sequence. The standard Smith-Waterman algorithm for aligning two sequences can be extended to handle these situations as follows. Let c (n, m) be the penalty for a n-length homopolymeric stretch of the reference appearing as an m-length stretch in the read. Then, the dynamic programming algorithm was modified to consult the c matrix also when computing the optimal alignment of the sequences. The entries of the c (n, m) matrix needed to be defined heuristically. In the current study, we set c (n, m) such that in homopolymer stretches of length ≥5, two gaps were ignored and the remaining penalized using the standard affine gap penalty model of BLASTN. In homopolymeric stretches of length 4, one gap was ignored. Since the largest homopolymeric stretch in the mitochondrial sequence is only 8 bases long, these values in the c (n, m) table were sufficient to yield good results. For performance reasons, we carried out alignment first using BLAST. Portions of the resultant alignment that were likely to benefit from our homopolymer-aware aligner were identified and refined using a Perl implementation of the model described above. We developed an independent de novo assembly of each mtgenome. In our approach, we initially populate a database comprising all unique n-mers (n = 27 here) and the frequency of each n-mer in the raw read data. To populate the database we slide a window, n bases long, along each read and record the sequence within the window as the read is traversed. Starting at the first base position, the n-mer comprising the first base and the subsequent n-1 bases is recorded. The window position is then incremented 1 base at a time until all n-mers from the read have been entered into the database. If an n-mer sequence already exists in the database, the number of occurrences (multiplicity, m) is incremented by 1. As an example, the distribution of m over all 454 reads for sample NA06993 is shown in Supporting Figure S9. The distribution is multimodal. The peak at multiplicity m = 1 comprises all n-mers that contain one or more 454 sequencing errors and that are not repeated as a group in any other read of the particular region of the genome. The peak near m = 50 is the mode of the local, n-mer matched, consensus coverage of the genome. The high multiplicities in the tail of the distribution are due to genomic regions where the long PCR segments overlap. To de novo assemble the mtgenome using the n-mer database data, we make the following minimal set of assumptions: 1) there are no duplicated n-mers within the genome; 2) there are no palindromic n-mers, i. e. , an n-mer on the L-strand of the mtGenome is not found in reverse complement form on the H-strand and vice versa, and 3) for a short n-mer drawn from the genome, the sequence read of this n-mer is more likely to be correct than contain an error. The third assumption depends on the sequence-context-dependent error rate of the 454 platform. If we consider as a characteristic value, λ = 0. 005, for the average 454 error rate per base, then for any n-mer, the probability that the n-mer is error free is given by p = (1−λ) n. If we choose n = 27, this gives p = 0. 87. This means that a majority of database n-mers are correct given that most of the mtgenome sequences are “average in content”. This calculation assumes errors are uncorrelated along the n-mer, which is not the case for the sequence context of long homopolymeric runs (see below). Our choice of n = 27 is a compromise value that seeks to ensure the validity of assumptions 1–3: the shorter an n-mer is, the more likely it is to be repeated in the mtgenome or be a palindrome; on the other hand, if the n-mer is chosen to be too long, the majority of n-mers derived from reads at a given genome position will contain an error somewhere within the n-mer. The satisfaction of assumption 3 allows us to apply a “majority base wins” criterion as our basis for selecting sequences in our de novo consensus assembly. The de novo assembly initially proceeds by searching the database for the n-mer matching at position 1 on the L-strand of the rCRS and ensuring the multiplicity m for L- strand and H- strand sequences at this position exceed 10. This starting condition was satisfied for all mtgenomes assembled (i. e. , no genome contained a polymorphism with respect to the rCRS in this portion of the genome, otherwise successive positions along the rCRS could readily be probed until this condition was satisfied). First, de novo assembly proceeds in the antegrade direction (with increasing rCRS position). We form the four possible candidates for the successive n-mer in the sequence and their respective reverse complements by dropping the first base of the n-mer at rCRS position 1 and adding A, T, C, or G to the end. The database is then searched for each candidate n-mer and its reverse complement, and the sum of the respective forward and reverse n-mer multiplicities is recorded for each candidate n-mer. According to assumption 3), the appropriate choice of the subsequent n-mer is the one that is the most abundant in the database. The selected new base is then added to the de novo assembly and the process is repeated until the starting n-mer sequence at rCRS position 1 is again encountered (exploiting the circular nature of the mtgenome). The antegrade de novo assembly is then complete. To assign consensus coverage at each base position we form the n-mer from the antegrade assembly in which the position in question is at the center, with (n_mer-1) /2 bases on either side. The database is then searched for this n-mer and the sum of the L-strand and H- strand multiplicites, m, is recorded as the consensus coverage. As a check on the antegrade de novo consensus assembly, the entire assembly process above is repeated by sequencing from rCRS position 1 in the retrograde direction using the same database. Here, the base at the end of the L-strand n-mer is dropped and the candidate n-mers for the next position in the retrograde direction are formed by adding A, T, C, or G to the beginning of the n-mer. The alternative assemblies in the antegrade and retrograde directions are subsequently compared to identify discrepancies for further investigation and curation. Substitution heteroplasmy candidates, and their respective fractions with respect to the consensus sequence, can then be determined by replacing the central base at each position with the other three possible bases, and then summing the L-side and H-side multiplicities of the n-mers in the database. Indel heteroplasmys with respect to the consensus can also be determined using a method aligning the unused n-mers in the database against the consensus.

This manuscript details a novel algorithm to evaluate high-throughput DNA sequence data from whole mitochondrial genomes purified from genomic DNA, which also contains multiple fragmented nuclear copies of mtgenomes (numts). 40 samples were selected from 2 distinct reference (HapMap) populations of African (YRI) and European (CEU) origin. While previous technologies did not allow the assessment of individual mitochondrial molecules, next-generation sequencing technology is an excellent tool for obtaining the mtgenome sequence and its heteroplasmic sites rapidly and accurately through deep coverage of the genome. The computational techniques presented optimize reference-based alignments and introduce a new de novo assembly method. An important contribution of our study was obtaining high accuracy of the resulting called bases that we accomplished by quantitative filtering of reads that were error prone. In addition, several sites were experimentally validated and our method has a strong correlation (R2 = 0. 96) with the NIST standard reference sample for heteroplasmy. Overall, our findings indicate that one can now confidently genotype mtDNA variants using next-generation sequencing data and reveal low levels of heteroplasmy (>10%). Beyond enriching our understanding and pathology of certain diseases, this development could be considered as a prelude to sequence-based individualized medicine for the mtgenome.

Abstract Introduction Results Discussion Materials and Methods

population genetics sequence assembly tools algorithms genome sequencing genome analysis tools population biology mitochondrial diseases sequence analysis genetic polymorphism biology computer science genetics genomics computational biology genetics and genomics human genetics

2012

Next-Generation Sequencing of Human Mitochondrial Reference Genomes Uncovers High Heteroplasmy Frequency

The heptavalent pneumococcal conjugate vaccine (PCV7) was introduced in the United States (US) in 2000 and has significantly reduced invasive pneumococcal disease; however, the incidence of nonvaccine serotype invasive disease, particularly due to serotype 19A, has increased. The serotype 19A increase can be explained in part by expansion of a genotype that has been circulating in the US prior to vaccine implementation (and other countries since at least 1990), but also by the emergence of a novel “vaccine escape recombinant” pneumococcal strain. This strain has a genotype that previously was only associated with vaccine serotype 4, but now expresses a nonvaccine serotype 19A capsule. Based on prior evidence for capsular switching by recombination at the capsular locus, the genetic event that resulted in this novel serotype/genotype combination might be identifiable from the DNA sequence of individual pneumococcal strains. Therefore, the aim of this study was to characterise the putative recombinational event (s) at the capsular locus that resulted in the change from a vaccine to a nonvaccine capsular type. Sequencing the capsular locus flanking regions of 51 vaccine escape (progeny), recipient, and putative donor pneumococci revealed a 39 kb recombinational fragment, which included the capsular locus, flanking regions, and two adjacent penicillin-binding proteins, and thus resulted in a capsular switch and penicillin nonsusceptibility in a single genetic event. Since 2003,37 such vaccine escape strains have been detected, some of which had evolved further. Furthermore, two new types of serotype 19A vaccine escape strains emerged in 2005. To our knowledge, this is the first time a single recombinational event has been documented in vivo that resulted in both a change of serotype and penicillin nonsusceptibility. Vaccine escape by genetic recombination at the capsular locus has the potential to reduce PCV7 effectiveness in the longer term. Streptococcus pneumoniae (the “pneumococcus”) is one of the most important bacterial pathogens worldwide, especially among children. Pneumococcal pneumonia, meningitis, and septicemia result in 1 million deaths annually among children <5 y of age [1]. PCV7 protects against seven pneumococcal capsular types (serotypes) —4,6B, 9V, 14,18C, 19F, and 23F [2]—and has been used to vaccinate children in the United States since 2000. PCV7 has been remarkably effective in reducing disease among vaccinated children, and even among unvaccinated children and adults as a result of a striking herd immunity effect, due to the disruption of pneumococcal transmission from young children to older children and adults [2–6]. PCV7 is increasingly being used to vaccinate children in other countries. There are 91 known pneumococcal serotypes, the last of which was discovered recently [7,8]. The ecological niche for the vast majority of the pneumococcal population is the nasopharynx of healthy children [9]; thus, any serotype-specific vaccine that is of limited valency and affects nasopharyngeal carriage will perturb the composition of the circulating pneumococcal population, with unknown consequences. The capsule is the principal known virulence factor with respect to invasive pneumococcal disease [10], and population biology studies indicated that certain serotypes have a greater potential to cause invasive disease than others [11,12]. Another important aspect of pneumococcal biology is that there is a strong association between genotype as defined by multilocus sequence typing (MLST), and serotype—that is, strains with the same MLST sequence type (ST) are usually of the same serotype [11,13,14]. Since the capsule is the principal invasive disease determinant, and is the target for serotype-specific prevention of disease by vaccination, there are two events that are especially important with respect to understanding vaccine effects: serotype replacement and capsular switching. At a population level, serotype replacement simply refers to a decrease in the prevalence of vaccine serotype pneumococci in the nasopharynx accompanied by a corresponding increase in nonvaccine serotype pneumococci, as they fill the newly vacant ecological niche. Serotype replacement in the nasopharynx of a healthy child may or may not be problematic; the public health concern is whether or not replacement serotypes also cause disease. Serotype replacement invasive disease among children and adults has significantly increased in the US post-vaccination [15–18], and invasive pneumococcal disease among children <5 y of age is now predominantly due to serotype 19A [17]. A recent JAMA report described a significant overall increase in nonvaccine serotype disease among Native Alaskan children, most frequently due to serotype 19A [18]. Genotypic characterisation of these serotype 19A strains by MLST showed that most of the serotype 19A replacement disease can be explained by clonal expansion of one genotype, ST199, which existed prior to vaccination [15,18]. The second major vaccine-related concern is the possibility of capsular switching, when the genes encoding one type of capsule are exchanged, via transformation and recombination, for the genes encoding a different type of capsule. Capsular switches from one vaccine serotype to another were first described 16 y ago [19–21], but it is the vaccine-to-nonvaccine serotype switch that is of primary concern, because it contributes to serotype replacement and allows for the possibility of vaccine escape. Acquisition of a nonvaccine capsule by a pneumococcal strain capable of causing invasive disease has been a serious concern related to the use of any serotype-specific vaccine [11,22]. The capsular locus of the pneumococcus is located between two genes, dexB and aliA ([7]; Figure 1). (Serotype 37 is one exception, as it has a defective capsular locus and the serotype is determined by the type 37 synthase gene [tts] located elsewhere in the pneumococcal genome [7,23].) Two of the six penicillin-binding proteins (PBPs) possessed by the pneumococcus are near the capsular locus: pbp2x is upstream of dexB and pbp1a is downstream of aliA. Alterations in PBPs confer penicillin resistance, and alterations in pbp2x, pbp1a, and pbp2b are the most important [24–26]. Penicillin-resistant pneumococci are a major problem throughout the world [27]; prior to use of PCV7 in the US, nearly one-quarter of invasive pneumococci were penicillin-nonsusceptible [28]. Initially, use of PCV7 in the US reduced the incidence of invasive disease due to antimicrobial-resistant vaccine serotypes, but more recently, antimicrobial resistance has increased with the increase of serotype 19A disease [28,29]. The Centers for Disease Control and Prevention (CDC) has been monitoring invasive pneumococcal disease since 1995 through the Active Bacterial Core (ABC) surveillance program [6,14,30] and as a result, the post-vaccination increase in nonvaccine serotype 19A disease in the US was quickly detected. Serotype 19A strains collected by the CDC through 2005 were genotyped by MLST, which revealed that vaccine escape strains had begun to emerge in 2003 [14,15]. These strains possessed an MLST genotype, ST695, that had always been associated with vaccine serotype 4 (ST6954), but now expressed a serotype 19A capsule (ST69519A). These strains were detected only 3 y after vaccine implementation, but rapidly increased in prevalence. The first three strains were detected in 2003; two strains were detected in 2004; and 32 strains were detected in 2005, some of which had evolved further. Moreover, in 2005, two new types of serotype 19A vaccine escape strains emerged, ST236519A (n = 4) and ST89919A (n = 1); these appeared to represent new recombinational events that also occurred between serotype 4 recipients and serotype 19A donors. The aim of this study was to sequence the regions upstream and downstream of the capsular locus, including both PBPs, to identify the putative recombinational event (s) that resulted in these vaccine escape strains. Strains were selected based upon serotyping and MLST genotyping data and are described in Table 1. Between 1998 and 2005,31,669 cases of invasive pneumococcal disease among all ages were identified from ABC sites, from which 28,363 pneumococcal strains were available for serotyping. 15,736 (55%) strains were of nonvaccine serotypes, and 1,935 of those were serotype 19A, 88% of which were recovered from 2000 to 2005 ([17] and ABC surveillance program, unpublished data). The following serotype 19A strains were genotyped by MLST: 82 of 113 (73%) strains from 1999 [14,31] and 779 of 1,597 (49%) strains from 2001 to 2005 ([14,15] and ABC surveillance program, unpublished data). No serotype 19A strains from 1998 or 2000 were genotyped by MLST. Possible serotype 19A capsular switches were identified among 57 of 779 (7%) of 2001–2005 genotyped strains: ST271,690,899,1092, and 1790 (n = 1 strain each); ST236,338, and 557 (n = 2 strains each); ST230 and 292 (n = 3 strains each); ST156 (n = 4 strains); and ST695 (n = 37 strains) [14,15,31]. Among these strains, the ST69519A plus related ST89919A strains were characterised in this study. All other possible capsular switch strains were present in very low numbers and/or were also found outside the US; vaccine and nonvaccine serotype variants of ST156 are recognised in the global database (http: //www. mlst. net/). Recipient strains were those with the background genotype into which the 19A capsular locus recombined: ST6954 (n = 3), recovered in 1999, and ST8994 (n = 1), recovered in 2002. Based upon pulsed-field gel electrophoresis, selected MLST data, and eBURST analysis, it is likely that strains of ST8994 were also common among children <5 y of age before vaccine introduction [14]. Putative donor strains of the 19A capsular locus were selected from pneumococci recovered in 1999 and 2003. The most likely donor of the 19A capsular locus was ST19919A, because ST19919A existed as a prevalent strain pre-vaccine implementation and was the major clone responsible for serotype replacement disease post-vaccine implementation [15,17]. Thus, four pre- and post-vaccine implementation strains of ST19919A were selected, plus one strain of ST64519A, which is a single-locus MLST variant of ST19919A. All serotype 19A strains collected through 2005 that were ST695 or closely related MLST genotypes were characterised as progeny strains. This resulted in the characterisation of 42 unique strains: ST69519A (n = 36), ST236519A (n = 4), ST236319A (n = 1), and ST89919A (n = 1). Pneumococci were recovered from patients aged <1 to 96 y with invasive pneumococcal disease (Table 1). All pneumococci were recovered from blood, apart from one strain (cdc46) that was recovered from cerebrospinal fluid. Among vaccine escape strains, 40 of 42 (95%) were collected in the northeastern US (note that 37% of all 20,196 available isolates since 2000 were from the Northeast; ABC surveillance program, unpublished data); and two strains were from Minnesota (cdc27) and Colorado (cdc21). Three putative donor strains were from Georgia (cdc8, cdc9, and cdc11); all other putative donor and recipient strains were collected in the Northeast. All vaccine escape strains were recovered from 2003 to 2005. Strains from 2003 to 2004 were recovered from children ≤4 y of age, although only pediatric isolates were genotyped [15]; 2005 strains were recovered from children and adults. Regions upstream and downstream of the capsular locus were sequenced in all 51 pneumococci (Figure 1). Sequence alignment comparisons showed that recipient, donor, and progeny strains could be grouped based upon the relatedness of the regions flanking the capsular locus, considered as one joined sequence of 20. 6 kb (Figure 1). The capsular locus has not yet been sequenced, but serotypes have been confirmed phenotypically, and the serotypes of a subset of the vaccine escape strains were confirmed by a PCR assay [32]. The capsular locus is considered here only as it encodes either a serotype 4 or 19A capsule. Progeny strains consisted of three groups, P1, P2, and P3 (Table 1). Twenty-seven ST69519A progeny strains (P1) were identical to each other in the capsular locus flanking regions and were the major group of vaccine escape strains. There were five additional P1 variants, all of which differed from the P1 strains by only one to two single nucleotide polymorphisms (SNPs) over the entire region of sequence. Furthermore, one strain, P1var6, was a single-locus MLST variant of the P1 strains at spi, but the capsular locus flanking regions of P1 and P1var6 were identical. P2 progeny strains and strain P3 were 0. 3% and 1%, respectively, divergent from P1 progeny over the entire region of sequence. Among recipient strains, the flanking sequences of cdc1, cdc3, and cdc4 (R1) were identical; recipient strain cdc7 (R2) was 0. 5% divergent from R1 strains (Table 1). Among 19A donor strains there were four unique sequences: cdc2 and cdc8 (D1), cdc9 (D2), cdc11 (D3), and cdc6 (D4). Sequence alignments of the capsular locus flanking regions revealed the upstream and downstream recombinational crossover points in the ST69519A strains. Three SNPs (heterologous to donor and homologous to recipient) marked and confirmed the upstream crossover point, just upstream of pbp2x; similarly, two SNPs marked and confirmed the downstream crossover point, at the 3′ end and just beyond the 3′ end of pbp1a (Figures 2A and S1), resulting in a recombinational fragment of 38. 6 kb. To our knowledge, this is the first reported in vivo example of a recombinational event in pneumococcus in which the capsular locus and both adjacent PBPs recombined in what appeared to be a single event. This was shown to be experimentally possible by Trzcinski and colleagues [33]. The likely donor of the serotype 19A capsular locus and flanking regions in ST69519A was ST19919A. Although the actual strain that donated the DNA cannot be identified with certainty, it is represented by the pre- and post-vaccination D1 strains from 1999 and 2003, respectively (Figures 2A and S1). Sequence alignments also revealed a second recombinational event around the capsular locus in four strains of P2 progeny, all of which were first identified in 2005. Three P2 strains were identical (or one SNP different, P2var1) to D4 in the entire flanking region sequenced (Figures 2B and S1), suggesting that D4 was the likely donor of the 19A capsule among P2 progeny strains. The third type of 19A progeny strain identified in this study was P3 (ST89919A). ST899 appears to have circulated at least since 1999 as a serotype 4 strain [14], but in 2005 appeared as a 19A serotype strain (Table 1). None of the sequenced 19A donors in this study share distinctive regions of homology to the P3 progeny; thus, the donor of this 19A capsular locus is still unknown. Figure 3 further clarifies the evolution of these vaccine escape strains by depicting the relatedness of the background MLST genotypes. The P1 progeny (including P1 variants) are all in the cluster with ST695 as the founder genotype. ST695 is a single-locus variant of ST899, differing only at xpt (allele 113) as a result of a 39-codon deletion in the middle of the xpt gene. This truncated xpt allele was first identified in 1999 among US strains [31] and was present in ST236319A and ST22844. ST2363 and ST2284 differ from ST695 at spi and recP, respectively. The truncated xpt allele is identical to xpt allele 10 apart from the large deletion, thus allele 113 probably derived from the full-length allele 10 (Table 1; http: //www. mlst. net/). The P2 progeny (including the P2variant) appeared to evolve from the founder ST2474 clonal complex. The oldest known strains of ST2474 were identified in The Netherlands in 1978, and ST2474 and the related single-locus variants in the ST2474 cluster have been detected in other European and South American countries through at least 2005 (Figure 3; http: //www. mlst. net/). The P2 progeny were first identified in 2005 in the US. ST2474 strains have not yet been characterised, thus recombinational crossover points in P2 progeny have not yet been identified. The third serotype 19A capsular change was the P3 progeny (ST89919A); Figure 3 depicts the relatedness of this ST to other STs in the clonal complex. Alterations in PBP genes pbp2x, pbp1a, and pbp2b confer penicillin resistance [24–26]. Allele assignments were given to each unique pbp2x and pbp1a sequence among this collection of strains; the pbp2x and pbp1a sequences from TIGR4, a serotype 4 penicillin-susceptible strain, were used for comparison and designated allele 1 (Table 1; Figure 4). All R1, R2, and P3 strains possessed pbp2x and pbp1a alleles similar to those in TIGR4, which correspond to the basal penicillin minimum inhibitory concentration (MIC) of 0. 03 μg/ml that is typical of all ST6954 isolates. Among D3, D4, and P2 strains, pbp2x was altered while pbp1a was largely conserved; both pbp2x and pbp1a were altered in the D1, D2, P1, and P1variant strains. Phenotypically, strains with altered pbp2x and pbp1a genes resulted in a MIC to penicillin of 0. 06–0. 12 μg/ml (Table 1). The breakpoints for penicillin resistance are ≤0. 06 μg/ml (susceptible), 0. 12–1. 0 μg/ml (intermediate resistance), and ≥2 μg/ml (high-level resistance), but the variation in susceptibility testing is ±1 doubling dilution [34]. All donor strains and all progeny strains apart from the P3 strain had MIC values at the penicillin-susceptible breakpoint; however, based on the pbp2x and pbp1a sequences, D1, D2, P1, and P1variants had divergent PBPs typical of penicillin-nonsusceptible strains ([24–26]; Figure 4). All pbp sequences in this study were unique to the National Center for Biotechnology Information database. Among all strains, four donor and five vaccine escape strains were erythromycin resistant, possessing the mef (A) efflux phenotype [29,35]; all five vaccine escape strains and two donors (cdc8 and cdc9) were confirmed to be mef (A) -positive by PCR ([36,37]; Lesley McGee, Emory University, personal communication). The other donor strains were not tested by PCR. One erythromycin-resistant donor strain (D4) was also clindamycin resistant (Table 1). D1 and D2 donor strains were co-trimoxazole resistant, and one vaccine escape strain (cdc13) was resistant to tetracycline. One P1var2 strain (cdc44) was resistant to chloramphenicol; R2 was resistant to chloramphenicol and levofloxacin. To our knowledge, this is the first report of vaccine escape events in the US subsequent to national pneumococcal vaccination. The main event resulted in nonvaccine serotype, penicillin-nonsusceptible ST69519A pneumococci as a result of recombination between the recipient ST6954 and donor ST19919A (D1). A second event resulted in the emergence of ST236519A from the well-established serotype 4 clonal complex of ST247, and these strains had a different serotype 19A donor, ST64519A. No recipient has yet been identified, but recipient strains of ST247 will be characterised to elucidate the recombinational crossover points in these progeny. Finally, a third event resulted in the emergence of ST89919A, but the serotype 19A donor is not yet known. The most plausible explanation for the emergence of these strains is that one main recombinational event, ST69519A, occurred around 2003, and the fact that it was a nonvaccine serotype variant provided the selective advantage by which these strains could evade the immune pressures invoked by the vaccine. Furthermore, with the additional selective advantage of penicillin nonsusceptibility, this progeny strain increased in frequency over the next 2 y, disseminating through the northeastern US to become the fourth most common serotype 19A genetic complex in the US (ABC surveillance program, unpublished data). The recombinational event did not appear to result in a decrease in fitness among these strains, as strains increased over time within the population. More worryingly, all of these strains were recovered from patients with bacteremia and meningitis. We cannot be absolutely certain that ST69519A strains never existed pre-vaccination, but extensive surveillance pre- and post-vaccination in the US failed to reveal any such strains [14,31], and no such strains have been reported to the MLST database from other parts of the world. Hence, even if these strains did exist pre-vaccination they were likely to be very rare, and it could still be maintained that the immune pressure resulting from PCV7 use selected for the emergence of such strains. The two additional progeny strain types, ST236519A and ST89919A, were identified in 2005, and this may be the very early detection of these new recombinants. Whether or not these strains will increase in prevalence remains to be seen. Since serotype 19A was the major replacement disease serotype to emerge post-vaccination [15,17], our efforts to understand the genetics of this event have been focused on serotype 19A strains. This will be expanded to include other putative capsular switching events to better understand the phenomenon of capsular switching. Although it has been known for many years that capsular switching can and does occur, it is not at all clear how often it occurs within the pneumococcal population. An exciting component of the current study is that recombination in the pneumococcus has been revealed almost as it occurred. This presents a unique opportunity to measure recombination in nature as a result of vaccine-induced changes, and may shed light on how much recombination occurs in general within the pneumococcal genome. Studies to explore these strains in detail are ongoing. What are the implications for vaccine escape? Clearly, vaccine escape by recombination at the capsular locus has the very real potential to reduce PCV7 effectiveness in the longer term. This will almost certainly be true for any serotype-specific pneumococcal vaccine, given the diversity and complexity of serotypes possessed by the pneumococcus. These US data reinforce two key points: i) the importance of surveillance pre- and post-vaccination in countries preparing to implement PCV7, to detect changes as and when they occur; and ii) the importance of understanding the genetic events that result in vaccine escape. Discerning the genetic event is crucial to understanding vaccine escape and pneumococcal recombination in general. Such knowledge will provide guidance about the design and use of future pneumococcal vaccines. Invasive pneumococcal strains were collected as part of the ABC surveillance program at the CDC. Microbiological testing and molecular characterisation by MLST was performed as previously described [14,31]. Strains for this study were selected from pre- and post-vaccine implementation ABC collections and sent to the University of Oxford. Sequence alignments of the TIGR4 [38], R6 [39], and Spain23F-1 (http: //www. sanger. ac. uk/Projects/S_pneumoniae/) pneumococcal genomes were used to design 51 sets of PCR primers (available upon request) in regions upstream and downstream of the capsular locus (Figure 1). DNA was extracted from pneumococcal strains using the DNeasy Tissue Kit (Qiagen UK). PCR assays were identical for all combinations of PCR primers: 2. 5 μl of PCR buffer (Qiagen), 0. 5 μl of dNTPs (200 μM stock), 0. 5 μl of forward primer (100 μM stock), 0. 5 μl of reverse primer (100 μM stock), 0. 3 μl of Taq polymerase (Qiagen UK), 19. 7 μl of nuclease-free water (Invitrogen UK), and 1 μl (∼1 μg) of extracted DNA. Thirty-five cycles of PCR amplification were performed: denaturation at 94 °C for 30 s, annealing at 55 °C for 30 s, and elongation at 72 °C for 60 s. Ten-microliter sequencing reactions were prepared with 4 μl primer (1 μM stock of PCR primers), 1. 75 μl sequencing buffer (Applied Biosystems), 0. 5 μl Big Dye (Applied Biosystems), 1. 75 μl nuclease-free water, and 2 μl cleaned PCR amplicons; followed by 25 sequencing cycles of 96 °C for 10 s, 50 °C for 5 s, and 60 °C for 2 min. PCR amplicons and sequencing products were cleaned as described previously [15]. Sequences were assembled and edited using Pregap4 and Gap4 (Staden Package, http: //staden. sourceforge. net/). Consensus sequences were aligned using Multalin [40] and trimmed to the longest possible consensus sequence shared by all strains. Trimmed sequences were uploaded into MEGA version 3. 1 [41], and compared in an iterative process to ascertain which strains were donors of the 19A capsule and reveal recombinational crossover points in progeny strains. The eBURST algorithm was used to compare all progeny strains to all pneumococcal strains represented in the MLST database [42]. All unique pbp2x and pbp1a sequences in this study were submitted to GenBank (http: //www. ncbi. nlm. nih. gov/Genbank/index. html) under accession numbers EU034013–EU034023.

The 7-valent pneumococcal conjugate vaccine is a remarkable public health success story. It has significantly reduced invasive pneumococcal disease in the United States not only by protecting vaccinated children, but also by protecting unvaccinated older children and adults by herd immunity. However, there was always a concern that use of a limited-valency vaccine would result in an increase in disease due to nonvaccine serotypes, and this has now occurred in the US. The predominant nonvaccine serotype causing invasive disease is 19A, and this increase is partially explained by “vaccine escape” pneumococci, strains that have exchanged the genes that encode a vaccine serotype 4 capsule for genes that encode a nonvaccine serotype 19A capsule. These strains are then able to escape vaccine-induced immunity. Characterisation of the genetic event that resulted in these vaccine escape strains was the focus of our study and the results were surprising. The results of this study have important relevance to the long-term effectiveness of the current vaccine and to the development of future pneumococcal vaccines.

Abstract Introduction Results Discussion Materials and Methods Supporting Information

infectious diseases microbiology eubacteria

2007

Vaccine Escape Recombinants Emerge after Pneumococcal Vaccination in the United States

Continuous cultures of mammalian cells are complex systems displaying hallmark phenomena of nonlinear dynamics, such as multi-stability, hysteresis, as well as sharp transitions between different metabolic states. In this context mathematical models may suggest control strategies to steer the system towards desired states. Although even clonal populations are known to exhibit cell-to-cell variability, most of the currently studied models assume that the population is homogeneous. To overcome this limitation, we use the maximum entropy principle to model the phenotypic distribution of cells in a chemostat as a function of the dilution rate. We consider the coupling between cell metabolism and extracellular variables describing the state of the bioreactor and take into account the impact of toxic byproduct accumulation on cell viability. We present a formal solution for the stationary state of the chemostat and show how to apply it in two examples. First, a simplified model of cell metabolism where the exact solution is tractable, and then a genome-scale metabolic network of the Chinese hamster ovary (CHO) cell line. Along the way we discuss several consequences of heterogeneity, such as: qualitative changes in the dynamical landscape of the system, increasing concentrations of byproducts that vanish in the homogeneous case, and larger population sizes. Recombinant protein production requires suitable cell hosts and culture conditions [1]. For this purpose mammalian cells are often grown in chemostat-like cultures where a continuous flow of incoming fresh media replaces culture liquid containing cells and metabolites. Alternative processes such as batch or fed-batch are also adopted by many industrial facilities, but the advantages of the continuous mode have been predicted to drive its wide adoption in the near future [2–7]. However, experiments have demonstrated that continuous cultures exhibit hallmark phenomena of nonlinear dynamics, such as multiple steady states under identical external conditions [8–11] and hysteresis loops [8,12,13]. Sophisticated control strategies are then required to drive the system towards desired steady states. In this context, mathematical modeling has been used with some success [13–15]. Already in Ref. [13], we have shown how a model of a homogeneous continuous cell culture can explain these phenomena in the context of a detailed metabolic model, while predicting numerous metabolic transitions as a function of the ratio between cell density and dilution rate (also known as the inverse cell-specific perfusion rate [16]). However, most of these works deal with simple cell populations, consisting of identical cells (as in Ref. [13]), or at most of a few competing species [15,17]. Although it is known that no two cells in culture are alike [18], the effects of individual cell-to-cell variability are seldom considered [19]. Attempts to model heterogeneity in cell cultures are often based in population balance models [20] or similar approaches, which require prior postulation of the mechanism driving the heterogeneity and depend on more quantitative parameters than homogeneous modeling. These models are affected in part by the limited availability of quantitative data [21], but also by an incomplete understanding of the role played by different mechanisms driving heterogeneity. Indeed, many complex processes contribute to heterogeneity in a cell population, including gene expression noise [22], partition errors at cell division [23,24], mutations [25], size variability [26], as well as environmental gradients in the culture [27]. It is then important to understand what features change or are missed in a model of continuous culture by treating the cell population as an homogeneous system. In this work we propose to apply the maximum entropy principle [28–34] to model cell heterogeneity in a continuous bioreactor. This approach necessitates much less parameters than a microscopically detailed model based on population balance. We show how the model can be resolved computationally in large metabolic networks by exploiting a recent implementation of the Expectation Propagation algorithm [35,36]. To highlight the phenomenology of the heterogeneous cell population, we establish a comparison to the chemostat model of Ref. [13] which considers a homogeneous population, but we note that the framework presented here can easily accommodate other models of continuous cell cultures and metabolic networks. Although some of the predictions of Ref. [13] remain valid in the presence of heterogeneity (such as the importance of the ratio between cell density and dilution rate as a control parameter), the heterogeneity introduced here leads to important qualitative differences that we discuss in detail below. These include changes in the kinds of metabolic transitions observed (e. g. , diverging concentrations of byproducts that should be zero if the population were homogeneous or depletion of different nutrients), higher cell densities in the presence of toxicity, and drastic modifications of the dynamical landscape of the system such as disappearance of a bistable regime. We study the steady states of a population of cells inside a well-mixed bioreactor, where fresh medium continuously replaces culture fluid at a given dilution rate D. Each cell will be described by vectors r _ (h) and u _ (h), giving the flux of every reaction in the metabolic network of the cell line under study and the metabolite uptake rates, respectively. Here and in what follows, the super-index h denotes an individual cell. If Nik denotes the stoichiometric coefficient of metabolite i in reaction k (Nik > 0 if metabolite i is produced in the reaction, Nik < 0 if it is consumed, and Nik = 0 if it does not participate in the reaction), then cell h produces metabolite i at an overall rate ∑ k N i k r k (h). This production must balance the cellular demands for metabolite i. In particular we consider a constant maintenance demand e i (h) which is independent of growth [37,38], and the requirements y i (h) of each metabolite for the synthesis of biomass [39,40]. If biomass is synthesized at a rate z (h), we obtain the following overall balance equation for each metabolite i: ∑ k N i k r k (h) + u i (h) = e i (h) + y i (h) z (h) (1) For simplicity we will assume that the overall macromolecular composition of the cell is constant, i. e. that e i (h) = e i and y i (h) = y i are independent of h. The net growth rate of a cell depends both on the rate of biomass synthesis, z, and on the concentration of metabolites in the media. We assume that some metabolites are toxic (such as lactate and ammonia) and affect the growth rate of cells. In order to accommodate the most common dependencies found in the literature, we assume: λ (h) = z (h) × K (s _) - σ (s _) (2) where s _ is the vector of metabolite concentrations in the culture, λ (h) is the net growth rate, K (s _) is a growth inhibition factor, and σ (s _) a death rate. In addition to Eq 1, the fluxes r k (h), u i (h) must respect physico-chemical constrains arising from thermodynamics and molecular crowding. These constrains will also depend on the extracellular conditions of the culture, specifically the concentrations s _ of metabolites. They can be summarized as: ∑ k α k | r k (h) | ≤ C (3) lb k ≤ r k (h) ≤ ub k (4) - L i ≤ u i (h) ≤ U i (5) where αk are the enzymatic flux costs per unit of flux through reaction k, C is the maximum enzymatic cost available to the cell, lbk, ubk are the lower and upper bounds of reaction k, Li = 0 for metabolites that cannot be excreted from the cell and Li = ∞ otherwise, Ui is the maximum uptake rate of metabolite i, and X the total number of alive cells. The value of Ui will depend on the conditions of the culture and will be specified below. The reaction bounds lbk, ubk are −∞ and ∞ respectively for reversible reactions, while lbk = 0, ubk = ∞ for irreversible reactions. The constrains (1–5) define a convex polytope of feasible metabolic states [41,42] that we denote by P s _. Each point within this space consists of coordinates v _ = (r _, u _) which fully specify the metabolic state of a cell in the model. Let P s _ (v _) be the density of cells with metabolic fluxes v _. To determine the form of P s _ (v _) d v _, we adopt the Principle of Maximum Entropy (MaxEnt), which in this context can be stated as follows [28,32]: For β = 0 every point in the space P s _ is equally likely. In this case cells explore uniformly the space of allowed solutions defined by Eqs (1–5). For β = ∞ the distribution is a Dirac delta with infinite mass on the points of maximum growth rate. In the language of Statistical Mechanics this is the ground state of the system, also known in the Systems Biology literature as the Flux Balance Analysis (FBA) solution [41]. To gain some insight, we analyze first a simplified metabolic model that exhibits a switch from oxidative pathways at low growth rates to fermentative pathways at higher growth rates [43]. The model consists of the following stoichiometric constrains: 2 v g + v l - v o = 0 (6) 2 v g + 18 v o = v a t p (7) v g ≤ min { V g, c g D / X } (8) v l ≤ 0,0 ≤ v o ≤ R (9) where vg is the velocity of consumption of glucose, vl the excretion of lactate, vo the velocity of oxidative phosphorylation, and vatp the synthesis of ATP (see Fig 1). We only consider the ATP requirements of growth and maintenance and ignore additional biomass components. Therefore the rate of biomass synthesis (z) is an affine function of the rate of ATP synthesis, z = (vatp − e) /y, where y is the biomass yield and e is the constant maintenance demand for ATP [37]. The rate of oxidative phosphorylation is caped by the limited capacity of mitochondria, R [44]. This simplified model or close variations have been extensively discussed in the literature, e. g. [13,43,45,46]. We assume that the secreted lactate is toxic, so that the net growth rate is given by λ = z − σ, where the death rate σ is some increasing function of the extracellular concentration of lactate, w. For simplicity we use a linear form, σ = τw, where τ is the death rate per unit of lactate. One can think of τ as the slope of a more realistic description of the dependence of the death rate on the concentrations of toxic compounds, thus avoiding the introduction of too many parameters into this simplified model. Parameters are based on experimental values obtained for mammalian cells in the literature. The ATP maintenance demand e = 1mmol/gDW/h has been measured for mouse LS cells [37] (see [47] for similar values in other cell types). The value R = 0. 45mmol/gDW/h was calculated in [13], based on measurements of the glucose uptake threshold that triggers a switch to fermentation in mammalian cells [43]. The maximum velocity of glucose uptake, Vg = 0. 5mmol/gDW/h is based on a value measured for HeLa cells [48]. The value y = 348mmol/gDW was then adjusted so that the maximum growth rate matches typical duplication rates of mammalian cells of 1/day. From a linear approximation of the death rate dependence on lactate concentration measured in a mammalian culture [49] we obtain τ = 0. 0022h−1mM−1. The concentration of glucose was set at a value typical of mammalian cell culture media, cg = 15mM. Motivated by the fact that most therapeutic proteins requiring complex post-translational modifications are produced in Chinese hamster ovary (CHO) host cell-lines [1], our second example is a metabolic model of this cell line. We employed a CHO-K1 line-specific metabolic model, based on the latest reconstruction of CHO metabolism available at the time of writing [50]. The network recapitulates experimental growth rates, essential enzyme requirements and cell line specific amino acid auxotrophies. In order to enforce (3), we complemented this network with a set of reaction costs of the form αi = MWi/ai, where MWi is the molecular weight of the enzyme catalyzing reaction i and ai the specific activity. These values were mined by T. Shlomi et. al [51] from public enzyme data repositories for use in a human metabolic network. Unfortunately there is no experimental data available to determine a set of reaction costs specific for CHO-K1, but those of a human network should give a good approximation of the cost coefficients of generic mammalian cells. For reactions where the corresponding value could not be found, we set its flux cost to the median of all the flux costs available. Finally C = 0. 078mg/mgDW is the mass fraction of metabolic enzymes in the dry weight of a typical mammalian cells, which can be estimated from protein abundance measurements [51,52]. The predictions of combining flux balance analysis with a crowding constrain obtained in this manner have been shown to exhibit significant correlation with enzyme mRNA expression levels [51]. A constant maintenance demand was added as an ATP hydrolysis flux at a rate 1mmol/gDW/h, based on measurements for mouse LS cells [37]. The maximum glucose uptake was set at Vg = 0. 5mmol/gDW/h, a value measured in HeLa cells [48]. Unfortunately kinetic parameters to estimate Vi for many other metabolites are not available. However, based on multiple reports in the literature [49,53,54] we estimated that amino acid uptakes are typically tenfold slower than sugar uptake, and therefore set Vi = Vg/10 for amino acids. In the simulations we used Iscove’s modified Dulbecco’s medium (IMDM) to set the values of ci, and set Vi = ∞ and infinite concentrations for water, protons and oxygen (see Supplementary Materials for full specification). Since lactate and ammonia are the most commonly recognized toxic byproducts in mammalian cell cultures, we set λ = K × z with K = (1 + snh4/ Knh4) - 1 (1 + s lac / K lac) - 1 (10) with Knh4 = 1. 05mM and Klac = 8mM, based on the values and functional form reported by [55]. In order to compute the steady state concentrations si, it is necessary to compute the expected value of ui under the MaxEnt distribution. Although in the simple model this poses no difficulty, in a genome-scale metabolic network such as the CHO-K1 considered in this paper the vector v _ has hundreds of components and an exact computation [56] becomes intractable. A hit and run Monte Carlo approach has been used for moderately sized networks [32,57]. Although these methods are guaranteed to converge to an uniform sample, this is only true in the asymptotic limit of an infinite number of steps. Unfortunately the geometrical shape of the metabolic flux space is highly elongated in some directions but very compressed in others [57]. In practice it becomes very hard to determine how long the Monte Carlo computation should run to achieve convergence, particularly so for large metabolic networks. A better approach is to use message-passing algorithms [58]. Recently, Expectation Propagation (EP) [35] has been successfully employed to compute a very good approximation of the marginal flux distributions in absence of selection (β = 0) [36]. In [36] the reader can find an exhaustive assessment of the quality of this approximation in a variety of metabolic networks. In Supplementary Materials we describe how the same method can be used to approximate the marginal flux distributions for non-zero β. All model simulations were carried out in Julia [59]. The CHO-K1 metabolic network [50] was loaded and setup using the COBRApy package [60]. The expectation propagation implementation was taken from https: //github. com/anna-pa-m/Metabolic-EP [36]. Since taking the absolute value in (3) is not a linear operation, we must replace reversible reactions in the model by two reactions, one in the forward and another in the backward direction. This ensures that all reaction fluxes are non-negative variables and (3) becomes a linear inequality. However, this almost doubles the number of reactions in the CHO-K1 model, which makes Expectation Propagation extremely slow to converge or in some cases it even fails to do so. In order to obtain a reduced tractable network, we first solved our model at β = ∞, which can be done with standard linear programming packages [61]. In cases where FBA produces multiple solutions maximizing biomass production [62], we select the one that minimizes the enzymatic cost in (3) [63]. This guarantees that the solution is uniquely determined, because the cost coefficients are sufficiently heterogeneous that the contours of constant cost are almost surely not parallel to the contours of constant biomass production. Moreover from the biological viewpoint it is reasonable that the cell minimizes the enzymatic mass required to support a certain metabolic goal [63–65]. Then we removed from the CHO-K1 model all reactions that were identically zero for all values of ξ. This results in a reduced model with 380 metabolites and 401 reactions (provided as supplementary materials), where Expectation Propagation converges robustly and fast. In order to determine P s _ we need to know the concentrations si, the function λ s _ (v _) giving the growth rate for each feasible metabolic state, and the average growth rate 〈λ〉. The later is easy in the chemostat: 〈λ〉 = D (the dilution rate) [15]. It is much more difficult to obtain experimental data for all the relevant si. In the worst case in which no information about the si is available, we notice that when the chemostat is in steady state, the input flux of metabolite i must balance its consumption by the cells and its output flow. The input and output fluxes per culture volume in the chemostat are given by Dci and Dsi, respectively, where ci is the concentration of metabolite i in the external feed. Since the rate of consumption of individual cells is u i (h), the total consumption by the population of cells can be estimated as: ∑ h u i (h) ≈ X ∫ u i (v _) P s _ (v _) d v _ (11) where X is the total number of cells. In the limit of a large number of cells this equation becomes exact. In steady state, the concentration si is constant, which leads to the following mass-balance equation: D (c i - s i) - X ∫ u i (v _) P s _ (v _) d v _ = 0 (12) To obtain the steady state values of si, Eq (12) must be solved self-consistently together with the MaxEnt form of the distribution P s _ (v _). Moreover since the steady state value of si must be non-negative, we can extract from this equation an approximate form for the maximum uptake rate that further simplifies the computation (see Supplementary Materials for a derivation of this equation from Michaelis-Menten kinetics), U i = min { V i, c i X / D } = min { V i, c i ξ } (13) where Vi is the maximum uptake rate of metabolite i. The solution of the model in the case β = ∞ (FBA) is described in detail in Ref. [13], where it was argued that the ratio ξ = X/D determines the steady state of the culture and links steady states in the chemostat with those in perfusion systems. It has units (cells × time / volume), and can be interpreted as the number of cells maintained alive for a period of time per unit of volume of fresh media supplied into the culture. In the case of finite β, the value of ξ only determines the shape of the polytope of phenotypic states that cells can adopt. According to the maximum entropy principle, the distribution of cells within this polytope is of the form: P (v _) = e β ′ z (v _) ∫ P e β ′ z (v _) d v _ (14) where β ′ = β K (s _) and the parameter β quantifies the level of heterogeneity in the population of cells. It is a macroscopic representation of the underlying noisy processes that sustains cell-to-cell variability in the population. Small values of β lead to an almost uniform distribution P (v _) over all possible states. This corresponds to a highly heterogeneous population. In contrast, at larger values of β the population concentrates around the FBA solution maximizing the growth rate. This corresponds to a highly homogenous population. From (14), we compute the expected values of the exchange fluxes 〈ui〉 using the Expectation Propagation algorithm described in the appendix [36]. Next, the values of the metabolite concentrations are obtained from (12): s i = c i - ⟨ u i ⟩ ξ (15) Given s _, we compute K (s _), σ (s _) and then β = β ′ / K (s _). Similarly, 〈 λ 〉 = 〈 z 〉 K (s _) - σ (s _), which then determines the value of the dilution rate consistent with this solution, D = 〈λ〉. Finally, the total number of cells in the steady state is given by X = ξD. Within our framework the simple metabolic model admits an exact solution that provides important clues about the role of heterogeneity in more realistic models. For a given value of ξ, in this case the MaxEnt distribution takes the form: P (v a t p, v g; ξ) = e - β v a t p / y / Z (ξ) (16) for (v a t p, v g) ∈ P ξ, where P ξ is the polytope defined by the constrains (6) – (9) (after eliminating vl, vo), and Z (ξ) = ∫ P ξ d v a t p d v g e - β v a t p / y (17) Notice that constant terms, including the additive death rate σ and the maintenance demands, cancel upon normalization. Due to the low-dimensionality of this model, the moments of (16) can be evaluated by numerical integration to any desired accuracy. Then the steady state concentrations of glucose and lactate can be calculated using (15). Fig 2 shows typical flux distributions of vatp and vglc in the population of cells, for different values of ξ and β. As β increases (more homogeneous populations), cells cluster around the maximum feasible rate of ATP production, which also coincides with maximum glucose consumption. In the β → ∞ limit (FBA) the distribution becomes a Dirac delta (purple line) localized in this optimal point. On the other hand, when β → 0 cells distribute uniformly in the space of feasible metabolic states, favoring states with low growth rate. This observation has been interpreted as implying that higher growth rates require active regulation in the cell [32] (see Supplementary Material for a study of an evolutionary model consistent with this explanation within our framework). Fig 3 shows the solution of the model as a function of ξ, for selected representative values of β. For comparison the β = ∞ solution is shown in purple. As the crossing curves in Fig 3a indicate, the effect of decreasing β is not simply to decrease the average growth rate of the population of cells. At first sight this seems to contradict the fact that the β = ∞ solution is where cells adopt a phenotype with maximum growth rate (the FBA solution). However, due to the accumulation of toxic byproducts in the culture, maximizing z may result in higher toxicities, and the net effect is to decrease the overall growth rate of the population. This translates into the fact that the heterogenous population may have higher cell numbers than the homogenous one (where the red curve is higher than the purple in Fig 3b). This explanation is confirmed in Fig 3d, which shows that the concentration of lactate reaches the highest levels when β = ∞. A second feature connected to heterogeneity is that even for high values of β and ξ, the concentration of lactate in the culture increases with ξ. However in the strictly homogeneous limit (β = ∞) it goes to zero. This striking difference has important implications in the interpretations of bulk metabolic measurements in populations of cells. For example, the observation of an increasing concentration of lactate could be interpreted as the result of a selective pressure, pushing cells towards fermentation. On the contrary, our model shows that in a chemostat this is a natural consequence of the heterogeneity of the population. Moreover notice that, for high enough β, the curve of D versus ξ (Fig 3a) displays multistability. To the same value of D there may correspond more than one value of ξ. A theorem proved in Ref. [13] establishes that a steady state is stable if D (ξ) is decreasing at ξ. The theorem also holds in the present model, where the heterogeneity only redefines the function D (ξ) to which the theorem was originally applied. In this case, an important consequence of the heterogeneity is that it reshapes the dynamical landscape of the system, which is tightly connected to the decrease of the accumulation of lactate in the system as β decreases (cf. Fig 3d). For example, increasing the heterogeneity (decreasing β) abolishes the bistable regime. This is shown clearly in Fig 4, where the steady state values of X are plotted against the dilution rate. Heterogeneity also has the undesirable consequence of decreasing the medium depth (ξm), the maximum value of ξ with a non-zero steady state concentration of live cells. It is important to realize that even if the heterogeneity may help to increase the number of cells in the presence of toxicity (as discussed in the previous paragraphs), the medium depth never decreases with β. A plot of ξm as a function of β is shown in Fig 5 (continuous line). For highly heterogeneous systems (low values of β), ξm is almost insensitive to changes in β. Then there is a sharp slope change after which ξm steadily increases with β. Since the stability of the system depends on the level of heterogeneity, in Fig 5 we show how the values of ξ corresponding to unstable steady states depend on β (discontinuous lines). This way the ξ, β plane is divided into three regions: infeasible, where no steady state is possible, stable and unstable, according to the type of steady state. This confirms the dramatic effect of heterogeneity on the dynamic landscape of the system, as unstable steady states disappear below a certain threshold value of β, i. e. after a critical level of heterogeneity. For large β, the system becomes more robust in the sense that the range of values of ξ defining unstable states is constant, while the stable regime becomes wider. Fig 5 also shows the density of cells in steady state for each pair of values (β, ξ) (color gradient). Notice that in this model, the highest cell densities occur near unstable states. This should be interpreted as a word of caution, since trying to increase the number of cells in the industrial setting can have the undesired effect of washing the culture. Next, we study a reconstruction of the CHO-K1 cell line metabolic network [50]. Fig 6 shows the steady state concentrations of selected metabolites as functions of ξ and for certain representative values of β. There are several differences between the homogeneous (shown in purple in the plots) and heterogeneous regimes. Formate, a byproduct of mitochondrial oxidative metabolism secreted by normal tissues and especially cancer cells [66], stabilizes at a constant concentration for large ξ when β = ∞. The selective pressure for growth predicted by an FBA calculation only favors a mild secretion of this metabolite that is not enough to support its accumulation at larger ξ. In contrast, even mild levels of heterogeneity result in an increasing accumulation of formate at larger values of ξ. Ammonia shows a similar behavior. An heterogeneous population shows an accumulation of formate and ammonia that does not originate from the selective pressures for faster growth acting on individual cells. These differences resemble the behavior of lactate in the simple model. In the case of histidine and valine, we have the opposite situation, where the homogeneous model predicts non-vanishing concentrations at large ξ, but even mild levels of heterogeneity drive the concentrations of these metabolites to zero. Other metabolites show less significant differences, such as glutamate, serine and glycine, where the level of heterogeneity only controls the rate of depletion as ξ increases but does not seem to produce any qualitative differences. Glucose and glutamine are almost insensitive to variations in β. This means that the structure of the metabolic network itself favors maximal consumption of these metabolites, even in absence of active regulation. In Fig 7 (a) and 7 (b) we plot the dilution rate and the cell concentration in steady state as functions of ξ, for representative values of β. The crossing of curves corresponding to different values of β indicates that under certain conditions, heterogeneity may enable larger quantities of cells or faster dilution rates. This is surprising because larger values of β mean that the population of cells concentrate nearer the point of maximum growth rate. The explanation is that in this case larger β also leads to higher secretion of toxic byproducts. For example, in Fig 7 the red curve has a higher β than the black curve, but at ξ = 0. 1 the black curve has higher cell counts. Comparing with Fig 6, we see that the red curve at this point also has higher concentrations of the toxic byproducts ammonia and lactate, explaining the reduction in cell growth. This also confirms a prediction already made in the simpler model. Fig 7 (c) plots the steady state values of the cell density for each dilution rate, with unstable steady states shown in dashed lines. As in the simple metabolic model, we find that the system admits more than one steady state for some dilution rates. The dynamical landscape of the system depends on the value of β. In particular, if β is too low the unstable regime disappears, again recapitulating the behavior found in the simple metabolic model. Since many of the enzymatic flux costs used in these simulations (cf. Eq (3) ) are unknown or poorly annotated, we repeated these simulations after doing random perturbations on all these coefficients of up to 25% relative to the original values. Figures S3 and S4 in the Appendix show that all of the qualitative features discussed here are preserved under these perturbations. In this work we have developed a framework to model cell heterogeneity based on the Maximum Entropy principle (MaxEnt) [28]. In contrast to previous applications of MaxEnt to model metabolic heterogeneity [32], our work has focused on continuous cell cultures in steady state. As we have shown, in this case it is possible to write down a self-consistent set of equations that determines the distribution of cells in phenotypic space as a function of the dilution rate. The dependence has a non-trivial character due to the nonlinear nature of the chemostat, in some cases including multistable regimes. We applied this framework in a simple metabolic network first, where all the computations can be carried out exactly. In this toy model we were able to discuss in detail many qualitative features of the model. We found that heterogeneity enables larger populations of cells because of reduced toxicity, that the dynamical landscape includes includes a multistable regime which shrinks or even disappears entirely as the level of heterogeneity increases, and that a byproduct predicted to be zero if heterogeneity is ignored exhibits increasing concentrations when heterogeneity is included in the model. In this work we have also shown how the Maximum Entropy approach to metabolic modeling can be applied to large metabolic networks, by taking advantage of a recent implementation of the Expectation Propagation algorithm [36]. By exploiting this algorithm we were able to obtain a numerical solution of our model, applied to a CHO-K1 genome-scale metabolic network reconstruction. Although this metabolic model is much more complicated than the toy model, it nonetheless recapitulates all the qualitative findings made in the simpler network, but with a richer phenomenology. In order to obtain a tractable model, we have made some simplifications. We chose to ignore cell-to-cell heterogeneity in mass composition (the parameters ei, yi in the notation of Methods). Indeed, although cells can vary widely in their metabolic arrangements, the mass fractions of major constituents such as total protein, lipids, and nucleotides, are approximately invariant for a given cell type. This approximation is not essential for the formal statement of the model, but it greatly simplifies calculations because then the exponent in the Maximum Entropy distribution is a linear function of metabolic fluxes. We have assumed that the availability of nutrients depended only on the total cell concentration. In a more realistic model cells compete for certain metabolites and their concentrations determines the uptake bounds. In the literature models of this type have been studied [67,68] and they have an interesting phenomenology of their own. Our approach can be seen as a “mean-field” approximation where each cell interacts with the entire population instead of with specific nutrients. We believe that our main conclusions will remain valid if this approximation is relaxed, but further work is needed in this direction. Finally, incorporating the crowding constrain entails replacing reversible reversible reactions by two reactions in the forward and backward direction, almost doubling the total number of reactions. We therefore applied a reduction where reactions found to carry no flux identically for all ξ in the β = ∞ regime were removed from the model. Obviously this modifies the solution space of the model. On the other hand, an extensive model such as the genome-scale CHO-K1 model [50] contains many reactions that are inactive under most conditions and therefore including them would only add noise to our analysis. Since FBA has been a successful tool in the detection of relevant metabolic pathways in multiple organisms [50,51,69–71], we believe that this reduction improves the quality of our predictions. Indeed, all the byproducts predicted by this model have been observed experimentally to be secreted by mammalian cells [13]. Some general qualitative features are common to both the homogeneous regime (β = ∞) and the heterogeneous regime (finite β). In both cases, multistability is a consequence of negative feedback by toxicity accumulation, although extremely heterogeneous populations become monostable as shown in Figs 4,5 and 7. For a given combination of cell-line and feed media, the ratio of cell density to the dilution rate (ξ = X/D, inverse of the cell-specific perfusion rate [16]) determines the steady state of a continuous culture and therefore can be used to connect different modes of continuous cultures (chemostat and perfusion). This conclusion was obtained in [13] in the context of an homogeneous model (β = ∞). In the presence of heterogeneity the steady state does not consist of single flux values for every reaction in the network. Instead, it must be described by a global probability distribution representing the fraction of cells adopting each metabolic phenotype. The parameter β quantifies the spread of this distribution. Although following the standard practice of the MaxEnt principle, β should be determined to match experimental data on the average growth rate of a population, for the sake of generality, in this work β was treated as a free parameter. This is analogous to studying different temperatures in statistical physics. Therefore the selected values of β used in the figures carry no special significance, except that they are representative of the most salient aspects of the general phenomenology of the model. In the Appendix we show how β can be connected to a simple underlying model of the evolutionary dynamics of a cell population. Interestingly, toxicity also leads to the paradoxical result that more heterogeneous populations can achieve larger sizes. This happens because at finite β less cells are secreting the toxic byproduct at maximal rate. Thus, a prediction of our model is that inducing heterogeneity might be beneficial in an industrial setting where cell numbers are limited by the accumulation of toxic byproducts. We also showed that heterogeneity might be responsible for features of the bulk population not derivable from selective pressures on individual cells. The CHO-K1 reconstruction exhibits divergence of formate accumulation, while for the simplified metabolic model it is lactate that accumulates. This difference is not surprising because in the toy model lactate is the only allowed byproduct, while the CHO-K1 model has many possible byproducts. In both cases an homogeneous model would predict zero or constant concentrations. Compared to population balance models of heterogeneity, our approach has several advantages. First, by using MaxEnt, assumptions about the detailed mechanisms behind cellular heterogeneity are not required. As a consequence, the description of heterogeneity is simple and only one additional parameter β suffices for this purpose. This simplicity enables the study of genome scale metabolic networks such as the CHO-K1, which are not accessible to detailed population balance models.

While the advantages of continuous culture in the biotechnological industry have been widely advocated in the literature, its adoption over batch or fed-batch modes stalls due to the complexities of these systems. In particular, continuous cell cultures display hallmark nonlinear phenomena such as multi-stability, hysteresis, and sharp transitions between metabolic phenotypes. Moreover, the impact of the heterogeneity of a cell population on these features is not well understood. We use the maximum entropy principle to model the phenotypic distribution of an heterogeneous population of cells in a chemostat. Given the metabolic network and the dilution rate, we obtain a self-consistent solution for the stationary distribution of metabolic fluxes in cells. We apply the formalism in two examples: a simplified model where the exact solution is tractable, and a genome-scale metabolic network of the Chinese hamster ovary (CHO) cell line widely used in industry. We demonstrate that heterogeneity may be responsible for qualitative changes in the dynamical landscape of the system, like the disappearance of a bistable regime, the increase of concentrations of byproducts that vanish in the homogeneous system and larger number of cells. We explain the causes behind these phenomena.

Abstract Introduction Materials and methods Results Discussion

cell physiology carbohydrate metabolism protein metabolism biological cultures metabolic networks enzymology cell metabolism glucose metabolism metabolites network analysis enzyme metabolism cell cultures thermodynamics enzyme chemistry research and analysis methods computer and information sciences entropy physics biochemistry cell biology biology and life sciences physical sciences metabolism

2019

Maximum entropy and population heterogeneity in continuous cell cultures

Caenorhabditis elegans CEP-1 and its mammalian homolog p53 are critical for responding to diverse stress signals. In this study, we found that cep-1 inactivation suppressed the prolonged lifespan of electron transport chain (ETC) mutants, such as isp-1 and nuo-6, but rescued the shortened lifespan of other ETC mutants, such as mev-1 and gas-1. We compared the CEP-1-regulated transcriptional profiles of the long-lived isp-1 and the short-lived mev-1 mutants and, to our surprise, found that CEP-1 regulated largely similar sets of target genes in the two mutants despite exerting opposing effects on their longevity. Further analyses identified a small subset of CEP-1-regulated genes that displayed distinct expression changes between the isp-1 and mev-1 mutants. Interestingly, this small group of differentially regulated genes are enriched for the “aging” Gene Ontology term, consistent with the hypothesis that they might be particularly important for mediating the distinct longevity effects of CEP-1 in isp-1 and mev-1 mutants. We further focused on one of these differentially regulated genes, ftn-1, which encodes ferritin in C. elegans, and demonstrated that it specifically contributed to the extended lifespan of isp-1 mutant worms but did not affect the mev-1 mutant lifespan. We propose that CEP-1 responds to different mitochondrial ETC stress by mounting distinct compensatory responses accordingly to modulate animal physiology and longevity. Our findings provide insights into how mammalian p53 might respond to distinct mitochondrial stressors to influence cellular and organismal responses. Mitochondria are major sites of numerous metabolic processes, in particular electron transport and ATP production, and are essential for life. Not surprisingly, mitochondria also play central roles in aging and disease [1]. In model organisms such as worms, flies, and mice, specific point mutations or RNAi knockdowns directly affecting the electron transport chain (ETC) result in varying effects on development and longevity, ranging from developmental arrest and shortened survival to extended lifespan. The extended lifespan associated with moderate mitochondrial ETC dysfunction was surprising and further highlights the complex relationship between mitochondrial function and aging. An emerging model posits that a moderate reduction in mitochondrial ETC function can lead to compensatory responses that lengthen lifespan [2]–[5], whereas a more severe reduction in mitochondrial ETC function, beyond an innate threshold, will lead to developmental arrest and/or early death [6]–[7]. How different degrees of mitochondrial dysfunction result in opposing effects on longevity remains largely unknown. Caenorhabditis elegans represents a powerful model to study the genetic basis of cellular and organismal changes in response to mitochondrial dysfunction. Previous findings in C. elegans have revealed a number of long-lived and short-lived ETC mutants. The nuo-6 (qm200) mutant, which harbors a point mutation in the NADH-ubiquinone oxidoreductase of complex I, the isp-1 (qm150) mutant, which harbors a point mutation in the rieske iron sulphur subunit of complex III, and the clk-1 (e2519) mutant, with a point mutation in a coenzyme Q biosynthesis enzyme, exhibit substantial lifespan extension [8]–[9]. In contrast, the mev-1 (kn-1) mutant, with a point mutation in the succinate dehydrogenase subunit c of complex II, and the gas-1 (fc21) mutant, with a point mutation in the NADH: ubiquinone oxidoreductase NDUFS2 subunit of complex I, live significantly shorter than wild-type worms [10]. Furthermore, large-scale RNAi screens have revealed that RNAi-mediated inactivation of many of the ETC subunits result in prolonged or shortened lifespan [11]. Studies using genetic mutants and RNAi-mediated knockdown of ETC components in worms have begun to reveal the mechanistic basis of the longevity outcomes associated with mitochondrial dysfunction. Reactive oxygen species (ROS) have emerged as an important signaling intermediate in the ETC mutants. Specifically, nuo-6, isp-1, and mev-1 mutants have been shown to exhibit elevated levels of mitochondrial superoxide, and antioxidant treatment of these worms was able to revert their longevity phenotype [12]–[13]. Interestingly, the long-lived clk-1 mutant was not found to exhibit a higher level of mitochondrial superoxide, and antioxidant treatment had no impact on its lifespan, suggesting that the clk-1 mutation influences lifespan independent of ROS. In addition to increased ROS levels, the ETC mutants also exhibit an altered metabolism. The long-lived nuo-6, isp-1, and clk-1 mutants share similar metabolic profiles that are distinct from that of the mev-1 short-lived mutant [14]–[15]. An elevated production of metabolites, such as α-ketoacids and α-hydroxyacids, has been proposed to act as a pro-longevity signal in the long-lived ETC mutants. Furthermore, studies that have largely employed RNAi-mediated knockdown of various ETC subunits demonstrated that an imbalanced stoichiometry of the ETC protein subunits triggered a strong mitochondrial unfolded protein response (mtUPR). In this scenario, processed peptides in the mitochondria are thought to serve as the signal that activates several transcriptional regulators, including UBL-5 and ATFS-1, to induce transcriptional responses necessary to restore proteostasis in the mitochondria, which contributes to longevity determination [16]. Lastly, several RNAi and candidate screens have identified additional transcription factors, such as CEP-1 [17], CEH-23 [18], and TAF-4 [19] that mediate the lifespan of various ETC mutants. The transcription factor p53 has recently emerged as a key regulator of metabolic balance [20]–[22]. Despite its importance, how p53 senses metabolic stress and accordingly regulates molecular changes that determine the physiological outcomes of an organism remains poorly understood. C. elegans CEP-1, the sole homolog of the mammalian p53 family [23] (p53, p63 and p73), is known to mediate the lifespan changes in worms with mitochondrial dysfunction. Inactivation of cep-1 has been shown to partially suppress the extended longevity of isp-1 mutant worms [24]. Furthermore, using different concentrations of RNAi to cause different degrees of knockdown of several ETC components demonstrated that CEP-1 is required for the increased longevity under mild mitochondrial disruption as well as the shortened lifespan when mitochondrial damage is more severe [25]. Therefore, CEP-1 exerts opposite effects on lifespan that likely depend on the levels of mitochondrial stress experienced. The underlying mechanism governing this intriguing duality of CEP-1 function is not known. In this study, we sought to further characterize the role of CEP-1 in the longevity of several mitochondrial ETC mutants. Our results indicate that CEP-1 is a critical mediator of the lifespan of several mitochondrial mutants, suggesting that CEP-1 plays a central role in sensing mitochondrial distress and coordinating physiological outcomes accordingly. We also evaluated the CEP-1-regulated transcriptomes in the long-lived isp-1 and the short-lived mev-1 mitochondrial ETC mutants. Despite the opposing roles that CEP-1 appears to play in determining the lifespan of these mutants, the CEP-1-regulated transcriptional profiles were largely similar in these mutants. Nevertheless, the expression of a small group of genes was differentially regulated by CEP-1 between the long-lived isp-1 and the short-lived mev-1 mutants. Interestingly, this small group of genes is enriched for the Gene Ontology functional group “aging”, indicating they are over-represented by genes previously known to have a role in aging in worms. We functionally validated one of these differentially regulated genes, ftn-1, which encodes ferritin in C. elegans, by demonstrating that its RNAi-mediated depletion significantly impacted the lifespan of the isp-1 but not mev-1 mutant. This result supports our hypothesis that CEP-1 can differentially regulate a small subset of target genes to achieve distinct longevity outcomes in response to mutations in different ETC components. Previous results suggested that cep-1 is required for the lifespan extension associated with mild mitochondrial dysfunction and the shortened lifespan associated with severe mitochondrial dysfunction [25]. We confirmed these results and demonstrated that inactivation of cep-1 partially but consistently suppressed the extended lifespan of isp-1 mutant animals (Fig. 1A), (Fig. S1A, Table S1, S2). We further showed that inactivation of cep-1 largely restored the lifespan of mev-1 mutants to wild-type (Fig. 1B), (Fig. S1A, Table S1, S2). Our data are therefore consistent with previous findings suggesting that CEP-1 can respond to different degrees of mitochondrial dysfunction and modulate lifespan accordingly. In addition to lifespan changes, mitochondrial ETC mutants also develop slowly and display reduced brood sizes. To assess whether CEP-1 participates in the development of mitochondrial ETC mutants, the development time of isp-1 and mev-1 mutants with or without cep-1 was compared to that of wild-type (wt) worms. Synchronized embryos of the various strains were allowed to develop at 20°C for 60 hours, and the number of adults and larvae were counted (Fig. 1C, Table S3). The data showed that wt and cep-1 mutant worms developed at similar rates, and 100% of the populations had reached adulthood by 60 hr. As expected, the isp-1 mutant worms grew slowly, and the majority were in the L1 (37%) and L2 (62%) stages after 60 hr. However, cep-1; isp-1 double mutants developed noticeably faster, and the majority were L3s (40%), L4s (40%), and adults (12%) after 60 hr. These data suggest that cep-1 inactivation partially recues the slow development of the isp-1 mutant. Interestingly, cep-1 inactivation exerted an opposite effect on mev-1 mutant development. The mev-1 mutants were slightly developmentally delayed, where the majority of mev-1 mutant worms were in the L3 (6%) and L4 (93%) stages at 60 hr. The development rate of the cep-1; mev-1 double mutant worms was heterogeneous and further delayed (L2 (11%), L3 (19%) and L4 (69%) at 60 hr) compared to mev-1 single mutant worms. To examine a possible role for CEP-1 in the reproduction of mitochondrial ETC mutants, the average brood size of isp-1 and mev-1 mutants with or without cep-1 was compared to that of wt animals. The brood size of the cep-1 mutant did not significantly differ from wt (p = 0. 1), and, as expected, the mitochondrial mutants (isp-1, mev-1) displayed significantly lower brood sizes compared to wt (p<0. 0001). The double mutants cep-1; isp-1 and cep-1; mev-1 exhibited a further brood size reduction compared to isp-1 and mev-1 single mutants, respectively (p≤0. 05) (Fig. 1D,). These results suggest that cep-1 inactivation further exacerbates the reproductive defect associated with mitochondrial dysfunction. Taken together, cep-1 appears to participate in multiple physiological outcomes of ETC mutants. cep-1 activity promotes, at least partially, both the slower development and the longer lifespan of the isp-1 mutant but is required to prevent further reproductive deterioration. In contrast, cep-1 activity promotes a shortened mev-1 mutant lifespan but is required for a relatively normal developmental rate. Similar to its role in isp-1 reproduction, cep-1 prevents reproductive decline in mev-1 mutants. Since cep-1 loss similarly impacts the reproductive success of isp-1 and mev-1 mutants, but its inactivation has opposing effects on the lifespan and developmental rate of these mutants, the role of cep-1 in reproduction might be independent of its role in lifespan and development in ETC mutants. CEP-1 is a well-established key regulator of stress-induced apoptosis [24]. Since CEP-1 is a crucial mediator of the longevity outcomes of isp-1 and mev-1 mutants, we asked whether CEP-1 does so by modulating apoptosis in these mutants. While apoptosis occurs throughout embryonic and larval development in C. elegans, we reasoned that monitoring apoptosis in adults would be more relevant in investigating adult lifespan. In C. elegans adults, physiological and stress-induced apoptosis occurs in the germline. We monitored physiological apoptosis in the germline of wild-type, isp-1 and mev-1 mutants with or without cep-1 inactivation. CEP-1 is best characterized for its role in stress-induced apoptosis, and our data indicated that loss of cep-1 alone only mildly affected physiological apoptosis in the germline. The short-lived mev-1 mutant and mev-1; cep-1 double mutant exhibited wt levels of physiological germline apoptosis (Fig. 2). Interestingly, the long-lived isp-1 mutant displayed a significantly lower level of physiological germline apoptosis, which was completely rescued in the cep-1; isp-1 double mutant (Fig. 2). These data suggest that CEP-1 protects against physiological germline apoptosis in the isp-1 mutant. As CEP-1 has largely been demonstrated to promote apoptosis in C. elegans, this new role of CEP-1 in protecting against apoptosis merits further investigation. CEP-1 appears to dually mediate the lifespan of mitochondrial mutants. We hypothesized that moderate mitochondrial dysfunction initiates a CEP-1-dependent defensive response that promotes longevity in long-lived mitochondrial mutants. Conversely, decreasing mitochondrial function beyond a threshold in short-lived mitochondrial mutants likely engages CEP-1 in a different way that results in a shortened lifespan. Since mammalian p53 is a well-established transcription factor, we sought to investigate the transcriptional response induced by CEP-1 in both a long-lived, isp-1, and a short-lived, mev-1, mitochondrial mutant. We hypothesized that the genes differentially regulated by CEP-1 between isp-1 and mev-1 mutants may mediate the distinct CEP-1 lifespans of the isp-1 and mev-1 mutants (see below). We compared the transcriptional profiles of synchronized isp-1 (qm150) and cep-1 (gk138); isp-1 (qm150) young adults, as well as mev-1 (kn1) and cep-1 (gk138); mev-1 (kn1), using the Agilent 4×44K oligonucleotide microarray (Table S4). Although CEP-1 exerts opposing effects on the lifespans of isp-1 and mev-1 mutants, hierarchical gene cluster analysis revealed that the CEP-1-regulated transcriptional profiles in long-lived isp-1 and short-lived mev-1 mutants were largely similar (correlation coefficient of 0. 58) (Fig. 3A). Only a small number of genes exhibited cep-1-dependent differential regulation. Data analysis using the statistical tool SAM (Significance Analysis of Microarray) with a FDR (false discovery rate) of 0. 5% revealed that CEP-1 regulated the expression of 3,404 genes (Table S5) in a similar manner in the long-lived isp-1 (qm150) and short-lived mev-1 (kn1) mutants (Fig. 3B, Fig. S2A). SAM analysis with a FDR of 1% followed by a gene list comparison also revealed 71 genes (Table S6, see Materials & Methods for details on gene list filtering) that were differentially regulated by CEP-1 in the isp-1 (qm150) and mev-1 (kn1) mutants (Fig. 3B, Fig. S2B). The expression of these genes largely differed quantitatively rather than qualitatively, i. e. , they generally showed a greater CEP-1-mediated regulation in the isp-1 or mev-1 mutant backgrounds. To validate our microarray findings, we selected six genes (F15E6. 8/dct-7, C37A5. 2/fipr-22, F28C6. 1/AP-2 like TF, C52D10. 6/skr-12, C08A9. 1/sod-3 and F57B9. 9/abu-13) based on their classification by SAM analysis and performed quantitative reverse transcription PCR (qRT-PCR) using an independent biological sample. The expression changes of all six genes echoed the microarray results (Fig. S3). To examine the biological processes of the CEP-1-regulated genes identified by SAM, we performed Gene Ontology (GO) analysis using DAVID (Database for Annotation, Visualization, and Integrated Discovery). We focused on the GO term categories that were most significantly enriched in our dataset compared to the distribution in the C. elegans genome. Interestingly, the small group of CEP-1-regulated genes that differentially changed in isp-1 and mev-1 mutants were enriched for genes with known roles in aging and metabolism (Fig. 3C, Table S7C). This finding supports our hypothesis that the genes differentially regulated by CEP-1 in isp-1 and mev-1 mutants are likely particularly important for the opposing effects on lifespan that CEP-1 exerts in these mutants. The genes that were similarly regulated by CEP-1 in isp-1 and mev-1 mutants likely account for the effects CEP-1 has on development, reproduction, and other physiological changes in these mutants and might also contribute to the final longevity (Fig. 3C, Tables S7A, S7B). The microarray results helped narrow our analysis to a small number of CEP-1-regulated genes (Fig. 3C) that might contribute to the dual effects CEP-1 exerts on the longevity of mitochondrial mutants with varying degrees of dysfunction. One of these genes, ftn-1, encodes the C. elegans homolog of the ferritin heavy chain. Using qRT-PCR, we observed that ftn-1 expression was repressed ∼2 fold in the cep-1 mutant and induced ∼1. 5 fold in the isp-1 mutant compared to wt. The cep-1; isp-1 double mutant displayed a similar repressed level of ftn-1 expression as observed for the cep-1 single mutant (Fig. S4A). While ftn-1 expression was induced in the mev-1 mutant, similar to the isp-1 mutant, this induction was not changed in the cep-1; mev-1 double mutant (), consistent with the microarray results. Inspection of the ftn-1 upstream sequence did not reveal a known p53 binding motif, suggesting that CEP-1 might not directly regulate ftn-1 transcription. C. elegans harbors another ftn-1 homolog, ftn-2, but its expression was not changed in the ETC mutants compared to wt (Fig. S4B). Lastly, we used the Pftn-1: : gfp strain, where a GFP reporter is fused to the ftn-1 promoter [26], to assess whether the microarray and qRT-PCR results translated into observably meaningful ftn-1 expression changes. Pftn-1: : gfp was substantially induced in isp-1 mutants and was completely suppressed in cep-1; isp-1 double mutants (Fig. 4A), consistent with our microarray and qRT-PCR results. Additionally, Pftn-1: : gfp expression was slightly induced in mev-1 mutants and unchanged in cep-1; mev-1 double mutants. Taken together, our data indicate that CEP-1 is important for ferritin regulation in wt and isp-1 mutants but is dispensable in the mev-1 mutant. Ferritin regulates the storage and release of iron. As iron homeostasis is known to play an important role during mitochondrial dysfunction, we wanted to investigate whether ferritin regulation was responsible for the dual roles of CEP-1 in mitochondrial ETC mutant longevity. To assess whether ferritin upregulation promoted isp-1 mutant longevity, we examined the lifespan of isp-1 and cep-1; isp-1 mutants after ftn-1 and ftn-2 knockdown by RNAi. We knocked down ftn-1 and ftn-2 simultaneously to prevent possible functional redundancy, which may preclude observable phenotypes when either gene is knocked down alone. Double RNAi-mediated knockdown of ftn-1 and ftn-2 significantly suppressed the extended lifespan of isp-1 mutants, although not to the same degree as cep-1 inactivation. Importantly, ftn-1/2 RNAi did not further suppress the lifespan of the cep-1; isp-1 double mutant (Fig. 4B, Fig. S5), suggesting that cep-1 and ftn-1/2 act in the same genetic pathway to mediate isp-1 longevity, corroborating our model that ftn-1/2 are downstream targets of CEP-1. Consistent with our expression results suggesting that ftn-1/2 are not regulated by CEP-1 in the short-lived mev-1 mutant, the mev-1 mutant lifespan was not affected by ftn-1/2 RNAi (Fig. 4C, Fig. S5). Given the important role CEP-1 plays in determining the lifespan of isp-1 and mev-1 mutants, we next tested whether cep-1 is generally required for the longevity of additional ETC mutants that have been well characterized in C. elegans. We explored the long-lived nuo-6 (qm200) mutant, which harbors a point mutation in the NADH-ubiquinone oxidoreductase of complex I, the long-lived clk-1 (e2519) mutant, which harbors a point mutation in a coenzyme Q biosynthesis enzyme, and the short-lived gas-1 (fc21) mutant, which has a point mutation in the NADH: ubiquinone oxidoreductase NDUFS2 subunit of complex I. The cep-1 null mutation largely suppressed the long-lived phenotype of the nuo-6 mutant but did not affect the longevity of the clk-1 mutant, as the cep-1; clk-1 double mutant lived as long as the clk-1 single mutant (Fig. 5A, 5B, Fig. S1A). Furthermore, cep-1 deletion rescued the short-lived phenotype of the gas-1 mutant, similar to its effect on the mev-1 mutant lifespan (Fig. 5C, Fig. S1A). Taken together, the data suggest that CEP-1 crucially participates in the longevity outcome of multiple ETC mutants. The fact that cep-1 abrogation did not affect the clk-1 mutant lifespan was intriguing. All of the well-characterized mitochondrial ETC mutants in C. elegans (isp-1, nuo-6, mev-1, and gas-1), with the exception of clk-1, have been shown to harbor elevated levels of mitochondrial superoxide. In fact, antioxidant treatment of all of these mutants, again with the exception of clk-1, has been shown to suppress (the long-lived) isp-1 and nuo-6 or rescue (the short-lived) mev-1 and gas-1 mutant lifespans. Therefore, the cep-1-mediated lifespans of these mitochondrial ETC mutants parallel the proposed roles that elevated ROS is exerting in these mutants, i. e. , CEP-1 mediates the lifespans of mutants with elevated ROS but not of mutants without. This observation suggests the interesting possibility that CEP-1 might somehow be linked to the ROS-mediated longevity increases observed in the long-lived mitochondrial ETC mutants. Future studies aimed at thoroughly investigating the relationship between CEP-1 and ROS-mediated longevity will likely yield fruitful insights. As cep-1 impacted the development rates of isp-1 and mev-1 mutants, we tested whether cep-1 similarly affected the development of nuo-6 and gas-1 mutants. Our data indicated that cep-1 deletion had little impact on nuo-6 and gas-1 mutant development (Fig. 5D), unlike what we found for isp-1 and mev-1 mutants (Fig. 1C). Therefore, whereas cep-1 mediates the lifespans of all four ETC mutants tested here (isp-1, nuo-6, mev-1, and gas-1), its involvement in the development of each of the mutants is mixed, suggesting that the role of CEP-1 in development may be distinct from its role in longevity in these various mutants. Given that we observed an important role for CEP-1 in ftn-1 regulation and iron homeostasis in ETC mutant longevity, we further tested whether ftn-1 & ftn-2 are regulated by CEP-1 in nuo-6 and gas-1 mutants and whether ftn-1/2 are required for their longevity. Strikingly, ftn-1 was greatly induced in the nuo-6 mutant, even more so than in the isp-1 mutant (Fig. 6C). Somewhat surprisingly, cep-1 was not required for this induction, unlike in the isp-1 mutant. We also demonstrated that RNAi-mediated depletion of ftn-1/2 substantially suppressed the extended lifespan of the nuo-6 mutant, similar to their depletion in the isp-1 mutant (Fig. 6A). Although CEP-1 is essential for ftn-1 induction in the isp-1 mutant, it is dispensable in the nuo-6 mutant; however, both cep-1 and ftn-1/2 are important for the longevity of nuo-6 and isp-1 mutants. Therefore, while iron homeostasis and cep-1 are both important for longevity determination in the long-lived isp-1 and nuo-6 mutants, CEP-1 regulates ftn-1 only in the isp-1 mutant. We hypothesize that another transcription factor likely regulates ftn-1 expression in the nuo-6 mutant. Interestingly, ftn-1 was induced in the short-lived gas-1 mutant independently of cep-1, similar to that in the short-lived mev-1 mutant (Fig. 6C). Additionally, depleting ftn-1/2 did not rescue the shortened lifespan of the gas-1 mutant, similar to that observed in the mev-1 mutant (Fig. 6B). Therefore, iron homeostasis might be a determinant of isp-1 and nuo-6 longevity, but it does not appear to be important for mediating the lifespans of gas-1 and mev-1 mutants. Lastly, ftn-1 was not induced in the clk-1 mutant, another indication that clk-1 mutant might engage a different mechanism to extend lifespan compared to the other mitochondrial ETC mutants studied here. The distinct ftn-1 results led us to consider more broadly whether isp-1 & nuo-6 and gas-1 & mev-1 share some common molecular signatures, which could account for their similar longevity (extended or shortened, respectively). We examined the expression of a handful of CEP-1-regulated genes that we identified earlier in isp-1/mev-1 mutants. Using qRT-PCR, we first analyzed the expression of nine genes that were differentially regulated by CEP-1 in isp-1 and mev-1 mutants (Fig. 7). Overall, we observed very different patterns for how these genes responded to the absence of cep-1 in the two long-lived mutants (isp-1 and nuo-6). The majority of genes that showed substantial cep-1-dependent induction in isp-1 mutants barely changed when cep-1 was depleted in nuo-6 mutants; only two genes (cyp-35A5, F09C8. 1) showed consistent, but moderate, cep-1-dependent expression changes between isp-1 and nuo-6. Interestingly, the response of these target genes to cep-1 depletion was more similar between the two short-lived mutants (mev-1 and gas-1). However, it is worth noting that the expression of the majority of these genes barely changed in mev-1 and gas-1 mutants in response to cep-1 deletion (Fig. 7). We also assayed the expression of genes that were similarly regulated by CEP-1 in isp-1 and mev-1 mutants and observed a similarly discordant pattern (Fig. 8). The regulation of these genes by CEP-1 was strikingly different between isp-1 and nuo-6 mutants but was much more similar between mev-1 and gas-1 mutants. In the absence of further genome-wide analysis, it is difficult to estimate the extent of common targets that are shared between (the long-lived) isp-1 and nuo-6 and between (the short-lived) mev-1 and gas-1 mutants. However, based on the small number of genes tested here using qRT-PCR, it appears that CEP-1 regulates some common genes between mev-1 and gas-1 mutants, which might account for its ability to restore a normal lifespan in these short-lived mutants. For the long-lived mutants, the situation is more complex; CEP-1 appears to regulate largely distinct genes between isp-1 and nuo-6 even though it is required for the extended lifespan of both mutants. The best-characterized role of mammalian p53 is its ability to respond to DNA damage by inducing cell cycle arrest and repair or apoptosis. We wondered whether the CEP-1-regulated transcriptome changes in mitochondrial mutants would resemble changes observed in response to DNA damage. The global transcriptional profiles of wt or cep-1 mutant worms treated with UV, gamma, or x-ray irradiation have previously been published [27]–[28]. Using clustering analysis, we compared the CEP-1-dependent global expression profiles in response to mitochondrial dysfunction (in isp-1 or mev-1 mutants) to profiles 4 hr after exposure to UV irradiation, 6 hr after gamma-ray, or 2 hr after X-ray (Fig. S6A, Table S8). We observed a higher similarity between the CEP-1-regulated gene expression profiles in response to mitochondrial dysfunction and after UV treatment (correlation coefficient of 0. 36) but less after gamma-ray treatment (correlation coefficient of 0. 20). The published gene expression profiles after 2 hr of X-ray treatment did not display significant differences from the no treatment control and thus were not included in the downstream analysis. We analyzed the CEP-1-regulated genes in response to mitochondrial dysfunction and UV and gamma irradiation. We focused on the gene sets that exhibited ≥1. 4-fold changes under at least one of the stressed conditions (i. e. , isp-1 mutant, mev-1 mutant, UV treatment, or gamma irradiation) and performed K-mean clustering to identify the gene sets that were either similarly regulated across all of the stressed conditions or differentially regulated under one or more conditions. The K-mean analysis revealed six distinct clusters (Fig. S6B, Table S9). We employed DAVID to examine the GO terms generated from genes representing each cluster (Fig. S6C, Table S10). Cluster-‘a’ (579 genes) represents genes that were upregulated by CEP-1 in response to all of the stressors. This group is particularly enriched for genes that function in phosphate metabolism, including kinases and phosphatases. Cluster-‘f’ (1,357 genes) represents genes that are also upregulated across all of the stressors but to a lesser extent than in cluster-‘a’. Cluster-‘f’ is enriched for ribosomal and ion transport proteins. Cluster-‘e’ (901 genes) is enriched for growth regulation and represents genes that are repressed by CEP-1 in response to mitochondrial dysfunction and to UV and gamma irradiation. Clusters-‘a’, ‘e’ and ‘f’ together suggest that CEP-1 regulates common transcriptional programs in response to stress, which might in turn induce key signaling pathways to counter the stress and simultaneous suppression of growth. Clusters-‘b’, ‘c’, ‘d’ represent genes that are similarly regulated by CEP-1 in response to mitochondrial dysfunction and UV but are different after gamma irradiation. These gene groups are significantly enriched for nematode cuticle and collagen proteins, cell cycle regulators, glycoproteins and signaling proteins. In summary, both mitochondrial dysfunction and UV irradiation are known to induce ROS, which might reflect the common CEP-1-regulated transcriptional changes observed here. The major function of p53 is to integrate stress signals and to orchestrate appropriate cellular responses. Under normal, unstressed conditions, p53 is maintained at low levels via proteosome-mediated degradation. Upon stress, p53 levels stabilize and activate stress response programs that range from cell repair to cell death [29]. Our genetic data indicate that mutations in the C. elegans ETC subunits isp-1 and nuo-6 engage CEP-1 in initiating a stress response that results in a longer lifespan. Conversely, mitochondrial defects, due to mutations in the ETC subunits mev-1 and gas-1, act through CEP-1 to confer a shorter lifespan. Taken together, our data suggest the intriguing possibility that CEP-1 can sense distinct dysfunctional mitochondrial processes and modulate overall longevity accordingly. How CEP-1 is able to sense mitochondrial dysfunction caused by different ETC mutations remains unclear. ROS have been proposed to be important regulators of p53 [30]. ROS can induce DNA damage, which leads to p53 activation via DNA damage checkpoint pathways. ROS are also known to engage p53 directly through modifying the redox-sensitive Cystein (Cys) residues on p53 [31]. p53 contains several critical Cys residues located within the DNA-binding domain. Importantly, the Cys residues that are required to coordinate zinc and maintain the protein structure that enables interaction with the minor groove of target DNA are conserved between CEP-1 and human p53 [32]. As discussed earlier, many C. elegans ETC mutants have been shown to produce elevated levels of mitochondrial ROS. Altered ROS production in the mitochondrial ETC mutants might be coupled to CEP-1 via regulation of the conserved Cys residues. Upon activation, p53 is well-known to regulate the transcription of genes involved in ROS metabolism, including both antioxidant and pro-oxidant genes [33]. Therefore, depending on upstream signaling, CEP-1/p53 can either alleviate ROS stress or promote further ROS accumulation. Deficiencies in the ETC can also affect cellular energy homeostasis. The isp-1 mutant has been shown to exhibit higher AMP: ATP ratios compared to wt worms, and aak-2, the AMPK α subunit of C. elegans, is partially necessary for the extended lifespan of isp-1 mutants [34]. In mammals, the cellular energy sensor AMP kinase (AMPK) is known to directly phosphorylate p53 at Ser15, leading to stabilization and transcriptional activation of p53 [35]. Intriguingly, this Serine residue is conserved between CEP-1 and human p53, and thus, it is possible that an altered AMP: ATP ratio in ETC mutants engages AMPK to regulate CEP-1 function. Further experiments are necessary to definitively identify the signals generated by C. elegans ETC mutants that lead to CEP-1 activation. Although CEP-1 plays opposite roles in isp-1 and mev-1 longevity, CEP-1-regulated genes in these two mutants are strikingly similar. This observation suggests that CEP-1 induces similar compensatory responses to restore cellular homeostasis regardless of the specific mitochondrial ETC defect. Our GO analyses suggested, upon mitochondrial dysfunction, that CEP-1 likely promotes a kinases/phosphatase- and/or neuropeptide-mediated signaling cascade, activates metabolic processes poised for defense and detoxification and represses the energy demanding cell cycle program. Although mtUPR has emerged as a key pathway that mediates the physiological outcomes of ETC mutants, we did not observe changes in major mtUPR response genes, such as atfs-1, hsp-6, or hsp-60. Our data suggest that CEP-1 likely does not act through mtUPR to affect the lifespan of ETC mutants. We were excited to identify a small number of genes that were differentially regulated by CEP-1 in the isp-1 and mev-1 mutants, and this gene set was over-represented for genes previously known to modulate aging. We demonstrated that one of these genes, ftn-1, was indeed functionally important for the cep-1-mediated alterations to longevity in the isp-1 mutant but not in the mev-1 mutant, further underscoring CEP-1' s functional duality. Ferritin functions to store iron in a non-toxic form, to deposit it in a safe form, and to transport it to areas where it is required. Mitochondria are the sites of iron-sulfur cluster synthesis, which are critical catalytic and structural components of many cellular proteins [36]. Conversely, the presence of iron in mitochondria must be tightly regulated, as free iron can react with ROS and further produce hydroxyl radicals through the Fenton reaction. As isp-1 encodes an iron-sulphur cluster protein and can bind to iron, iron levels may accumulate in the isp-1 mutant and thus upregulate ftn-1 as a compensatory response to restore iron homeostasis. However, this hypothesis requires a thorough investigation of iron homeostasis in isp-1 and the other ETC mutants to be validated. Additionally, further functional analysis of the small group of genes differentially regulated by CEP-1 in isp-1 and mev-1 mutants will likely reveal new genes important for longevity and illuminate how CEP-1 modulates the lifespans of animals with different ETC mutations. Whereas CEP-1 is essential in mediating the extended lifespan of both nuo-6 and isp-1 mutants, our results, albeit based on only a handful of target genes, suggest that CEP-1 regulates distinct genes in each case. This is quite surprising, especially considering that CEP-1 appears to regulate largely similar genes in gas-1 and mev-1 mutants. These differences might be due to the specific complex that is compromised and the precise point of electron transport that is defective in the various ETC mutants. During normal electron transport, electrons from complex I or complex II can be passed onto complex III. In the nuo-6 mutant, where complex I is compromised, fuels can still enter the ETC via complex II, allowing for some degree of electron transport. The situation might be quite different in the isp-1 mutant, which impairs complex III and would be expected to block electrons transferred from either complex I or complex II. In contrast, a defect in complex I, such as in gas-1, or in complex II, such as in mev-1, could have a similar consequence on electron transport if fuels are not limiting, as either likely partially compromises electron flow to complex III. Further experiments are necessary to elucidate whether and how different ETC mutations influence CEP-1 activity and gene regulation. The best known function of CEP-1 in C. elegans is its ability to induce apoptosis upon stress. We, however, did not observe expression changes in egl-1, the key CEP-1 target for initiating apoptosis [37]. Intriguingly, our results suggest that CEP-1 may protect against physiological apoptosis specifically in the long-lived isp-1 mutant. This is an aspect of CEP-1 function that remains poorly characterized. Interestingly, analysis of our microarray results revealed that ced-8, ced-9, egl-38, genes known to regulate apoptosis [37], were specifically regulated in the isp-1 mutant in a cep-1-dependent manner, but their expression was not changed in the mev-1 mutant. Whether these genes might be central to the ability of CEP-1 to confer protection from physiological apoptosis will warrant further investigation. Future experiments to delineate whether repressed physiological apoptosis is required for the isp-1 lifespan extension and which CEP-1 target genes might contribute to protection from physiological apoptosis will likely provide important new insights into the function and physiological role of CEP-1/p53. DNA damage is a major p53-activating stressor, so we compared the CEP-1-mediated transcriptional profiles in mitochondrial mutants and worms treated with UV or gamma irradiation. Our results demonstrated a considerably greater overlap between the CEP-1-regulated transcriptional response induced by ETC disruption and UV irradiation than by gamma irradiation. This might seem surprising given that UV irradiation and gamma irradiation are both genotoxic stressors that cause DNA damage, whereas mitochondrial ETC dysfunction might be considered a metabolic stress. However, UV irradiation is known to produce ROS, which might induce CEP-1 activation in a manner similar to mitochondrial ETC dysfunction, which is also known to induce ROS. Interestingly, the common set of genes that are upregulated by CEP-1 in mitochondrial mutants and in worms exposed to UV and gamma irradiation are highly enriched for kinases and phosphatases but not for DNA damage response genes. Therefore, it is likely that in response to mitochondrial dysfunction and different genotoxic stresses, CEP-1 can induce a core group of signal transduction molecules that initiate a downstream signaling cascade to mount a general stress response. In addition, CEP-1 appears to be able to sense specific damage and induce distinct responses associated with gamma irradiation and UV irradiation or mitochondrial ETC inhibition. Future dissection of the core and damage-specific responses of CEP-1/p53 will enhance our knowledge of the mechanisms that govern the function and regulation of p53, arguably one of the most important and ubiquitous tumor suppressors. All strain stocks were kept at 16°C and grown under standard growth conditions. The following strains were used: Wild-type N2, isp-1 (qm150), nuo-6 (qm200), mev-1 (kn-1), gas-1 (fc21), clk-1 (e2519), cep-1 (gk138), and Pftn-1: : gfp (GA641). Standard genetic methods were utilized to construct the following strains: cep-1 (gk138); isp-1 (qm150), nuo-6 (qm200) cep-1 (gk138), cep-1 (gk138); clk-1 (e2519), cep-1 (gk138); mev-1 (kn1), cep-1 (gk138); gas-1 (fc21), Pftn-1: : gfp; isp-1 (qm150), and Pftn-1: : gfp; mev-1 (kn-1). All lifespan assays were performed at 20°C on Nematode Growth Media (NGM) plates seeded with E. coli OP50 or RNAi bacteria. A detailed experimental procedure is described in the Supplementary Materials and Methods (Text S1). The survival function of each worm population was estimated using the Kaplan-Meier method, and statistical analysis was performed using a log-rank test (SPSS software). P≤0. 001 was considered as significantly different from the control population. The independent trials were analyzed separately, and representative experiments are shown in the figures. All of the data from all the trials are shown in Supplementary Table S1. Percent mean lifespan differences from controls were plotted from multiple experiments in Fig. S1, and the data are shown in Supplementary Table S2. Worms of each strain were synchronized by picking, and the numbers of apoptotic corpses were counted 48 hours post L4. The corpses were assessed using Differential Interference Contrast (DIC) microscopy under 63× magnification as described in Lant and Derry (2013) [38]. For each strain, at least 3 independent experimental replicates were performed with n≥15, where n = number of gonad arms, for each replicate. Total RNA was purified from synchronized young adult worms grown at 20°C on OP50 bacteria. Total RNA was isolated using Tri-reagent (Molecular Research Center, Inc.) and purified with the RNeasy kit (Qiagen). cRNA synthesis/amplification, Cy3/Cy5 dye labeling, and hybridization onto Agilent 4×44K C. elegans oligonucleotide microarrays were performed as previously described [39]. One of three replicate arrays was dye-flipped. The normalized expression data were uploaded onto the Princeton University MicroArray database (PUMA [http: //puma. princeton. edu]). The raw data were retrieved by SUID (Sequence Unique IDentifier) then averaged by SEQ_NAME with any remaining SUIDs removed. Log2-transformed fold-change data were acquired after setting spot filter criteria, where genes with >80% good data were used. The data were analyzed and visualized using Cluster 3 and TreeView [40]–[41]. The log2 ratios of wt vs. cep-1 (gk138) with or without UV treatment were obtained from Derry et al. (2007). The log2 ratios of wt vs. cep-1 after gamma and X-ray irradiation data were obtained from Greiss et al. (2008). For the UV, gamma, and X-ray datasets, we averaged the intensity values of all three wt arrays for each treatment and used it as a reference. Then, we compared the results for each cep-1 array for the same treatment to the reference to obtain the log2 ratio. SAM analysis [42] was used to identify gene sets that were similarly and differentially regulated in isp-1 and mev-1 mutants in a cep-1-dependent manner from our microarray data. Log2-transformed fold-change data with no cutoff were submitted to SAM. One class analysis was used to identify genes that similarly changed significantly and consistently in isp-1 vs. cep-1; isp-1 and mev-1 vs. cep-1; mev-1 datasets. To identify genes differently changed between isp-1 vs. cep-1; isp-1 and mev-1 vs. cep-1; mev-1 datasets, SAM two-class unpaired analysis with a FDR = 1% was performed. The resulting gene list was compared with the gene list obtained from SAM one-class to exclude any duplicate genes. The unique 71 genes (Table S6) that were present only in SAM two-class analyses were considered differentially regulated between isp-1 and mev-1 mutants in a cep-1-dependent manner. Worm Base IDs (WBID) of genes identified in SAM and K-mean clusters were input into the Functional annotation-clustering tool in DAVID (http: //david. abcc. ncifcrf. gov/) [43] for gene annotation enrichment analysis. Functional annotation clustering was performed with the default criteria, and the enrichment score for each annotation cluster was determined. Total RNA was isolated from synchronized young adult worms using Tri-reagent (Molecular Research Center, Inc.). cDNAs were synthesized with oligo-dT using the SuperScript III First-Strand Kit (Invitrogen). qRT-PCR reactions were performed using iQ SYBR Green Supermix (BIO-RAD) and the MyiQ Single Color Real-Time PCR Detection System (BIO-RAD). act-1 was used as the internal control. The qRT-PCR experiments were performed at least in triplicate using independent RNA/cDNA preparations. For GFP fluorescence images, worms at the L1-L2 stage were paralyzed with levamisole on an agar pad. The GFP signal was visualized at 60× magnification using a Leica DM 5000B microscope. All images were captured with the same intensity and exposure time using Open Lab software.

Perturbing different components of the mitochondrial electron transport chain (ETC) can affect the lifespan of Caenorhabditis elegans in divergent ways. ETC dysfunction can either attenuate or extend C. elegans lifespan. Here we demonstrate that the C. elegans homolog of mammalian p53, CEP-1, is key in mediating differential survival of distinct ETC mutants. Inactivating cep-1 in two long-lived ETC mutants shortened their lifespans, whereas cep-1 deletion in two short-lived ETC mutants restored their lifespans. Because how CEP-1 does this is unknown, we compared the genome-wide expression profiles of the long-lived isp-1 and short-lived mev-1 ETC mutants upon cep-1 removal. We found that CEP-1 is able to differentially regulate a small subset of genes depending on the specific ETC mutations. We validated one of the genes, the iron transporter ferritin (ftn-1), to be an important target through which CEP-1 mediates distinct lifespan outcome in specific ETC mutants. We also identified numerous other candidate CEP-1 targets that merit future analysis, which will further elucidate how CEP-1 differentially responds to the distinct ETC dysfunction of various mitochondrial mutants. Our study provides new insights that will inform how mammalian p53, a critical tumor suppressor, might respond to mitochondrial and metabolic stresses.

Abstract Introduction Results Discussion Materials and Methods

genetics biology gene function

2014

CEP-1, the Caenorhabditis elegans p53 Homolog, Mediates Opposing Longevity Outcomes in Mitochondrial Electron Transport Chain Mutants

We report the first genome-wide association study (GWAS) whose sample size (1,053 Swedish subjects) is sufficiently powered to detect genome-wide significance (p<1. 5×10−7) for polymorphisms that modestly alter therapeutic warfarin dose. The anticoagulant drug warfarin is widely prescribed for reducing the risk of stroke, thrombosis, pulmonary embolism, and coronary malfunction. However, Caucasians vary widely (20-fold) in the dose needed for therapeutic anticoagulation, and hence prescribed doses may be too low (risking serious illness) or too high (risking severe bleeding). Prior work established that ∼30% of the dose variance is explained by single nucleotide polymorphisms (SNPs) in the warfarin drug target VKORC1 and another ∼12% by two non-synonymous SNPs (*2, *3) in the cytochrome P450 warfarin-metabolizing gene CYP2C9. We initially tested each of 325,997 GWAS SNPs for association with warfarin dose by univariate regression and found the strongest statistical signals (p<10−78) at SNPs clustering near VKORC1 and the second lowest p-values (p<10−31) emanating from CYP2C9. No other SNPs approached genome-wide significance. To enhance detection of weaker effects, we conducted multiple regression adjusting for known influences on warfarin dose (VKORC1, CYP2C9, age, gender) and identified a single SNP (rs2108622) with genome-wide significance (p = 8. 3×10−10) that alters protein coding of the CYP4F2 gene. We confirmed this result in 588 additional Swedish patients (p<0. 0029) and, during our investigation, a second group provided independent confirmation from a scan of warfarin-metabolizing genes. We also thoroughly investigated copy number variations, haplotypes, and imputed SNPs, but found no additional highly significant warfarin associations. We present power analysis of our GWAS that is generalizable to other studies, and conclude we had 80% power to detect genome-wide significance for common causative variants or markers explaining at least 1. 5% of dose variance. These GWAS results provide further impetus for conducting large-scale trials assessing patient benefit from genotype-based forecasting of warfarin dose. Warfarin is the most widely prescribed anticoagulant for reducing thromboembolic events that often give rise to stroke, deep vein thrombosis, pulmonary embolism or serious coronary malfunctions [1]. A combination of genetic and non-genetic factors cause Caucasians to exhibit 20-fold interindividual variation in required warfarin dose needed to achieve the usual therapeutic level of anticoagulation as measured by the prothrombin international normalized ratio or INR [2]–[4]. Thus, in the absence of information (genotypic, clinical, etc.) for predicting each patient' s required warfarin dose, initial prescribed doses may be too low (risking thrombosis) or too high (risking over-anticoagulation and severe bleeding). Warfarin' s risk of serious side effects, narrow therapeutic range, and wide interindividual variation in warfarin dose have focused attention on the need to better predict dose in the initial stage (s) of treatment. We and others have shown that the warfarin drug target VKORC1 (vitamin K epoxide reductase complex, subunit 1) contains common polymorphisms that account for a major portion (∼30%) of the variance in required warfarin dose [5], [6], and we have recently evaluated ∼1500 Swedish patients of the Warfarin Genetics (WARG) cohort in the largest study to date showing likely patient benefit from genetic forecasting of dose [3]. The study confirmed that SNPs in VKORC1 and in the warfarin-metabolizing gene CYP2C9 (cytochrome P450, family 2, subfamily C, polypeptide 9) predict ∼40% of dose variance while non-genetic factors (age, sex, etc.) jointly account for another ∼15%. The robust and now widely replicated associations of warfarin dose with VKORC1 and CYP2C9 have provided one of the most successful applications of pharmacogenetics to date [7] and offer promise for genetic predication of required dose in a clinical setting [3]. Knowledge of major predictors of warfarin dose also impacts the methodology for finding further dose-related genes. In early candidate gene work with a small sample of 201 patients [8], we noted that univariate regression (with tested SNP as the only dose predictor) could statistically detect warfarin association with VKORC1 and with one of two non-synonymous CYP2C9 SNPs (*3) known to influence warfarin dose (Table 1 in [8]). However, a second non-synonymous CYP2C9 SNP (*2) with known but weaker influence on warfarin dose was not detected by univariate regression, but *2 was statistically significant in multivariate regression adjusted for the other known genetic and non-genetic predictors of dose (Table 3 in [8]). These empirical results in a small warfarin sample provided a signpost underscoring the potential importance of multivariate regression for detecting weak effects in studies now searching for additional warfarin genes across the genome. A genome-wide association study (GWAS) enables a systematic search of the entire genome for genetic factors that cause any inherited trait. This method has successfully identified susceptibility loci for common diseases [9], and is beginning to be applied to pharmacogenomics. A recent warfarin GWAS in 181 patients did not detect other genetic factors with major effects on warfarin dose beyond VKORC1 [10] but was underpowered for identifying loci with a moderate contribution. We have now genotyped 325,997 SNPs in 1053 patients of the WARG cohort and here report the first GWAS that is sufficiently powered to detect additional genetic factors that may only modestly influence warfarin dose. Figure 1A and the first line of Table 1 summarize results of testing 325,997 GWAS SNPs for association with warfarin dose by univariate regression. The strongest associations were at multiple SNPs in and near VKORC1 (Figure 1A) with the lowest p-value given by rs9923231 (P = 5. 4×10−78). In prior fine-mapping of the VKORC1 locus [8], we identified rs9923231 as one of three SNPs located in introns or immediately flanking VKORC1 that exhibit almost perfectly concordant genotypes yielding pairwise linkage disequilibrium (LD) r2≈1 and which define the warfarin-sensitive A-T-T haplotype at rs9923231-rs9934438-rs2359612 (see also [11]). These highly concordant SNPs were the best predictors of warfarin dose in our previous study and in this GWAS analysis (p<5. 4×10−78) and completely accounted for the dose variance explained by all other fine-mapping SNPs near VKORC1 [8]. The group of SNPs with the second lowest univariate p-values clustered around CYP2C9 which contains two non-synonymous exonic SNPs whose minor alleles (*2, *3) impair warfarin metabolism and are well known to be associated with warfarin dose. In our previous work [8], we discovered an unusual SNP (rs4917639) whose minor allele is almost perfectly associated with the “composite” CYP2C9 allele formed by combining *2 and *3 into a single allele. Indeed, the GWAS results (1053 subjects) confirmed that LD is nearly perfect (pairwise r2≈1. 0) between rs4917639 and the composite of *2 and *3. Thus, the highly significant univariate result for rs4917639 (R2 = 0. 121, p<3. 1×10−31) reflects the combined effect of CYP2C9*2 rs1799853 (R2 = 0. 038, p<8. 8×10−13) and CYP2C9*3 rs1057910 (R2 = 0. 080, p<4. 5×10−17). Figure 1A therefore indicates p-values for this composite SNP as well as for *2 and *3. Figure 1B and Table 1 (lines 2 to 5) show the results of multivariate regression analysis in which individual SNPs were tested for association with warfarin dose after adjustment for established genetic and non-genetic predictors of dose. The only SNP reaching genome-wide significance (p<1. 5×10−7) was a non-synonymous SNP (rs2108622) in exon 2 of CYP4F2 (cytochrome P450, family 4, subfamily F, polypeptide 2) introducing a Val to Met amino acid change at position 433 (V433M). SNP rs2108622 predicts additional dose variance (∼1. 1%) that is independent of the variance already explained by VKORC1 and CYP2C9. As noted in the Introduction, our early studies with a small sample of 201 Swedish patients failed to detect the weak CYP2C9*2 effect on dose by univariate regression but *2 was significant in multiple regression [8]. The results in Table 1 with rs2108622 of CYP4F2 show the same phenomenon with a p-value of 1. 6×10−5 in univariate regression (line 1) but progressively lower p-values as known predictors are added to the multivariate model so that for the full model a p-value of 8. 3×10−10 is achieved which is far below genome-wide significance (p<1. 5×10−7). The CYP4F2 association was further confirmed by testing an independent replication panel of 588 Swedish warfarin patients who gave a multivariate p-value of 0. 0029 and a total overall p-value of 3. 3×10−10 when combined with the GWAS subjects (Table 2). During preparation of this paper, a candidate gene study of drug-metabolizing and transporter genes independently discovered the association of rs2108622 and CYP4F2 with warfarin dose, providing further confirmation [12]. To increase the power of our multivariate regression model and possibly detect additional weak effects, we added CYP4F2 (rs2108622) to the model as a predictor and conducted further analyses. First, we retested the GWAS SNPs, but no new SNPs reached genome-wide significance and there was also no apparent excess of SNPs at lower significance thresholds (Figure S1). We also tested warfarin association with haplotypes and with ungenotyped SNPs imputed at 2. 2 million HapMap SNPs, but no haplotype or imputed SNP approached genome-wide significance in a genomic region not containing VKORC1, CYP2C9 or CYP4F2. To explore whether copy number variations (CNVs) detectable by the HumanCNV370 array might influence warfarin dose, we used rigorous quality control and retained 879 samples calling 2530 CNVs (see Materials and Methods). None of the CNV loci were significantly associated with dose after correction for multiple testing (lowest CNV p-value was 1. 1×10−4 which exceeds 0. 05/2530≈2. 0×10−5). We note that probe density in many of the detected CNVs is not optimal for conducting association analyses and these results should therefore be viewed as preliminary. Finally, after excluding SNPs near VKORC1, CYP2C9 and CYP4F2, we identified 40 other loci containing one or more GWAS SNPs with p-values below 2. 0×10−4 and we genotyped 40 SNPs representing these loci in a follow-up sample of 588 Swedish warfarin patients. However none of the 40 loci replicated for association with warfarin dose, the lowest p-value being 0. 04 which is not significant after correction for 40 tests (Table S1). Having not found evidence for any additional genetic modulators of dose, we examined the entire data set (GWAS plus followup samples) for evidence of statistical interaction between pairs of the established dose predictors (VKORC1, CYP2C9, CYP4F2, age, sex). None of the pairs exhibited statistically significant interaction after p-values were corrected for the 15 interaction tests (Table S2). We also performed a GWAS for a secondary trait (“over-anticoagulation”) which we previously found was associated with VKORC1 and CYP2C9 in a candidate gene study [3]. By titrating warfarin dose, physicians attempt to achieve a target level of anticoagulation determined by a reading of 2. 0 to 3. 0 for the prothrombin international normalized ratio (INR), which is the ratio of time required for a patient' s blood to coagulate relative to that of a reference sample. However over-anticoagulation (defined as an INR above 4. 0) sometimes occurs and, using Cox regression, our GWAS tested for SNP association with the occurrence of over-anticoagulation in patients during the first 5 weeks of treatment (see Materials and Methods: Association testing of SNPs and haplotypes). We observed genome-wide significant association (p<1. 5×10−7) at several SNPs in and around VKORC1 including rs9923231 (P = 8. 9×10−9), but no other SNPs achieved genome-wide significance including CYP2C9*3 (p<4. 0×10−5), CYP2C9*2 (p = 0. 93), or the “composite” *2*3 SNP rs4917639 (p<0. 007) (Figure S2). However we note that our previous candidate gene study evaluated a larger sample set (1496 WARG subjects) which yielded genome-wide significant association with over-anticoagulation for both VKORC1 rs9923231 (P = 5. 7×10−11) and CYP2C9*3 (P = 1. 5×10−9) [3]. To explore whether these SNPs might cause over-anticoagulation independent of altering the required (i. e. , administered) warfarin dose, we added required dose to the Cox regression model as a predictor of over-anticoagulation, and found that both VKORC1 and CYP2C9*3 have a significant effect independent of dose (P<0. 05) (Table S3). We conducted the first GWAS sufficiently powered to detect DNA variants with a modest influence on the warfarin dose needed to achieve therapeutic anticoagulation. In univariate analysis of GWAS SNPs (Figure 1A), we identified extremely strong association signals (p = 10−78 to 10−13) at SNPs in and near VKORC1 and CYP2C9, two genes already known to explain ∼30% and ∼12% of warfarin dose variance, respectively. By applying multivariate regression adjusting for known genetic and non-genetic predictors of dose (Figure 1B), we also detected genome-wide significance of p<8. 3×10−10 at CYP4F2 (rs2108622) that accounted for approximately 1. 5% of dose variance. The increased power of multivariate regression to detect this modest effect is nicely illustrated in Table 1 which shows a higher univariate p-value for CYP4F2 (p<1. 6×10−5) but progressively lower multivariate p-values as known predictors of dose are added to the regression model. We confirmed the CYP4F2 association in a second large sample set and the association was also reported by another group [12] during preparation of our work, thus fully establishing the genuine effect of CYP4F2 (see also [10] where CYP4F2 explained ∼1% dose variance with nominal p<0. 043 significance). Although multivariate regression has not been widely used to increase power in other GWAS analyses because known genetic variants usually explain little phenotypic variance, the potential for power increase is perhaps obvious if known predictors do explain substantial variance. Thus multiple regression has, for example, been previously advocated for linkage analyses of line crosses [13], [14]. To estimate the multivariate regression power of our GWAS (1053 subjects), we used Equation 1 (see Materials and Methods) to calculate power to detect SNPs explaining specific magnitudes of variance () for warfarin dose (see Table 3). The table shows that power to achieve genome-wide significance (p<1. 5×10−7) is essentially 100% for VKORC1 rs9923231 (), CYP2C9*3 () and CYP2C9*2 (), but power falls to ∼48% for CYP4F2 rs2108622 (). The table also shows that when CYP4F2 is added to the multivariate model, a SNP accounting for 1. 5% or 1. 0% of the dose variance would have ∼82% or ∼41% power of being detected, respectively. Therefore we estimate that our GWAS had at least 80% power to detect warfarin-associated variants explaining at least 1. 5% of the dose variance but 40% or less power to detect genome-wide significance if a variant accounts for 1% or less dose variance. However it is important to emphasize that these power estimates assume that the dose-altering DNA variant is genotyped and tested directly or is indirectly detected through a marker in sufficiently high LD to the dose variant that the marker' s magnitude is detectable (Table 3). The assumption of directly testing the dose-altering variant is accurate for CYP2C9*2 and *3 which are each known to alter warfarin metabolism [15], [16] and is likely correct for CYP4F2 rs2108622 which, like *2 and *3, changes protein coding sequence. However, to explore whether other dose-altering variants might be undetected due to insufficient LD with genotyped GWAS SNPs, we determined the relationship between the variance observed at a marker () and at the causative variant () assuming pairwise LD of r2 between the two polymorphisms (see Materials and Methods: How Much Does Linkage Disequilibrium Attenuate Association with a Quantitative Trait?). The relationship is given by Equation 3 in Materials and Methods () which is analogous to Pritchard and Prezworski' s relationship () for the number of cases () providing equal power in a case-control study that tests either the disease-causing SNP or a nearby marker [17]. To use the equation to estimate magnitudes for variants that might be undetected by our GWAS, we note that ∼90% of the GWAS SNPs had a minor allele frequency (MAF) above 10% in our warfarin subjects implying that a “rare” dose-altering variant (MAF≈1%–5%) would be covered at a likely maximum r2 of only ∼0. 1 to ∼0. 5. This low r2 coverage implies that rare variants could have values (0. 05 to 0. 02) easily detected by regression testing of the variant itself, but unlikely to be detected through a GWAS marker since maximum values could drop to 0. 01 or much lower (see Equation 3 and Table 3). By contrast, “common” SNPs (MAF≥5%), which might also be dose variants, are covered by GWAS SNPs of this study at reasonably high r2 values in most instances (r2>0. 8 or r2>0. 5 for ∼60% or ∼80% respectively of common SNPs [18] and r2>0. 9 for ∼90% of non-synonymous common SNPs [19] in HapMap Caucasians). We therefore conclude that our GWAS probably detected most common SNP variants explaining 1. 5% or more of the warfarin dose variance, but may have failed to detect rarer variants that could individually explain up to 5% of dose variance. We further note that the HumanCNV370 array used in this study does not have the required marker complement to undertake a comprehensive GWAS of common CNVs. As noted in the Introduction, the widely replicated warfarin dose associations with VKORC1 and CYP2C9 represent one of the most successful applications of pharmacogenetics to date. Our study together with that of Caldwell et al. [12] now also clearly demonstrates that CYP4F2 (rs2108622) is a third gene that influences warfarin dose, but our GWAS and statistical analysis also implies that additional common SNP variants that influence dose may not exist in Caucasian populations. However, Caucasians might carry common variants with effects smaller than CYP4F2 or rare variants whose effects are substantially larger than the ∼1% of dose variance explained by CYP4F2. Furthermore, other unidentified genes may influence warfarin dose in other ethnicities such as Asians or Africans, and some rare dose-altering variants in known genes such as VKORC1 may exist in only a subset of populations of European descent [20]. Hence, future research could address ethnic differences in the genetic variants that influence warfarin dose as well as subtle intra-ethnic differences and admixture that may exist in European or other populations. In a recent study [3], we highlighted the potential benefit of pre-treatment forecasting of required warfarin dose based on patient genotypes at VKORC1 and CYP2C9 together with non-genetic predictors of dose. Indeed, in August 2007, the US Food and Drug Administration (FDA) updated warfarin labeling to recommend initiating lower warfarin dose in some patients based on VKORC1 and CYP2C9 genotypes. However this recommendation is not a requirement due to a lack of large trials demonstrating warfarin patient benefit from dose forecasting (though two small trials [21], [22] do support such benefit; see also [23]–[27] for reviews and other trials). The results of our GWAS provide further impetus for conducting large-scale dose-forecasting trials by identifying CYP4F2 as a third genetic predictor of dose and also by showing that additional major genetic predictors may not exist in Caucasians or may not emerge in the near-term. Hence, large-scale trials of patient benefit from dose forecasting based on VKORC1 and CYP2C9 (with possible inclusion of CYP4F2 as a minor predictor) are likely to provide state-of-the-art clinical benchmarks for warfarin use during the foreseeable future. The study subjects were 1053 Swedish patients collected for the WARG study [3] (http: //www. druggene. org/). This is a multi-centre study of warfarin bleeding complications and response to warfarin treatment [28]. Anticoagulant response is measured by INR, which is the ratio of the time required for a patient' s blood to coagulate relative to that of a reference sample. By titrating warfarin dose, physicians aim for a therapeutic INR reading between 2. 0 and 3. 0; thus the primary quantitative outcome for the GWAS was the mean warfarin dose (mg/week) given to a patient during a minimum series of three consecutive INR measurements between 2 and 3 [3]. As a secondary GWAS outcome, we also catalogued each patient for the occurrence or non-occurrence of “over-anticoagulation” during the first 5 weeks of treatment (defined as an INR reading above 4. 0) and tested for genetic association which adjusted for the treatment day (1 to 35) of the over-anticoagulation event (see “Association testing” below). The clinical data collected by the WARG protocol included gender and age since each is a known non-genetic predictor of warfarin dose but did not include bodyweight and dietary information (e. g. vitamin K intake). Regression analysis of prescribed medication which can potentiate or inhibit warfarin action was not a statistically significant predictor of warfarin dose in the 1053 WARG GWAS subjects and hence was not included as a predictor variable in the multivariate regression analyses. The WARG study samples were previously described elsewhere [3], [4], [28], [29] as were the Uppsala followup samples [8]. The WARG and Uppsala studies received ethical approval from the Ethics Committee of the Karolinska Institute and the Research Ethics Committee at Uppsala University, respectively. From approximately 1500 WARG samples [3] examined for non-degradation and appropriate concentration of DNA (∼50 ng/µl), we randomly selected 1208 subjects for genotyping SNPs and CNV probes using the HumanCNV370 BeadChip array (Illumina). We excluded SNPs with MAF below 1%, call rate below 95%, or if call rate fell below 99% when MAF was below 5%. SNPs that departed from Hardy-Weinberg equilibrium (P<10−6) were also excluded. Subjects with genotyping call rate below 95% were also eliminated. Using iPLEX (Sequenom), subject identity (and associated phenotypic data) was cross-checked by genotyping four gender markers and 47 SNPs also carried on the HumanCNV370 array, enabling us to exclude ∼136 misidentified subjects. Sample quality (contamination) was further assessed by plotting each subject' s genome-wide heterozygosity and eliminating outliers (with heterozygosity above or below the range of 0. 312–0. 372). After these quality control steps, a total of 1053 warfarin patients and 325,997 GWAS SNPs were retained for analysis. The GWAS SNPs included two SNPs not on the HumanCNV370 array but which are highly predictive of warfarin dose [rs9923231 (VKORC1) and rs1799853 (CYP2C9*2) ] which we genotyped by TaqMan assay (Applied Biosystems). Although we retained 325,997 GWAS SNPs for association testing of SNPs, it should be noted that all ∼370,000 probes on the Human CNV370 array were used to define CNVs. Log R ratio values of probes were output from the BeadStudio software [30]. A loess correction was applied to each sample to remove local correlations or genomic wave [31]. The resultant genomic copy number profiles were then segmented using Circular Binary Segmentation [32]. Some samples displayed abnormally high numbers of segments indicating problems in DNA quantity or quality or hybridization. Samples were removed until the number of segments across all samples was approximately normal. Using this technique, 143 (14%) of samples were flagged as problematic. These samples were excluded when CNV regions were defined but included for association testing. Putative CNV were defined from segments by applying a threshold on the segment log R ratio. This threshold was asymmetric allowing for a differing response for deletions and duplications. The central peak of the segment log R ratio distribution was fitted and the threshold values obtained by taking values at ±5 standard deviations from the centre. In order to define regions for association testing, merging of CNV across samples was performed. This was achieved by merging two putative CNV into a region if there was greater than 40% reciprocal overlap. This procedure defined 2530 CNV regions in total. Of these, most were singletons (54%) or low frequency, <3% (93%), while 820 (70%) of the non-singleton regions overlapped CNVs from the Database of Genomic Variants [33]. We tested all 2530 CNVs for association, because a CNV discovered as a “singleton” might well include multiple copies of a rare CNV allele in the study samples. At each SNP, genotypes were coded 0,1 or 2 and the SNP was tested for association with the square root of warfarin dose [8] by either univariate or multivariate linear regression analysis conducted in PLINK [34] (http: //pngu. mgh. harvard. edu/~purcell/plink/) or in R software (http: //www. r-project. org/). We used the same regression analysis to test association with all HapMap SNPs not on the HumanCNV370 array by imputing ∼2. 2 million SNPs using Beagle software [35] trained from genotypes of the 60 HapMap CEU parents [36]. We excluded SNPs whose imputed MAF was below 5% or differed by more than 5% with MAF of the CEU parents. We also tested haplotypes for association with warfarin dose by two approaches: (1) each subject' s warfarin dose residual (difference between observed and predicted dose based on the full multivariate regression model containing CYP4F2) was considered a quantitative trait value and tested for association with haplotypes defined across the genome in sliding windows of 2,3 or 4 consecutive SNPs as implemented by PLINK software; (2) by scanning GWAS genotypes, Beagle software groups genetically related haplotypes into clusters which it then resolves into diallelic (SNP-like) “pseudo-markers” optimized for detecting phenotypic association. To test haplotypes, we evaluated the pseudo-marker genotypes of warfarin patients at 1. 97 million pseudo-markers covering the genome by testing each pseudo-marker in the same multivariate regression framework used to test individual SNPs (as described in the preceding paragraph). We tested for statistical interaction in modulating warfarin dose for each pair of established dose predictors (VKORC1 rs9923231, CYP2C9*2 and *3, CYP4F2 rs2108622, Age, Sex) using multivariate regression and R software as described above. An interaction term formed by multiplying the pair of predictor variables was added to the multivariate regression equation which contained only main effects of the 6 predictors, and standard ANOVA compared this main-effect model with the enhanced interaction model by testing for a statistically significant increase in explained dose variance. Interaction test p-values were considered statistically significant if below the Bonferroni cutpoint determined by correcting for the 15 interaction tests (i. e. p<0. 0033≈0. 05/15). To test for association with over-anticoagulation (INR>4. 0) during treatment days 1–35, we performed Cox proportional hazard regression on survival time (day of over-anticoagulation) using the survival library of R software. The GWAS data set of 1053 WARG subjects contained 215 subjects whose INR exceeded 4. 0 during days 1–35 while the entire dataset of 1489 WARG subjects contained 312 such subjects. For each CNV locus, association was tested with square root of warfarin dose by multivariate regression analysis in which subject copy number intensity was the CNV predictor of dose. This analysis differs from association testing with SNP genotypes since the two CNV alleles on homologous chromosomes generate one copy number intensity rather than a separate allele for each chromosome. As a QC strategy, we determined each subject' s rank in the dataset for copy number intensity at each CNV on chromosome 17. This enabled us to differentiate the majority of subjects (whose individual distribution of ranks were approximately random and uniform) from 174 obvious outliers due to poor quality DNA (whose ranking distributions were “U-shaped” since their intensities strongly clustered at both high and low ranks). These 174 subjects were excluded from the primary CNV association analysis (with further confirmation of lower quality DNA for these subjects being their rough correspondence to the subjects with lower (<99%) SNP call rates). However, we also crosschecked the primary CNV analysis by conducting association testing on the dataset without excluding the 174 subjects and found no statistically significant association with warfarin dose at any CNV whether the dataset excluded or included the subjects. Association testing of the CNVs was executed using R software [37]. For the replication of CYP4F2 rs2108622, we genotyped a panel of 588 warfarin patients consisting of 410 subjects from the WARG cohort [3] and 178 from the Uppsala cohort [38]. Table 2 shows regression on this pooled sample of 588 subjects. Separate results for each of the two panels are given in Table S4. To possibly identify SNPs with genuine but weak associations to warfarin dose, we excluded VKORC1, CYP2C9, CYP4F2 and identified 40 other GWAS loci for follow-up genotyping exhibiting multivariate regression p-values below 0. 0002, and selected 40 SNPs representing these loci for genotyping. Only genotyped (not imputed) SNPs were chosen for follow-up. We genotyped the same 558 patients as in the CYP4F2 replication using the iPLEX MassARRAY. Suppose multiple regression analysis is conducted in N total samples by testing a SNP with coefficient of determination (i. e. , explained variance) R2test after adjustment for known predictors whose total of coefficient of determination is R2knw. The probability (power) to detect the tested SNP at a significance level α equals: (1) where F′ (1, N–2, θ2) is the probability density function for an F distribution with 1, N-2 degrees of freedom and non-centrality parameter θ2 (Section 28. 28 in [39], Example 8. 4 in [40]). Here the constant c satisfies the equation: (2) where α is the significance level, and F (1, N–2) is the probability density function for a F-distribution of degree of freedom one and N–2. Association with a quantitative trait (QT) becomes weaker for a marker SNP in LD with a SNP that alters the QT, and hence the association becomes more difficult to detect at the marker than at the QT-altering SNP. Here we quantify the LD attenuation for a QT when testing for association by linear regression (which includes the Cochran-Armitage trend test for dichotomous traits), and we obtain a result analogous to the LD attenuation for the Pearson Chi-square test for allelic association to dichotomous traits as in cases and controls [17]. If a causative QT-altering SNP has a coefficient of determination (i. e. , explained variance) and is in pairwise LD of r2 with a marker SNP, then the coefficient of determination for the marker SNP () is approximated by: (3) In other words, when testing a marker, the proportion of explained variance decreases by a factor of r2. To begin the proof of Equation 3, let the QT be represented by the random variable “q”, and let “m” and “x” be SNP genotypes (coded 0,1, or 2) representing the marker and causative (QT-altering) SNP, respectively. The coefficients of determination are equal to the square of two correlation coefficients (denoted by “Corr”) measuring the correlation of m or x with q: (4) (5) Also note that correlation between genotypes at the marker and causative SNP is given by another correlation coefficient: (6) It is well known that the partial correlation coefficient of m and q conditioned on x is (equation 16. 20, p. 649 in [41]): (7) However, conditional on genotype at the causative SNP, marker m and the QT q would be uncorrelated (assuming m is not in LD with a second causative polymorphism) and thus the numerator of Equation 7 would be zero implying that: (8) Based on prior work [42]–[44], we show in Text S1 that the squares of the genotypic correlation coefficient and LD correlation coefficient r2 are approximately equal if the population is in Hardy-Weinberg equilibrium. Therefore, substituting r2 for in Equation 8 gives Equation 3.

Recently, geneticists have begun assaying hundreds of thousands of genetic markers covering the entire human genome to systematically search for and identify genes that cause disease. We have extended this “genome-wide association study” (GWAS) method by assaying ∼326,000 markers in 1,053 Swedish patients in order to identify genes that alter response to the anticoagulant drug warfarin. Warfarin is widely prescribed to reduce blood clotting in order to protect high-risk patients from stroke, thrombosis, and heart attack. But patients vary widely (20-fold) in the warfarin dose needed for proper blood thinning, which means that initial doses in some patients are too high (risking severe bleeding) or too low (risking serious illness). Our GWAS detected two genes (VKORC1, CYP2C9) already known to cause ∼40% of the variability in warfarin dose and discovered a new gene (CYP4F2) contributing 1%–2% of the variability. Since our GWAS searched the entire genome, additional genes having a major influence on warfarin dose might not exist or be found in the near-term. Hence, clinical trials assessing patient benefit from individualized dose forecasting based on a patient' s genetic makeup at VKORC1, CYP2C9 and possibly CYP4F2 could provide state-of-the-art clinical benchmarks for warfarin use during the foreseeable future.

Abstract Introduction Results Discussion Materials and Methods

genetics and genomics/pharmacogenomics pharmacology/personalized medicine mathematics/statistics

2009

A Genome-Wide Association Study Confirms VKORC1, CYP2C9, and CYP4F2 as Principal Genetic Determinants of Warfarin Dose

In 2010 and 2011, Haiti was heavily affected by a large cholera outbreak that spread throughout the country. Although national health structure-based cholera surveillance was rapidly initiated, a substantial number of community cases might have been missed, particularly in remote areas. We conducted a community-based survey in a large rural, mountainous area across four districts of the Nord department including areas with good versus poor accessibility by road, and rapid versus delayed response to the outbreak to document the true cholera burden and assess geographic distribution and risk factors for cholera mortality. A two-stage, household-based cluster survey was conducted in 138 clusters of 23 households in four districts of the Nord Department from April 22nd to May 13th 2011. A total of 3,187 households and 16,900 individuals were included in the survey, of whom 2,034 (12. 0%) reported at least one episode of watery diarrhea since the beginning of the outbreak. The two more remote districts, Borgne and Pilate were most affected with attack rates up to 16. 2%, and case fatality rates up to 15. 2% as compared to the two more accessible districts. Care seeking was also less frequent in the more remote areas with as low as 61. 6% of reported patients seeking care. Living in remote areas was found as a risk factor for mortality together with older age, greater severity of illness and not seeking care. These results highlight important geographical disparities and demonstrate that the epidemic caused the highest burden both in terms of cases and deaths in the most remote areas, where up to 5% of the population may have died during the first months of the epidemic. Adapted strategies are needed to rapidly provide treatment as well as prevention measures in remote communities. The cholera epidemic in Haiti, which began in 2010 spread rapidly in both urban and rural areas. One month after confirmation of the first case in Mirebalais, in the department of Centre, the whole country had been affected [1,2]. During the first few days, the focus was on case management in hospitals, which were quickly overwhelmed [1]. Gradually, the Ministry of Health (MSPP), together with partners including non-governmental organizations (NGOs), started setting up dedicated treatment facilities ranging from large specialized centers to decentralized oral rehydration points (ORP) in more isolated communities, cholera-specific health education messages, and water and sanitation activities [1,3]. A national training program for cholera management was developed to train clinical staff, nearly all of whom were unfamiliar with the disease [4]. Despite these efforts, over 600,000 cases of cholera and 7,000 deaths were reported by the national health-structure based surveillance system within two years of the first case [5], and, at the time of writing this article, cases are still being reported (http: //mspp. gouv. ht/). In the Nord department, the first cholera cases were officially reported on October 22nd, 2010 (week 42). The first cholera treatment center (CTC) was opened on October 23rd in Cap Haitien, the administrative center of the department, and cholera treatment units (CTUs) were gradually opened in November in the main communes of the department. ORPs only started operating in remote areas in December 2010 and January 2011 (Fig. 1). Since national surveillance data were based on reports from the health structures and were likely to miss community cases, large retrospective population-based surveys were conducted by Médecins Sans Frontières (MSF) in April and May 2011 to estimate the cholera burden during the first weeks of the epidemic and get insight into health-seeking behavior. Here, we present the results of a survey that was conducted in a large rural, mountainous area across four districts of the Nord department, chosen to facilitate comparison between regions with good versus poor accessibility by road, and with rapid versus delayed response to the outbreak. We also present results of a risk factor analysis carried out in the same area, looking for potential explanatory factors for geographic differences in mortality with the aim of providing information to improve future outbreak response strategies in similar settings. The procedures followed were in accordance with the ethical standards of the Helsinki Declaration. The National Ethics Committee of Republic of Haiti granted ethical approval and the Ministry of Public Health of the Republic of Haiti gave authorization to perform the survey. Written informed consent for study participation was obtained from all participants. The Nord department is located on the northern coast of Haiti and encompasses coastal and mountainous areas, with limited road infrastructure (Fig. 1). Health structures are located in urban centers with catchment areas of hundreds of square kilometers. Many houses are not accessible by road and some hamlets are located more than a 10-hour walk from the nearest health structure. The study took place in 2011 in four districts (“communes”) of the Nord department: Plaisance, Pilate, Borgne and Port Margot (Fig. 1). Villages in the districts of Plaisance and Port Margot are mainly accessible by roads, while villages in the more mountainous districts of Pilate and Borgne are more difficult to reach due to their mountainous terrain. Each district is divided into 6 to 8 sections. According to a 2009 estimate [6], the total population in the four districts was 218,649 inhabitants, of whom 173,903 lived in rural areas targeted by this survey. From the beginning of the cholera outbreak and until the time of the survey, 29,295 cases and 654 deaths were reported in the Nord department [Rapport journalier MSPP du 22 mai 2011], for an estimated attack rate (AR) of 2. 9% and case-fatality rate (CFR) of 2. 2%. MSF, one of the main partners working with the MSPP to treat cholera patients in Haiti, intervened early in Plaisance, where a CTU was opened in epidemiological week 44,2010, followed by a CTU in week 47 and 5 ORPs in weeks 49 and 50. In Pilate, the intervention was slightly delayed, with a CTU opening in week 47, followed by 2 ORPs in week 49 and another 6 ORPs in week 1,2011. In Borgne, a CTC opened in week 47, followed by a CTU in week 50,5 ORPs in week 52,9 ORPs in week 1,2011 and 6 ORPs in week 2. In Port Margot, a CTU was opened by the Catholic Church before week 48 and an ORP by the MSPP in week 48, while MSF started late, with one ORP in week 52,2 ORPs in week 1,2011 and a CTU and one additional ORP in week 2. A two-stage, household-based cluster survey was conducted in the study area. The sample size was 16,000 individuals, calculated to estimate an expected crude mortality rate of 0. 5 per 10 000 persons per day with a precision of 0. 1, an anticipated design effect of 2 and a recall period of 170 days (from October 17th, 2010 to the earliest survey date). In total, 140 clusters of 23 households, with an expected average size of five members per household, were selected in the four districts. For the first sampling stage, clusters were attributed to each communal section proportionally to the size of the rural population (37 in Plaisance, 34 in Pilate, 42 in Borgne and 27 in Port-Margot). Villages of more than 5,000 inhabitants were considered urban and therefore excluded from the sampling frame. For the second stage, a large number of random geographic points was generated using the R statistical package. These randomly selected points were then mapped using Google Earth; only points with a house found visually within a 50m radius were retained. For each section, the number of points allocated was then randomly selected from the remaining points. The corresponding GPS points were used in the field to locate the initial house of each cluster. The next house was selected by proximity, i. e. , next closest house, until 23 households had been visited in each cluster. Households in which no adult was present at the time of the first visit were revisited at least once before the study team left the village. Data were collected using a standardized questionnaire. After providing written consent, the head of household was asked to provide the age and sex of all household members (defined as persons living under the same roof and sharing meals). For each household member present at the beginning of the recall period, the head of household was asked about episodes of diarrheal illness (defined as at least three watery stools within a 24-hour period) and deaths that occurred during the recall period. More detailed information was collected on the diarrheal episode, or on the most severe one if multiple episodes were reported for the same individual. Information collected included duration and symptoms of the episode, health-seeking behavior (i. e. , type (s) of health structure (s) visited or reason for not visiting a health structure), and outcome (i. e. , death or survival). Severe cases were defined as those in which patients reported lethargy or altered consciousness during the diarrheal illness. Death was considered related to diarrhea when it was reported as the outcome of the most severe diarrheal episode. In each cluster, the time and type of transport to the closest village with a health structure (excluding ORPs) was documented. Double data entry was done in Epidata 3. 1 (EpiData, Odense, Denmark) by four trained data entry clerks. Data validation and statistical analysis were performed using Stata 11 (StataCorp, College Station, Texas, USA) and R Statistical Software. As not all clusters achieved a sample of 23 households, weighted analysis was used to adjust for the probability of each household being selected, by dividing the expected household number per cluster (23) by the actual number of households included. In all analyses, we accounted for the clustering of households within the cluster and applied the selection weights. Design effects are reported where relevant. Each principal outcome was presented as a percentage with its associated 95% confidence intervals. Results were then extrapolated to the overall rural population by applying a weight which multiplies the selection weight by the total rural population of each district divided by the sample size in this district. A geographical representation of each principal indicator was done using a generalized additive model assuming a Poisson distribution and using an isotropic spline to describe the spatial variation of the different indicators [7]. The level of smoothness of the spatial terms was selected using the restricted maximum likelihood method. Finally, we used a Poisson regression model for the univariate and multivariate analyses of risk factors for cholera morbidity and mortality, and present here crude and adjusted relative risks (RR, ARR) and associated 95% confidence intervals. The district of Plaisance was considered as the reference for comparisons among districts. The survey took place from April 22nd to May 13th 2011. In total, 138 randomly selected clusters were visited, and information on 3,187 households and 16,946 individuals collected, which corresponded to approximately 9% of the area’s estimated rural population (173,903 inhabitants). Of these, 46 individuals were subsequently excluded from the analysis due to incorrect inclusion criteria (n = 28) or missing data (n = 18). The median number of individuals per household was 5 (range 1 to 20 persons). The male/female ratio was 0. 91 and the median age was 21 years (IQR: 11–40). In total, 2,034 persons (12. 0%) reported at least one episode of watery diarrhea during the recall period (Fig. 2). Among them, 1,979 (97. 3%, 95% CI: 96. 0–98. 2) reported a single episode during the recall period (range 1–4). The median length of the most severe episode investigated in each individual who reported watery diarrhea was 3 days (IQR: 2–4; range 1–15). Of those individuals reporting diarrhea, 68. 9% (95% CI: 63. 1–74. 1) reported vomiting, and 38. 3% (95% CI: 33. 5–43. 4) lethargy or altered consciousness during the diarrheal illness, and were therefore considered as severe cases. The attack rate of watery diarrhea in the area during the recall period was 12. 0% (95% CI: 10. 8–13. 2), with a design effect of 5. 9. Extrapolated to the rural population in the four districts, this translated into an estimate of 21,681 individuals (95% CI: 19,440–23,922) suffering from watery diarrhea during the recall period. The geographical distribution of attack rates showed marked disparities, with attack rates estimated at more than 20% in some sections in the west of Borgne and Pilate and lower than 10% in most sections of the Plaisance and Port Margot districts (Fig. 3). This was reflected in the estimated attack rates by district, which ranged from 8. 6% in Port Margot to 16. 2% in Borgne (Table 1). In total, 275 individuals were reported to have died during the recall period, leading to a crude mortality rate estimate of 0. 82 deaths per 10,000 persons per day (95% CI: 0. 64–1. 05), which represented 1. 62% (95% CI: 1. 26–2. 07) of the population during the recall period or 2,925 (95% CI: 2199–3651) deaths of all causes when extrapolated to the rural population of the four districts (Plaisance: 393; Port Margot: 246; Pilate: 746; Borgne: 1540). Most of these deaths (84. 8%; 95% CI: 77. 5–90. 0) were attributed to diarrhea. Of the 2,034 diarrhea cases, the outcome of the episode was death in 224, for a CFR of 11. 0% (95% CI: 8. 6–13. 9) with a design effect of 3. 8. Extrapolated to the rural population of the four districts, this represents 2,375 (95% CI: 1,710–3,040) deaths due to diarrhea during the recall period (Plaisance 256; Port Margot 155; Pilate 609; Borgne 1,355). The overall CFR in both Borgne and Pilate was significantly higher than in Plaisance, the reference district (Table 1). The highest CFR (up to 30–40%) were found in western Borgne and Pilate (Fig. 3). Of 2,030 individuals reporting diarrhea and for whom information on health-seeking behavior was available, 1,447 (71. 2%, 95% CI: 66. 3–75. 6) sought care in a health structure. More than 50% of those who sought care visited a specialized CTC or CTU, and only 3% reported using the ORPs (Table 2). Overall, the main reasons for not seeking care were that the health structure was too far or that the diarrhea was not perceived as requiring care or not perceived as cholera. Among the most severe cases, almost two-thirds reported distance as the main reason for not seeking care (Table 2). The lowest proportion of individuals seeking care was in the remote areas of western Borgne and Pilate (Fig. 3). Of the four districts, the highest proportion of patients seeking care was in Port Margot (83. 2%) and lowest in Borgne (61. 6%) (Table 1). The reasons for not seeking care also varied by district: distance was cited as a barrier by 52. 7% of patients who did not seek care in Borgne but was less cited in Pilate (20. 8%), Plaisance (14. 5%) and Port Margot (13. 5%). In the latter two districts, the main reason for not seeking care was a combination of no perceived need and illness not perceived as cholera (Plaisance: 59. 7%; Port Margot: 55. 7%, Pilate: 41. 4%, Borgne: 42. 7%). A stratified analysis of risk factors by district showed that similar factors contributed to higher CFR across all districts: older age (> = 60 years), greater severity of illness, living in remote areas, and not seeking health care (Table 3). These factors were also found to have a significant association in the univariate analysis (Table 4). Factors associated with highest risk were severity of disease (RR = 8. 1) and not seeking care (RR = 5. 1). There was no significant difference in case fatality between males and females. Stepwise introduction of risk factors in a multivariate analysis showed that the differences between districts remained significant when adjusted for age and severity of disease. Due to colinearity between the two variables, remoteness and health-seeking behavior were introduced separately in the model. In each model, older age, severity of disease, and health-seeking behavior or remoteness were associated independently with a higher risk of death (Table 4). Interestingly, when remoteness was introduced in the model, the differences between districts were no longer significant. The results of this large community-based survey on the burden of cholera during the first six months of the outbreak in a rural and mountainous area in the northern part of Haiti show very high attack rates and case fatality rates. It highlights important geographical disparities in the four districts investigated, and in particular, the higher risk of both disease and death in the most remote areas. Both the attack rate and case fatality rate found through the survey were more than four times higher than those calculated using data recorded by the national surveillance system in the same period in the Nord department. Moreover, the extrapolated number of cases in the rural populations of these four communes only (21,681 for a population of 173,903) almost reached the total number of cases reported in the national surveillance for the whole department until May 22nd (29,295 for a population of 1,004,247), while the extrapolated number of diarrhea-related deaths in the four communes (2,375) was 3. 5 times higher than the total number of deaths (654) reported in the whole department over the same period. This acute underreporting of cases and deaths through the national surveillance system derived from health facility-based cases highlights the importance of community data to better estimate disease burden in areas where national surveillance system may encounter major limitations due to the limited access of the population to health structures. Such data are crucial for targeting the most urgent responses to the highest-priority areas. To achieve this goal, local social leaders (head of villages, religious leaders, etc.) and associations should be mobilized early on to participate in both sensitization and community-based surveillance. Very remote areas were particularly affected by the outbreak, in terms of both number of cases (high attack rates) and diarrhea-related deaths (high CFR). This led to extremely high mortality rates estimates which suggest that up to 5% of the populations in these areas may have died during the first months of the epidemic. Rural areas are generally thought to show lower attack rates than urban and more crowded areas. For example, MSF generally projects attack rates of 0. 2%-1 in rural and 1–5% in urban settings, based on a review of MSF programs in previous cholera epidemics [8]. Our data, as well as others suggest that these estimates should be revised [9,10]. In all districts, CFR were particularly high in elderly people (> 60 years old), in patients with severe diarrhea, those living in remote areas accessible only by foot, and those who did not seek care. In a multivariate analysis, older age, severe diarrhea and not seeking care were independently associated with an increased risk of death. We did not find any association with sex, as reported in other studies, while other risk factor identified elsewhere, such as larger household sizes and being in poor health at onset of disease were not investigated here [11,12]. Not seeking care, in contrast, was reported in all studies with a similar adjusted odds ratio of 5. 4 in Guinea Bissau. Health-seeking behavior was influenced by the type of information received in Zimbabwe, with person-to person communication by village health workers being more efficient than other sources of information such as friends, family, NGOs or radio [12]. Here, distance rather than lack of information seemed to be the main barrier to health-care seeking in remote areas. Accordingly, in a separate multivariate model, remoteness was also independently associated with an increased risk of death, with an adjusted risk ratio of 2. 20. In addition, in this model, the differences between districts became non-significant, while the risk of death remained higher in Pilate than Plaisance in the multivariate analysis including health-seeking behavior. This finding suggests that in Borgne and Pilate the mortality risk for diarrhea patients (which was more than twice that in Plaisance), could be explained mostly by the fact that these districts have more remote areas accessible only by foot. In addition to reducing the proportion of people seeking care, the delay to reach the treatment center probably also influenced the risk of dying as suggested by the higher, though not significantly, risk of dying in patients consulting more than 12 hours after onset of diarrhea in Guinea Bissau [11]. Considering their high vulnerability, it is important to improve response strategies for remote populations. Rapid implementation of ORPs in remote settings might be a good option. Here, only a small proportion (3%) of diarrhea cases reported using them. We did not investigate reasons for this low attendance, but their late implementation certainly did reduce their efficacy. Other factors such as low awareness of their purpose and location or lack of confidence in the quality of care provided could have participated. Early community involvement could probably improve all these aspects [13]. Mass vaccination campaigns have shown good efficacy to prevent cholera cases [14] and these would be particularly relevant in remote areas where other prevention or treatment strategies are difficult to sustain. Finally, these data illustrate the lack of adequate general health system in rural areas of Haiti, as well as in many other low and middle-income countries. Improving general access to care in these areas would probably be the best step towards reducing the high burden of cholera outbreaks as well as other diseases. The main limitation of this retrospective survey may have been recall bias, particularly due to the long recall period. In contrast to mortality data, reporting of diarrhea is highly prone to recall bias in infants, and long recall periods are generally not recommended to assess diarrhea [15,16]. However, diarrhea in adults, particularly severe diarrhea, is rare and thus less prone to recall bias and we believe that it was not a major bias in the reported diarrhea cases. This belief is reinforced by the shape of the epidemic curve obtained through the survey, which is similar to those reported by the national surveillance system (http: //mspp. gouv. ht). However, information bias might have had a more important impact on respondents’ report of the health structures visited and could have led in particular to an under-estimation of the number of visits to an ORP if patients also visited a higher-level health structure. Another limitation of our risk factor analysis was that it was a post-hoc analysis and we did not explore all risk factors, such as access to water and sanitation, socio-economic status, or access to health information. In conclusion, we show here that attack rates and case fatality rates of the first cholera epidemic peak were much higher than reported by the national surveillance system, and that people living in very remote areas in the Nord department were particularly at risk for both disease and death during the early phase of the outbreak. Although an initial response focusing on urban and more densely populated areas was appropriate considering the large number of patients treated, this analysis shows that rural areas with poor access to health care and to cholera prevention and treatment information were at the greatest risk. Adapted strategies to rapidly provide access to preventive activities and treatment in remote communities are urgently needed to prevent this disproportionate impact in future cholera outbreaks.

In October 2010, a large cholera outbreak was declared in Haiti and rapidly spread throughout the country, quickly overwhelming the existing health system. Specialized treatment structures were opened rapidly, generally in cities or large villages, and decentralized treatment units or rehydration points were gradually opened later on. To gain insight into the true burden of the cholera outbreak in the community and on potential geographical differences due to accessibility, we conducted a survey in April–May 2011 in a large rural area across four mountainous districts in the Nord department. We interviewed 3,187 households, corresponding to 16,900 individuals, of whom 2,034 (12%) had had diarrhea, probably cholera, since the beginning of the outbreak. The two most remote districts showed higher proportions of population affected by the disease, up to 16. 2%, and higher proportions of deaths among patients with probable cholera, up to 15. 2%, than the two districts with better accessibility. Remote populations, older patients, severe cases and those not seeking care were at increased risk of dying of the disease. These results show the very high burden of the cholera outbreak in remote areas, emphasizing the need to develop strategies to rapidly provide treatment and prevention measures in remote communities.

Abstract Introduction Methods Results Discussion

2015

Geographic Distribution and Mortality Risk Factors during the Cholera Outbreak in a Rural Region of Haiti, 2010-2011

The manner in which different distributions of synaptic weights onto cortical neurons shape their spiking activity remains open. To characterize a homogeneous neuronal population, we use the master equation for generalized leaky integrate-and-fire neurons with shot-noise synapses. We develop fast semi-analytic numerical methods to solve this equation for either current or conductance synapses, with and without synaptic depression. We show that its solutions match simulations of equivalent neuronal networks better than those of the Fokker-Planck equation and we compute bounds on the network response to non-instantaneous synapses. We apply these methods to study different synaptic weight distributions in feed-forward networks. We characterize the synaptic amplitude distributions using a set of measures, called tail weight numbers, designed to quantify the preponderance of very strong synapses. Even if synaptic amplitude distributions are equated for both the total current and average synaptic weight, distributions with sparse but strong synapses produce higher responses for small inputs, leading to a larger operating range. Furthermore, despite their small number, such synapses enable the network to respond faster and with more stability in the face of external fluctuations. Experiments analyzing the distribution of synaptic weights impinging onto neurons typically observe low-amplitude peaks with only few, large-amplitude excitatory (EPSPs) or inhibitory post-synaptic potentials (IPSPs) [1]–[14]. These have been fitted by lognormal [1], truncated Gaussian [10], [11] or highly skewed non-Gaussian distributions [4], [6], [9], [13]. This raises the question of the functional role of such relatively rare but powerful synaptic inputs. The functional implications of such strong synapses can be very significant [15]. Circuits of the Mind [16] proposes a powerful computational brain architecture (the neuroidal model) to explain the brain' s remarkable flexibility to quickly memorize new events and associate them with previous stored ones. It is based on a very small fraction of powerful excitatory synapses. Furthermore, a recent study [17] of strong but sparse synapses, combined with weak and probabilistic synaptic amplitude distributions provided both computational justification as well as empirical support for the role of these rare yet powerful synaptic events in supporting low-frequency, spontaneous firing in neuronal networks at rest. This question has become more acute following several reports that individual cortical pyramidal neurons from human tissue recovered during surgery are sufficiently powerful to drive other neurons by themselves [18], [19], unlike equivalent cells in rodent cortex. The relation between probabilistic synaptic weight distribution and population dynamics can be studied using simulations [17]. However, the large parameter space to be explored and the need to repeat simulations many times makes this an impractical first method to apply, in particular when modeling the activity in large regions or even the entire mammalian brain. An alternative is the analytical Fokker-Planck method, based on a continuous stochastic process (a brief review of stochastic processes following [20] and [21]–[23] has been provided in Text (S1) ) that models the dynamics of homogeneous neuronal populations with a single partial differential equation (PDE). It is able to quickly explore parameter space and provide analytical insights [24] and characterizes any distribution of synaptic weights by just quantities-the drift and diffusion terms in the equation, corresponding to what the mean and variance of the membrane potential would be if the neuron did not have a threshold. But this method cannot reproduce dynamics in the presence of large synapses. Models based on jump stochastic processes [25], [26] treat synaptic input as being composed of pulses with finite amplitudes. The population dynamics in the presence of such synapses have been investigated using analytical and numerical techniques [27]–[38]. Richardson and Swarbrick [39] characterized analytically an integral formulation in the case of exponentially distributed stochastic jump processes. We here present a fast, semi-analytical approach to study the integral formulation of stochastic jump processes with arbitrary distributions. We apply this method to study the role of synaptic distributions on neuronal population dynamics. The equation for the membrane potential relative to rest for a generalized leaky integrate and fire (gLIF) neuron with normalized capacitance is, (1) is a random variable characterizing the shot-noise synaptic current. It takes the value with probability and with probability. is the input synaptic event rate so that is the mean number of inputs in a time. represents the distribution of synaptic weights, with and and being the minimum and maximum synaptic weights respectively. When the synaptic input is sufficient to cause the membrane potential to exceed a threshold value, it is reset so that (2) This reset implementation is used to account for the shot-noise nature of the synaptic input. For nearly instantaneous synapses, represents the peak of the post-synaptic potential (either EPSPs or IPSPs) (for non-instantaneous synapses, see Methods: Non-instantaneous synapses). represents the sum of all non-synaptic currents, which can be voltage-dependent but not explicitly time-dependent. For the standard LIF neuron, , where is the membrane time constant. For an exponential integrate-and-fire neuron, . is the spike detection threshold and is the slope factor. The resulting equation for the probability of a neuron to have a voltage in at time -the master equation-is an integro-partial differential equation with displacement (DiPDE), (3) with is the unit step function and the threshold membrane potential. The first term on the RHS in equation (3) incorporates the drift due to non-synaptic currents. The second term removes the probability for neurons which previously were at potential and received a synaptic input. The third term adds the probability that a neuron away in potential receives a synaptic input such that its potential is changed to. The last term represents a probability current injection of the neuron which previously spiked. includes the effect of any excess synaptic input above the threshold at which spiking occurs, due to large super-threshold synaptic events. To account for the effect of the excess input with instantaneous synapses, the probability current is injected between the resting potential and, where represents the membrane potential that would be reached due to the excess input. This serves the purpose of ‘remembering’ the excess input, whose effect would have been held by the synaptic variables in the case of slow synapses, without losing it on resetting the membrane potential after spiking. The output firing rate is given by (4) In the Fokker-Planck formalism, is part of a boundary condition. If, then there is no probability current through threshold due to continuous processes for this system. The alternative case is discussed in (Methods: Boundary conditions). Since we include non-infinitesimal synaptic inputs in an infinitesimal time interval, it is possible for a neuron to cross the threshold by a large amount. In equation (3), this additional depolarization is accounted for in the term as outlined above. The case in which this additional depolarization is ignored is treated in (Methods: Boundary conditions). Simpifications obtained with -function distribution of synaptic weights are described in (Methods: Simplifications). It is also possible to include the effect of synaptic delays and distributions of synaptic delays within the DiPDE formalism, as discussed in (Methods: Non-instantaneous synapses). The stationary solution for equation (3) can be obtained as the solution to the following equation, (5) where is the stationary probability distribution for the membrane potential. Although the model can include both excitatory and inhibitory connections, all the results presented below, except for those in Figure S6 c) and d), are for feed-forward networks with excitatory connections. Intuitively, this model quickly diverges from the Fokker-Planck formalism when the synaptic strength is large. Consider the hypothetical case of a neuron starting at rest which has all the synapses equal and large. After a short time step, the Fokker-Planck formalism produces a narrow Gaussian distribution in membrane potential near rest, while the DiPDE formalism yields a membrane potential distribution which is a sum of two scaled delta functions: a large one at rest, and a small one at the synaptic weight. Over time, the Fokker-Planck equation converges to a single broader Gaussian, while the DiPDE formalism leads to a larger coefficient for the delta function at the synaptic weight value. A generalization to conductance-based synapses is presented in (Methods: Conductance-based synapses), while a generalization to the case of exponential integrate-and-fire neurons [40] is presented in (Methods: Exponential integrate-and-fire neurons). We solve equation (3) without the displacement terms using the method of characteristics. The characteristic equations can be solved to obtain a non-uniform discretization of the membrane potential. For the standard LIF neuron, the characteristic equations are solved analytically. We then numerically add the effects of the displacement terms at every time step (see Methods: Numerical Solutions). This semi-analytic technique has the advantage of reducing errors due to numerical diffusion at each time-step. The solution of the DiPDE equation (3) is in good agreement with simulations of 10,000 leaky integrate-and-fire neurons for both low frequency, large amplitude (Figure (1) ) and for high frequency, small amplitude distributions (Figure (S1) ). Differences between continuous and discontinuous stochastic processes can be seen in the transient behavior of the probability distribution of membrane potential in the top panel (Figure 1a, b, c) and are statistically significant. p-values for differences between the sub-threshold steady-state membrane potential distributions have been provided in Text (S4), Table (S2) and Table (S3). The steady state firing rate obtained from DiPDE is Hz and from simulations is Hz. The small discrepancy is caused in part by simulating synapses which are not instantaneous but rather have a time constant of ms (for more details see (Methods: Non-instantaneous synapses) ). Since the stochastic input is Poisson-distributed, expected 95% intervals for spike counts can be directly computed from DiPDE (see Methods: Expected 95% intervals for spike counts and Figure (1f) ). The transient time to firing, defined as the time taken to reach 10% of the equilibrium firing rate, is ms and provides a good estimate of how quickly the neuronal population responds to a given input. By comparison, the Fokker-Planck formalism results in a lower steady state firing rate of Hz and a slower transient time ms. The higher number of neurons closer to resting potential at equilibrium in Figure (1e) reflects the nature of the ‘jump’ stochastic process. For higher-frequency, low-amplitude inputs, all three methods converge to the same results (Text (S2) and Figure (S1) ). Thus, the formalism presented in equation (3) is in good agreement with simulations for instantaneous synaptic input and does not depend on the choice of time-step for the numerical solution (Figure (S2) ). For non-instantaneous synapses, upper and lower bounds on the steady state output firing rates can be obtained (Methods: Non-instantaneous synapses, Figure (S4) and Figure (S5) ). The DiPDE implementation also matches equivalent simulations for conductance-based synapses (Figure (1g), Figure (S3) ) and exponential integrate and fire neurons (Figure (1h) ). We used the DiPDE formalism to investigate the effect of generic synaptic weight distribution on the steady-state, subthreshold membrane potential distribution and output firing rates in feed-forward networks. Gaussian synaptic weight distributions with different mean weights whose input firing rates are adjusted to produce the same average synaptic current, result in different transient and steady-state firing rates as well as different equilibrium voltage distributions (top row in Figure (2), balanced excitation/inhibition Figure (S6) ). We then examined six different distributions tuned to have both the same mean (1 mV) synaptic weight and input rate (1,000 Hz), so that the average synaptic current was the same (see Methods: Matched synaptic distributions for the exact forms of the distributions and their respective variances). We find that the average synaptic current is not sufficient to accurately determine either the output firing rate or membrane voltage distributions (bottom row in Figure (2) ). We also tested distributions that were matched to have input synaptic current with the same mean and variance (Figure (S7) ). The undulating voltage distribution for -function input (Figure (2e) ), reflects the nature of the ‘jump’ stochastic process. Heavier-tailed distributions generate faster transient responses to inputs and monotonically reach steady-state output firing rates. For example, the power-law distribution leads to an equilibrium firing rate Hz and a transient time of ms, while the Gaussian distribution results in Hz and ms. Numerical results for all simulations are provided in Table (1). Even changing a small fraction of the synaptic weights can have a significant effect. For example, the -function distribution converges to the lowest steady state firing rate of Hz (for computation of 95% confidence-intervals, see Table (1) ). In contrast, the bimodal distribution which differs from the -function by only 3. 4% of synapses having a much higher weight, results in the highest steady state firing rate of Hz. The tail of the synaptic weight distribution has an even larger effect on the transient times starting from rest - it is 16. 2 ms for the -function, but only 2. 6 ms for the bimodal distribution. Since the network we study is a feed-forward network, the mean synaptic delay results in a simple time translation of the responses. However, the overshoot of steady state seen in Figure (2f) decreases as the variance in the distribution of synaptic delays increases (see Methods: Synaptic delays and Figure (S8) ) relative to the membrane time constant. To analyze what characteristic of the synaptic distribution is most important for the fast response observed for our heavy-tailed distributions, we generated more than 1000 random distributions matched to have the same mean synaptic weight and input rate (Methods: Tail weight numbers) and analyzed the responsiveness of a neuronal population (time to a fraction of the equilibrium firing rate) as a function of the moments of the synaptic distribution (Table (2) and top row of Figure (S10) ). Higher order moments explain part of the variance observed in the responsiveness. However, a larger fraction of the variance can be explained by introducing a set of measures specifically designed to quantify the number of strong synapses (Table (2) and bottom row of Figure (S10); see also Figure (S9) for our six chosen distributions). These are the tail weight numbers, the density above threshold of convolutions of the effective synaptic weight distribution: (6) where is related to the synaptic weight distribution by, (7) and represents a convolution. represents a Poisson process with mean and events occurring in a time-step. By definition, , , and so on. thus represents the average distribution of depolarization of a single neuron, when each neuron in the population receives excitatory inputs on average. then represents the fraction of neurons that spike in a neuronal population starting at rest when each neuron in the population receives excitatory inputs on average. Having established a measure of population activity when all neurons start at rest, we examined the dynamics resulting from an equilibrium different from rest (see Methods: Input-output curves). Mathematically, this amounts to analyzing the effect of synaptic weight distribution on the transient dynamics from one stationary solution to another stationary solution, when the input synaptic event rate in equation (3) is instantaneously changed by a constant from. The different synaptic distributions result in an overshoot of the population' s eventual equilibrium firing rate. The overshoot becomes smaller for heavier-tailed distributions. For the power-law, there is no overshoot present. The amount of overshoot is a measure of the stability of the neuronal population to sudden changes in input firing, given a distribution of synaptic weights. The weight of the tail in the synaptic distribution correlates with the stability of the system (Figure (3) and Figure (S12) ). The DiPDE formalism also enables insights into the influence of short-term synaptic depression (STSD), which is known to play a key role in neural network homeostasis and in the generation of multiple network states [41] via a Tsodyks-Markram mechanism [42]. With the inclusion of synaptic depression (see Methods: Implementation of short-term synaptic depression), the output firing rates for all six chosen synaptic weight distributions begin to saturate when the effective input synaptic events per second (integrated over all the synapses impinging onto a neuron) are around Hz (Figure (3f) ). Heavier-tailed synaptic weight distributions lead to a higher dynamic range (Table (S4) ). In the preceding analysis, we saw how perturbations of the input event rate in a population with different synaptic weight distributions affect the entire time-course of the population' s evolution from an initial to a final equlibrium. In contrast, we also analyzed how fluctuations in inputs (see Methods: Fluctuation Analysis) affect the instantaneous change in equilibrium firing rate of neuronal populations with different synaptic weight distributions (Figure (4), Figure (S13) and Figure (S14) ). To quantify the instantaneous change in the equilibrium firing rate, we make use of a closely-related variant of the tail weight numbers defined in equation (6). Instead of starting with an initial probability distribution for the membrane potential corresponding to all neurons at rest, we use the stationary probability distribution obtained from the solution to equation (5) with input rate to define, (8) where is as defined in equation (7). thus represents the fraction of neurons that spike in a neuronal population starting from the stationary distribution, when each neuron in the population receives additional inputs on average. This means that, effectively in a time, the total number of inputs becomes. Using, we can define the relative excitability of a neuronal population with a given synaptic weight distribution as (9) This is a measure of susceptibility of the neuronal population. If is the increase in input fluctuations for a population with additional inputs per neuron on average, is the corresponding increase in the output firing rate of this population. A relative excitability implies that the additional instantaneous response to an external input is independent of other inputs received at the same time. Starting from the equilibrium of Figure (2), for all synaptic weight distributions, the relative excitability initially rises (Figure (4a) ). In this case, it can be partially compensated by synaptic depression (Figure (4b) ). Table (S5) lists the maximum relative excitabilites for all the distributions, without and with synaptic depression. Results for an equilibrium obtained with a higher input firing rate have been presented in Figure (S14). Heavier-tailed distributions of synaptic weights cause smaller changes in relative excitability of the neuronal population. A few strong synapses can exert an undue large influence on the mean response of the population. Consider the unimodal -function synaptic distribution, whose weight is set at 1 mV and the bimodal distribution, where 96. 6% of the synapses have 0. 5 mV weight, while 3. 4% of its synapses are much larger at 15 mV (an even more extreme case with the two distributions differing in only 0. 1% of their synapses is presented in Figure (S11) ). Even though both the average synaptic input and the total current are the same for both distributions, their steady-state firing rates are 19. 6 and 28. 7 Hz, their transient times are 16. 2 and 2. 6 ms and their dynamic ranges (see Table (S4) ) are 10. 22 and 31. 83 respectively. More dramatic is the manner in which these two distributions react to synaptic fluctuations. When each neuron in the population receives additional inputs on average the relative excitability for the -function distribution is, while that for the bimodal distribution is (see Figure (4a) ). Our study provides an advance on two fronts, computationally and neurobiologically. First, we developed and validated a semi-analytic method to model the sub-threshold membrane potential probability distribution and the firing rate of homogeneous neuronal populations with finite synaptic inputs. Second, we apply this method to explore the effect of varying synaptic weight distributions on equilibrium and transient population characteristics. From a methodological standpoint, the DiPDE formalism reproduces population behavior from aggregate simulations of identical point neurons, without the need to run the large-scale simulations themselves with the attendant computational costs (see Text (S3) and Table (S1) ). The Fokker-Planck equation works best in the regime where the synaptic distribution and input firing rates approach a continuous process. It does not accurately model the response to even rare high-amplitude (‘jump’) synaptic inputs. By contrast, the ability of the DiPDE formalism to model both jump and continuous processes makes it a powerful framework for modeling single as well as multiple interacting neuronal populations. This population statistics method makes possible the modeling and characterization of the membrane potential distribution and the spiking statistics of very large networks of distinct but homogeneous populations of hundreds of neuronal cell types in numerous brain regions. This would be relevant, for instance, when modeling the resting state activity at the cellular level throughout the awake or the sleeping brain. Second, in order to understand the biological significance of powerful but rare EPSPs [7], [18], [19], we used the DiPDE formalism to compare distributions which contain some very strong synapses against distributions with a preponderance of small EPSPs (but that are matched in their mean and variances). We will refer to the distributions as heavy-tailed, but a more quantitative description is presented in Methods: Tail weight numbers. We showed that power law and related heavy-tail distributions lead to faster transient behavior than non-heavy-tailed distributions (Figure (2f) and Figure (3f) ). Second, heavy-tailed synaptic distributions lead to a higher dynamical range (Figure (3f) ) than non-heavy-tailed distributions. Third, heavy-tailed distributions are much less sensitive to random fluctuations in synaptic activity (Figure (4) ). All three properties associated with such heavy-tailed distributions can be functionally advantageous compared to matched synaptic input with no such large synapses. For example, faster transient responses would be desirable when a potential prey needs to detect the presence of a predator and escape. Higher dynamical ranges would be useful when sensory systems need to respond external stimuli over a large range, without the need to have multiple networks designed to respond over smaller ranges. The lesser sensitivity of heavy-tailed distributions to random fluctuations in synaptic activity can be useful when the response to a stimulus needs to be independent of its context. Teramae and colleagues [17] recently published a joint computational and physiological investigation into the role of strong but sparse excitatory EPSPs, superimposed onto a very large pool of weak EPSPs and discovered the critical role that the former play in generating and sustaining long-term, low-frequency spontaneous firing activity in mixed excitatory-inhibitory neuronal networks in the absence of sustained external input or NMDA synapses. Such states allow the neurons to often reside one synaptic input away from the threshold, thereby leading to high correlation between the pre-synaptic and post-synaptic neurons of the strong synapses. Such investigations highlight the need to focus the attention of electrophysiologists studying synaptic transmission in vivo onto the critical role that such rare, “Black Swan”-like, events can play in the day-to-day life of the brain. Given the large biological and instrumental noise present in synaptic measurements, in particular under in vivo conditions, distinguishing between these distributions in practice would not be easy as it would require recordings of long duration to collect the relevant statistics. For experimental purposes, the numbers of independent recordings of sub-threshold steady state membrane potential required to differentiate between different synaptic weight distributions have been provided in Table (S2) and Table (S3). Yet as shown here, large but rare excitatory synaptic inputs can exert undue influence on population dynamics and robustness. These conclusions are very important for the emerging field of connectomics: a weighted graph description for neuronal networks in which each node represents a homogeneous population and each connection is characterized by the mean postsynaptic current is not sufficient. The network has to be either refined to individual neurons or expanded to include knowledge of the entire distribution of postsynaptic currents. The method further developed here facilitates the modeling and characterization of the membrane potential distribution and the spiking statistics of very large networks of distinct but homogeneous populations of hundreds of neuronal cell types in numerous brain regions. A simplified case can be obtained if all EPSPs are assumed to have the same value; that is, if. Equation (3) becomes: (10) This is a first order partial differential equation with displacement (DiPDE). For numerical solutions of equation (10), the matrix obtained by discretizing is as sparse as the matrix needed to solve the Fokker-Planck equation. If, then equation (5) reduces to a delay differential equation (DDE) (11) In equation (3), if the additional depolarization after the neuron crossed threshold is neglected, the equation becomes (12) where is the two-dimensional Dirac -function and corresponds to the resting potential. If, then there is an additional contribution from the continuous processes to the probability flux through threshold. The output firing rate is now given by, (13) Numerical simulations, against which the Fokker-Planck (for current-based synapses) and DiPDE formalisms were compared, were performed by invoking the NEST simulator after writing the code in PyNN. For current-based synapses, simulations were performed for a population of independent and identical leaky integrate-and-fire (LIF) neurons with decaying exponential post-synaptic current. With the resting and reset potential being equal and denoted by, the neuron parameters were chosen to be mV, mV, membrane time constant ms, membrane capacity nF, refractory period ms, input resistance M and decay time ms for excitatory synapses. All neurons in the population were supplied with a Hz Poisson excitatory input of amplitude nA (although this is a huge current, with our choice of parameters, the effective charge deposited on the neuron is given by pC) and the membrane potential was recorded over a simulation time of ms with a time-step of ms. The recorded voltages were then analyzed to obtain the probability distributions shown in Figure (1). The recorded spike-times were binned with an interval of ms and the resultant output firing rate was obtained as shown in Figure (1d). For simulation results presented in Figure (S1), we used Hz Poisson excitatory input of amplitude nA. For conductance-based synapses, simulations were performed for a population of independent and identical leaky integrate-and-fire (LIF) neurons with decaying exponential post-synaptic conductance. The neuron parameters were chosen to be mV, mV, excitatory reversal potential mV, membrane time constant ms, membrane capacity nF, refractory period ms and decay time ms for excitatory synapses. All neurons in the population were excited by external Poisson input with input firing rate Hz and peak synaptic conductance µS and their membrane potential was recorded over a simulation time of ms with a time-step of ms. The recorded voltages were then analyzed to obtain the probability distributions shown in Figure (S3). The recorded spike-times were binned with an interval of ms and the resultant output firing rate was obtained as shown in Figure (1g) and Figure (S3c). Simulations were also performed for a population of independent and identical exponential integrate-and-fire (EIF) neurons with conductance-based synapses without adaptation. The neuron parameters were chosen to be mV, mV, mV, slope-factor mV, excitatory reversal potential mV, membrane time constant ms, membrane capacity nF, refractory period ms and decay time ms for excitatory synapses. The adaptation parameters were chosen to be ms, and, such that adaptation was absent. All neurons in the population were excited by external Poisson input with input firing rate Hz and peak synaptic conductance µS and their membrane potential was recorded over a simulation time of ms with a time-step of ms. The recorded spike-times were binned with an interval of ms and the resultant output firing rate was obtained as shown in Figure (1h). To solve the evolution equation (3) for the probability density, we first solve the advection (leak) portion of the equation, and then include the synaptic terms to calculate the overall time derivative. We then march the solution for forward in time using an explicit first-order-in-time scheme with constant time step. Based on the general explicit scheme criterion, we use a sufficiently small time step to ensure stability. This scheme is conservative; the integral of the probability distribution is not affected by numerical errors. For the standard LIF neuron with membrane time-constant, the leak term is linear with. The ODE has an analytical solution and the method of characteristics provides the full solution for the advection term. In order to ensure an exact implementation of the leak term, we use a geometric binning scheme for the membrane potential, with the bin ratio determined by the product of leak and time-step. The bin-edges are determined, starting from threshold membrane potential, using (14) A total of bin-edges are generated until the first bin between the resting potential and is at least as small or smaller than the first bin generated from. Mathematically, the lower bound on the number of bin-edges is given by the condition (15) since we have chosen and the second equality follows from equation (14). Increasing the numbers of bins beyond this lower bound increases the accuracy, but also increases the computational cost involved (see Text (S3) ). For more general forms of the leak term (for e. g, the exponential integrate-and-fire (EIF) neuron; see Methods: Exponential integrate-and-fire neurons) the ODE can be solved numerically starting from the threshold membrane potential in order to generate the bins. At each time-step, the probability distribution for membrane potential (which is a vector) is first evolved with the leak term and then the synaptic input. With our non-uniform binning scheme, the evolution of the probability distribution for the membrane potential due to the leak reduces to a single-index shift towards the resting potential and reduces the error due to numerical diffusion at each time step. To implement the effect of instantaneous synaptic input, including the effect of any excess input above the threshold the input distribution of synaptic weights is first used to construct a transition matrix. The additional columns per row in are used to keep track of the effect of excess synaptic input above. If this additional depolarization due to super-threshold inputs is ignored and the membrane potential is reset to zero upon exceeding, then is just a matrix. This is then used to construct an effective transition matrix (16) where (17) represents the probability for the neuronal population to receive inputs in a time from an external homogeneous Poisson process with mean input rate. The Poisson process is truncated at a sufficiently high value of such that. For inhomogeneous Poisson inputs, an effective transition matrix would have to computed at each time step, depending on the input rate. Each off-diagonal element in the effective transition matrix represents the proportion of neurons in the population receiving a specific synaptic input. The probability distribution for the membrane potential at each time-step is then multiplied by this transition matrix to generate the new probability distribution after synaptic input. The case for non-instantaneous synapses has been discussed below in Methods: Non-instantaneous synapses. In order to compare results with simulations, we need to match the total synaptic input to the neuronal population with that from simulations. The DiPDE formulation is exact when implementing instantaneous synapses. The synaptic weight is equated with the maximum depolarization of membrane potential (peak EPSP) achieved after input from an exponentially decaying current-based synapse, (18) This value is reached after a time (19) For our choice of simulation parameters outlined above mV provides good agreement with simulations as shown in Figure (1). The time-step was chosen to be ms to match with simulations. Another constraint imposed by the nature of the numerical solution is that the discretization of the membrane potential means that it is impossible to have the initial probability distribution to be a sharp -function at mV. The initial probability distribution is therefore spread uniformly over the width of the first voltage bin in both DiPDE and simulations. The underlying stochastic process in equation (3) assumes Poisson-distributed inputs. If an equivalent numerical simulation with input firing rate involves a finite population of neurons and the resultant spikes are binned in time intervals, then there are inputs per bin, with and. The relative variance in the number of inputs per bin is. Expected 95% intervals for output spike counts can then be obtained as follows. At each time step, the solution obtained from equation (3) for a given distribution of synaptic weights with a given input rate is additionally subjected to input rates and the corresponding output rates are calculated. These give the expected 95% intervals to the mean output firing rate. The 95% intervals for the expected spike counts per time-bin for a population neurons are in the interval between. Figure (1f) shows the output firing rates corresponding to the expected 95% intervals for results presented in Figure (1d). For a population of neurons and a bin size of ms, the 95% interval for expected output spike counts per bin obtained from DiPDE is (90. 3,99. 6) with corresponding output firing rates of Hz and Hz respectively. For simulations, the 95% interval for spike counts corresponds to (90. 5,104. 5). For non-instantaneous synapses, upper and lower bounds on the output firing rate can be obtained. The exponentially decaying current-based synapses used here in simulations (see (Materials and Methods: Current-based synapses) ) result in an exact EPSP given by for synaptic input at time. is the unit step function. If is the total charge deposited by an instantaneous, excitatory current-based synapse on a neuron with normalized capacitance at time, then the corresponding EPSP is given by, is the synaptic weight used in equation (3). Setting corresponds to equating the total charge, while setting corresponds to equating the maximum depolarization. Figure (S4) shows the output firing rates obtained from DiPDE after equating the total charge, the maximum depolarization, an intermediate estimate obtained by setting and that obtained by setting, along with the corresponding numerical simulation. A lower bound on the output firing rate can obtained by equating the maximum depolarization, with corresponding to. Equating the total charge with corresponding to provides an estimate of the output firing rate. An upper bound can be obtained by taking with corresponding to. Figure (S5) shows a plot of the different EPSPs obtained with mV, ms and ms for synaptic input at time. For, a formal proof for the upper and lower bounds can be provided as follows. Let, and represent the EPSPs due to synaptic input at time. The maximum depolarization attained due to inputs can be represented as: (20) For any, . Therefore, for all times after the synaptic input, so that. Hence, using in equation (3) an upper bound for the output firing rate is obtained. For the lower bound, consider the difference between and. For all, since, . For, . Using this, it is straightforward to show that when. Thus for all times and hence. The time translation by does not affect the result since' s have been defined to be the maximum depolarization over all times. For current-based synapses, the change in membrane potential resulting from synaptic input is independent of the initial membrane potential. For a conductance-based synapse however, this change is proportional to the difference between the initial membrane potential and the reversal potential for the channels present in a particular synapse. Equation (3) becomes, (21) Because of the additional dependence on the membrane voltage and synaptic reversal potential, modeling an instantaneous change in voltage through a conductance-based synapse is not as straightforward as with a current-based synapse. To model this type of synapse efficiently, we introduce the non-dimensional quantity as in equation (21) above, which represents the instantaneous voltage change due to synaptic activation as a fraction of the charge needed to reach the reversal potential from a given membrane potential. For example, indicates that a single synaptic event would shift the neuron from its current membrane potential to halfway between the current voltage and the synaptic reversal potential. Mathematically, if synaptic events have already occurred, then the change in membrane potential induced by the -th synaptic input can be represented by, (22) For such synaptic events, it can be verified that this reduces to: (23) This non-dimensionalization of synaptic weight for conductance-based synapses thus automatically rescales all weights to lie between zero and 1, and provides a simple way to model these synapses as generating instantaneous changes in membrane voltage. For our choice of simulation parameters, the maximum depolarization achieved by a neuron starting from rest due to synaptic input with and mV is mV. This is obtained from the numerical solution of the equation for a LIF neuron with exponentially decaying conductance-based synapses (with time-step chosen to be 0. 1 ms to match with simulations). Figure (S3) shows that numerical simulations (see Methods: Simulations for parameters used in simulations) are in good agreement with numerical solutions of DiPDE for conductance-based synapses. The equilibrium output firing rate is Hz and the transient time to firing is ms. For neurons and a bin size of ms, the 95% confidence interval for the spike counts per bin is (43. 0,47. 6) corresponding to output firing rates of Hz and Hz. For multiple neuronal populations connected to each other, one can generalize equation (21) to a system of equations, (24) while defining two synaptic weight matrices, The DiPDE formalism (equation 3) can be generalized to other types of neurons such as for e. g - the exponential integrate-and-fire (EIF) neuron. The membrane potential dynamics for the EIF neuron with excitatory conductance-based synapses is governed by (25) where the first two terms on the RHS contribute to the leak and the third term corresponds to the synaptic input. represents the leak reversal potential, is the spike detection threshold, is the slope-factor, is the membrane capacitance, is the membrane time-constant and is the excitatory reversal potential. In the absence of synaptic input, the leak portion can be solved numerically to obtain the discretized non-uniform bins for the membrane potential. For the exponential integrate-and-fire (EIF) neuron, the numerical solution of the equation with exponentially decaying conductance-based synapses gives. With mV, this means that the maximum depolarization achieved by a neuron starting from rest is mV (with time-step chosen to be 0. 05 ms to match with simulations). Figure (1h) shows that the output firing rate obtained from numerical simulations (see Methods: Simulations for parameters used in simulations) is in good agreement with that obtained from the numerical solution of DiPDE. The equilibrium output firing rate is Hz and the transient time to firing is ms. For neurons and a bin size of ms, the 95% confidence interval for the spike counts per bin is (48. 5,53. 1) corresponding to output firing rates of Hz and Hz. The Fokker-Planck equation for current-based synapses was solved using an explicit Forward-Time Centered-Space (FTCS) scheme. As in the DiPDE, the leak term was solved analytically, although the voltage discretization was kept uniform in order to be compatible with the standard Centered-Space discretization used for the diffusion term. To ensure that the discretization was sufficiently fine (especially since FTCS is only first-order accurate), we compared the equilibrium firing rate generated by this scheme against the analytical solution [43] given by (26) where is the analytical firing rate, is the refractory period and is the membrane time constant. and are parameters related to the drift and diffusion coefficients in the Fokker-Planck equation, and are given by (27) where and are excitatory and inhibitory synaptic weights respectively, while and are the corresponding input firing rates. The effect of various matched synaptic weight distributions on the population dynamics in feed-forward networks was investigated using the DiPDE formalism. The top panel of Figure (2) showed that two self-similar Gaussian distributions matched to produce the same average synaptic current resulted in different steady-state output firing rates and sub-threshold membrane potential distributions. In the regime when the firing is primarily driven by drift, the difference in output firing rates is small as shown in the top panel of Figure (S6). The higher-mean Gaussians result in an steady state output firing rate of Hz and Hz, while the lower mean Gaussians lead to Hz and Hz. If the firing is driven exclusively by variations in input (Gaussian distributions with balanced excitation and inhibition), the differences in output firing rates are large as shown in the bottom panel of Figure (S6). For the Gaussians with 6 mV mean and either 1 mV or 2 mV standard deviations, the steady state output firing rates are Hz and Hz respectively. The steady state output firing rates for the Gaussians with 3 mV mean and either 0. 5 mV or 1 mV standard deviations are Hz and Hz respectively. These results imply that population response is determined not only by the total current, but also by the mean synaptic weights. For the bottom row of Figure (2), six different synaptic weight distributions were tuned to have the same mean (1 mV). Using the same input rate (1,000 Hz) for the all the distributions, the average synaptic current was the same. The corresponding expressions for these distributions normalized between and are: The' s represent the normalization constants. The corresponding zero-centered second moments are: Delta (1. 0 mV2), Gaussian (1. 3 mV2), Exponential (1. 8 mV2), Lognormal (2. 2 mV2), Bimodal (8. 0 mV2) and Power-law (8. 7 mV2). The firing rates and transient times corresponding to distributions for which the mean input current was matched (Figure (2) is provided in Table (1). An alternate way to match synaptic weight distributions is to match the drift and diffusion, corresponding to what the mean and variance of the membrane potential would be if the neuron did not have a threshold, while allowing the input rates to vary. In the Fokker-Planck formalism, such distributions matched for drift and diffusion would have led to the same results. Figure (S7) shows the differences in output firing rates and equilibrium membrane potential distributions obtained from eq. (3) for synaptic weight distributions matched for drift and diffusion. The input rates are: Delta (555. 5 Hz), Gaussian (745. 98 Hz), Exponential (952. 38 Hz), Lognormal (1,062. 15 Hz), inverse power-law (4,166. 72 Hz) and Bimodal (1,825. 33 Hz). The corresponding expressions for these distributions normalized between and are: The' s represent the normalization constants. The heavier-tailed distributions still lead to faster transients, however the steady-state firing rates now decrease with increasing heaviness of the tail in the distributions. This is in contrast to the case when the input currents were matched (Figure (2). In order of increasing heaviness of the tail distributions, the equilibrium output firing rates and the transient firing times for the different distributions are provided in Table (1). We also implement synaptic delays within the DiPDE formalism. This is done by a using a queue to store the output firing rate, which is then accessed and updated depending on the distribution of synaptic delays. Figure (S8) shows the effect of synaptic delays on the output firing rate of a feed-forward network with different synaptic weight distributions in the presence of -function and Gaussian distributions of synaptic delays with the same mean delay. Since the network is feed-forward, the mean delay simply translates the responses in time. Higher variance in the distribution of delays relative to the membrane time constant leads to lesser overshoot of the steady state. The tail-heaviness of the synaptic weight distribution can be characterized by either constructing the moments of the distribution or a set of tail weight numbers. These numbers were defined in equations (6) and (7) and represent the fraction of neurons that spike in a neuronal population starting at rest when each neuron in the population receives excitatory inputs on average. While higher order moments of the synaptic weight distributions explain part of the variance observed in the responsiveness, a larger fraction of the variance can be explained by the tail weight numbers. Figure (S9) shows the values of the tail weight numbers as a function of (the average synaptic weight) for distributions matched for mean input current (top left) and for drift and diffusion (top right). These numbers start increasing much quicker for heavier-tailed distributions as is evident from the lower panel of Figure (S9). To test which key characteristic of synaptic weight distributions better describes the transient times, we generated 1222 random distributions between 0 and, each with a mean of 2 mV. With an input firing rate of 500 Hz, we computed the output firing rates obtained from DiPDE and the corresponding transient times taken to reach different fractions of the steady state firing rate (time taken to reach % of equilibrium firing rate denoted as here) for all these matched distributions. Figure (S10) shows a scatter plot of as a function of the first few moments and response numbers respectively for all these distributions along with the best-fit exponential curves (shown in red). Scatter plots as a function of tail weight numbers provide a better fit to the transient times than the moments. This can be seen from Table (2), which lists the sum of squared errors of the residuals for each of the best fit exponentials for different' s as a function of the first few moments and tail weight numbers respectively. Plots for the other look similar to the plots for shown in Figure (S10). Input-output curves in Figure (3) and Figure (S12) were generated using the following protocol: for a given fraction and base input rate, the DiPDE was solved first with an input firing rate. After 500 ms (at which time all runs had reached an equilibrium voltage distribution), the input rate was instantaneously changed to. The resulting output firing rate was calculated as the maximum output firing rate achieved after the instantaneous input rate change. We quantified the spread in the input-output relationship due to different values of the fraction for a given base input rate by measuring the slope, defined as follows: (28) The most comprehensive treatment of synaptic utility would involve a single 3-D integro-differential equation where the probability distribution depends additionally on the synaptic utility. However, since this formalism is computationally more expensive and given that the two separate 2-D integro-partial differential equations provide adequate results in cases where the full joint probability can be expressed as a product of the marginals over the voltage and synaptic utility, we opted to model synaptic utility as a separate differential equation via a Tsodyks-Markram mechanism [42]. This additional equation describes the probability of a synapse to have a particular utility. The synaptic utility ranges from 0 to 1, with 1 indicating maximum availability of synaptic vesicles. Its recovery is governed by (29) In the pre-synaptic population, for all probability at the membrane potential threshold, the synaptic utility is decremented using, (30) where thus represents the fraction of synaptic vesicles available after a spike. For our analysis, we use and ms respectively. Given the exponential form of equation (29), the synaptic utility axis (ranging from 0 to 1) is also discretized geometrically, similar to the voltage discretization for linear leak (equation (14) ). Thus, at each time step, the pre-synaptic probability matrix undergoes a single-index shift towards full synaptic utility (). When a non-zero proportion of the probability for a given pre-synaptic neuronal population is at threshold (i. e. there is spiking in the population), the output synaptic weight distribution from that population to the target population is convolved with the synaptic utility distribution, (31) and the utilities are subsequently modified according to equation (30). To analyze the effect of fluctuations generated by different synaptic distributions, the DiPDE numerical solution was evolved with a given synaptic distribution for 300 ms, long enough for the voltage distribution to reach equilibrium. At this point, to simulate fluctuations which occur on timescales, extra synaptic inputs are added per neuron on average to the population in a single time step. This is done by convolving the stationary probability distribution obtained from a given external input rate with the corresponding synaptic weight distribution and the instantaneous change from the equilibrium firing rate is quantified using equation (9). This same analysis was then conducted with synaptic depression during the fluctuation stage, starting with the stationary distribution obtained without synaptic depression. Unlike in the no-depression case, the synaptic distribution was scaled by the synaptic utility (using the method outlined in (Methods: Implementation of short-term synaptic depression) ) after each additional synaptic event. The utility did not recover (). The utility fraction, which represents the decrement in pre-synaptic utility after an input (equation (30) ), was calculated for each synaptic distribution separately. Denoting the peak of the relative excitability curve generated from the no-depression case by after additional inputs per neuron on average, we chose the decrement to be: (32) Intuitively, with this choice, after additional inputs per neuron on average, the relative excitability in the presence of synaptic depression gets closer to unity. All the results for DiPDE simulations were obtained using MATLAB. The entire code base is available for download at http: //download. alleninstitute. org/publications/the_influence_of_synaptic_weight_distribution_on_neuronal_population_dynamics/.

Neurons communicate via action potentials. Typically, depolarizations caused by presynaptic firing are small, such that many synaptic inputs are necessary to exceed the firing threshold. This is the assumption made by standard mathematical approaches such as the Fokker-Planck formalism. However, in some cases the synaptic weight can be large. On occasion, a single input is capable of exceeding threshold. Although this phenomenon can be studied with computational simulations, these can be impractical for large scale brain simulations or suffer from the problem of insufficient knowledge of the relevant parameters. Improving upon the standard Fokker-Planck approach, we develop a hybrid approach combining semi-analytical with computational methods into an efficient technique for analyzing the effect that rare and large synaptic weights can have on neural network activity. Our method has both neurobiological as well as methodological implications. Sparse but powerful synapses provide networks with response celerity, enhanced bandwidth and stability, even when the networks are matched for average input. We introduce a measure characterizing this response. Furthermore, our method can characterize the sub-threshold membrane potential distribution and spiking statistics of very large networks of distinct but homogeneous populations of 10s to 100s of distinct neuronal cell types throughout the brain.

Abstract Introduction Results Discussion Methods

2013

The Influence of Synaptic Weight Distribution on Neuronal Population Dynamics

Metabolic manipulation of host cells by intracellular pathogens is currently recognized to play an important role in the pathology of infection. Nevertheless, little information is available regarding mitochondrial energy metabolism in Leishmania infected macrophages. Here, we demonstrate that during L. infantum infection, macrophages switch from an early glycolytic metabolism to an oxidative phosphorylation, and this metabolic deviation requires SIRT1 and LKB1/AMPK. SIRT1 or LBK1 deficient macrophages infected with L. infantum failed to activate AMPK and up-regulate its targets such as Slc2a4 and Ppargc1a, which are essential for parasite growth. As a result, impairment of metabolic switch caused by SIRT1 or AMPK deficiency reduces parasite load in vitro and in vivo. Overall, our work demonstrates the importance of SIRT1 and AMPK energetic sensors for parasite intracellular survival and proliferation, highlighting the modulation of these proteins as potential therapeutic targets for the treatment of leishmaniasis. Visceral leishmaniasis (VL) is a potentially fatal vector-borne disease caused by protozoan Leishmania donovani and L. infantum parasites. Infection of the mammalian host is initiated with the inoculation of the flagellated promastigote forms during the sand fly bloodmeal. Once inside the host, Leishmania parasites are phagocyted mainly by macrophages, where they reside inside the phagolysosomal compartment and differentiate into obligate intracellular amastigotes. The interplay between parasite factors and host responses is crucial for the final outcome of infection, thereby for disease pathogenesis [1]. Extensive studies have focused on the characterization of Leishmania virulence factors and the strategies developed by the parasite to manipulate host intracellular signaling pathways towards immune evasion and survival [2,3]. Yet, scarce attention has been paid to the manipulation of host nutrient and energy sources by Leishmania parasites despite the competition of both organisms for identical resources. Mitochondrion plays a crucial role during apoptotic cell death [4], is the site of ATP synthesis and is where essential metabolic pathways take place. These include the citric acid cycle, fatty acid oxidation, the synthesis and degradation of amino acids and the synthesis of iron–sulfur clusters and heme. Mitochondria are dynamic compartments that rearrange in response to stress and changes in nutrient availability or oxygen concentration. Metabolic reprogramming of cells is an integral view of cell metabolism in order to satisfy cell proliferation and survival requirements. Despite its recognized importance in disease pathogenesis, there is limited understanding of various aspects of mitochondrial bioenergetics in the context of host-pathogen interactions. From a bioenergetic and metabolic perspective, intracellular pathogens may benefit from existing resources and can manipulate the host for their own profit to fulfill their requirements. The intracellular growth of Trypanosoma cruzi was recently shown to rely predominantly on the energy production, nucleotide metabolism and fatty acid oxidation of the host [5]. Although the metabolic manipulation of host cells is recognized to play an important role in the pathologic processes of infection [6], the puzzle is even more complex when the invading pathogen shares many metabolic pathways with the host, a frequent case for protozoan infections. AMP-activated protein kinase (AMPK), a central cellular signaling hub involved in the regulation of energy homeostasis, has been suggested as a potential attractive target for pathogen manipulation [6]. As a paradigm, hijacking of AMPK pathway by viruses has been documented [7], with its activation or inhibition being strictly dependent on the species involved but invariably satisfying viral interests. From a metabolic point of view, AMPK protein complex is activated upon changes in the AMP/ATP ratio mirroring shortage of nutrients [8]. In addition to allosteric regulation by AMP, AMPK activation is controlled by phosphorylation on conserved threonine 172 (Thr172) residue namely by two upstream kinases, the tumor suppressor kinase LKB1 (liver kinase B1) [9,10] or the calcium/calmodulin-dependent protein kinase kinases, whose activity is dependent on Ca2+ levels [11,12]. As a final output, AMPK activation tips the energetic balance by switching on catabolic pathways. Akin, sirtuin proteins deacetylate key targets in response to intracellular NAD+ levels fluctuations, playing a role as cellular energetic sensors [13]. In particular, the partnership between SIRT1 and AMPK in mediating the cellular response to nutrient availability has been described [14]. Nevertheless, the promiscuous relation of AMPK and SIRT1 appears to be cell specific and their putative role during infection remains elusive. In this work, we dissected the macrophage metabolic pathways engaged by Leishmania parasites and elucidated the consequences of this metabolic hijacking for parasite growth and persistence. Bioenergetic flux analysis of L. infantum infected macrophages revealed a two-step infection process, including an initial transient aerobic glycolytic phase followed by a metabolic shift towards mitochondrial metabolism. This metabolic switch requires the catalytic activities of SIRT1 and LKB1 as well as the downstream AMPK energetic sensor. While allowing the metabolic recovery of the host cell, the activation of the SIRT1/AMPK axis ultimately contributes to parasite survival in vitro and in vivo. Our findings point out that the macrophage metabolism could be a potential therapeutic target against leishmaniasis diseases. To address how Leishmania infantum infection impacts on macrophage metabolism and bioenergetic state, we quantified the extracellular acidification rate (ECAR), a consequence of lactate production, and the mitochondrial oxygen consumption rate (OCR), to monitor the rate of oxidative phosphorylation, using live cell extracellular flux analysis. Six hours post-infection, higher basal ECAR levels were observed in infected bone marrow derived macrophages (BMMo) compared to uninfected cells (Fig. 1A and S1A Fig.), corresponding to a higher glycolytic capacity (Fig. 1B and S1A Fig.). At the same time point, a significant reduction of the OCR values was observed on infected BMMo (Fig. 1A and S1B Fig.). Accordingly, the spare respiratory capacity (SRC), as calculated by the difference between the maximal OCR determined in the presence of the uncoupler FCCP and basal value was significantly reduced on infected BMMo indicating lower levels of mitochondrial respiration (Fig. 1B and S1B Fig.). In contrast, this pattern was altered after 18 hours post-infection (p. i.), where higher respiration levels were observed in infected BMMo as evidenced by increased SRC (Fig. 1B and S1B Fig.). Concomitantly to the increase in OCR values, ECAR was reduced at 18 hours p. i. in infected BMMo (Fig. 1A and S1A Fig.). These variations were reflected by a significant variation on the OCR/ECAR ratio (Fig. 1C). To understand if live intracellular parasites by themselves contribute to increase the glycolytic metabolism, BMMo were challenged with irradiated L. infantum promastigotes. Irradiated promastigotes maintain membrane integrity for 6 hours as shown by 7-Aminoactinomycin D (7-AAD) staining (S2A-B Fig.) and were phagocyted in a similar manner as live parasites (S2C Fig.). Intra or extracellularly, the irradiated promastigotes were unable to grow; in fact a dramatic increase in membrane permeabilization associated with cell death was observed when measured 18 hours post-irradiation (S2A-D Fig.). Our results showed no difference in the ECAR parameter between uninfected and irradiated L. infantum BMMo (Fig. 1A-C and S1A-B Fig.), while the levels of OCR and SRC were always lower in the latter, yet without affecting the OCR/ECAR ratio at 18 hours p. i. These results indicate that the observed phenotype is mostly due to the manipulation of the host by parasite and not caused by the presence of the parasite. Altogether, our results demonstrated that infection of BMMo with live L. infantum is associated with a transient bioenergetic profile towards aerobic glycolysis with a concomitant reduction of mitochondria function early after infection followed by a metabolic shift towards mitochondrial metabolism. Glycolysis, despite its lower energetic efficiency as compared with to oxidative phosphorylation, is far quicker than oxidation of pyruvate in the mitochondria. Pyruvate derived from glycolysis is either reduced to lactate or enters into the tricarboxylic acid (TCA) cycle in processes involving several kinases. We assessed the expression of hexokinase 1 (Hk1), hexokinase 2 (Hk2), pyruvate kinase M1 (Pkm1), phosphofructokinase (Pfk), pyruvate dehydrogenase kinase 1 (Pdk1), pyruvate kinase M2 (Pkm2) and lactate dehydrogenase a (Ldha) transcripts (S3A Fig.). The transcripts were analyzed in BMMo challenged with live or irradiated L. infantum to ascertain the mechanisms responsible for the bioenergetic profile observed. In the first hours of infection (2 and 6 hours), live parasites induce a significant increase of Pfk, Pdk1, Pkm2 and Ldha transcripts as compared to either uninfected cells or challenged with irradiated promastigotes (Fig. 2A). All up-regulated genes returned to their initial levels approximately at 10–14h p. i. No differences were found in the transcripts of Hk1, Hk2 and Pkm1 glycolytic enzymes (S3B Fig.). Further experiments performed in sorted infected and bystander BMMo demonstrated that the increase of Pfk, Pdk1 and Ldha transcripts was specific to infected cells (Fig. 2B), with the exception of Pkm2 that is equally transcribed in both populations. Importantly, the raise of Ldha transcript paralleled the increase of LDH enzymatic activity and concomitant lactate secretion (Fig. 2C). The LDH enzymatic activity or lactate secretion was not due to the presence of intracellular parasites since only live but not irradiated promastigotes displayed the referred phenotype (Fig. 2C). In order to determine whether the in vitro profile observed is reminiscent to in vivo situation, macrophages sorted from the spleens of mice infected with CFSE-labelled L. infantum were analyzed. As before, an increase of glycolytic genes was observed early after infection (Fig. 2D and S3C). In order to evaluate the impact of the differentiation process from promastigotes to amastigotes, we infected BMMo with axenic amastigotes. Unlike promastigote-challenged macrophages, infection with axenic amastigotes was not associated with increased LDH activity or lactate secretion (S3E Fig.). In agreement, amastigote infection failed to upregulate the mRNA levels of glycolytic enzymes as compared with uninfected macrophages, with the exception of Pdk1, a regulator of the activity of the pyruvate dehydrogenase complex (S3D Fig.). Altogether, these results demonstrated that the burst of aerobic glycolysis is associated with early upregulated expression of key glycolytic kinases. To gain insight into the regulation of mitochondrial adaptations during the late phase of macrophage infection by L. infantum, we analyzed the peroxisome proliferator-activated receptors (PPARs), in particular the levels of PPAR-γ coactivator-1α (PGC-1α) and PPAR-γ coactivator-1β (PGC-1β), known to induce oxidative metabolism and mitochondrial biogenesis [15]. BMMo infected with live, but not with irradiated parasites, showed increased levels of Ppargc1a, but not Ppargc1b, transcripts peaking at 18 hours (Fig. 3A). Consistent with the transcriptional analysis, only BMMo infected with viable parasites presented higher PGC-1α protein levels at 18 and 24 hours (Fig. 3B). The increase of Ppargc1a transcripts was also observed in splenic macrophages recovered from mice at 48 hours post-infection (Fig. 3D). Accompanying the increase PGC-1α levels, we detected increased mitochondrial biogenesis in infected BMMo, as shown by an increment of mitochondrial DNA/nuclear DNA ratio upon 14 hours post-infection (Fig. 3E), and of nuclear genes encoding for mitochondrial complexes namely, Ndufa9 (complex I) and Cox4 (complex IV) both in in vitro infected BMMo or macrophages recovered from the spleen of infected mice (S4A-S4C Fig.). As found with promastigotes, amastigotes infected cells displayed an increase of the transcription levels of Ppargc1a (S4D Fig.), and higher levels of mitochondrial DNA/nuclear DNA ratio (S4E Fig.). In conclusion, the metabolic regimen towards oxidative phosphorylation is a common mechanism for both L. infantum forms. Although both glycolysis and oxidative phosphorylation produce ATP, the energetic yield in the latter is higher. Therefore, we quantified the intracellular pools of ATP and AMP throughout the infection. Surprisingly, a significant increase in the AMP/ATP ratio at 10 and 14 hours post-infection was observed (Fig. 4A), in agreement with the variations in the total content of AMP and ATP (S5A-S5B Fig.). The variation on the nucleotide pool was dependent not only on the presence of live parasites, as no significant difference was observed between uninfected BMMo and those exposed to irradiated parasites (S5A-S5B Fig.), but also on the parasite dose (S5C Fig.). ATP reduction was exclusively observed in infected cells (S5D Fig.). Furthermore, amastigote-infected cells displayed a drop in the total ATP pool, although to a lower extent than macrophages infected with promastigotes (S5E Fig.). A central metabolic sensor in response to nutrient and energetic restriction is AMPK (Austin and St-Pierre, 2012). Concomitantly with the increase on AMP/ATP ratio, AMPK is phosphorylated at Thr 172 (AMPK-P-Thr172) in cells infected with live promastigotes (Fig. 4B and 4C) or amastigotes (S5F Fig.). AMPK activation by phosphorylation is mediated by the upstream kinase LKB1 as Thr 172 is not phosphorylated in infected LKB1 KO BMMo (Fig. 4D and S5G Fig.). Aside PGC-1α, AMPK activation is also associated with induced expression of glucose transporters (GLUTs) encoded by Slc2a genes. Slc2a4, but not Slc2a1 (Fig. 4E) or Slc2a2–3 (S5H Fig.) transcripts were upregulated in cells infected with live parasites. An increase of Slc2a4 and Slc2a1 transcripts in splenic macrophages were also upregulated (S5I Fig.) which rules out a potential in vitro artifact. In opposition, irradiated parasites were unable to induce the expression of Slc2a4 (S5J Fig.). To gain insight on the role of the LKB1/AMPK axis, we infected BMMo devoid of AMPK or LKB1 with L. infantum. Our data demonstrated that in infected LKB1 KO BMMo, AMPK is not phosphorylated (Fig. 4D). Upon Leishmania infection of AMPK or LKB1 KO BMMo, both Slc2a4 and Ppargc1a genes were not induced demonstrating that LKB1/AMPK signaling pathway controls their transcript levels in the context of Leishmania infection (Fig. 4F). As such, we next assessed whether AMPK impact on glucose uptake. In contrast to WT BMMo, infected AMPK KO BMMo failed to increment glucose uptake (Fig. 4G and S6A Fig.). Infected AMPK KO BMMo presented a higher fold increase in ECAR values than for WT BMMo when compared to uninfected cells at 18 hours p. i. (Fig. 4H and S6B Fig.). Importantly, no significant change on OCR measurement and SRC was noticed in infected vs. uninfected AMPK KO BMMo in comparison to WT cells (Fig. 4H-I). Consequently, infected AMPK KO BMMo presented a lower OCR/ECAR ratio compared to WT cells (S6C Fig.). Measurements at real time of the bioenergetic profile corroborated the higher aerobic glycolytic flux and unchanged respiratory capacity for infected AMPK KO BMMo (S6B Fig.). Altogether, our data demonstrated that L. infantum induces the activation of AMPK in macrophage-infected cells controlling glucose uptake and respiration. A link between SIRT1 and AMPK has been previously described in several experimental models, allegedly acting as energetic sensors to sustain the metabolic homeostasis of the cell. Nevertheless, a clarification of the actual position of SIRT1 in relation to AMPK protein was needed. The aforementioned results prompted us to evaluate the levels of NAD+ and NADH nucleotides. BMMo infected with live parasites displayed a significant and transitory NADH increase at 6 hours p. i. (Fig. 5A). In contrast, NAD+ increased significantly in infected cells at 18 and 24 hrs (Fig. 5B), which reflected the higher NAD+/NADH ratio at these time points (Fig. 5C). Most importantly, a transient decrease in SIRT1 expression at the transcription and translational levels was observed both in vitro and in vivo infections (Fig. 5D-F). We also evaluated other members of the sirtuin family modulating metabolic pathways. However, we did not observe significant differences at the transcription levels of SIRT3 and SIRT6 during L. infantum infection (S7A Fig.). Due to the lack of specific chemical inhibitors for SIRT1, we used BMMo from mouse deficient or expressing an inactive form of SIRT1 [16]. Infected macrophages from these mouse strains failed to upregulate Slc2a4 and Ppargc1a genes, resembling infected AMPK KO BMMo (Fig. 5G), and presented unchanged glucose uptake in comparison to uninfected cells (Fig. 5H and S7B Fig.). In order to dissect the cascade of events associated with SIRT1 and AMPK, we infected BMMo from WT and SIRT1 KO with L. infantum. The first evidence that AMPK is at the bottom of SIRT1 came from the absence of increase levels of AMPK-P-Thr172 in infected SIRT1 KO BMMo when compared to uninfected cells (Fig. 6A). Furthermore, the absence of SIRT1 activity influenced the up-regulation of PGC-1α protein expression (Fig. 6B). The addition of AICAR, an activator of AMPK, to infected WT or SIRT1 KO BMMo led to the activation of AMPK (Fig. 6A-B) and a significant increase in Slc2a4 and Ppargc1a gene expression (Fig. 6C), which were abrogated when the AMPK inhibitor compound c was added simultaneously (Fig. 6A-C). Finally, the analysis of AMPK activation on splenic macrophages from uninfected and infected mice at 18 hours p. i. showed an increase of AMPK-P-Thr172 in splenic macrophages of infected WT mice (Fig. 6D-E). To note, AMPK-P-Thr172 was significantly reduced on splenic macrophages of SIRT1 KO infected mice (Fig. 6D-E). Overall, our results demonstrated the crucial role of SIRT1 in regulating AMPK activation in L. infantum-infected macrophages. Our data demonstrated that L. infantum parasites exploit the SIRT1-LKB1-AMPK axis to shift macrophage metabolism towards mitochondrial oxidation. To unravel the role of SIRT1-LKB1-AMPK triad in the infection outcome, BMMo derived from AMPK, SIRT1 or LKB1 knockout mice were infected. All BMMo with a faulty active member of the triad were significantly more resistant to infection than the WT counterparts (Fig. 7A-B). Of note, the absence of AMPK leads to a shift on infected macrophages towards an inflammatory M1 profile, as represented by iNOS/Arg1 ratio (S8A-B Fig.). As Leishmania-infected BMMo AMPK can be activated with AICAR in the absence of SIRT1 (Fig. 6A-B), the treatment with AICAR led to a significantly increase on infection levels, while the concomitant addition of compound c abrogated this infection rise to levels of untreated cells (Fig. 7B). To further validate this link, the percentage of infected BMMo from WT and SIRT1 KO cells treated with AICAR or either AICAR plus compound c was determined. The numbers of infected SIRT1 KO BMMo were significantly lower than their WT counterparts (Fig. 7B). Nevertheless, AMPK activation precluded this effect with infection rates similar to WT cells (Fig. 7B). Importantly, in the presence of compound c the infection returns to levels similar of untreated KO cells (Fig. 7B). Pre-treatment of L. infantum promastigotes with AICAR did not significantly altered the parasite growth curve, viability or the capacity to infect BMMo (S9A-S9C Fig.) discarding a direct effect on a potential Leishmania AMPK functional ortholog. An additional control was the observation of the lack of effect when infected AMPK or LKB1 KO BMMo were treated with the SIRT1 activator SRT1720 in comparison to WT cells (Fig. 7A). Ultimately, a definitive demonstration of the biological role of AMPK and SIRT1 proteins was provided by infection of WT, myeloid restricted (Mac) -AMPK KO and Mac-SIRT1 KO mice with L. infantum, with evaluation of the parasite load in the spleen, liver and bone marrow 10 days post-infection. The absence of AMPK and SIRT1 led to a significantly reduction of the parasite load in all tested organs, except for the spleen of SIRT1 KO mice that showed a decreasing trend (Fig. 7C-E). Overall, the high correlation of AMPK and SIRT1 activities with parasite burden in in vivo infections underlines a new and high biologically relevant role for a metabolic control by the parasite. Subversion of host cell energy metabolism by intracellular pathogens has been proposed to play a key role in microbial growth and persistence [6,17]. This topic is of extreme importance especially when lower eukaryotes as intracellular pathogens rely on a dramatic metabolic reprogramming to adapt to the challenges set by the new host nutritional status. Herein, we demonstrated that L. infantum infection of BMMo induced a switch towards mitochondrial oxidation phosphorylation favoring its own growth. We found that infection is associated with the activation of AMPK downstream to SIRT1 and LKB1. Indeed, BMMO defective in either SIRT1 or AMPK have reduced parasite growth. A representation of our proposed model is shown in Fig. 8. Previous observations suggested that Leishmania can modulate metabolic pathways in their host cells to enhance parasite access to essential nutrients [17]. The transcriptomic signature of L. major infected macrophages identified anaerobic glycolysis as one of the major pathways regulated by the parasite early after infection [18]. Additionally, L. major was shown to inactivate mTORC1 decreasing macrophage global translation [19]. Although these studies suggest that subtle changes in the transcription and activities of host metabolic enzymes may have profound effects on the infection outcome, the mechanisms involved in the subversion of host metabolism and bioenergetic remodeling remain largely elusive. The extracellular flux analysis with living cells allowed the bioenergetic evaluation of host macrophage upon challenge with L. infantum. We found that live parasites induced a rapid shift of macrophage metabolism towards aerobic glycolysis (Warburg effect) with a concomitant decrease of mitochondrial function. The observed increase on glycolytic pathways occurs specifically on infected cells although the transcription of Pkm2 glycolytic enzyme is still induced on bystander macrophages (exposed to the parasite but not infected). Pkm2 transcription has been related with macrophage [20,21] and NF-kB activation [22]. This phenomenon suggests that bystander macrophages could be activated in a similar fashion as it was demonstrated to occur with bystander dendritic cells [23,24] or even by infected macrophage secreted extracellular vesicles [25]. From the early energetic and metabolic settings responsible for a glycolytic environment resulted a nutrient and energetic deficit. Essential metabolites for the parasite including amino acids, purines, vitamins and haem have to be acquired from the host milieu [17]. This dependence on host purines could partially account for the ATP levels detected. Uninfected macrophages are known to rely on a glycolytic metabolism that could be explained by its presence in hypoxic inflamed tissues [26]. The yield of ATP produced by glycolysis is quite low. Macrophages, facing impaired respiration, respond by increasing their glycolytic rate in order to preserve the ATP pool and the glycolytic intermediates necessary to preserve mitochondrial membrane potential and viability [27]. Even though, the high dependency of infected macrophages on aerobic glycolysis with a concomitant impairment of mitochondria function in the early steps after infection reflects an even higher imbalance between these two pathways that may account the changes in ATP/AMP pool observed. As a consequence of the higher intracellular AMP/ATP ratio, we detected the subsequent activation of AMP-activated protein kinase (AMPK), governing a known molecular stress response pathway regulating energy utilization and production [28]. AMPK activation stimulates a myriad of host catabolic processes to restore intracellular energy and nutrients, which can then be used to nourish the obligate intracellular pathogen. The activation of AMPK activity was followed by the raise of GLUT4 and PGC-1α, but not PGC-1β. PGC-1α induces a positive control over mitochondria biogenesis, increasing mitochondria functions and minimizing the buildup of its by-products, ensuring a global positive impact on oxidative metabolism [15]. In L. infantum infected cells the enhanced mtDNA/nDNA ratio supports an increase in abundance and functionality of mitochondria in the cell corroborating PGC-1α role. SIRT1 and AMPK proteins share several downstream targets as PGC-1α [29], FOXO1 [30] and PPARα [31] as well as reciprocal regulation [32]. Some reports establish a functional link between these proteins demonstrating on the one hand a regulation of AMPK by SIRT1 through LKB1 deacetylation [33,34] and in contrast an activation of SIRT1 by AMPK dependent [35] or independent [28] induction of NAM phosphoribosyltransferase (Nampt) activity. Here, we demonstrated that in the absence of a catalytically active SIRT1, infected macrophages display a similar transcription and infection phenotype as AMPK KO cells. Our data suggests that both enzymes act synergistically to favor parasite growth. To elucidate their respective contribution, we activate SIRT1 in AMPK KO macrophages as well as the inverse strategy. Our data demonstrated that AMPK activation with AICAR is feasible in Leishmania infected SIRT1 KO/mutant cells, but not inversely, ranking SIRT1 upstream AMPK. A likely explanation is that SIRT1 exerts its activity over AMPK via the tumour supressor LKB1 kinase [33,34]. Indeed, LKB1 KO macrophages presented the same infection phenotype as SIRT1 KO BMMo. AMPK activation led to the reversion of Ppargc1a and Slc2a4 transcription and consequently to increase infection burdens. AMPK plays a crucial role in the set-up of a mitochondria microenvironment, as suggested by the establishment of an aerobic glycolytic profile with a reduction on infection rate in its absence. The regulation of Slc2a4 transcript by PGC-1α correlates AMPK activation with PGC-1α activity and place AMPK in the core of the transcriptional program of mitochondria function and biogenesis during Leishmania infection. Moreover, the absence of AMPK that impede the metabolic shift impacts the polarization profile of infected macrophages establishing a M1 inflammatory phenotype. The change of metabolism during infection likely underlines the modification of cellular redox potential (NAD+/NADH ratio). The higher levels of NADH found in infected macrophages concomitantly to the establishment of aerobic glycolysis are potentially a consequence of reduced complex I activity, as suggested by the reduced Ndufa9 transcript levels. NAD+ levels do not decrease significantly suggesting that NADH is at same extent being converted to NAD+ in the formation of lactate and reused by glycolysis to maintain the glycolytic flux. The decrease of NAD+/NADH ratio could be enough to reflect a reduction on SIRT1 energetic sensor activity that could actually contribute to the bioenergetic profile displayed by the infected cells in early time points. Of note, Leishmania parasites are NAD+ auxotroph using nicotinamide (Nam), among other precursors, for NAD+ synthesis [36]. This led us to hypothesize that the potential consumption of Nam by the parasite may lower its intracellular levels below the threshold necessary to induce a SIRT1 inhibitory effect [37]. Infections performed in mice deficient in the myeloid lineage for AMPK or SIRT1 demonstrated that the absence of each protein led to a significantly decrease on the liver, bone marrow and spleen parasite burden. The establishment of an in vivo microenvironment with an enhancement of mitochondria functionality in the presence of L. infantum is not only suggested by the described parasite burden but also by the phosphorylation of AMPK found in macrophages from the spleen of infected WT mice. Nevertheless, in an in vivo microenvironment, we cannot exclude other mechanisms/variables that could interfere with the host immune response against L. infantum infection. It is tempting to hypothesize that the manipulation of AMPK activation could be a common mechanism advantageously used by intracellular pathogens. In our model of infection, AMPK develops a crucial role for the persistence of the parasite inside the host. However, the acute silencing of AMPK catalytic or regulatory subunits favors intracellular T. cruzi growth [5]. Thus, even for parasites displaying a close phylogenetic relationship such as Leishmania and T. cruzi may exploit AMPK in a diametrically opposite manner. Several Leishmania virulence factors, namely lipophosphoglycan and glycoprotein gp63, at the surface or released as exosomes [38,39], are known to modulate host signaling pathways interfering with the function of transcription factors and consequently altering host gene expression [40]. Herein, we found that energetic/metabolic alterations only occur in infected but not bystander cells. Moreover, irradiated parasites were unable to induce a similar phenotype, indicating that the observed variations are not due to the presence of the parasite itself but to the manipulation of host bioenergetics status by the parasite. Our results are also consistent with earlier observation that Leishmania has also the capacity to delay programmed cell death (PCD) induction in the infected macrophages by modulating apoptosis through mitochondrial permeabilization [41]. Thus, mitochondria represent for Leishmania one the main organelles targeted by the parasite not only to modulate cellular viability but also to control mitochondrial bioenergetics, as shown in the current work. Importantly, the increase of oxidative phosphorylation in infected macrophages is not exclusive for promastigote infection. Amastigotes also induced increased levels of Pdk1 transcripts, however at lower extent, suggesting also an inhibition of the TCA cycle and mitochondrial respiration, which was supported by the decrease of ATP levels. Although we did not observed an early change in the levels of LDH activity/lactate secretion in, amastigote infection cells, the observed phenotype lead us to hypothesize that the possible inhibition of host PDH1 enzyme in our model of amastigote infection suggests a preventive role, developed by the parasite, in the production of reactive oxygen species (ROS) by host mitochondria as was already described under hypoxic conditions [42] and [43]. This is in agreement with the described capacity of Leishmania amastigotes to resist to the macrophage microbicidal activity in order to persist inside the host. Similarly to the promastigote infection, we detected increased transcript levels of Pparg1c and mitochondrial biogenesis as well as increased levels of AMPK phosphorylation. Overall, our data places energetic deficit, AMPK activation and increased mitochondrial metabolism as a common mechanism induced by L. infantum irrespective of the infective parasite form. In this study, we uncovered a subversion mechanism employed by L. infantum impacting host metabolic homeostasis. L. infantum interferes positively with AMPK pathway predisposing the host microenvironment for parasite growth. Our work underlines the potential of macrophage metabolism as a new therapeutic target to modulate leishmaniasis infection. Balb/c, Mac-Sirt1 KO mice with 98% C57BL/6 background (myeloid cell-specific Sirt1 knockout mice), Mac-AMPKα1 KO, Mac-LKB1 KO and the respective littermate lox controls (Lysozyme-Cre+/+ Sirt1flox/flox) mice were maintained at the Instituto de Biologia Molecular e Celular (IBMC, Porto, Portugal) laboratory conditions, in sterile cabinets and allowed food and water ad libitum. All animals used in experiments were aged from six to twelve weeks. The enzyme-dead Sirt1-H355Y mice and the respective littermate lox controls were maintained at the animal facilities of Ottawa Hospital Research Institute. RS has an accreditation for animal research given from Portuguese Veterinary Direction (Ministerial Directive 1005/92). A cloned line of virulent L. infantum (MHOM/MA/67/ITMAP-263) were maintained by weekly subpassages at 26°C in RPMI 1640 medium (Lonza, Swtzerland) supplemented with 10% heat-inactivated Fetal Bovine Serum—FBS (Lonza, Switzerland), 2 mM L-glutamine, 100 U/ml penicillin, 100 mg/ml streptomycin and 20 mM HEPES buffer (BioWhittaker, Walkersville, MD). Only L. infantum promastigotes under four to ten passages were used in the experiments. Promastigote to amastigote differentiation was achieved by culturing 107 stationary phase promastigotes/ml at 37°C in a cell free culture medium (MAA20). After a complete differentiation (3 days) the parasites were maintained by weekly subpassages. Bone marrow precursors were recovered with DMEM medium after flushing femurs and tibias from the hind legs of Balb/c, AMPK KO mice, LKB1 KO, Sirt1 KO and Sirt1 mutant as described previously [44] with minor modifications. The hind legs of Sirt1 KO, Sirt1 mutant mice were kindly provided by Dr. Michael McBurney from Ottawa Hospital Research institute. The bone marrow cells obtained were suspended in complete macrophage medium (DMEM medium with glucose (4,5g/L) (Lonza, Switzerland) and HEPES buffer supplemented with 10% heat-inactivated Fetal Bovine Serum—FBS (Lonza, Switzerland), 2 mM L-glutamine, 100 U/ml penicillin and 100 mg/ml streptomycin (BioWhittaker, Walkersville, MD) ) and 5% of L-929 cell conditioned medium (LCCM) was added. After 4h of incubation non-adherent cells were recovered and seeded at 5x105cells/ml in 96,24 and 6-well plates, in complete medium and 5% of LCCM, to continue bone marrow differentiation. Renewal of LCCM was made at day 4 of culture. Macrophages acquired a definitive differentiation status at day 7 of culture with a purity superior to 90%. For the recovery of peritoneal macrophages, Balb/c mice were injected intraperitoneally (i. p.) with ice-cold PBS. The inflated peritoneum was carefully shaken and the PBS solution removed. The macrophages purity was confirmed and then seeded in the same concentrations as BMMo. After an overnight incubation the cells were ready to use. L. infantum promastigotes were used for CFSE (Invitrogen Molecular probes, Eugene, Oregon) and eFluor670 (eBioscience) labelling at a concentration of 6x107 promastigotes/ml. Promastigotes were washed two times with PBS and labelled with 5 μM CFSE for 10 min or 1 μM eFluor670 for 5 min at 37°C followed by 5 min incubation at 4°C to stop the reaction. The parasites were then washed twice and suspended in RPMI supplemented medium before proceeding to infections as described before [45]. Seven-days differentiated BMMo were incubated with CFSE/eFluor670-L. infantum promastigotes at a 1: 10 ratio. At same co-culture ratios, we performed experiments with irradiated (3000 Gy; Gammacell 1000 Elite) parasites and with unlabelled L. infantum amastigotes. After 4 hours of incubation, cells were washed to remove the non-internalized parasites, except for analyses made at 1 and 2 hours of infection. At defined time points the cells were recovered and the percentage of infected BMMo was determined by flow cytometry evaluation of CFSE+/eFluor670+ cells in a FACSCanto II cytometer (BD Biosciences) and analysed with FlowJo software (Tree Star, Ashland, OR). Growth curve and viability (1 μg/ml of 7-AAD) were assessed for both viable and irradiated promastigotes. BMMo were treated at 6 hours post-infection with Aicar (440μM), Aicar + compound c (5μM), SRT1720 (1μM), Sirtinol (10μM), Resveratrol (1μM) and nicotinamide (1mM) that were then used for further analysis. At defined time points post-infection, BMMo were recovered and the percentage of infected cells was determined by flow cytometry evaluation of eFluor670+ or CFSE+ cells in a FACSCanto II cytometer (BD Biosciences) and analyzed with FlowJo software (Tree Star, Ashland, OR). L. infantum promastigotes were also treated during 18h with Aicar (440μM), Aicar + compound c (5μM) and compound c. The parasites were washed intensively to remove the excess of drug. Viability and growth curves were determined as well as BMMo infection rate. All the compounds used were obtained from Sigma-Aldrich (St. Louis, MO). Uninfected or eFluor-labelled L. infantum infected cells were incubated with 90 μΜ of 2-[N- (7-nitrobenz-2-oxa-1,3-diazol-4-yl) amino]-2-deoxy-D-glucose (2-NBDG) (Cayman Chemical, Michigan), a fluorescent analogue of glucose, in DMEM supplemented medium without glucose for 1 hour at 37°C. Cells were twice washed with cold PBS, harvested and stained with 1 μg/mL of 7-Aminoactinomycin D (7-AAD) (Sigma, Saint Louis, Missouri). The 2–NBDG uptake was separately quantified on live eFluor- and eFluor+ BMMo in a FACSCanto II cytometer and analyzed with FlowJo software. Bone marrow macrophages: BMMo were infected with CFSE-labeled parasites and sorted according to their F4/80+ CFSE+ or F4/80+ CFSE- for infected and bystander cells, respectively or F4/80+ staining in the case of non-infected cells. Splenic macrophages: Mac-SIRT1 KO and littermate lox controls were infected intraperitoneally with 1 × 108 CFSE-labeled L. infantum promastigotes. Naïve and infected mice were euthanized at 12,18 or 48 hours post-infection and the spleens removed. Splenic T and B lymphocytes were depleted using the CD3ε and the CD19 microbeads coupled with LD columns (Miltenyi Biotec). The remaining cell suspension was labeled with anti-CD11b-PE, anti-Ly6C-PerCP/Cy5. 5 and anti-Ly6G-Pacific Blue and sorted according the surface expression of CD11b+Ly6Cint/highLy6Glow and CFSE expression gated on infected (CFSE+CD11b+Ly6Cint/highLy6Glow) or bystander (CFSE-CD11b+Ly6Cint/highLy6Glow) splenic macrophages. For all the experiments, CD11b+Ly6Cint/highLy6Glow cells from the spleen of non-infected mice were sorted as a control. Sorting experiments were performed in a FACSAria I using FACSDiva software (BD Biosciences). The purity of the separation was always higher to 90% as confirmed by flow cytometry. WT, Mac-Sirt1 KO and Mac-AMPK KO mice were infected intraperitoneally with 1 × 108 L. infantum promastigotes resuspended in sterile PBS. Ten days post-infection, the animals were euthanized and the DNA from spleen, liver and bone marrow extracted with DNAzol Reagent (Invitrogen, Barcelona, Spain). The parasite burden was evaluated as previously described [46]. Total RNA was isolated from cells with TRIzol reagent (Invitrogen, Barcelona, Spain) or RNeasy micro kit (Qiagen), according to the manufacturer instructions. To determine mitochondrial DNA (mtDNA) /nuclear DNA (nDNA) ratios, mitochondrial and genomic DNA was extracted by QIAmp DNA micro kit (Qiagen). DNA from spleen, liver and bone marrow were extracted by DNazol according to the manufacture instructions. The RNA and DNA concentration was determined by OD260 measurement using a NanoDrop spectrophotometer (Thermo Scientific, Wilmington, DE). Total RNA (10–200ng) was reverse-transcribed using the iScript Select cDNA Synthesis Kit (BioRad, Hercules, CA, USA). Real-Time quantitative PCR (qRT-PCR) reactions were run in duplicate for each sample on a Bio-Rad My Cycler iQ5 (BioRad). The mitochondrial DNA (mtDNA) /nuclear DNA (nDNA) ratio was quantify by qPCR as previously described [47]. Primer sequences were obtained from Stabvida (Portugal) and thoroughly tested. The resulting RT product was expanded using the Syber Green Supermix (BioRad). The results were then normalized to the expression of a housekeeping gene Rps29. After amplification, a threshold was set for each gene and cycle threshold-values (Ct-values) were calculated for all samples. Gene expression changes were analysed using the built-in iQ5 Optical system software v2. 1 (BioRad). The complete list of primers used is represented in the S1 Table. Adenine nucleotide concentrations were determined in cell extracts by an enzymatic method [48]. As alternative, the ATP levels in macrophages were measured at each time point of infection. To this end, cells were washed with PBS and suspended in Glo-lysis buffer (Promega, Madison, Wisconsin). After 15 min of incubation, supernatants were recovered after centrifugation and used to quantify ATP by a luciferin–luciferase method using an ATP determination Kit (Molecular probes, Eugene, OR) according to manufacturer instructions. NAD and NADH levels were determined using Fluorescent NAD/NADH detection kit (Cell Technollogy, Inc, Montain View, CA) using 2x105cells following protocols from the manufacturer. For the determination of the bioenergetic profile of infected BMMo, oxygen consumption rate (OCR) and extracellular acidification rate (ECAR) were determined at 6 and 18 hours post-infection using an XF-24 Extracellular Flux Analyzer (Seahorse Bioscience). BMMo after 7 days of differentiation were seeded at 2x105cells/well in 400μl of complete macrophage medium in XF-24 cell culture plates. After an overnight period the cells were incubated with irradiated or not L. infantum promastigotes at a 1: 10 ratio. One hour before the defined times of infection, the cells were washed and the medium change to XF medium (unbuffered DMEM supplemented with 4. 5g/L of glucose, 2% of FBS, 2 mM L-glutamine, 100 U/ml penicillin and 100 mg/ml streptomycin). The real time measurement of bioenergetic profile was obtained under basal conditions and in response to oligomycin (1μM), fluoro-carbonyl cyanide phenylhydrazone (FCCP—1μM), Rotenone (1μM) and Antimycin A (1μM). The non-mitochondrial respiration was obtained by subtracting the Rotenone/Antimycin A values. The spare respiratory capacity (SRC) was obtained by subtracting FCCP from basal OCR values, and the glycolytic capacity defined as the variation between oligomycin and basal ECAR values. The procedure used in the experiments was established according to Seahorse manufacturer instructions. Lactate dehydrogenase (LDH) and lactate were measured on an AutoAnalyzer (PRESTIGE 24i, PZ Cormay S. A.) using reagents from the same provider. LDH catalysis the reduction of pyruvate by NADH, according the following reaction: Pyruvate + NADH + H+ → L-lactate + NAD+. The rate of decrease in concentration of NADH was measured photometrically at 340 nm, and it is proportional to the catalytic concentration of LDH present in the sample accordingly to the method described by Elliot and Wilkinson. Lactate is oxidized by lactate oxidase to pyruvate and hydrogen peroxide, which, in presence of peroxidase (POD), reacts with Nethyl-N- (2-hydroxy-3-sulfopropyl) -3-methylaniline (TOOS) forming a red compound, which colour intensity measured a t 546 and 700 nm is proportional to the concentration of lactate in the examined sample. In vitro and ex vivo macrophages (approximately 1×106) and L. infantum promastigotes (1×107), were lysed in ice-cold lysis buffer containing 50 mM Tris, pH 7. 4,1% Triton X-100,150 mM NaCl, 10% glycerol, 50 mM NaF, 5 mM sodium pyrophosphate, 1 mM Na3VO4,25 mM sodium-β-glycerophosphate, 1 mM DTT, 0. 5 mM PMSF, and protease inhibitors (Complete Protease Inhibitor Cocktail; Roche), for 30 min at 4°C. The lysates (twenty to fifty micrograms of protein) were subjected to SDS-PAGE electrophoresis and the proteins were transferred to mini nitrocellulose membranes (Biorad) by the Trans Blot Turbo Transfer System (Biorad). The membranes were then incubated with primary antibodies and with horseradish peroxidase-coupled secondary reagents (Jackson ImmunoResearch). The membranes development was made by Super Signal West Pico or West Dura chemiluminescence substrate (Thermo Scientific). Primary antibodies were directed against: total AMPKα (23A3), AMPKα phosphorylated at Thr172 (#2531; both from Cell Signaling), total SIRT1 (H-300) and total PGC1β (E-9; both from Santa Cruz), total PGC1α (4C1. 3; Merck Millipore), and β-actin (C4; Antibodies-online). Statistical analyses were performed using the Student’s t test for paired observations or one-way ANOVA test with a Bonferroni multiple-comparison posttest for multiple group comparisons. Statistically significant values are as follows: *p < 0. 05, **p < 0. 01, ***p < 0. 001. The experimental animal procedures were approved by the local Animal Ethics Committee of Institute for Molecular and Cell Biology, University of Porto, Portugal and licensed by DGV (Director of Veterinary, Ministry of Agriculture, Rural Development and Fishing, Govt. of Portugal in December 29,2010 with reference 25406. All animals were handled in accordance with the IBMC. INEB Animal Ethics Committee and the DGV General guidelines and the principles and guidelines established in the European Convention for the Protection of Vertebrate Animals Used for Experimental and Other Scientific Purposes (Council of Europe, ETS no. 123,1991).

Leishmania infantum, a causative agent of visceral leishmaniasis, is able to infect host macrophages and modulate a myriad of signalling pathways that contributes to the disease outcome. In order to survive, L. infantum must compete with the host for the same metabolic resources, however scarce attention has been dedicated to clarify the potential interference of the parasite with the host metabolic pathways and its impact for the infection outcome. We analysed the macrophage metabolic alterations induced by L. infantum focusing on host energetic players exploited by the parasite. We describe that L. infantum induced a metabolic switch from an early aerobic glycolytic environment to a later mitochondrial metabolism. In this process, L. infantum modulates important energetic sensors of the host, such as the SIRT1-LKB1-AMPK axis. This triad is important for the recovery of the host energetic status and also for the parasite survival. With this work, we demonstrate that the host SIRT1-LKB1-AMPK axis has a crucial impact on the parasite survival in vitro and in vivo.

Abstract Introduction Results Discussion Materials and Methods

2015

Leishmania infantum Modulates Host Macrophage Mitochondrial Metabolism by Hijacking the SIRT1-AMPK Axis

Decision-making in the real world presents the challenge of requiring flexible yet prompt behavior, a balance that has been characterized in terms of a trade-off between a slower, prospective goal-directed model-based (MB) strategy and a fast, retrospective habitual model-free (MF) strategy. Theory predicts that flexibility to changes in both reward values and transition contingencies can determine the relative influence of the two systems in reinforcement learning, but few studies have manipulated the latter. Therefore, we developed a novel two-level contingency change task in which transition contingencies between states change every few trials; MB and MF control predict different responses following these contingency changes, allowing their relative influence to be inferred. Additionally, we manipulated the rate of contingency changes in order to determine whether contingency change volatility would play a role in shifting subjects between a MB and MF strategy. We found that human subjects employed a hybrid MB/MF strategy on the task, corroborating the parallel contribution of MB and MF systems in reinforcement learning. Further, subjects did not remain at one level of MB/MF behaviour but rather displayed a shift towards more MB behavior over the first two blocks that was not attributable to the rate of contingency changes but rather to the extent of training. We demonstrate that flexibility to contingency changes can distinguish MB and MF strategies, with human subjects utilizing a hybrid strategy that shifts towards more MB behavior over blocks, consequently corresponding to a higher payoff. For optimal decision-making, animals must learn to associate the choices they make with the outcomes that arise from them. Classical learning theories suggest that this problem is addressed by habitual or goal-directed strategies for reinforcement learning [1,2]. These strategies differ in that habitual behavior seeks simply to reinforce responses based on environmental cues, whereas goal-directed behavior considers action-outcome relationships – that is, contingencies–in the environment. Habitual and goal-directed strategies have been implemented in model-based (MB) and model-free (MF) reinforcement learning algorithms, respectively. Both algorithms make decisions by estimating action values and choosing the actions that maximize reward in the long term [3,4]. The MF system achieves this retrospectively, caching past rewards using a reward prediction error signal [5] whereas the MB system achieves this prospectively, planning using a learned internal model of the state transitions and rewards in the environment [6,7]. Recent studies have emphasized that MB and MF systems work in parallel rather than in isolation [4,8–10]. Early studies discerned MB and MF contributions using manipulations of reward values, such as in reward devaluation paradigms, but did not seek to quantify their relative contributions [1]. A recent study [8] addressed this by developing the hallmark “two-step” task in which each trial following rare or common outcomes was informative of the MB/MF tradeoff, thereby permitting model-fitting analyses to quantify their relative influence in decision-making. Human subjects showed a hybrid MB/MF strategy in the task, a result that has been widely replicated under different manipulations [11,12,13] and extended to the non-human animal literature (Groman et al. Soc. Neurosci. Abstracts 2014,558. 19, Miranda et al. Soc. Neurosci. Abstracts 2014 756. 09, Akam et al. Cosyne Abstracts 2015, II-15; Hasz & Redish, Soc. Neurosci. Abstracts 2016 638. 08). All these studies measured MB/MF contributions in terms of behavioral flexibility following reward updates, whereby “rare” (as opposed to “common”) observations of a rewarded or unrewarded outcome was informative to the MB system, but not the MF system, and thus these observations disclosed which system controlled participants’ choices. Theory predicts that flexibility to transition contingency changes can – like flexibility to reward structure–determine the relative influence of MB and MF strategies [4,14]. Two studies have examined the flexibility of MB and MF systems to global contingency changes [15,16]. However, quantification of the MB/MF tradeoff was limited as these studies manipulated contingency and tested flexibility to the change of contingency in separate phases; at these timescales, it becomes difficult to exclude the effect of adaptation on MB/MF weights. Thus, we developed a novel two-level contingency change task containing multiple, frequent and interleaved transition contingency changes that elicit different consequent actions by the MB and MF systems. Our design, like the two-step task [8] and its variants, therefore permits model-fitting analyses to robustly determine the relative influence of the MB/MF systems. The contingency change task is structured such that actions following frequent contingency changes are distinctly attributed to either a MB or MF strategy; this then permits quantification of the degree to which each system is in control. On top of a hybrid MB/MF strategy, subjects may not remain at one level of MB/MF control but instead shift their relative weight in accordance with environmental factors. In general, animals show habit formation with time, a robust effect reported since early reward devaluation studies [17] in which extensive training stamped in habits, resulting in insensitivity to reward devaluation; in contrast, limited training retained goal-directed behavior. Sensitivity to contingency degradation (the omission of a previously-learned contingency between actions and outcomes) also decreases with overtraining, likewise reflecting a trend towards habitization with time [18]. In the original two-step task, the MB/MF trade-off was designed to be stable [8], but will shift under manipulations such as limited time [10] or cognitive load [11,19,20]. However, habits are not guaranteed to form with time; even after extended training, rats can show residual flexible responding following outcome devaluation, indicating that they retained goal-directed behavior despite overtraining [21]. In another study using the two-step task [20], the level of MB/MF control in fact increased in favour of more MB control (i. e. towards less habitual behavior) over three days of training. However, general shifts in MB/MF control should be disentangled from the effects of environmental volatility, which are known to affect the MB/MF balance [22]. Thus, in this study, we examined whether the MB/MF relationship is affected by environmental stability, or whether it shifts more generally over time. We found that human subjects indeed showed a hybrid strategy in reacting to contingency changes in our task, with an increased influence of MB control over the first two blocks. However, relative MB/MF control did not significantly differ across rates of contingency changes; thus, the increase in MB control may be a more global effect of “anti-habitization” over time. Subjects (N = 16) performed a two-level contingency change task that consisted of 600 trials (Fig 1). Each trial began at either the first level (S0) with 50% probability, or the second level with 50% probability – 25% for each of the two states at this level (S1 or S2). If a trial started at the first level, a two-alternative choice was possible between two abstract stimuli. Each first-level action always deterministically led to the same second-level state, i. e. A1 to S1 and A2 to S2. Critically however, transitions from the second-level states to the terminal states flipped between two contingencies every 3–14 trials. Each of the two terminal states was then associated with either high or low reward, with the exact reward values drifting across trials (see Methods for details). Thus, flexibility to contingency changes was essential for maximizing reward. If a contingency change occurred, subjects always experienced the new transition structure regardless of whether they started at the first or second level, as contingency could only change between second-level and terminal states. Therefore, provided that an action was possible at the next trial (i. e. that the next trial started at the first level) the MB system would plan using the updated causal structure and thus would take the action that led under the new transition contingencies to the high reward terminal state. However, if a contingency change trial started from the second level, the MF system would not choose the optimal action on the next trial, as neither the received reward nor the new contingency would update the cached values of first-level actions, simply because no first-level action was experienced on those trials. As a result, the relative contribution of MB and MF systems can be measured by the degree of behavioral flexibility on first-level trials following contingency change trials starting from the second level. To examine the effect of environmental volatility on the contribution of the two systems, the frequency of contingency changes was varied–from 3–6 trials for 200 trials, to 7–10 trials for another 200 trials, and then 11–14 trials for the final 200 trials. The order of fast and medium contingency-change blocks was counterbalanced across two subject groups (n = 8 each). Every 40 trials, assignment of the high and low reward states also flipped to prevent formation of habits over an extended state representation, which could masquerade MF as MB behavior [23]. Simulated choices on the task were implemented according to MB and MF reinforcement learning algorithms (see Methods for details). For each system, we measured a “stay probability” index that followed the logic of contingency change trials described above. This index differs from classic stay probabilities [8] as trials starting from the second level do not have any choices to “stay”. Instead, stay probability in our task was defined as the probability of choosing the first-level action that results in the same second-level state as the previous trial. Stay probability was measured for four different conditions: whether the reward received in the previous trial was “high” or “low”, and whether the transition experienced in the previous trial, relative to the trial before that, was “changed” or remained “fixed”. Our analyses of stay/switch choices were restricted to trials that started from the first level and thus allowed the participants to make a choice. Furthermore, we restricted our analyses to first level trials following trials starting at the second level, since only these could distinguish MB and MF strategies. Across these conditions, MB and MF systems showed different stay probability patterns. The MF system, having no experience of the action that led to the new contingency, was more likely to stay on the action leading to the high reward state, and shift on the action leading to the low reward state, under “fixed” than “changed” conditions (p < 0. 01), indicating it was not flexible to changes in contingencies (Fig 2A). However, the MB system could immediately adapt with the correct next action, staying on the action if it would lead to the high-reward state but shifting if it would lead to the low-reward state, with a main effect of reward (p < 0. 01) regardless of contingency condition (Fig 2B). As expected, for contingency changes from the first level, MB and MF systems did not differ in stay probability patterns, as the MF system was able to update its action values accordingly, given that it directly experienced the action leading to the new contingency (S1 and S2 Figs). In addition to pure MF and pure MB strategies, we simulated a hybrid model that linearly weights MB and MF action values according to a parameter wMB. The stay probability pattern produced by this hybrid system reflected a mixture of the effects observed for the pure MF and MB stay probabilities–that is, showing a main effect of reward (p < 0. 01), but also an interaction between reward and contingency (p < 0. 01) (Fig 2C). Subjects showed hallmarks of both MB and MF strategies in reacting to contingency changes (Fig 3A), showing a main effect of reward, F (1,60) = 24. 65, p < 0. 01, as well as a reward/contingency interaction, F (1,60) = 13. 60, p < 0. 01. Therefore, subjects did not solely use a MB or MF strategy when reacting to contingency changes, but rather displayed a hybrid MB/MF strategy. While stay probabilities ruled out a purely MB or purely MF strategy, this measure could not quantify the degree to which subjects used the hybrid strategy; therefore, we used a hierarchical Bayesian method to fit candidate models of behavior to the subjects’ data, to determine which model best explained subjects’ choices and to obtain parameter estimates for the MB/MF weighting used by the subjects. The models tested included a pure MB model, a pure MF model, a hybrid model with one constant weight wMB across the session, a hybrid model with three separate wMB weights for each range of contingency change rates (fast, medium, or slow), and a hybrid model with three separate wMB weights for the three experimental blocks. The last two models served to test whether the relative contribution of the two systems depended on volatility of transition structure, or instead on block order, as contingency change rates were counter-balanced across the first two blocks. Model-fitting was confirmed to be able to recover true parameter values, as simulations showed that median estimated parameter values from model-fitting (see Methods for details) were well-correlated to known simulated parameter values, r ≥ 0. 99, p < 0. 01. Model-fitting results supported the existence of a hybrid MB/MF strategy in our task. Candidate models were compared using two criteria – integrated Bayesian Information Criterion (iBIC) which controls for number of parameters [24] and exceedance probabilities [25] (S2 Table). The hybrid model with three wMB weights over blocks outperformed the other candidate models under both criteria, with the lowest iBIC and a probability of 89. 4% that it was the most common of the four models across subjects. Thus, from here we only discuss the results of best-fit model, the three-block hybrid model. The median fitted wMB weights in the three-block hybrid increased across the three blocks (Fig 3B–3D), indicating some extent of “anti-habitization” rather than habit formation. The increase of wMB from block 1 to block 2, but not the increase from block 2 to block 3, was significant according to permutation tests, p < 0. 01. Stay probability analyses were not conducted on the three separate blocks, as slower contingency changes meant that the later blocks had fewer samples of contingency changes for comparison. The increase in wMB across blocks was not attributable to differences in quality of fit from the model-fitting procedure, as the log-likelihood of parameter estimates did not differ significantly across blocks, F (2,45) = 1. 42, p > 0. 05. Strength of correlations between simulated wMB weights and wMB weights recovered from simulation were also similar across blocks (block 1: r = 0. 99, block 2: r = 1. 00, block 3: r = 0. 99; p < 0. 01 for all blocks). Therefore, the significant increase in wMB from the first to second block was not caused by differences in quality of model fit. To confirm that the increase in model-based weight was not due to differences in the rate of contingency changes, we further analysed the fitted weights from the three-frequency hybrid model, which had a different wMB assigned to each range of contingency change rates, i. e. fast (every 3–6 trials), medium (every 7–10 trials) and slow (every 11–14 trials) contingency change blocks. The estimated wMB weights (S3 Fig) were not significantly different between fast vs. medium, or medium vs. slow frequency of contingency change blocks in permutation tests, p > 0. 05. Thus, the increase in wMB in our study seemed to be an effect of block order rather than environmental volatility from differences in contingency change rates. In summary, subjects became more model-based across the first two blocks but did not differ in MB influence between different rates of contingency changes; therefore, it seems that block order, but not contingency change volatility, affects wMB in our task. As subjects became more model-based, high reward choices and consequently reward rate also increased. Choice probabilities for the high reward action differed over blocks, F (2,45) = 5. 77, p < 0. 01, with post-hoc tests finding a significant increase between the first and third blocks (p < 0. 01) and the second and third blocks (p < 0. 05). Additionally, there was a significant difference in reward rate across blocks, F (2,45) = 3. 83, p < 0. 05, specifically increasing between the first and third blocks (p < 0. 05). Mean reaction time and number of missed trials due to timeout did not significantly change across blocks, p < 0. 05; therefore, the increase in high reward choices over blocks was not necessarily because subjects were worse at the task to begin with. Two analyses were performed to rule out the possibility of practice effects driving the association between reward rate and model-based weight. Within each block, there was a significant correlation of each subject’s median wMB and reward rate (block 1: r = 0. 66, p < 0. 01, block 2: r = 0. 65, p < 0. 01, block 3: r = 0. 56, p < 0. 05), indicating that on an individual subject basis, the extent of MB control was related to reward earned. Since these analyses were conducted within blocks, the association with reward rate could not be accounted for by block order. Additionally, the hybrid model was simulated using a range of MB weights (0,0. 2,0. 4,0. 6,0. 8 and 1) using the one-weight hybrid model for simplicity. Other free parameters were set to values fitted to the participants’ data. There was a significant effect of MB weight on reward rate, (F (5,90) = 8. 5, p < 0. 01). Therefore, MB influence in this task corresponds to a better “payoff” in terms of reward gained. However, lack of wMB adaptation to contingency-change rate suggests that using the cognitively-demanding MB system, at least in this task, is not motivated by its higher payoff. In other words, although being model-based does increase the payoff, it is not the reason for participants showing MB behavior. To further investigate this, we also computed the effect of wMB on reward rate, separately for different contingency change frequencies (fast/medium/slow) and found a significant effect of frequency on reward rate in a one-way ANOVA (F (2,45) = 5, p = 0. 01). This shows that contingency change volatility did not affect wMB, despite having a significant effect on how much a MB strategy pays off. However, this absence of evidence should not be taken as evidence of absence, given the sample-size in this study. Alternatively, this absence of evidence could be because the effect of wMB on reward rate, though statistically significant, is in fact very small (S4 Fig). Therefore, this difference may not be apparent enough to discern or motivate a higher engagement of the MB system as contingency change frequency increases. We developed a novel two-level contingency change task in which flexibility to frequently-changing transition contingencies between states could determine the extent to which subjects were using a model-based or a model-free strategy. Subjects showed a hybrid strategy when reacting to contingency changes, corroborating recent evidence of the parallel contribution of MB and MF systems in reward-guided decision-making. Importantly, this finding confirmed that changes to transition contingencies can elicit a balance of MB and MF behavior akin to changes to reward structure. Model-fitting analyses indicated that a hybrid model with three MB weights best explained subjects’ choices, with relative MB control increasing over blocks. The rate of contingency changes did not significantly shift the MB/MF balance; rather, MB control increased over the first two blocks of trials. This increase in MB control was concurrent with an increased proportion of high reward choices and consequently increased reward rate; individually, each subject’s wMB was also correlated with reward gained in the same block. In all, these results illustrated that not only do subjects use a mixed MB/MF strategy, but within this hybrid strategy, the trade-off shifts towards “anti-habitization” across the first two blocks. This agrees with a previous study [20] that used the two-step task over three days, reporting that their subjects’ MB weight increased across days. One distinction between our findings is that in [20], subjects started relatively model-based (i. e. median wMB > 0. 5) whereas in our case, subjects began relatively model-free (i. e. median wMB < 0. 5). This difference in starting MB weight simply may be due to individual differences, which is evident even within our subject pool. Alternatively, differences could be accounted for by the relatively short reaction time limit in our task compared to theirs (750 ms in ours vs. 2000 ms). A shorter reaction time limit is known to provide a depth-of-planning pressure and favor more MF control [10]. Hence, our subjects may have started more model-free and only become more model-based once they mastered prospective planning of the task structure. This is supported by the lack of significant changes in reaction time across blocks, suggesting that subjects may have used the full extent of their time and eventually learned more efficient planning under time pressure, therefore showing increased MB influence over blocks. These findings of an increase in MB control over blocks, however, goes against another study [26] using a similar task to the two-step task, that found an exponential decay in MB weight over the experimental session, or habit formation. This difference in findings is likely because they used a fixed rather than drifting amount of reward; in stationary environments such as these, habit formation can occur from overtraining, manifesting in an increase in MF rather than MB behavior [22]. Thus, these results point to the importance of maintaining a changing environment, as subjects can otherwise adapt to the change and become habitized. Manipulations of the rate of contingency changes did not seem to affect MB/MF control. While it has been shown that environmental volatility can influence MB/MF levels in the context of common or rare updates of reward structure [22], in our case, the kind and range of contingency change volatility did not elicit a significant difference in relative MB/MF control. Further work is certainly needed to definitively rule out the possibility that environmental volatility in the form of the rates of contingency changes does not affect MB weight, but in the present study, we find that subjects did not change their use of MB control with contingency change volatility, but rather increased MB influence more generally with block order. In conclusion, in a two-level contingency change task, subjects showed a hybrid MB/MF strategy, emphasizing their parallel contribution in reacting to changes in transition contingencies. The inclusion of multiple, frequent changes allowed us to perform model-fitting; by doing so, we found an increase in MB control over the first two blocks, a result not detectable in model-agnostic analyses alone. Our results build on the literature reporting the use of a hybrid MB/MF strategy in reacting to changes in information about reward structure, here demonstrating a mixture of strategies in reacting to multiple, frequent contingency changes that has yet been unexplored. In addition to MB and MF systems, a third reinforcement learning algorithm known as the successor representation (SR) [15,27] caches transitions (i. e. , the probability of occupying a state after performing an action in a previous state) in a model-free fashion, but learns reward values in a model-based fashion. The SR algorithm is therefore flexible to changes in rewards (like the MB system), but inflexible to changes in contingencies (like the MF system). As a result, the MB behavior seen in the original two-step task with non-stationary reward structure [8] could be equally explained with a SR model. In other words, wMB and wSR are conflated into wMB. Similarly, the MF behavior in our task with non-stationary transition structure could be interpreted under a SR framework, whereby wMF and wSR are conflated into wMF. In this sense, the two-step task and the task presented here complement each other in providing evidence that humans use both MB and MF strategies. This novel paradigm therefore provides another avenue for exploring the relationship between MB and MF control for future studies in neuropsychiatric disorders that may differentially implicate this balance between changes in transition contingencies and changes in reward values. Sixteen subjects (nine males, mean age 24 years) took part. The study was approved by the University College London Research Ethics Committee (Project ID 3450/002). All subjects provided written informed consent. Subjects performed 600 trials of three blocks (200 each) which differed in frequency of contingency changes: fast (contingency change every 3–6 trials), medium (every 7–10 trials) or slow (every 11–14 trials). Each subject was assigned to one of two groups (n = 8 each), which differed by the order of presentation of fast and medium contingency change blocks, i. e. half of the subjects had fast, medium, then slow contingency changes, and the other half started with medium, fast, then slow frequency of contingency changes. To ensure subjects understood the task structure, they were first trained with practice trials (N = 35) which followed the same structure of the task but used practice stimuli. After this, a training session (55 trials) started which used the test stimuli but without reward. This phase was intended to introduce some familiarity to the transition relationships between states before participants were allowed to make reward-guided decisions. This was then followed by the test session using the same stimuli, but now rewarded, for 600 trials. Subjects were informed that contingency changes would occur, but did not know the frequency of changes nor that those rates would vary across the session. At the first level, subjects had a two-alternative forced choice between two actions (pressing ‘S’ for the action available on the left side of the screen, ‘L’ for the right) with the presentation of stimuli randomized for the left/right side of the screen. To ensure that subjects recognized second-level states, they had to press ‘D’ if they encountered one of these states, and ‘K’ for the other. Both responses had a time limit of 750ms, following which the trial would end with no reward. Missed trials were not repeated. Payoff at the high-reward terminal state varied according to a Gaussian random walk (N (μ = 0. 5, σ = 0. 2), with a drift rate of 0. 15), bounded between £0 and £1. The payoff at the low-reward terminal state was £1 minus the reward of the high-reward terminal state. In practice, this resulted in an unambiguously large reward in one terminal state and an unambiguously small reward in the other terminal state. Therefore, the overall reward structure was stationary (until it flipped after every 40 trials, as noted). Subjects received a fixed proportion of their total reward gained, with payoff bounded between £5 and £25. To make the task adequately difficult and prevent formation of complex state-space representations [23], high- and low-reward assignments switched every 40 trials between the two terminal states. This change was designed never to co-occur with contingency changes. Both model-free and model-based algorithms seek to estimate the values of state-action pairs in order to choose the actions which can maximize expected future rewards. The state space was modelled as having a first-level state s0 with two actions a1 and a2, two possible second-level states s1 and s2, and two possible terminal states s3 and s4. There was only one action available on second-level and terminal states, as the subject did not have any choices at these levels. The model-free algorithm updates values of state-action pairs using temporal difference Q-learning [3,28]. The reward rt is used to compute a reward prediction error δt which updates action values for that state s and action a at time t, QMF (st, at). At the first level rt is set to be 0 as there is no reward at this level. The reward prediction error updates existing action values according to a learning rate αMF, and is modified by the eligibility trace λ, which governs how much credit past actions are given for outcomes. In a TD (λ = 0) algorithm, first-stage actions are updated only by the second-level action values, which in turn are updated by terminal state rewards. In contrast, in a TD (λ = 1) algorithm, first-level actions are directly updated using the reward from the terminal state reached on that trial. The model-based algorithm learns both transition probabilities PT and reward probabilities RT. The transition probabilities track the transition contingencies PT between states s and subsequent states s′. Upon encountering a contingency change, the model-based system always updated its knowledge of both transitions. The reward probabilities RT use the reward rt to update its subjective reward R for that state s and action a at time t. These learned transition and reward functions are then used to update the action values for the model-based system, QMB. Other parameters from the simulated models included learning rates for model-based and model-free systems, αMB and αMF, and a stay bias which temporarily increased the action value for the previously-selected action regardless of outcome, to quantify a perseveration bias. These additional parameters improved fit even when controlling for model complexity (S3 Table). For both systems, values for the non-selected action were updated as well, assuming that subjects knew that the reward for the selected action and reward for the non-selected action were negatively related, according to proposals of fictive reward [29]. Action values were updated for both visited and non-visited states, with the action values of non-visited states corresponding to 1 − Q (st, at) of the visited states. The inclusion of fictive reward updates resulted in a better fit to the subjects’ choices (S3 Table). The hybrid model weighted MB and MF action values according to a parameter wMB, with wMB = 1 indicating fully MB control: Qhybrid (st, at) =wMB∙QMB (st, at) + (1−wMB) ∙QMF (st, at) Action selection was then determined for all models according to a “softmax” rule which computes action probabilities as proportional to the exponential of the action values. The inverse temperature β determined the extent to which action selection was stochastic or deterministic from action values, quantifying an exploration/exploitation trade-off. To best replicate the subjects’ data of 600 trials for 16 subjects, each simulation was run for 16 initializations of 600 trials each. All reported simulations used fitted parameters from the three-block hybrid model for the learning rates αMF and αMB, inverse temperature β, eligibility trace λ and stay bias (S1 Table). wMB values were 1 for pure MB and 0 for pure MF models. Subjects’ data were fit to the models using mixed effects hierarchical model fitting. Expectation-maximisation was used which iteratively generates group-level distributions over individual subject parameter estimates, choosing the parameters that maximizes the likelihood of the data given those estimates. In each iteration, parameters were estimated by minimizing the negative log-likelihood of parameter estimates using fminunc in Matlab (MathWorks). The group-level distributions over all free parameters were assumed to be Gaussian, with no constraint. To then impose sensible constraints (0 ≤ wMB ≤ 1; 0 ≤ α ≤ 1; β ≥ 0), the original free parameters were passed through a logistic function (with slope parameter equal to one) for computing wMB and α, and through an exponential function for computing β. Consequently, although the original parameters were Gaussian, the resulting parameters of the models (wMB, α, β) were not necessarily Gaussian (hence the skewed distributions in Fig 3). This method is preferred to imposing hard constraints on parameters because it avoids parameters hitting boundary conditions and also remains loyal to the Gaussian assumption required for hierarchical Bayesian modelling. To ensure the efficacy of wMB parameter estimation for the candidate model, each block wMB was simulated for 11 different parameter values: 0,0. 1,0. 2, … 1. These resulted in a total of 33 parameter settings for wMBblock 1, wMBblock 2, wMBblock 3, with 16 iterations per setting. All other parameters in the simulations were set constant as the median parameter estimates taken from the hybrid three-block model from model-fitting on the subjects’ data. The same model-fitting procedure was performed on the simulated data and estimated parameter values were extracted. The integrated Bayesian information criterion (iBIC) [24] was used to compare the fits of candidate models to the data, with lower scores indicating better fit; this criterion penalizes more complex models. Finally, Bayesian model selection [25] was used to examine the prevalence of each model in the participant population. This quantifies an exceedance probability, the probability that each model is the most common in the subject pool. Permutation tests were run to evaluate the probability that wMB could differ across blocks by chance. Subjects’ blocks were randomly permuted such that each “block” contained a mixture of true first, second and third blocks. Model-fitting was run on each permutation to extract parameter estimates of wMB for each new “block”. The probabilities p (wMBblock 2 > wMBblock 1), and p (wMBblock 3 > wMBblock 2) were then evaluated for each permutation. The occurrences of the random permutations which had a smaller p (wMBblock 2 > wMBblock 1), and p (wMBblock 3 > wMBblock 2) than the true permutation were then tallied. Likewise, to evaluate the effect of frequency of contingency changes, permutation tests were run to compare wMB for fast, medium and slow contingency change blocks. Each subject was randomly assigned to one of the two groups (which differed in the order of fast and medium contingency change blocks) then wMB of each frequency block was computed for each permutation. Both the aforementioned one-tailed permutation test and a two-tailed Hellinger distance permutation test were used. To rule out the possibility that the effective learning rate of the MB and MF systems – rather than their fundamental differences – produced the behavior, we conducted several further analyses. A hybrid model composed of two MF systems with small (0. 25) and large (0. 75) learning rates could not replicate the stay probability patterns observed from subjects and the MB+MF hybrid system. Even more extreme, a hybrid model composed of two MF systems, one with a small (0. 25) learning rate and no eligibility trace (λ = 0), and another with a large (0. 75) learning rate and with eligibility trace (λ = 1) could not replicate those patterns. Furthermore, a hybrid model composed of two MB systems with small (0. 25) and large (0. 75) learning rates could not replicate the patterns either. When fitted to the behavioral data, all three hybrids of MF (α = 0. 25) +MF (α = 0. 75), MF (α = 0. 25, λ = 0) +MF (α = 0. 75, λ = 1), and MB (α = 0. 25) +MB (α = 0. 75) performed significantly worse than the real hybrid MF+MB model, according to our measures of model comparison. Together, these results rule out the possibility of just different effective learning rates of the two systems having produced the observed behavior. We further fitted a hybrid MF+MB with a free learning rate parameter, αMBTransition for updating transitions by the MB system (rather than assuming the parameter was 1). This model, in terms of model comparison, fit the data better than pure MB and MF models, and even slightly better than the hybrid MB (αMBTransition = 1) +MF model. In all previous analyses, the non-diagonal elements of the covariance matrix were set to zero (i. e. , assuming no correlation between free parameters). However, for the hybrid MB+MF with a free αMBTransition, we observed strong correlations between parameters when the covariates were allowed to change freely. The αMBTransition parameter was highly negatively correlated with αMF, and positively correlated with stay-bias, wMBblock 1, and wMBblock 2. We therefore decided not to rely on this model and instead, for the sake of parsimony, to use the original hybrid MB (αMBTransition = 1) +MF model throughout the paper.

To make good decisions, we must learn to associate actions with their true outcomes. Flexibility to changes in action/outcome relationships, therefore, is essential for optimal decision-making. For example, actions can lead to outcomes that change in value – one day, your favorite food is poorly made and thus less pleasant. Alternatively, changes can occur in terms of contingencies–ordering a dish of one kind and instead receiving another. How we respond to such changes is indicative of our decision-making strategy; habitual learners will continue to choose their favorite food even if the quality has gone down, whereas goal-directed learners will soon learn it is better to choose another dish. A popular paradigm probes the effect of value changes on decision-making, but the effect of contingency changes is still unexplored. Therefore, we developed a novel task to study the latter. We find that humans used a mixed habitual/goal-directed strategy in which they became more goal-directed over the course of the task, and also earned more rewards with increasing goal-directed behavior. This shows that flexibility to contingency changes is adaptive for learning from rewards, and indicates that flexibility to contingency changes can reveal which decision-making strategy is used.

Abstract Introduction Results Discussion Methods

learning decision making applied mathematics social sciences neuroscience habits learning and memory simulation and modeling algorithms cognitive psychology mathematics statistics (mathematics) cognition discrete mathematics combinatorics research and analysis methods behavior statistical models psychology permutation biology and life sciences physical sciences cognitive science

2017

Flexibility to contingency changes distinguishes habitual and goal-directed strategies in humans

The RNA exosome is the major 3′-5′ RNA degradation machine of eukaryotic cells and participates in processing, surveillance and turnover of both nuclear and cytoplasmic RNA. In both yeast and human, all nuclear functions of the exosome require the RNA helicase MTR4. We show that the Arabidopsis core exosome can associate with two related RNA helicases, AtMTR4 and HEN2. Reciprocal co-immunoprecipitation shows that each of the RNA helicases co-purifies with the exosome core complex and with distinct sets of specific proteins. While AtMTR4 is a predominantly nucleolar protein, HEN2 is located in the nucleoplasm and appears to be excluded from nucleoli. We have previously shown that the major role of AtMTR4 is the degradation of rRNA precursors and rRNA maturation by-products. Here, we demonstrate that HEN2 is involved in the degradation of a large number of polyadenylated nuclear exosome substrates such as snoRNA and miRNA precursors, incompletely spliced mRNAs, and spurious transcripts produced from pseudogenes and intergenic regions. Only a weak accumulation of these exosome substrate targets is observed in mtr4 mutants, suggesting that MTR4 can contribute, but plays rather a minor role for the degradation of non-ribosomal RNAs and cryptic transcripts in Arabidopsis. Consistently, transgene post-transcriptional gene silencing (PTGS) is marginally affected in mtr4 mutants, but increased in hen2 mutants, suggesting that it is mostly the nucleoplasmic exosome that degrades aberrant transgene RNAs to limit their entry in the PTGS pathway. Interestingly, HEN2 is conserved throughout green algae, mosses and land plants but absent from metazoans and other eukaryotic lineages. Our data indicate that, in contrast to human and yeast, plants have two functionally specialized RNA helicases that assist the exosome in the degradation of specific nucleolar and nucleoplasmic RNA populations, respectively. Efficient processing and degradation of RNA is a key process for the post-transcriptional control of gene expression. The main 3′-5′ RNA degradation machine of eukaryotic cells is the exosome, a multi-subunit complex found in both cytoplasm and nuclear compartments [1], [2]. The exosome participates in a plethora of processing and degradation reactions, including the processing of ribosomal RNAs, snoRNAs and snRNAs, the turnover and quality control of mRNAs and the efficient elimination of RNA maturation by-products and diverse RNA species generated from non-genic regions [3]–[5]. In vivo, exosome activity requires the interaction of the exosome complex with associated RNA helicases. In yeast, cytoplasmic and nuclear exosomes are activated by the RNA helicases SKI2 and MTR4, respectively [6]–[8]. In both yeast and human MTR4 is an essential protein required for all functions of the nuclear exosome [9], [10]. Interestingly, Arabidopsis thaliana has two MTR4 homologues, designated AtMTR4 and HEN2 [11]–[13]. We have previously shown that AtMTR4 (encoded by At1g59760) is a predominantly nucleolar protein required for the efficient degradation of misprocessed 5. 8S rRNA precursors and specific fragments of the 5′ external transcribed spacer (5′ ETS), a by-product released during processing of three rRNAs from their common precursor transcript [13]. The requirement for AtMTR4 in efficient rRNA production is reflected by the phenotype of mtr4 mutants, which show a characteristic combination of developmental growth defects also observed in ribosomal protein mutants and in other Arabidopsis mutants lacking putative ribosome biogenesis factors such as nucleolin [14]–[19]. HEN2 (HUA enhancer 2, At2g06990) was originally identified in a genetic screen for mutations that enhance the flower morphology defects observed in hua1 and hua2 mutants [12]. A follow-up study showed that hen2 single mutants accumulate, as compared to wild type plants, slightly higher levels of a polyadenylated transcript comprising the two first exons and a large portion of the second intron of the AGAMOUS gene product, suggesting that the HEN2 protein could be involved in the degradation of misprocessed AGAMOUS transcripts [20]. These data and the strong homology with the exosome activator MTR4 prompted us to examine possible roles of HEN2 in exosome-mediated RNA degradation in Arabidopsis thaliana. In contrast to AtMTR4, HEN2 is not required for processing or degradation of 5. 8S rRNA precursors or the elimination of the 5′ ETS [13]. We show here that HEN2 is a nucleoplasmic protein that is associated with the Arabidopsis exosome core complex and has a specific role in the exosome-mediated degradation of non-coding RNAs, misprocessed mRNAs, introns and transcripts derived from retrotransposons and non-genic regions. Interestingly and as recently reported for human MTR4 [9], [21], HEN2 associates with homologues of the NEXT (for Nuclear Exosome Targeting) complex components and also co-purifies with the cap-binding complex. MTR4, by contrast, is associated with a distinct set of proteins, many of which appear to be involved in ribosome biogenesis. Our results indicate a high degree of spatial and functional specialization of exosome activating RNA helicases in Arabidopsis. Several co-factors of the Arabidopsis exosome such as RRP6L1 (RRP6-LIKE 1), RRP6L2, MTR4 and DIS3/RRP44 have been identified based on sequence homology with yeast counterparts and are genetically linked with nuclear exosome functions [13], [22]–[25]. However, none of them has yet been shown to physically interact with the exosome. To better define the composition of plant exosome complexes, we used myc-tagged and GFP-tagged versions of the exosome core subunit RRP41 as baits in co-immunoprecipitation (IP) experiments. RRP41 fusion proteins were expressed under the control of the 1 kb genomic region upstream of the endogenous RRP41 gene. Both myc-tagged RRP41 and GFP-tagged RRP41 constructs were able to complement the otherwise lethal rrp41-X null mutation. A similar full phenotypic complementation was previously reported for TAP-tagged RRP41 [5]. Complementation of the null mutation by both RRP41-myc and RRP41-GFP suggested that both fusion proteins can be integrated in the core exosome complex. To test this hypothesis and identify potential exosome co-factors, tagged-RRP41 and associated proteins were affinity-purified using superparamagnetic particles coated with anti-myc or anti-GFP monoclonal antibodies, respectively. As shown for myc-tagged RRP41 IP, a specific group of proteins were visualized on silver-stained SDS-PAGE gel as compared with mock IP (Fig. 1). The proteins co-purifying with RRP41 were identified by mass spectrometry (nano LC-MS/MS) analysis. A final list of 14 proteins was established by excluding proteins present in mock purifications and by crossing the datasets of three biological repeats (Table 1). All 14 proteins were identified with both Mascot and PEAKS DB algorithms with a false discovery rate <1%. An exhaustive list of the peptides shared by the three RRP41 IPs is shown in Table S1. As expected, all nine canonical core subunits of EXO9 (RRP41, RRP42, RRP43, RRP45B, RRP46, MTR3, CSL4, RRP4 and RRP40A) were identified which confirms that we indeed purified intact exosome complexes (Table 1, Table S1). In addition to the nine exosome subunits that were previously characterized [5], five novel proteins were detected, albeit with lower number of spectra reflecting a lower abundance as compared to the canonical EXO9 subunits (Table 1, Table S1). AtMTR4 and HEN2 share 43% identity and 59–60% similarity with yeast MTR4 and with each other, but only 24% identity/39% similarity with SKI2, a cofactor of the cytoplasmic exosome [7], [32], [33]. Structural modeling of HEN2 and AtMTR4 confirmed that both possess an arch domain, a characteristic feature of MTR4/SKI2 RNA helicases [34]–[37]. While the modeled structure of HEN2 matches closely to the structure of yeast MTR4, AtMTR4 has an insertion of nine amino acids in the RNA binding part of the arch domain, the KOW-motif [34], [35]. Interestingly, a similar insertion is present in the KOW motifs of all plant MTR4 proteins investigated (Fig. S1). Other characteristic sequence differences between AtMTR4 and HEN2 concern RecA-domains, the arch domain and the C-terminal helix-loop-helix domain (Fig. S2), respectively, and allow the reliable discrimination of HEN2 and AtMTR4 homologues by sequence alignment algorithms. A search for homologues of AtSKI2, AtMTR4 and HEN2 in all genomes available at www. phytozome. net shows that all three RNA helicases are conserved throughout green algae, mosses and land plants. A phylogenetic analysis of related proteins from animals, fungi and other eukaryotic clades revealed that most organisms possess both a single MTR4 and a single SKI2 protein; however, HEN2 homologues are restricted to the green lineage (Fig. 2). Despite of the short insertion in the KOW motif, plant MTR4 proteins cluster with the MTR4 proteins of other organisms, while HEN2 proteins form a separate clade. Taken together, these data suggest that HEN2 is a plant-specific isoform of the nuclear exosome activator MTR4. To extend previous localization studies [13], we transiently expressed HEN2 and AtMTR4 GFP fusion proteins in Nicotiana benthamiana leaves, alongside with RFP-labeled XRN2 and Fibrillarin as nucleolar markers, and with SRP34 as a nucleoplasmic marker [38]–[43]. Similar to XRN2-RFP and Fibrillarin-RFP, AtMTR4-GFP was detected in the nucleus, strongly enriched in nucleoli (Fig. S3). HEN2-GFP and SRP34-RFP were detected only in the nucleoplasm (Fig. S4). Next, we determined the intracellular localization of HEN2-GFP in root tips of stable Arabidopsis thaliana transformants. For comparison, we examined roots of plants expressing either AtMTR4-GFP or the exosome core subunits RRP4-GFP and RRP41-GFP. As expected, RRP4-GFP and RRP41-GFP were observed in both cytoplasm and nuclei, and enriched in nucleoli (Fig. 3, Fig. S5). As reported before, AtMTR4-GFP was observed predominantly in nucleoli, and only a faint signal was observed in the nucleoplasm (Fig. 3, Fig. S5) [13]. HEN2-GFP was observed in the nucleoplasm, was enriched in nuclear foci and appeared excluded from nucleoli (Fig. 3, Fig. S5, Fig. S6). These results show that HEN2 is a nucleoplasmic protein, and that AtMTR4 and HEN2 are for the most part located in distinct subnuclear compartments. To investigate whether AtMTR4 and HEN2 are associated with specific proteins reflecting their distinct localization and to confirm that both helicases interact with EXO9, we performed IP experiments using plant lines expressing GFP-tagged versions of AtMTR4 or HEN2 in their respective mutant backgrounds. The list of proteins co-purifying with AtMTR4-GFP or HEN2-GFP was established by considering only proteins that were not identified in mock purifications and common to replicate IPs for AtMTR4-GFP and HEN2-GFP experiments, respectively (Tables 2 and 3; Tables S2 and S3). All canonical nine subunits of EXO9 were identified in both AtMTR4-GFP and HEN2-GFP datasets, which confirmed that both RNA helicases interact with the Arabidopsis exosome complex. Remarkably, EXO9 subunits were the sole common proteins among the 43 and 16 significant proteins present in AtMTR4-GFP and HEN2-GFP IPs, respectively (Tables 2 and 3). A Gene Ontology (GO) analysis for the 34 proteins that were specifically co-purified with AtMTR4-GFP exposed that the most significant biological process GO term was ribosome biogenesis (Benjamini-Hochberg corrected p-value, 1. 3E-12), which tagged 9 out of 34 proteins. Further data mining revealed that additional 12 proteins have a proven or predicted role in ribosome biogenesis (Table 2). Nine out of the 13 remaining proteins corresponded to transducin/WD40 repeat proteins and/or proteins involved in nucleic acid metabolism (Table 2, Table S2). These results are in agreement with our previous results [13] and further substantiate the role of AtMTR4 in maturation and/or degradation of ribosomal RNA. HEN2-GFP co-purified the nine canonical subunits of EXO9, the alternative subunit RRP45A and 6 additional proteins (Table 3, Table S3). One of the six proteins that co-purified with HEN2-GFP was a homologue of the exon junction complex (EJC) component MAGO NASHI. Two proteins were the subunits of the cap binding complex (CBC), CBP80 (AT2G13540) and CBP20 (AT5G44200). Finally we identified three putative RNA binding proteins, two of which encoded by AT5G38600 and AT4G10110 had high spectral counts (Table 3, Table S3). AT5G38600 is a 532 amino acid protein containing the CX2CX4HX4C zinc-knuckle motif (Pfam14392), particularly found in plant proteins [44]. A BLAST analysis against the human proteome identified ZCCHC8 as the best sequence homologue (E-value 4e-26). ZCCHC8 is a zinc-knuckle protein that was recently identified as part of the human Nuclear EXosome Targeting (NEXT) complex [9]. At4G10100 is a small protein of 173 amino acids that shares some similarity is related to the second component of the human NEXT complex, RBM 7, although the similarity is restricted to the N-terminal two-thirds of the 266 amino acids of RBM 7 (26% identity, 27% similarity for the aligned sequence, Fig. S7). To further check whether HEN2, AT5G38600 and AT4G10110 form a NEXT-like complex in Arabidopsis, we slightly increased the stringency of HEN2 immunoprecipitation conditions. By a modest increase of ionic strength from 50 to 150 mM NaCl, the co-purification of EXO9 with HEN2-GFP was lost. However, both AT5G38600 and AT4G10110 were still present in duplicate immunoprecipitations (Table 3, Table S3). Interestingly, a third RNA-binding protein, AT1G67210, was identified in all four HEN2 IPs, albeit with a lower spectral count (Table 3, Table S3). As AT5G38600, AT1G67210 also contains the CX2CX4HX4C zinc-knuckle motif (Pfam14392) and both proteins share 49. 6% identities and 57. 6% similarities. These results support the existence of a NEXT-like complex in Arabidopsis and raise the interesting possibility that multiple NEXT-like complexes exist in plants. Taken together, our data show that AtMTR4 and HEN2 are associated with distinct sets of proteins. AtMTR4 co-purifies with the exosome, and with putative ribosome biogenesis factors, which highlights the function of AtMTR4 in pre-rRNA processing and degradation. HEN2 co-purifies with the exosome, the CBC complex and with two types of RNA-binding proteins to form a putative plant NEXT-like complex. These data suggest that the functional link between exosome, CBC and NEXT complexes that was recently established in human cells [21] may be conserved in plants. Furthermore, these results strengthen the evidence for a functional specialization of HEN2 and MTR4. So far, our data suggested that HEN2 might operate as cofactor of the nucleoplasmic exosome complex. In order to investigate the function of HEN2 for the degradation of exosome substrates, we tested the accumulation of a pseudogene and five non-coding RNAs selected from the list of known polyadenylated plant exosome substrates [5]. Targets comprised the pseudogene At1g79245, the non-coding RNAs MRP and 7SL, the dicistronic precursor of snoRNAs At3g58193 and At3g58196, a non-coding RNA encoded by At2g18440, and intergenic transcripts generated from a repeat region located on chromosome 5 [5]. Additional information and hyperlinks to visualize the selected regions on the SALK transcriptome/exosome website (http: //signal. salk. edu/) are provided in Fig. S8. Steady-state levels of the six selected exosome targets were determined by quantitative RT-PCR using oligo-dT primed cDNA samples prepared from seedlings of wild type, mtr4-1, mtr4-2, hen2-2 or hen2-4 mutant plants. We included also samples from RRP41 RNAi lines in which depletion of the exosome core subunit RRP41 is triggered by an inducible RNAi construct [5]. As shown in Fig. 4, all exosome substrates tested in this experiment were over-accumulated in hen2 samples as compared to wild type samples. By contrast, no or only a mild accumulation was observed in mtr4 mutants. These data provided a first indication that HEN2 is involved in the degradation of nuclear exosome targets that are not substrates of AtMTR4. To evaluate the respective contribution of HEN2 and AtMTR4 to the degradation of nuclear exosome substrates in an unbiased manner, we determined the accumulation of polyadenylated transcripts using full-genome (tiling) microarray arrays. For this experiment, cDNA was prepared from two biological replicates of wild type, mtr4-1 and hen2-4 mutants. Each mutant sample was co-hybridized against a wild type sample to NimbleGen A. thaliana 732K whole genome microarrays. The microarray chip contains 1,434,492 strand-specific probes covering both coding and non-coding regions with an average resolution of 175 nt. Probes are 45–85 nt long and designed to achieve a constant Tm of ∼76°C to enhance hybridization consistency across probes. For each biological replicate, expression of each mutant was compared to the expression of the wild type. The statistical analysis, based on a 4-state Hidden Markov Chain, classified probes into four clusters corresponding to over-expressed probes, under-expressed probes, probes with unchanged expression, and noise (not expressed probes), respectively. Interestingly, the analysis did not declare any probe as under-expressed in both biological replicates. This result is in line with the prediction that loss of HEN2 and AtMTR4 impairs RNA degradation, and therefore results predominantly in an increased accumulation of RNA substrates. Indeed, signals for 1860 unique probes were significantly increased in both biological replicates of hen2 samples. 499 probes were identified as overexpressed in both biological replicates of mtr4 samples. A file allowing the visualization of the upregulated probes aligned to the Arabidopsis genome can be found in dataset S1. For the further analysis, we sorted the probes according to their genome coordinates to identify upregulated regions. Only regions with at least two consecutive probes were considered for interpretation. Upregulated regions were then grouped with respect to annotated features taking into account both TAIR10 annotated genes and recently identified genes encoding snoRNAs, miRNAs and lincRNAs [45]–[47]. This procedure identified 387 regions, the majority of which was upregulated exclusively in hen2 samples (Tables S4, S5, S6, S7, S8, S9, S10, S11). 237 of the upregulated regions mapped to protein coding genes. However, for the majority of the cases, (149 regions, 112 of which were only observed in hen2 samples, Table S4), the upregulated transcripts were apparently not mature mRNAs. In fact, most of the upregulated regions corresponded to short portions of protein coding genes (Table S4). The upregulation of short regions located in the 3′ portion of protein coding genes was validated by qRT-PCR analysis for three examples (Fig. 5). As a positive control of exosome-mediated RNA degradation, we used the RRP41 RNAi line. In all three cases, we indeed observed the accumulation of transcripts corresponding to 3′ regions in both the hen2 and the rrp41 samples (Fig. 5). Another group of upregulated regions mapped within the body of the transcripts and beyond mature 3′ ends (Fig. 6, Table S4), indicative of alternative 3′ end processing or readthrough transcription. Furthermore, many upregulated regions contained both exonic and intronic sequences, suggesting the accumulation of incompletely spliced transcripts. To test this possibility, we compared by qRT-PCR the steady-state levels of individual exons, introns, regions comprising unspliced intron-exon junctions and correctly spliced transcripts from selected loci (Fig. 7, Fig. S9). These experiments confirmed the overaccumulation of transcripts comprising unspliced donor or acceptor sites in two independent T-DNA insertion alleles of HEN2 and in RRP41 RNAi plants (Fig. 7A, Fig. S9). For most loci, we detected the upregulation of both unspliced and spliced transcripts (albeit to different levels, please note the scales in Fig. 7 and S9), suggesting that heterogeneous transcripts are produced and targeted for degradation. In order to map the 3′ extremities of the unspliced transcripts, we amplified and sequenced transcripts derived from the At1g79270 locus by 3′ RACE-PCR (Fig. 7B). cDNA synthesis was initiated by oligo dT, and PCR products were amplified with nested forward primers situated in the 3′ region of the first exon, and the adapter sequence of cDNA synthesis primer as a reverse primer. All PCR products amplified from WT, mtr4 or RRP41 control plants corresponded to the fully spliced mature mRNA (Fig. 7B). By contrast, a smaller product of only about 500 bp was amplified from hen2 or RRP41 RNAi plants, and corresponded to a population of transcripts that comprised the unspliced donor site of the first intron (Fig. 7B, Fig. S10), while 3′ extremities were close to the acceptor site. Of the 20 clones that were obtained from the hen2-4 sample 18 were polyadenylated at or close to the intron acceptor site (Fig. S10), indicating that they are indeed marked for degradation by the nuclear exosome. The remaining 2 clones had polyadenylation sites 8 and 52 nt upstream of the acceptor site and likely represent degradation intermediates (Fig. S10). Hence, both qRT-PCR and 3′ RACE results confirmed the results of the microarray analysis and show that polyadenylated transcripts with incorrectly spliced introns accumulate in hen2 plants. The tiling data suggested that HEN2 also participates in the elimination of excised introns. In fact, 34 of the 237 protein coding regions that were detected in the microarray analysis corresponded exclusively to intron regions (Table S5), 28 of which were only observed upon loss of HEN2. As for incompletely spliced transcripts, qRT-PCR experiments confirmed the accumulation of intronic regions in two independent alleles of hen2 and in RRP41 RNAi samples (Fig. 8). Finally, 54 regions detected by the tiling analysis likely corresponded to mature mRNAs since they were regions with all probes in exons, and the detected regions spanned at least 50% of the mRNA. 33 mRNAs were detected in hen2,9 mRNAs in mtr4, and 12 in both hen2 and mtr4 samples (Table S6). The upregulation of the pseudogene At1g79245 (Fig. 4) and several other mRNAs in hen2 and mtr4 samples was validated by qRT-PCR (Fig. S11). However, not all of the tested mRNAs were also found upregulated in RRP41 RNAi samples (Fig. S11). Moreover, none of the positively tested mRNAs was previously identified as an exosome-regulated mRNA [5]. Finally, some mRNAs were only detected in the very same samples that have been used for the microarray, but not in other mutant plants grown in the same culture conditions (Fig. S12). This inconsistence was in sharp contrast to all other types of substrates that were tested in the course of the study, which were reproducibly detected in all replicates, in independent hen2 T-DNA insertion mutants, and in RRP41 RNAi lines. Therefore, we doubt that all of the mRNAs that were detected in our tiling analysis represent true substrates of exosome-mediated decay. Although nuclear degradation can probably contribute to mRNA degradation [3], [5], [48], the upregulation of mRNAs can also be explained by indirect effects, e. g. a differential response of WT and mutant plants to growth conditions. Indeed, data mining revealed that many of the mRNAs detected by the tiling arrays are linked to stress response (Table S12). Hence, the majority of the mRNAs that we detected by the tiling array are probably not bona fide substrates of HEN2 or the nuclear exosome. By contrast, the upregulation of short mRNA-derived transcripts (Fig. 5, Table S4), 3′ extended mRNAs (Fig. 6, Table S4), unspliced transcripts, (Fig. 7, Table S4, Fig. S9, Fig. S10), introns (Fig. 8, Table S5) is consistently detected in all replicates of both mutant alleles of hen2 and in RRP41 RNAi samples, and can be considered as bona fide substrates of the nuclear exosome and the RNA helicase HEN2. Taken together, our data indicate that loss of HEN2 results in the accumulation of short transcripts derived from mRNA regions, 3′ extended transcripts, incompletely spliced mRNA transcripts, and excised introns. The polyadenylated status of the accumulated transcripts indicates that they are tagged for degradation by the exosome, and indeed, all these classes of mRNA-derived transcripts have been described as exosome targets [5]. Hence, the most straightforward explanation for the accumulation of these transcripts in hen2 mutants is that HEN2 is required for the exosome-mediated elimination of different types of probably unfunctional RNAs that are generated from protein coding genes. Only a small number of such transcripts were observed in mtr4 mutants (Table S4, S5) and accumulated at lower levels (Fig. 5–8, Fig. S9), indicating that MTR4, as compared with HEN2, has a rather minor contribution to nuclear mRNA surveillance. Next, we examined the contribution of HEN2 and MTR4 to the degradation of non-coding transcripts. Of the 387 upregulated regions detected in the microarray analysis, 150 regions mapped to non-coding regions. 9 regions mapped to transposable elements (6 in hen2,3 in mtr4) and were not further investigated (Table S7). 45 regions, all of which were exclusively observed in hen2 mutants, contained one or more snoRNA genes (Table S8), including the regions encoding snoRNAs At3g58193 and At3g58196 (Fig. 4). In fact, almost all of the snoRNA regions detected in our tiling array have also been previously identified as exosome substrates (see Table S8 last column, and [5]). To further investigate the contribution of HEN2 to the degradation of snoRNA precursors, we tested two additional snoRNA regions by qRT-PCR (Fig. 9). The results indicated a strong accumulation of snoRNA precursors in both hen2 mutant lines and in RRP41 RNAi samples. The preferential accumulation of snoRNA precursor transcripts was further confirmed by transferring 3′ RACE-PCR products to membranes followed by hybridisation with radiolabelled probes (Fig. S13). These data strongly indicate that HEN2, but not MTR4, plays an important role for the degradation of snoRNA precursors. Other non-coding RNA regions were also unequally distributed between hen2 and mtr4 samples. 22 regions in hen2 samples encoded lincRNAs, putative miRNA precursors or other non-coding RNAs (Table S9), among them a portion of the region encoding GUT15 (for gene with unstable transcript 15) /At2g18440 (Fig. 4). Only 2 of such non-coding RNA regions were observed in mtr4 samples (Table S9), indicating that MTR4 plays a minor role for the degradation of non-ribosomal non-coding RNAs. Similarly, we detected 29 putative antisense transcripts in hen2 samples, while only 2 potential antisense regions were upregulated in mtr4 samples (Table S10). To confirm the polyadenylated status of antisense transcripts, we selected one of the potential antisense regions for 3′ RACE experiments. Cloning and sequencing of the PCR products revealed that 0 of 32 clones obtained from WT samples corresponded to the target sequence (Fig. 10, Fig. S14). By contrast, 28 of 32 clones obtained from hen2 samples and 3 of 32 clones obtained from mtr4 samples corresponded indeed to antisense transcripts derived from the target region (Fig. 10, Fig. S14). Antisense sequences were polyadenylated, a hallmark of exosome-mediated RNA degradation, and between 67 and 208 nt long (Fig. 10, Fig. S14). These data strongly suggest that the antisense transcripts derived from the At5g44306 locus are indeed substrates of polyadenylation-mediated decay facilitated by HEN2 and the RNA exosome. Finally, the microarray analysis detected 43 regions without any annotated genome features, including the intergenic repeat region on chromosome five that was already detected by our initial qRT-PCR experiments (Fig. 4). Similar to the distribution of non-coding RNA regions and potential antisense transcripts, the majority of the non-annotated regions (38 of 42) were exclusively observed in hen2 samples, while only 4 of 42 regions were found in mtr4 samples. These results indicate that the elimination of spurious transcripts generated from antisense or non-annotated regions that are usually described as the “dark matter” of the transcriptome relies mostly on HEN2. Taken together, the microarray analysis revealed that a large number of exosome targets, including short or incompletely spliced transcripts derived from mRNA genes, precursors and processing by-products of non-coding RNAs, and spurious transcripts generated from antisense and intergenic regions accumulate specifically in hen2 mutants (Fig. 11). This indicates that HEN2 has a major function in the elimination of many different types of nuclear exosome substrates. A much smaller number of such non-ribosomal exosome substrates accumulated in mtr4 plants (Fig. 11), and average accumulation levels in mtr4 samples were lower than in hen2 samples (Fig. S15). These data indicate that AtMTR4, though it can apparently contribute at least to the degradation of mRNA-derived transcripts, plays a rather minor role in nuclear RNA surveillance. Previous data indicated that mutations in 5′-3′ exoribonucleases in nucleoli (XRN2), nucleoplasm (XRN3) or cytoplasm (XRN4) enhance the efficiency of post-transcriptional gene silencing mediated by sense transgenes (S-PTGS) [49]. Mutations in the exosome core components RRP4 and RRP41, or mutations in the exosome co-factors RRP44/DIS3 and RRP6L1 also enhance S-PTGS, suggesting that both 5′-to-3′ and 3′-to-5′ RNA degradation pathways limit the entry of aberrant transgene RNAs into the S-PTGS pathway [50]. To determine in which nuclear compartment the exosome counteracts transgene PTGS, we analyzed the effect of mtr4 and hen2 mutations on S-PTGS using the Arabidopsis reporter line Hc1 [49]–[52]. Line Hc1 carries a 35S: : GUS transgene that triggers S-PTGS at a frequency of 20% at each generation, making this line ideal for identifying mutations that either increase or decrease S-PTGS efficiency. However, only EMS mutants or 35S CaMV promoter-free T-DNA insertion mutants are amenable for such analyses because copies of the 35S CaMV promoter present in the SALK, WISC or GABI T-DNA collections often interfere with the expression of 35S CaMV promoter-driven transgenes, which could report an impact on S-PTGS that is not directly related to the function of the mutated gene [53]. Only the T-DNA mtr4-2 [13] and the EMS hen2-1 mutant [12] fit this requirement. Hc1/mtr4-2 plants triggered S-PTGS at a frequency of 25% (n = 96), which is only slightly higher than S-PTGS frequency in Hc1 controls (Fig. 12A). These data indicate that compromising exosome activity in nucleoli has only a limited effect on transgene S-PTGS. For comparison, loss of the nucleolar exoribonuclease XRN2 was previously shown to trigger S-PTGS at a frequency of 47% [49] (Fig. 12A). In contrast, compromising the nucleoplasmic 5′-to-3′ exonuclease XRN3 had a stronger effect (Fig. 12A) [49]. To determine if loss of the nucleoplasmic protein HEN2 also affects S-PTGS, the hen2-1 mutation, which is in the Landsberg erecta (Ler) ecotype, was crossed to an Hc1/Ler line resulting from ten backcrosses of Hc1 to Ler. Remarkably, no Hc1/Ler plants exhibited S-PTGS (n = 96, Fig. 12B), suggesting that either Ler is less prone to trigger S-PTGS than Col or that Hc1 has lost its capacity to trigger S-PTGS after ten backcrosses to Ler. This later hypothesis was ruled out by crossing Hc1/Ler to xrn4-1 (in Ler). Mutations in the cytoplasmic 5′-to-3′ exonuclease XRN4 are known to enhance S-PTGS in Col [49] (Fig. 12A), so S-PTGS was expected to occur in Hc1/xrn4-1/Ler plants if the Hc1 locus has retained its ability to trigger S-PTGS in Ler. S-PTGS was observed in 100% of Hc1/xrn4-1/Ler plants (n = 96, Fig. 12B), indicating that Ler is less prone to trigger S-PTGS than Col, but that the Hc1/Ler line still is amenable to identify mutations that enhance S-PTGS. Indeed, 23% of Hc1/hen2-1/Ler plants (n = 83) exhibited S-PTGS (Fig. 12B), indicating that compromising exosome activity in the nucleoplasm strongly enhances transgene S-PTGS. The antagonistic effect of HEN2 on S-PTGS was confirmed in Col using an AGO1 transgenic reporter system. In this system, transformation of wild type Col with a T-DNA carrying an ectopic pAGO1: : AGO1 construct triggers cosuppression (S-PTGS) of endogenous AGO1 in 50% of the transformants [54] (Fig. 12C). Transformation of hen2-2 (in Col) with the same pAGO1: : AGO1 construct triggered AGO1 cosuppression in 72% of the transformants (Fig. 12C). This increase is almost comparable to the effect of xrn4 on AGO1 cosuppression (Fig. 12C), indicating that hen2 strongly affects S-PTGS in Col. Taken together, these data show that HEN2 counteracts S-PTGS in both Col and Ler, likely through its role as a co-factor of the nucleoplasmic exosome. We show here that two isoforms of MTR4, AtMTR4 and HEN2, assist the Arabidopsis exosome for the degradation of mostly distinct sets of nuclear RNA substrates. Both AtMTR4 and HEN2 co-purify with the Arabidopsis exosome core complex but AtMTR4 and HEN2 occupy primarily distinct intranuclear compartments. The main role of the nucleolar isoform AtMTR4 is to assist in the exosome-mediated degradation of misprocessed rRNA precursors and maturation by-products [13]. Our new data show that the main function of the nucleoplasmic isoform HEN2 is to assist the exosome-mediated processing and/or degradation of snoRNAs and snoRNA precursors, miRNA precursors, lincRNAs, and a large number of spurious transcripts derived from antisense and non-annotated regions. In addition, HEN2 is involved in the degradation of excised introns and incompletely spliced or otherwise mis-transcribed or mis-processed mRNAs. Of the 387 regions detected in this study, 100 have been previously identified as targets of the Arabidopsis core exosome (Tables S4, S5, S6, S7, S8, S9, S10, S11) using a different type of tiling microarray [5]. However, our qRT-PCR data indicate that majority of the HEN2 substrates accumulate also in RRP41 RNAi samples, even though some of them were not detected previously [5], such as At1g79270 (Fig. 6), At3g26510 (Fig. S9) and At1g58602 (Fig. S9). Vice versa, several known exosome substrates were not identified by our tiling analysis but easily detected by qRT-PCR (Fig. S16). These findings indicate that both tiling studies probably underestimate the contribution of HEN2 and the exosome to RNA surveillance. Only a small number of exosome substrates are observed in mtr4 mutants, which indicates that AtMTR4 participates, but plays only a minor role in nuclear RNA surveillance. This is in line with the finding that AtMTR4 and HEN2 have marginal and strong contributions, respectively, to the exosome activity that likely degrades aberrant transgene RNAs in the nucleus to limit their entry in the PTGS pathway [49], [50]. In a previous study, we have shown that hen2 mutants do not accumulate the 5. 8S rRNA precursors and the 5′ ETS that are observed upon down-regulation of AtMTR4 [13]. Accordingly, hen2 single mutants do not display any of the developmental defects linked to disturbed ribosome biogenesis or ribosome function that were observed in mtr4 mutants. These data were the first clues that AtMTR4 and HEN2 have rather distinct functions in plants. However, we were not able to obtain double mtr4 hen2 mutants, signifying that simultaneous loss of both AtMTR4 and HEN2 is lethal [13]. The data presented here suggest that AtMTR4 can perform some of the functions of HEN2. For instance, a limited number of mRNA-derived fragments, non-coding RNAs and spurious transcripts accumulated in mtr4 single mutants, indicating that AtMTR4 can contribute, even in presence of HEN2, to the degradation of non-ribosomal exosome targets. In line with this, a small fraction of AtMTR4-GFP can be detected in the nucleoplasm of stable Arabidopsis transformants (Fig. S17, see also [13]). Taking together, these results indicate that a limited overlap between AtMTR4 and HEN2 functions exists. However, our data show that most exosome functions are activated either by AtMTR4 or by HEN2 in the nucleolus and the nucleoplasm, respectively. It is interesting to note that Arabidopsis has also specific nucleoplasmic and nucleolar isoforms (RRP6L1 and RRP6L2, respectively) of RRP6, a catalytically active exoribonuclease associated with nuclear exosomes in yeast and human [28], [30]. So far, we and others [5] did not detect any of the plant RRP6-like proteins in plant exosome preparations. However, we have previously shown that the downregulation of the nucleolar isoform RRP6L2 leads to a mild accumulation of misprocessed 5. 8S rRNA precursors and the 5′ ETS, suggesting that RRP6L2 acts in the same degradation processes as AtMTR4 [13], [55]. By contrast, we did not detect a significant overaccumulation of HEN2 targets upon down-regulation of the nucleoplasmic isoform RRP6L1 (data not shown). A possible explanation is that RRP6L2 and RRP6L1 can substitute for each other in the degradation of HEN2 targets, since the two nuclear RRP6-like proteins appear to have both specific and common roles linked to the degradation of exosome targets [24], [55]. Interestingly, a recent study revealed that RRP6L1, but not RRP6L2 has also a role in transcriptional silencing by retaining PolV transcripts on chromatin, thereby promoting the production of 24 nt siRNAs that direct DNA methylation via the RdDM pathway [56]. Remarkably, this function of RRP6L1 is independent of the core exosome [56]. By contrast, transcriptional silencing at soloLTR loci is mediated by both RRP6L and the core exosome [24]. In addition, RRP6L1 and the exosome core complex have a common role in 21-nt siRNA-dependent posttranscriptional silencing (PTGS), since downregulation of either RRP41 or RRP6L1 alone is sufficient to enhance PTGS in the sensitive Hc1-GUS reporter system that was used in this study [50]. Hence, the role of RRP6L1 in PTGS is likely linked to exosome-mediated RNA degradation, suggesting that HEN2 and RRP6L1 are involved in at least one similar function. In animals and fungi, a single MTR4 protein is present in nucleoplasm and nucleoli, and essential for both processing/degradation of rRNA precursors and the elimination of all other nuclear exosome substrates [8]–[10], [57]. However, both yeast and human MTR4 proteins are incorporated in more than one exosome activator/adapter complex. Yeast MTR4 is detected in TRAMP4 and TRAMP5 (for TRF4/5 AIR1/2 MTR4 Polyadenylation), each of which comprises a RNA binding protein and a non-canonical poly (A) polymerase [58]–[61]. Although TRAMP4 and 5 have a similar composition and many redundant functions, TRAMP5 seems be more important for the polyadenylation of pre-rRNAs while TRAMP4 might be more important for the degradation of other non-coding RNAs and intergenic transcripts [62]–[64]. The functional specialization between nucleolar and nucleoplasmic exosome activator complexes is clearer in human. In nucleoli, hMTR4 is incorporated in a TRAMP-like complex which polyadenylates rRNA maturation by-products [9]. In the nucleoplasm, hMTR4 is associated with the NEXT (for Nuclear EXosome Targeting) complex, which targets PROMPTS (PROMoter uPstream TranScripts) for degradation by the exosome [9]. Hence, both yeast and animals possess nucleolar and nucleoplasmic exosome activators, which share MTR4 as a central component. By contrast, an exosome activating system with two specialized RNA helicases has evolved early in the green lineage. Interestingly, both the nucleoplasmic fraction of human MTR4 and the Arabidopsis nucleoplasmic-specific RNA helicase HEN2 appear associated with similar RNA binding proteins to form NEXT and NEXT-like complexes, respectively, and with the cap-binding complex [9], [21, this study]. These findings suggest a high degree of functional conservation between the nucleoplasmic fraction of human MTR4 and plant HEN2. By contrast, a TRAMP-like complex comprising a non-canonical poly (A) polymerase remains to be identified in plants. Hence, the emerging picture is that only the core exosome machinery is conserved in all eukaryotes, while exosome-associated activities and activating complexes show intriguing diversity and complexity in fungi, insects, animals and plants. With the exception of hen2-1 [12] used for the S-PTGS assay, all Arabidopsis thaliana plants were of Columbia ecotype (Col-0). T-DNA insertion lines were retrieved from NASC (http: //arabidopsis. info/). mtr4-1, mtr4-2, hen2-2 and hen2-4 lines are described in [13]. RRP41 RNAi lines are described in [5]. The S-PTGS reporter line Hc1 was first described in [51]. Hc1/xrn/col lines are described in [49]. Unless stated otherwise, plants were grown on soil at 20°C with cycles of 16 h light/8 h darkness. Sequences were retrieved from Phytozome (http: //www. phytozome. net), Metazome (http: //www. metazome. net) and JGI (http: //genome. jgi. doe. gov) genome databases, using AtMTR4, HEN2 and AtSKI2 proteins as BLAST queries. Structures of AtMTR4 and HEN2 were modeled with MODELLER Software (http: //modbase. compbio. ucsf. edu/ModWeb20-html/modweb. html) using the crystal structures of S. cerevisiae MTR4 (PDB 3L9O and 2XGJ) [34], [35] as templates. Alignments were performed with Chimera (http: //www. cgl. ucsf. edu/chimera, for structure-based sequence alignments) and ClustalX (http: //www. clustal. org, for phylogenetic analysis). The phylogenetic tree was calculated with the neighbor-joining algorithm built in ClustalX and 1000 bootstraps, and drawn with Figtree (http: //tree. bio. ed. ac. uk/software/figtree). For the expression of RRP4, AtMTR4 and HEN2 GFP fusion proteins under the control of the 35S CaMV promoter, the coding sequences of RRP4, AtMTR4 and HEN2 were amplified from cDNA and cloned into vector pK7FWG2 [65]. For expression of GFP-tagged or myc-tagged RRP41 the genomic sequence of RRP41 including 1 kb upstream of the RRP41 gene was cloned into vectors pGWB604 and pGWB616, respectively [66]. For immunoprecipitations (see below), AtMTR4-GFP was expressed under the control of its own promoter. To do so, a genomic region comprising 1 kb upstream of the MTR4 gene, the first two exons and the first intron were fused to the CDS downstream of the second exon and cloned into pGWB604 [66]. Constructs that allow the expression of RFP-tagged FIB1 [67], XRN2 and SRP43a [68] were kind gifts of Jane Brown and Martin Crespi, respectively. Infiltration of N. benthamiana leaves was performed as described in [50] except that P19 was used as a suppressor of silencing. Arabidopsis plants were transformed by the floral dip method [69]. Root tips of stable transformants were examined 8 days after germination by confocal microscopy. RRP41-myc, RRP41-GFP, AtMTR4-GFP and HEN2-GFP-fusion proteins were extracted from flowers of stable Arabidopsis transformants and purified using magnetic microparticles coated with monoclonal myc or GFP antibodies (MACS purification system, Miltenyi Biotech) according to the manufacturer' s instructions except that SDS was omitted from washing buffers. Co-IP experiments were carried out in triplicates for RRP41 and MTR4 and duplicates for HEN2 with 50 mM and 150 mM NaCl. For in-gel digestion, samples were separated by SDS-PAGE followed by trypsic digestion and peptide extraction as described in [70]. Otherwise, proteins were eluted directly from magnetic beads in 1× Laemmli buffer, precipitated with 100 mM ammonium acetate in methanol, and resuspended in 50 mM ammonium bicarbonate. After reduction and alkylation steps with 5 mM dithiothreitol and 10 mM iodoacetamide, respectively, proteins were digested overnight with trypsin 1/25 (w/w). Vacuum dried peptides were re-suspended in 15 µl 0. 1% FA (solvent A). One third of each sample was injected on a NanoLC-2DPlus system (nanoFlex ChiP module; Eksigent, ABSciex, Concord, Ontario, Canada) coupled to a TripleTOF 5600 mass spectrometer (ABSciex) operating in positive mode. Peptides were loaded on C18 columns (ChIP C-18 precolumn 300 µm ID × 5 mm ChromXP and ChIP C-18 analytical column 75 µm ID × 15 cm ChromXP; Eksigent) and were eluted using a 5%–40% gradient of solvent B (0. 1% FA in Acetonitrile) for 60 minutes at a 300 nl/min flow rate. The TripleTOF 5600 was operated in high-sensitivity data-dependant acquisition mode with Analyst software (v1. 6, ABSciex) on a 350–1250 m/z range. Up to 20 of the most intense multiply-charged ions (2+ to 5+) were selected for CID fragmentation, with a cycle time of 3. 3s (TOP 20 discovery mode). For protein identification, raw data were converted to Mascot Generic File format (mgf) and searched against a TAIR 10 database supplemented with a decoy database build from reverse sequences. Data were analyzed using Mascot algorithm version 2. 2 (Matrix Science, UK) through ProteinScape 3. 1 software (Bruker). Search parameters allowed N-acetylation (protein N-terminal), carbamidomethylation (C) and oxidation (M) as variable peptide modifications. Mass tolerances in MS and MS/MS were set to 20ppm and 0. 5Da, respectively. 2 trypsin mis-cleavages were allowed. Peptide identifications obtained from Mascot were validated with a FDR <1%. A second algorithm, PEAKS DB (version 5. 3, BSI Informatics) was used with the same search parameters to strengthen the identifications. Identified proteins were assessed by the total number of fragmented spectra per protein (spectral count). Data of three (RRP41, MTR4) or two (HEN2 50 mM NaCl, HEN2 150 mM) replicates were crossed. Protein partners were considered only if present in all co-IP replicates. All proteins observed in the corresponding control replicates, including same-set and sub-set proteins, were discarded. A second filter was set using controls from 15 independent other IP experiments in A. thaliana carried out by other laboratories in the same MS facility. All proteins observed in any of these negative controls were discarded from the final lists of partner proteins. Go-term analysis of proteins co-purified with MTR4-GFP was performed with DAVID (http: //david. abcc. ncifcrf. gov) [71], [72]. Plants were grown on MS agar plates supplemented with 0. 5% sucrose. Plates for induction of RRP41 RNAi contained 8 µM 17β-estradiol [5]. For each target, at least three biological replicates from WT, mtr4-1, mtr4-2, hen2-2, hen2-4, RRP41 non-induced (RRP41 Ctrl) and RRP41 induced (RRP41 RNAi) were analyzed. Total RNA was isolated from 7 day-old seedlings using TRI-reagent (MRC). cDNA was synthesized from 5 µg of total RNA with SuperScript III reverse transcriptase (Invitrogen) using 37. 5 pmol of oligo (dT) per 20 µl reaction according to the manufacturer' s instructions. Samples were analyzed as technical triplicates in a LightCycler 480 Real-Time PCR System (Roche). Each qRT-PCR reaction contained 1× LightCycler 480 SYBR Green I Master Mix (Roche), 5 pmol of each primer, 0. 5 µl of cDNA in a volume of 10 µl. ACT2, TIP41 and EXP were used as reference mRNAs. Wild type, hen2 and mtr4 plants were grown on MS agar supplemented with 0. 5% sucrose. Total RNA was extracted from two biological replicates using Nucleospin RNA plant columns (Machery & Nagel). cDNA synthesis and labeling with Cy3-dUTP or Cy5-dUTP (Perkin-Elmer-NEN Life Science Products) for fluorochrome reversal was performed as described previously [73]. Samples were hybridized to NimbleGen whole genome microrrays (Expression Omnibus (http: //www. ncbi. nlm. nih. gov/geo/), accession no. GPL17057 and GPL11005) as described in [73]. Topological positions and nucleotide sequence of the 1,434,492 strand-specific NimbleGen probes are available in the FLAGdb++ database at http: //urgv. evry. inra. fr/FLAGdb++ [45]. Two micron scanning was performed with an InnoScan900 scanner and raw data were extracted using Mapix software (Innopsys). Probes that mapped to repetitive sequences of the Arabidopsis genome were excluded from the analysis. Statistical analyses were performed with the software R. For each experiment, the raw data comprised the logarithm of median feature pixel intensities at wavelengths 635 nm (red) and 532 nm (green), respectively. The dye bias was corrected by a global intensity-dependent normalization for each chromosome and each array using the loess procedure [74] and then averaged over the technical replicates. The outputs of this procedure are two normalized intensity values per probe, one for each of the co-hybridized samples. Normalized raw data were deposited at Gene Expression Omnibus (http: //www. ncbi. nlm. nih. gov/geo/), accession no. GSE48178, and at CATdb (http: //urgv. evry. inra. fr/CATdb/), accession no. TIL-Ath-2011_3. For the statistical analysis, each mutant sample was compared to the co-hybridized wild type sample. Analyses were performed for each biological replicate and independently for each chromosome and each strand (2 mutants×2 biological replicates×2 strands×5 chromosomes = 40 analysis were performed). To determine probes that behave differently between a mutant and a wild type sample, we recast the question as an unsupervised classification problem for each biological replicate, for each chromosome and for each strand. A Hidden Markov Model was developed to model the joint distribution of the two normalized hybridization intensities in order to distinguish four different biologically interpretable clusters of probes: one cluster with similar behavior in both samples (identically expressed), one cluster with higher intensity in the first sample (over-expressed), a symmetric cluster with lower intensity in the first sample (under-expressed) and one cluster with low intensities in both samples (noise) corresponding to the non-transcribed probes. The emission distribution of the noise cluster was modeled by a spherical Gaussian. A Gaussian mixture, components of which were forced to be colinear along the main axis of the ellipse representing each cluster, modeled each of the three other clusters. Data projection on the main axes allowed us to work with unidimensional mixtures and to put a unique Gaussian distribution along the associated perpendicular axis for all components. Model parameters were estimated using an adapted version of the EM algorithm taking the model constraints and the spatial dependency between probes into account. A detailed documentation of the statistical analysis can be found in dataset S2. Probe classification into the four clusters was based on the conditional probabilities: a probe was assigned in the cluster for which the conditional probability was the highest (MAP rule). A probe was declared over-expressed in the mutant if this assignment was observed in the two biological replicates. 1860 probes were identified as overexpressed in hen2 mutants, and 499 probes were assigned as overexpressed in mtr4 mutants. For the majority of the estimated models, the main axis of the ellipse representing the cluster of probes under-expressed in the mutant was very closed to the main axis of the cluster with identically expressed probes. As a consequence, only a small number of probes were assigned under-expressed. Intersection of the lists with under-expressed probes across the two biological replicates revealed that no probes were underexpressed in both replicates of hen2 or mtr4, respectively. A file allowing the visualization of the upregulated probes aligned to the Arabidopsis genome using seqmonk software (http: //www. bioinformatics. babraham. ac. uk/projects/seqmonk/) can be found in dataset S1. Probes with a unique localisation in the genome and for which a significant over-expression was detected by the statistical analysis were sorted by genome coordinates to identify upregulated regions. Only regions with at least two consecutive probes were allowed. Bioinformatic analysis was performed with adapted perl scripts. Sequences of upregulated regions were annotated using TAIR10 genome database, FLAGdb++, and recent studies that identified snoRNA, miRNA genes and linc RNA genes [45]–[47], [75]. Sequence alignments were performed with BLASTn and sim4 tools (for annotation of spliced transcripts) [76], [77]. Only hits with 100% identity were considered. Annotation of upregulated regions was manually curated to remove double assignments (e. g. a region that matches to snoRNA genes located in an intron of a protein coding genes was assigned as snoRNA and removed from the list of introns). For comparison with a previously published list of exosome substrates, we compared the genome coordinates of upregulated regions with coordinates of upregulated regions extracted from [5], taking the difference between different versions of TAIR into account. ATMTR4: At1g59760; HEN2: At2g06990; RRP41: At3g61620.

Cells rely on a number of RNA degradation pathways to ensure correct and timely processing and turnover of both coding and non-coding RNAs. Another important function of RNA degradation is the rapid elimination of misprocessed RNA species, maturation by-products, and nonfunctional RNAs that are frequently produced by pervasive transcription. The main 3′-5′ RNA degradation machine in eukaryotic cells is the exosome, which is activated by cofactors such as RNA helicases. In yeast and human, processing, turnover and surveillance of all nuclear exosome targets depend on a single RNA helicase, MTR4. We show here that the Arabidopsis exosome complex can associate with two related RNA helicases, MTR4 and HEN2. MTR4 and HEN2 reside in nucleolar and nucleoplasmic compartments, respectively, and target different subsets of nuclear RNA substrates for degradation by the exosome. The presence of both MTR4 and HEN2 homologues in green algae, mosses and land plants suggest that the functional duality of exosome-associated RNA helicases is evolutionarily conserved in the entire green lineage. The emerging picture is that, despite a high degree of sequence conservation, intracellular distribution, activities and functions of exosome cofactors vary considerably among different eukaryotes.

Abstract Introduction Results Discussion Methods

rna interference brassica rna stability model organisms epigenetics plants research and analysis methods arabidopsis thaliana gene expression biochemistry rna plant and algal models rna processing cell biology nucleic acids genetics biology and life sciences molecular cell biology organisms

2014

The RNA Helicases AtMTR4 and HEN2 Target Specific Subsets of Nuclear Transcripts for Degradation by the Nuclear Exosome in Arabidopsis thaliana

Approaches to identify significant pathways from high-throughput quantitative data have been developed in recent years. Still, the analysis of proteomic data stays difficult because of limited sample size. This limitation also leads to the practice of using a competitive null as common approach; which fundamentally implies genes or proteins as independent units. The independent assumption ignores the associations among biomolecules with similar functions or cellular localization, as well as the interactions among them manifested as changes in expression ratios. Consequently, these methods often underestimate the associations among biomolecules and cause false positives in practice. Some studies incorporate the sample covariance matrix into the calculation to address this issue. However, sample covariance may not be a precise estimation if the sample size is very limited, which is usually the case for the data produced by mass spectrometry. In this study, we introduce a multivariate test under a self-contained null to perform pathway analysis for quantitative proteomic data. The covariance matrix used in the test statistic is constructed by the confidence scores retrieved from the STRING database or the HitPredict database. We also design an integrating procedure to retain pathways of sufficient evidence as a pathway group. The performance of the proposed T2-statistic is demonstrated using five published experimental datasets: the T-cell activation, the cAMP/PKA signaling, the myoblast differentiation, and the effect of dasatinib on the BCR-ABL pathway are proteomic datasets produced by mass spectrometry; and the protective effect of myocilin via the MAPK signaling pathway is a gene expression dataset of limited sample size. Compared with other popular statistics, the proposed T2-statistic yields more accurate descriptions in agreement with the discussion of the original publication. We implemented the T2-statistic into an R package T2GA, which is available at https: //github. com/roqe/T2GA. Great progress has been made toward the development of high-throughput technologies and their application to biological and clinical research. In a quantitative experiment, genes or proteins with significant changes in expression are potential to have important roles in a given phenotype or phenomenon. Therefore, the analysis of quantitative experimental data generally produces a list of differentially expressed genes or proteins in order. The list may share some insights if the aim of the experiments is restricted to few targets. As regards high-throughput data, the list hardly provides biological understanding of the mechanisms being studied, since the data involve complicated regulations among biomolecules and the number of biomolecules is too large to examine all candidates individually. Systematically investigating the underlying mechanisms from a high-throughput data therefore become a new challenge. To confront the challenge, one idea is to apply pathway analysis to identify the genes or proteins that are known to be involved in a biological process or interaction based on the existing knowledge. Many approaches of pathway analysis have been developed over years concerning different methodologies. Some general reviews of pathway analysis approaches can be found in [1–4]. These approaches can be broadly divided into two major factions: a competitive null with a gene sampling model and a self-contained null with a subject sampling model. The null hypothesis of a competitive test suggests that the target pathway (or a predefined gene set) is differentially expressed as well as the rest of all pathways. Practically a competitive null is closely tied to (although not necessarily) a gene sampling approach [5]. A gene sampling model principally implies the independence assumption; the assumption presupposes that genes are expressed independently of each other so that these genes can be manipulated as the sampling subject to produce the null distribution. The null hypothesis of a self-contained test, on the other hand, suggests that the target pathway is not differentially expressed between distinct phenotypes. Using the phenotypes of the experiments as the sampling subject is the setup of a subject sampling model. In other words, a competitive null with a gene sampling approach let pathways compete with each other in order to rank these pathways and use the number of genes as the sample size in the meanwhile; whereas a self-contained null with a subject sampling approach examine each single pathway to determine if the pathway is indeed differentially expressed between phenotypes and use the number of experiments as the sample size. Intuitively, a competitive null aims to find the pathway that is most significant among all pathways; a self-contained null aims to find the pathway that is the most significantly expressed between phenotypes. The classification of null hypotheses and sampling methods is firstly suggested by Geoman et al [5]. Even both categories have their own pitfalls and benefits, the authors suggested using a self-contained null with a subject sampling model in pathway analysis. Competitive null with gene sampling model usually implies the independence assumption which may produce inaccurate small p-values, cause serious false positives [6], and result misleading interpretations [5]. Further methodology issues of using a self-contained null can be found in [7–9], and of using a competitive null can be found in [10]. Another issue of pathway analysis comes from the construction of test statistics. These statistics can be divided into univariate tests and multivariate tests. Univariate statistics, such as a modification or a weighted summation of the t-scores [11–14], only focus on expression of genes or proteins and assume these biomolecules as independent units for statistical ease. The independence assumption ignores the associations among biomolecules with similar functions or cellular localization, as well as the interactions among them manifested as changes in expression ratios. In contrast, multivariate statistics take into consideration the associations among genes or proteins [15–20]. Some methodology studies [9,21–23] have evaluated univariate tests and multivariate tests with synthetic and experimental datasets. Compared with multivariate tests, univariate tests generally result a decrease in statistical power [21] and an increase in false positive rate [22] along with the rise of average correlation. Multivariate approaches usually incorporate the sample covariance matrix into calculation to address biological interaction. However, the sample covariance may not be precise enough to estimate the associations if the sample size is very limited. Compared with other gene expression data, proteomic data produced by mass spectrometry are more difficult to analyze systematically due to the limited number of experiments. This limitation causes current multivariate tests incompetent because the sample covariance will not be a robust statistic. The composition of pathway diagrams also become a challenge to pathway analysis. A pathway is a group of biomolecules that participate in a particular cellular process. The members of a pathway are usually defined by the tradition (i. e. , the history of pathway discovery) of molecular biology scientists. The structure of pathway diagrams is not standardized and therefore arises some issues to pathway analysis. First, the same pathway from different databases or other sources may have the same core members but different side members. Under different experimental conditions, the size of accessible biomolecules also changes. For example, the phosphorylation proteomic data may not provide information to the proteins not belong to phosphoproteome. Since the number of members within a pathway is not a constant, using this number as a parameter to determine if the pathway is significant or not may lead to inconsistent results. This issue arises with the assumption of a competitive null. Second, some molecules appear over pathways may play important roles as communication centers. For example, p53 appears in 38 pathways in the KEGG database. The shared members may interact or cooperate with each other and form a functional module. If this module is regulated (i. e. , the members within the module are differentially expressed as they cooperate together), the subsets of this module may present in abundant pathways and make these pathways seemingly significant. To identify the regulated pathway among the significant pathways of common modules, biologists usually utilize the distinctive molecules participating in that specific pathway; whereas pathways do not contain distinctive molecules may be irrelevant to the underlying mechanism. However, most of the current approaches do not take consideration of this issue. The suggestion of irrelevant pathways due to the redundancy over the significant pathways usually causes confusion to data description. A recent study [24] also focuses on the second issue. They demonstrate how the shared members affect p-values, and try to address this problem under a competitive null. In this study, we introduced a multivariate test, based on the Hotelling’s T2-statistic, to perform pathway analysis for quantitative proteomic data. The most serious problem of analyzing proteomic data produced by mass spectrometry is the limited number of experiments. We usually obtain only a few replicates (biological or technical) per experimental condition. To manage this issue, we had two special designs in our test. First, instead of using the sample covariance matrix (which is not robust when the sample size is limited), we use the covariance matrix that is constructed of the probabilistic confidence scores provided by the STRING and HitPredict databases. The proposed T2-statistic is then built of the protein expression profile and the covariance matrix to consider the expression level of individual proteins along with the associations among them. Second, we designed a self-contained model to produce a null distribution of altering protein expression while retaining the structure of protein associations. We are not capable of applying a subject sampling model because the number of experiments is too limited. In addition to the settlement of sample size issue, we designed an integrating procedure to categorize significant pathways as well as to avoid redundancy. The performance of the proposed T2-statistic is demonstrated using five public experimental datasets with different levels of biological complexity: the T-cell activation, the cAMP/PKA signaling, the effect of dasatinib on the BCR-ABL pathway, the differentiation process of myoblast, and the protective effect of myocilin via the MAPK signaling pathway. The first four datasets are proteomic data produced by mass spectrometry; the last dataset is a gene expression data of low sample size. We compared T2 with other popular statistics: DPA [25], GSEA [26], DAVID [27,28], and IPA [29]. For most of the situations, T2 yielded more accurate descriptions in agreement with the discussion of the original publication. We took four proteomic datasets of different biological complexity and experimental properties to demonstrate our approach. To be comprehensive, the testing datasets include the case of pathway activation and inhibition; also the case of signaling phosphoproteome and cellular proteome. We only used the final ratios provided by the datasets since they may have different integration approaches (e. g. to combine the results of biological and/or technical repeats, to handle missing data or outliers) under different experimental designs. The summarized ratio is also the most available format for quantitative proteomic data. In this situation, the sample size of data becomes only one, calculating a covariance matrix is not even possible. Our approach provide a solution to undertake this difficulty. We also applied our approach on a gene expression dataset of three samples to demonstrate that the general idea is applicable to other quantitative data of low sample size. To provide more generalized results, we choose two pathway databases and two protein-protein interaction databases: KEGG and Reactome provide pathway categories served as predefined gene sets, STRING and HitPredict contributes the confidence scores to estimate the covariance between protein expressions. As we mentioned in Introduction, subsets of one common active module may cause a lot of pathways statistically significant. These pathways may only have slight relevance to the target mechanisms (Fig 1). To avoid misinterpretation due to irrelevant pathways being reported, we identify delegates to categorize pathways into pathway groups. A pathway group is a set of pathways, in which the pathway being the superset of other pathways is defined as the delegate. Since other pathways do not show any distinctive proteins to support themselves, the significance of other pathways may simply originate from the regulation of the delegate. In other words, we want to avoid the situation that a pathway is enriched only because of some common proteins that are shared with other pathways. Our pathway integration procedure operates as follows. The set of the pathways being evaluated is denoted by P and each pathway P ∈ P is associated with a p-value. We iteratively perform the following steps until all the pathways in P are assigned to a specific pathway group M. A demonstrative example can be found in S1 Fig. The pathway integration procedure aims to find the pathways with the most sufficient information to represent current data. Please notice that this idea is not similar to pathway hierarchy in which higher level pathways are defined as the supersets over lower level pathways, in that case only the pathways of highest level are able to be delegates. One can only apply the integration procedure to pathways without hierarchy relationship. In addition, the procedure is only applicable to the statistical test using a self-contained null since the measure of significance is independent for each pathway. We discussed the flow of signal transduction in time order since the dataset is a time-series of 5 min, 15 min, and 60 min experiments. In the beginning, the treatment anti-CD3ϵ activated TCR signaling pathway. The TCR signaling pathway in KEGG and IPA depicts the signals from the TCR receptors all the way to the IL-2 expression. Usually, the response of signal transduction comes rapid, we expected that the TCR signaling pathway should be enriched in early time points; the downstream from the TCR signaling pathway to the IL-2 expression should be enriched in late time points. From Table 2 we found that T2×ST, T2×HP, DAVID, and IPA enriched the TCR signaling pathway in all experiments; GSEA enriched the TCR signaling pathway in 5 and 15 min experiments only. The ranks of the TCR signaling pathway are top in T2, DAVID, and IPA. The downstream of the TCR signaling pathway was illustrated in Fig 3, there are three possible routes directed from TCR signaling. The original publication focused on the TCR/Ras/MAPK route to the IL-2 expression and the cytoskeleton remodeling response. The TCR/PI3K-Akt/mTOR route, on the other hand, is also important and has been discussed as a cluster in literature [44–47]. From Fig 3 we found the enriched pathways by T2×ST were well-grounded for the following reasons: To sum up above, we found the results of T2×ST fit our expectation: T2×ST enriched the TCR signaling pathway in the early time data, the route TCR/Ras/MAPK and TCR/PI3K-Akt/mTOR in time order, and the response of actin regulation, in both pathway databases. T2×HP, DAVID, and IPA enriched most of the expected pathways in KEGG, although the ranks of these pathways are occasionally low in IPA; T2×ST, T2×HP and DAVID also enriched most of the expected pathways in Reactome (S1 Table); GSEA failed to enrich half of the expected pathways in both pathway databases. We also found that there is no distinguishable difference between the result using a self-contained null (T2×ST, T2×HP, DPA) or a competitive null (GSEA, DAVID, IPA) in the TCR dataset. The target of the original publication is PKA substrates. As the authors described in their results; PGE2 induced a rapid and maximal increase in phosphorylation level after 1 min, and the level remained high before the number of substrates gradually returned to near-basal conditions after 60 min (S2b Fig). From Table 2 we found T2×ST, T2×HP, DPA, and IPA enriched the cAMP signaling pathway for both 1 min and 60 min experiments. The downstream of the cAMP signaling pathway was illustrated in Fig 4, there are four possible routes directed from cAMP signaling. The proliferation route was suggested by T2×ST, T2×HP, and IPA; and the cytoskeleton remodeling route also, with an addition of DPA. The DNA repair route was enriched by T2×ST, T2×HP, and DPA, in both pathway databases. The glycogen synthesis route, on the other hand, was enriched only by IPA. The reason is that T2 takes expression ratios as an important feature, whereas the mapped proteins of the “Glycolysis / Gluconeogenesis” pathway are all of low ratios (min = −0. 23710, max = 0. 14950, mean = 0. 02742). Briefly, we found the results of T2 reasonable: T2×ST and T2×HP enriched the cAMP signaling pathway for both datasets; the PKA/Rap1/PI3K-Akt route in time order, and also the cytoskeleton remodeling route and the DNA repair route. DPA and IPA also enriched most of the expected pathways. GSEA and DAVID failed to enrich most of the expected pathways in both pathway databases. We also found that the result using a self-contained null (T2×ST, T2×HP, DPA) enriched more expected pathways than a competitive null (GSEA, DAVID, IPA). Since the pathways competes with each others under a competitive null, the success of some pathways will obstacle other pathways. For example, the top 1 enriched pathway provided by DAVID for the 1 min data is “Ribosome”. The pathway includes 87 proteins, and 51 of them were mapping by the data. The cAMP signaling pathway, on the other hand, includes 73 proteins, but only 8 of them were mapping by the data. The high mapping rate of “Ribosome” will makes it harder for DAVID to enrich the target cAMP signaling pathway. This study aims to characterize the changes in protein expression underlying the phenotype conversion from mononucleated muscle cells to multinucleated myotubes. According to the their analysis, five functional clusters were identified in this dataset: cell cycle withdrawal (72hr/0hr), cell adhesion and migration (24hr/0hr), RNA metabolism (both 24hr/0hr and 72hr/0hr), myofibril formation (72hr/0hr), and proteolysis, fusion, and ECM remodeling (both 24hr/0hr and 72hr/0hr). The corresponding KEGG pathways were illustrated in Fig 5. We chose the pathway “ECM-receptor interaction” as the target pathway in Table 2 because it is clearly stated to be differentially expressed in both experiments. From Table 2 we found the ECM-receptor interaction pathway was enriched by T2×ST, T2×HP and DAVID for both experiments. The original publication focused on the change of cellular phenotype accompanying myogenic differentiation and the development of myofibril. From Fig 5 we found the myofibril formation route was enriched by T2×ST, T2×HP, DAVID, and IPA. T2×ST and T2×HP further enriched muscle contraction related pathways, which were associated with myotube maturation as discussed in the original publication. The cell adhesion and migration play an essential role in the fusion of mononucleated myoblasts. Pathways related to adhesion and migration were enriched by T2×ST, T2×HP, and DAVID; those related to fusion were enriched by T2×ST, T2×HP, DAVID, and IPA. The elevation of lysosomal proteins contributed in remodeling intracellular components during the course of myotube formation. All tools enriched the lysosome pathway with an exception of IPA. The RNA metabolism and the cell cycle withdrawal routes represented the termination of proliferation since the growth factors and nutrition were removed from the medium. Related pathways were enriched by T2×ST, T2×HP and IPA. Briefly, T2×ST and T2×HP enriched all the pathways discussed in the original publication; the interpretation made by T2 fit the description of the dataset pretty well. In the meantime, DAVID and IPA also enriched most of the expected pathways, asides from muscle contraction pathways. DPA enriched most of the expected pathways in Reactome, but failed in KEGG; GSEA failed in both pathway databases. We also found that there is no distinguishable difference between the result using a self-contained null (T2×ST, T2×HP, DPA) or a competitive null (GSEA, DAVID, IPA) in the myogenesis dataset. This dataset, unlike the previous, is not a time series; it is a dose-comparison experiment. According to the original publication, two datasets shared nearly all identified proteins although 50nM dataset did down-regulate more phosphopeptides (S2d Fig). Consequently, the authors only paid attention to the proteins that are regulated by both 5nM and 50nM dasatinib. The main target of dasatinib is the BCR-ABL signaling pathway, which is described in the pathway “Chronic myeloid leukemia (CML) ” of KEGG. From Table 2 we found the CML pathway was enriched by T2×ST, T2×HP, DAVID and IPA for both experiments. The downstream of the CML pathway was illustrated in Fig 6, there are three possible routes directed from the inhibition of BCR-ABL. Since the two datasets shared nearly all proteins, we used the ranks of 5 nM dataset to represent the common pathways over two datasets. The original publication focused on the BCR-ABL/Ras/MAPK route and the connection between BCR-ABL signaling and apoptosis. From Fig 6 we found the enriched pathways by T2 were plausible for the following reasons: In short, we found the performance of T2 pleasant: T2×ST and T2×HP enriched the BCR-ABL/Ras/MAPK route, the BCR-ABL/PI3K-Akt/Apoptosis route, the JAK-STAT signaling pathway, and the pathways related to actin response. Both DAVID and IPA enriched some of the expected pathways. GSEA and DPA failed to enrich most of the expected pathways in both pathway databases. Generally there is no distinguishable difference between the result using a self-contained null or a competitive null, but T2 enriched more expected pathways than other methods. The dataset includes only one experiment, comparing the gene expression of myocilin expressed cells to control cells, under U0126 treatment. The authors concluded that myocilin has a protective effect to against apoptosis and further promotes cell survival and proliferation via the MAPK signaling pathways. They also experimentally confirmed that the Raf-MEK-ERK-MAPK cascade was activated by myocilin. From Table 2 we found T2×ST, T2×HP, DPA, and DAVID enriched the MAPK signaling pathway; all three methods under a self-contained null successfully enriched the target pathway, whereas only DAVID is under a competitive null. The upstream and downstream of the MAPK pathway was illustrated in Fig 7. There are two possible upstream receptors of the MAPK signaling pathway: the GPCR receptors and the Notch receptors. In KEGG, both upstreams were enriched by T2×ST, T2×HP and DAVID, DPA only enriched the GPCR/Ras/MAPK route and IPA only enriched the Notch/Ras/MAPK route. In Reactome, both upstreams were enriched by T2×ST and T2×HP, DAVID only enriched the Notch/Ras/MAPK route (S1 Table). The are two downstream routes, both are supported by the original publication. The results from both T2×ST and T2×HP suggested that the differentially expressed pathways are more upstream. This conclusion actually fit the discussion of the original publication, which suggests that myocilin may also regulate the upstream kinase of the MAPK signaling pathway. Briefly, both T2×ST and T2×HP successfully enriched the MAPK signaling pathway and its upstream; other tools enriched only some of the expected pathways. Generally speaking, we found that the proposed T2-statistic was able to enrich the pathways in agreement with the original publication, whereas the performances of DPA, GSEA, and DAVID were not stable. IPA, as a commercial software with high cost, also enriched most of the relevant pathways. Nevertheless some of the pathways are low-ranked and the numbers of enriched pathways are enormous. The results suggested that our multivariate design of the proposed T2-statistic does provide important information toward pathway analysis by considering the strength of interactions among proteins. In the meantime, our self-contained null hypothesis is capable of enrich relevant pathways by the significance of protein expression ratios, whereas the focus of the competition among pathways may neglect the clear distance between phenotypes. Both DAVID and IPA are based on the competitive null hypothesis, although DAVID performs a more stringent post hoc correction, they shared the failure of some relevant pathways. The tremendous numbers of enriched pathways also suggested IPA may report more false positive results. GSEA also applies a competitive null when the sample size is limited and its KS statistic is sensitive to small sample size. The unsatisfied results suggested that GSEA is not suitable for data of limited experiments. The design of DPA is similar to the proposed T2-statistic; both DPA and T2 are specifically designed for quantitative proteomic data, and they both use self-contained null hypotheses. The performance difference between DPA and T2 primarily comes from the aspect of statistic construction. The proposed T2-statistic outperformed DPA because it considers the strength of interactions among proteins. Briefly, in five testing datasets, the results using a self-contained null is generally more well founded than the results using a competitive null. The importance of applying the covariance matrix is to estimate accurate confidence interval. We illustrated an example in Fig 8 to demonstrate the situation that may cause inaccurate estimation. Both M1 and M2 are accurate null distributions since the data are normalized using proper covariance matrix and the distribution hence follows χ2. M4 indicates that situation that an independent data are misinterpreted as a correlated data. This happens when we have false positive protein-protein interactions in the databases. In order to minimize the risk, we only use the confidence scores derived from directly experimental evidence. M3 represents the case that a correlated data are misinterpreted as an independent data. This may actually happen due to our incomplete knowledge of the biology system. In this case, the null distribution will not follow χ2 and the estimation of p-value will be inaccurate. Even so, for pathways of vary high expression ratios, applying the covariance matrix or not does not change its p-value dramatically. Here we demonstrated the impact toward p-values using permuted and purged confidence scores. We performed 100 experiments with 30% and 60% permuted confidence scores (i. e. we reassign the score using the same score distribution) and another 100 experiments with 30% and 60% purged confidence scores (i. e. we randomly remove the scores from the PPI databases), and we checked if the expected pathways are still significant under current significance level (α = 0. 05). From Table 3 we observed that: The construction of the proposed T2-statistic showed T2 is heavily dependent on expression ratios. After all, the null hypothesis for T2 is to test if the mean vector equals to zero. The contribution of applying the covariance matrix is to estimate p-values in a more accurate manner: to rescue some pathways with moderate expression ratios but their regulation directions are consistent with current knowledge of protein interaction, and to discard some pathways with inconsistency. In this study, we presented a knowledge-based T2 approach to perform pathway analysis for quantitative proteomic data of a limited number of experiments. The proposed T2 is constructed as a multivariate statistic and the test of significance is under a self-contained null. We use the probabilistic confidence score provided by the STRING or HitPredict databases to approximate the covariance matrix of the protein profiles. The proposed T2-statistic is therefore able to reveal the influence of protein-protein interactions while performing the analysis. In addition, our pathway integration procedure is able to categorize pathways into pathway groups as well as to avoid redundancy. We performed the T2-statistic on five published quantitative proteomic dataset. In all cases, T2 was able to eliminate irrelevant pathways, as well as correctly identify relevant pathways that had been discussed in the original publication. The idea of incorporating biological evidence into conventional statistic can be widely applied to the analysis of quantitative proteomic data.

Pathway analysis is a common approach to quickly access the pathways being regulated in the experiments. There are numerous statistics to perform pathway analysis; most of them assume that the genes or proteins are independent of each other for statistical ease. This assumption, however, is unrealistic to the real biological system and may cause false positives in practice. A standard way to address this issue is to measure the associations among genes or proteins. Unfortunately, the estimation of associations requires sufficient sample size, which is usually not available for proteomic data produced by mass spectrometry. In this study, we propose a T2-statistic, which estimates the associations among gene products, to perform pathway analysis for quantitative proteomic data. Instead of calculating the associations directly from data, we use the confidence scores retrieved from protein-protein interaction databases. We also design an integrating procedure to reserve pathways of sufficient evidence as a regulated pathway group. We compare the proposed T2-statistic to other popular statistics using five published experimental datasets, and the T2-statistic yields more accurate descriptions in agreement with the discussion of the original papers.

Abstract Introduction Materials and methods Results and discussion Conclusion

phosphorylation random variables covariance protein expression mathematics tcr signaling cascade molecular biology techniques cellular structures and organelles mapk signaling cascades cytoskeleton research and analysis methods proteins gene expression biological databases proteomics molecular biology probability theory molecular biology assays and analysis techniques gene expression and vector techniques biochemistry signal transduction proteomic databases cell biology post-translational modification database and informatics methods genetics biology and life sciences physical sciences cell signaling signaling cascades

2017

A knowledge-based T2-statistic to perform pathway analysis for quantitative proteomic data

Leishmaniasis is a parasitic disease that is widely prevalent in many tropical and sub-tropical regions of the world. Infection with Leishmania has been recognized to induce a striking acceleration of Human Immunodeficiency Virus Type 1 (HIV-1) infection in coinfected individuals through as yet incompletely understood mechanisms. Cells of the monocyte/macrophage lineage are the predominant cell types coinfected by both pathogens. Monocytes and macrophages contain extremely low levels of deoxynucleoside triphosphates (dNTPs) due to their lack of cell cycling and S phase, where dNTP biosynthesis is specifically activated. Lentiviruses, such as HIV-1, are unique among retroviruses in their ability to replicate in these non-dividing cells due, at least in part, to their highly efficient reverse transcriptase (RT). Nonetheless, viral replication progresses more efficiently in the setting of higher intracellular dNTP concentrations related to enhanced enzyme kinetics of the viral RT. In the present study, in vitro infection of CD14+ peripheral blood-derived human monocytes with Leishmania major was found to induce differentiation, marked elevation of cellular p53R2 ribonucleotide reductase subunit and R2 subunit expression. The R2 subunit is restricted to the S phase of the cell cycle. Our dNTP assay demonstrated significant elevation of intracellular monocyte-derived macrophages (MDMs) dNTP concentrations in Leishmania-infected cell populations as compared to control cells. Infection of Leishmania-maturated MDMs with a pseudotyped GFP expressing HIV-1 resulted in increased numbers of GFP+ cells in the Leishmania-maturated MDMs as compared to control cells. Interestingly, a sub-population of Leishmania-maturated MDMs was found to have re-entered the cell cycle, as demonstrated by BrdU labeling. In conclusion, Leishmania infection of primary human monocytes promotes the induction of an S phase environment and elevated dNTP levels with notable elevation of HIV-1 expression in the setting of coinfection. Leishmaniasis has recently been recognized to be both one of the world' s most neglected and most important parasitic diseases, threatening an estimated 350 million people worldwide [1], [2]. Surveys have estimated that approximately 12 million people are currently infected with 2 million new cases reported yearly, primarily afflicting the world' s poorest populations in some 88 countries [3]. Leishmaniasis is transmitted to humans by the bite of the female Phlebotomine sandfly upon taking a blood meal [4]. Infection results in three basic clinical presentations. Cutaneous and mucocutaneous leishmaniasis are disfiguring and even mutilating diseases, while visceral leishmaniasis (VL) is characterized by fever, massive hepatosplenomegaly, pancytopenia, and a wasting syndrome called Kala-azar, which is nearly uniformly fatal without treatment [5], [6]. Early after the emergence of the global Human Immunodeficiency Virus Type 1 (HIV-1) epidemic, clinicians recognized that reciprocal activation of each pathogen by the other frequently occurred. It was noted, on the one hand, that infection with HIV-1 modifies the natural history of leishmaniasis, leading to 100–2,230 times increase in the risk of developing VL and reducing the likelihood of a therapeutic response [7]–[11]. At the same time, VL was shown to induce activation of latent HIV-1, increase viral load, and cause a striking acceleration in the progression of asymptomatic HIV-1 infection to AIDS that corresponded to a reduction of life expectancy in patients [12]–[15]. Similarly, it was recognized that monocytes and macrophages are the primary cell types coinfected with both HIV-1 and Leishmania. Initial studies demonstrated that Leishmania coinfection reactivated HIV-1 replication in latently infected monocytoid cell lines [16]. Subsequent studies in primary MDMs coinfected with L. infantum and HIV-1 also found enhanced HIV-1 replication associated with increased secretion of the pro-inflammatory cytokines TNF-α, IL-1α, and IL-6. In these experiments, HIV-1 replication, as measured by p24 ELISA, was reduced in the presence of either chemical inhibitors or blocking antibodies to these three cytokines [17]. Human monocytes circulate in the blood and reside in bone marrow and spleen and are generally believed not to proliferate in the steady state [18], [19]. However there is an emerging awareness that human monocytes possess far greater heterogeneity than originally perceived, and subpopulations of monocytes have recently been described that can re-enter the cell cycle in response to both Macrophage- and Granulocyte Macrophage-Colony Stimulating Factors (M-CSF and GM-CSF, respectively) [20]–[22]. Proliferation of these presumably immature peripheral blood monocyte subpopulations has been demonstrated by multiple techniques including uptake of 5-bromo-2′-deoxyuridine (BrdU) and CFSE labeling, leading to this population being termed “proliferative monocytes” [23], [24]. Such cellular proliferative capacity has important implications because cellular dNTP levels correlate directly with the replicative capacity of mammalian cells [25]. Consistent with this observation, a variety of studies, including those from our laboratory, have reported that dNTP levels are consistently higher in dividing versus non-dividing cells [25]–[31]. Among the retroviruses, HIV-1 possesses the unique ability to infect both dividing (activated CD4+ T cells) and non-dividing cells (macrophages). This ability is due, at least in part, to the evolutionary adaptation of its reverse transcriptase (RT) to function under conditions of extremely limited dNTP availability [32]. However, as noted for the replicative capacity of mammalian cells, HIV-1 replication efficiency is also directly correlated with cellular dNTP concentrations and proceeds with far greater efficiency in both tumor cells and PHA-stimulated CD4+ T cells, in which the average dNTP levels are 150–225 times higher than that of non-dividing MDMs [32], [33]. Several recent studies have shown that HIV-2 Vpx protein promotes the degradation of the SAMHD1, a host anti-viral restriction factor [34]–[37]. Recently, SAMHD1 was shown to function as a dNTP hydrolase [38], [39], limiting the cellular dNTP pool and restricting HIV-1 replication in cells of myeloid lineage [40]. Moreover, our recent paper shows a direct connection between SAMHD1 degradation, an increase in dNTP levels and enhanced transduction of HIV-1 in myeloid cells [41]. In the present study, we found that in vitro infection of freshly isolated, undifferentiated CD14+ primary human monocytes with Leishmania consistently led to maturation into macrophages and to higher cell numbers over time as compared to uninfected control cells. In addition to the inhibition of apoptosis previously reported in Leishmania-infected MDMs, we also report the unexpected finding that a sub-population of CD14+ human MDMs proliferate in response to Leishmania, as measured by BrdU incorporation at days 12–14 after infection. As the efficiency of HIV-1 RT DNA synthesis and subsequent viral replication are directly dependent on cellular dNTP concentration, we subsequently employed a highly sensitive single nucleotide incorporation assay that was recently developed in our laboratory to measure cellular dNTP concentration [32], [42], [43]. We found a marked increase in the content of dNTPs in Leishmania-maturated MDMs as compared to uninfected control cells. Consistent with this observation, elevated levels of ribonucleotide reductase (RNR), the rate-limiting enzyme for dNTP synthesis, was also found in Leishmania-maturated MDMs as compared to control cells. Finally, we found significantly enhanced expression and transcription of a GFP-expressing pseudotyped HIV-1 (HIV-1 D3 GFP) in Leishmania-maturated MDMs as compared to control cultures as assayed by FACS analysis of HIV-1 D3 GFP expressing cells and qPCR for 2 LTR-circle copy number. As noted above, previous studies have suggested a role for Leishmania infection of monocytes causing the induction of pro-inflammatory cytokines as a stimulus to HIV-1 replication in coinfected cells. Our data support a novel model whereby Leishmania infection stimulates monocytes' differentiation and cell division. Consistent with the increased proliferation capacity, Leishmania infection increases cellular dNTP concentrations that facilitate enhanced HIV-1 coinfection. The effect of Leishmania infection on cell survival of primary human monocytes was examined over a time course of 28 days from eleven individual donors. Preliminary experiments were performed to examine potential effects of both heat-inactivated Leishmania (also applied to monocytes at an MOI = 7) and day 7 conditioned medium from Leishmania-infected monocytes re-applied to freshly isolated monocytes. These experiments demonstrated no significant effects on monocyte cell survival, maturation, or proliferation (data not shown). In parallel experiments, Leishmania labeled with the vital dye PKH showed that at an MOI = 7 virtually all monocytes within the culture became infected (Figure S1). This MOI is well within the range of those previously published [16], [17]. Purified human monocytes were cultured at 1×106 cells/well in 6 well dishes, and three wells from each of three culture conditions were combined and counted: 1) RPMI media with 10% FBS (“control cells”), 2) RPMI media with 10% FBS plus 5 ng/ml human recombinant GM-CSF (“GM-CSF”), or 3) RPMI media with 10% FBS with Leishmania major (MOI = 7) at the time of plating (“Leishmania”). The GM-CSF-treated monocytes differentiate into MDMs and were used as a positive control for all the studies. Medium was changed at day 7 and then weekly, replating any non-adherent cells into their respective wells. As illustrated in Figure 1A, a marked decline in the cell numbers was seen at day 3 after initial plating in all three conditions, though more notably in the control monocytes as compared to either GM-CSF-treated or Leishmania-infected monocytes. Cell numbers fell from 3×106 at day 0 in all three conditions and were consistently lower in control cells as compared to either GM-CSF-treated (positive control) or Leishmania-infected cells at all times tested from day 3 to day 28. Control monocyte numbers declined until day 7, when their numbers stabilized through day 28. Cell numbers for GM-CSF-treated and Leishmania-infected groups remained significantly higher than control monocytes at all time points from day 3 to day 28 (Friedman test; p<0. 05). Next, we examined the different cell populations using light microscopy. The control cells largely retained a small, mostly rounded morphology (Figure 1B) at day 14 as compared to either GM-CSF-treated (Figure 1C; positive control) or Leishmania-maturated MDMs (Figure 1D). For both treatments, the monocytes were larger, more adherent and spread out with some processes, which is characteristic of mature macrophages. Using FACS analysis, the GM-CSF-treated and Leishmania-maturated MDMs were larger (as assayed by forward scatter) with greater cellular complexity (assayed by side scatter) as compared to control monocytes (Figure S2). These findings were further confirmed and quantitated by FACS analysis of cell surface CD14 expression from six independent donors. This demonstrated a decreased cell surface expression of CD14 (CD14low) in both day 14 GM-CSF-treated and Leishmania-infected MDMs as compared to control monocytes (CD14high), again consistent with monocytes to macrophages maturation in the GM-CSF and Leishmania-infected cultures (Figure S2). FACS analysis for both Annexin V and propidium iodide also showed pronounced reduction in cell death for the Leishmania-infected monocytes compared to uninfected controls (Figure S3) Collectively, these data suggest that Leishmania infection of monocytes leads to less cell death and increased cellular maturation towards a macrophage phenotype compared to control monocytes. While performing the kinetic studies of Leishmania-infected monocytes, we observed clusters of small cells lying on top of larger, more differentiated appearing macrophages in both the Leishmania-infected and GM-CSF-treated (positive control) cultures but not for the control cell culture. Although, as noted above, human monocytes are generally believed not to proliferate once released from the bone marrow [18], [19], it has been more recently recognized that these cells possess far greater heterogeneity than originally believed and subpopulations of monocytes have been recently described that can re-enter the cell cycle in response to M-CSF and GM-CSF [20]–[22]. Proliferation of these monocyte subpopulations has been demonstrated by multiple techniques including uptake of BrdU and CFSE labeling [23], [24]. Thus, we next asked whether their presence might also be induced in the setting of Leishmania infection. To address this, we did a time-course analysis at days 3,7, 10, and 14, examining BrdU uptake at 48 hours after treatment for the Leishmania-infected groups [24]. As expected, we detected a few cells that were uniformly BrdU+ (green) and nuclei counterstained with DAPI (blue) (Figure 2A). We detected a progressive increase in the numbers of BrdU+ cells over time, with maximal numbers of BrdU+ cells observed at day 14 of cell culture. Lastly, we co-labeled primary human monocytes with PKH-labeled L. major (orange) and then pulsed with BrdU (Figure 2A, bottom right panel Day 21). BrdU+ nuclei were seen in Leishmania-infected cells suggesting that infection may promote re-entry into the cell cycle for a sub-population of cells. This may be of importance to the dissemination of Leishmania within a host because macrophages are generally considered terminally differentiated, non-dividing cells [19]. We subsequently performed quantitative FACS analyses to compare the percentages of BrdU+ cells. As shown in a representative FACS plot, Figure 2B, a relatively large sub-population of BrdU+ cells was seen in both Leishmania-infected (13. 4%) and GM-CSF-treated cells (14. 8%) but not in control cells (<1. 0%). Figure 2C summarizes results for 48 hour BrdU incorporation for seven independent donors between days 12–14. Leishmania-maturated MDMs demonstrated highly statistically significant (p<0. 01) elevations of the percentage of BrdU+ cells as compared to control cells while GM-CSF-maturated MDMs were significantly (p<0. 05) higher. We also CSFE-labeled fresh monocytes and found at least one cell division in a small subpopulation of cells for the GM-CSF-treated and Leishmania-infected groups (data not shown). Collectively our results are consistent with previous studies of a proliferative monocyte sub-population that can be stimulated to enter cell division by the related monokine M-CSF [21], [24]. However, of greater relevance is the demonstration that L. major infection of monocytes can induce an S phase environment as assayed here by BrdU incorporation. Whether this promotes cell division in vivo, allowing for greater dissemination of Leishmania, remains unclear. We employed the highly sensitive HIV-1 RT based assay for measuring cellular dNTP content [32], [42]–[45]. As depicted in Figure 3A, HIV-1 RT is bound to a template/primer complex. HIV-1 RT can extend the primer by one nucleotide, depending on the template nucleotide (N) present at the 5′ end of the template. This assay allows for the determination of differences between cellular extracts for a specific cellular dNTP. Using this assay, we compared the cellular content of dGTP (purine) and dTTP (pyrimidine) for the different treatment groups. Figure 3B shows a representative result for primer extension of dGTP (left panel) and dTTP (right panel). Summary results for nine individual donors are presented in graph form in Figure 3C and are summarized below. In Figure 3B, left side panel, dGTP levels were assayed at days 7 and 13, while the right side panel shows dTTP analysis for the same days. In lanes 1 for both dGTP and dTTP analysis, no dNTPs were added to the reaction, leading to no extension product of the labeled primer (open arrow). In lanes 2, exogenous dNTPs were added as a positive control to show extension of all primers in the reactions (closed arrow). In lanes 3–8, days 7 and 13 cellular extracts were analyzed. Content of dGTP were notably higher in GM-CSF- and Leishmania-maturated MDMs as compared to untreated control cells at day 7 (lanes 4 and 5) and day 13 (lanes 7 and 8) after treatment. In comparison, dTTP concentrations at day 7 were slightly higher for the GM-CSF-maturated MDMs (lanes 4, positive control) as compared to the control and Leishmania-maturated MDMs. At day 13, we detected much higher dTTP concentrations in the GM-CSF and Leishmania-maturated MDMs at day 13 (lanes 7 and 8) as compared to the control group (lanes 6). These data demonstrate that Leishmania infection can lead to notable increases in cellular dNTP concentrations and this conclusion is fully validated by quantification of the assay results in nine individual donors (Figure 3C). Results for dGTP (Figure 3C, upper panels) demonstrated statistically significant increases for GM-CSF-matured and Leishmania-infected MDMs as compared to control monocytes at day 7; by day 13 dGTP increases were now highly significantly elevated in Leishmania and still significantly elevated in the GM-CSF-maturated MDM groups as compared to controls. The results for dTTP at day 7 (Figure 3C, lower left panel) trended higher in Leishmania-maturated MDMs as compared to monocyte controls but only reached significance in GM-CSF-maturated MDMs. However at day 13, (Figure 3C, lower right panel) Leishmania-maturated MDMs were significantly increased in dTTP concentrations as compared to monocyte controls. These data demonstrate that Leishmania infection of monocytes induces elevation of both purine and pyrimidine concentrations in the host cell. The finding of elevated purine levels is particularly intriguing in light of the fact that Leishmania species are entirely dependent on host cell synthesis for their supply of purine nucleotides [46]. Mammalian RNR is a dimeric enzyme essential for catalyzing the direct reduction of relatively large intracellular pools of ribonucleotides into the corresponding deoxyribonucleotides for DNA synthesis. The catalytic enzyme is a heterodimer, containing two subunits of R1 and either two subunits of R2 or p53R2. Expression of the R2 subunit is strictly limited to the S phase of the cell cycle [47]. As shown in Figure 4A, western blot analyses were done for R2 and p53R2 on cell extracts using freshly isolated monocytes, day 13 GM-CSF or Leishmania-maturated MDMs. As shown in Figure 4B, we quantitated the western blots for four independent donors and found that R2 was significantly (p<0. 05) increased in the Leishmania-maturated MDMs over control monocytes. For the p53R2, we found a significant increase in the GM-CSF-treated cells but the increase failed to reach significance for the Leishmania-infected cells when compared to monocytes, which were set to 1. Moreover, the R2 and p53R2 antibodies were specific for human ribonucleotide reductase and did not cross-react with L. major (data not shown). Collectively, these data show that 1) R2 subunit expression, which is S phase linked, is significantly increased upon Leishmania infection, and 2) that infection indirectly leads to an increase in the p53R2 subunit, which is involved in increasing cellular dNTP concentrations in non-dividing cells. Next, quantitative reverse transcriptase quantitative PCR (qRT-PCR) using Taqman analysis was performed in three individual donors to examine whether the observed increase in RNR R2 subunit and P53R2 protein expression showed transcriptional regulation (Figure 4C). Consistent with the significantly increased protein expression of the RNR R2 subunit seen by western blot, significantly increased transcription was seen in Leishmania-infected monocytes as compared to GM-CSF-treated MDM and control monocytes. It is also possible that these results may be due, at least in part, to an increase in RNR R2 transcript stability. In contrast, increased expression of p53R2 protein likely occurs due to post-transcriptional regulation as no significant elevation of transcription was seen in either the GM-CSF-treated or Leishmania-infected MDMs as compared to control monocytes. As noted above, cellular dNTP levels serve as a biomarker for the replicative capacity of mammalian cells, a finding corroborated by the presence of consistently higher dNTP levels in dividing cells as compared to non-dividing cells [25]–[30]. HIV-1 replication efficiency is also directly correlated with the cellular dNTP concentration, and we and others have reported that it proceeds with far greater efficiency in tumor cells or PHA-stimulated CD4+ T cells in which the average dNTP level is 150–225 times higher than in non-dividing monocytes/macrophages [32], [33]. Given our findings that Leishmania infection induces both significant elevation of dNTP levels and replication capacity in MDMs, we examined whether transduction of Leishmania-maturated MDMs with a VSV-g pseudotyped HIV-1 vector, designated HIV-1 D3 GFP, resulted in accelerated HIV-1 expression, as determined by GFP expression. Six days after isolation, control cells, GM-CSF maturated MDMs, and PKH-labeled (red) Leishmania-maturated MDMs were transduced in 6-well dishes with equal amounts of HIV-1 D3 GFP vector. We examined cells by bright field and fluorescence microscopy 24 hours later (Figure 5A). HIV-1 D3 GFP expression (“GFP” [green-top 3 panels]) was markedly enhanced, relative to control cells, in both the GM-CSF- and Leishmania-maturated MDMs (upper middle and right-sided panels, respectively), consistent with both strikingly increased intensity and numbers of HIV-1 D3 GFP transduced cells in these two conditions relative to control cells (Figure 5A; top left panel). Only a rare control cell appeared to express HIV-1 D3 GFP. GM-CSF- and Leishmania-maturated MDMs had many more cells expressing GFP as compared to control cells. Leishmania maturated MDMs labeled with PKH showed comparable numbers of GFP+ cells/field as compared to GM-CSF-maturated MDMs (Figure 5A; middle and top right panels). We next quantified the three different groups by FACS analysis (Figure 5B and Table 1). For these studies, the Leishmania were not labeled with PKH dye. As shown in Table 1, cells from four independent donors were examined at 24 and 48 h after the addition of the HIV-1 D3 GFP vector. The percent of GFP+ cells for Leishmania-maturated MDMs were consistently higher as compared to the control cell group with a somewhat weaker trend to higher percentage of GFP+ cells also found in the GM-CSF-maturated MDMs as compared to control cells. Next, we examined 2LTR circles, an indicator for the completion of DNA synthesis by HIV-1 reverse transcriptase but a failure of the DNA to integrate into the host genome. As shown in Figure 6, the 2LTR circles copy number ratio was significantly higher (*, p<0. 05) in the Leishmania maturated MDMs group as compared to control cell group (set to 1. 0). The 2LTR circle number ratio for GM-CSF-maturated MDMs group is higher than the controls cells, but did not achieve statistical significance. Collectively, these data indicate that Leishmania infection promotes a pro-HIV-1 environment within the cell, leading to higher dNTP concentrations that allow for more efficient viral infection. In mutually endemic areas of the world, Leishmania species and HIV-1 primarily co-infect mononuclear phagocytes of infected mammalian hosts. It is widely believed that Leishmania infection found concurrently with HIV-1 induces a state of chronic immune activation leading to subsequent increased HIV-1 viral load and accelerated progression to AIDS [48]. Although the mechanisms underlying this phenomenon are incompletely understood, in vitro studies to date have implicated a variety of Leishmania-induced pro-inflammatory cytokines including TNF-α, IL-1, and IL-6, in stimulating HIV-1 replication in both monocytoid cell lines and macrophages [17], [49]–[52]. For example, the induction of TNF-α is known to activate HIV-1 replication through mechanisms involving transcriptional activation of nuclear factors binding to NF-κB sequences in the HIV-1 LTR [49], while IL-6 and IL-1 appear to promote HIV-1 replication through less well-defined NF-κB-independent transcriptional and post-transcriptional mechanisms [51], [52]. In this study, a novel mechanism is described in which Leishmania infection of HIV-1 infected CD14+ primary human monocytes promotes accelerated HIV-1 expression by induction of MDMs RNR with subsequent elevation of intracellular dNTP concentrations. This same mechanism could explain numerous previous in vitro and in vivo observations of accelerated HIV-1 replication in AIDS clinical trials for patients treated with GM-CSF [53]. Soon after the recognition that HIV-1 was the etiologic agent of AIDS, it was recognized that physiological stimuli, including GM-CSF, could exert an inductive effect on HIV-1 replication in infected monocytoid cells, though the potential mechanisms for this induction have remained unknown [54]. Most subsequent studies have largely confirmed this original observation [55]–[61], although some others have demonstrated opposite results with the suppression of HIV-1 replication [62], [63]. In vivo, however, the results of four clinical trials using GM-CSF therapy in HIV-1 infected patients not treated with anti-retroviral drugs all demonstrated increased plasma levels of HIV-1 RNA and p24 antigen as compared to control patients [64]–[67]. Most recently, the results of the previous negative in vitro studies, in which treatment with GM-CSF may have lowered HIV-1 replication, may be reconciled: the majority of results showed that up-regulation of viral replication was generally enhanced in GM-CSF-maturated MDMs when grown at low densities, whereas more crowded cultures of MDMs and excessive acidification of the medium led to suppressed viral replication [68]. Although GM-CSF treatment promotes maturation of monocytes into macrophages, which are terminally differentiated, non-dividing cells, there is an emerging awareness that, although human monocytes do not proliferate in the steady state, a proliferative monocyte sub-population exists that can re-enter the cell cycle in response to both GM-CSF and M-CSF [18], [19], [21], [24], [69]. Leishmania-infected monocytes/macrophages have been found to be able to produce a variety of colony stimulating factors, most notably GM-CSF [70]–[72]. Here we confirm that monocyte sub-populations treated with GM-CSF are able to re-enter the cell cycle and show, for the first time, that Leishmania infection promotes an S-phase environment in normally quiescent monocyte sub-populations. Statistically significant elevated percentages of BrdU+ cells were found in Leishmania-infected MDMs compared to uninfected controls (Figure 2C). Further, both monocyte maturation and proliferation occurred through a mechanism independent of GM-CSF as treatment with a high concentration of neutralizing antibody, fully sufficient to block the effects of 5 ng/ml added GM-CSF had no effect on the Leishmania-infected cells (Figure S4). These findings are in accord with newly described rodent data demonstrating local in situ proliferation of tissue macrophages in response to infection with a rodent filarial nematode [73] and a previous study demonstrating in situ proliferation of macrophages in the lungs of hookworm-infected mice [74]. The promotion of monocyte proliferation by both GM-CSF treatment and Leishmania infection has profound implications for monocyte cell biology. We now report the quite novel finding that monocyte proliferation, induced by the presence of GM-CSF and, more potently by infection with L. major, also promotes significantly higher dNTP levels at days 7 and 13 in culture as compared to freshly isolated peripheral blood monocytes. Elevated synthesis of the purine, dGTP in particular, was highly statistically significant in day 13 Leishmania-infected MDMs compared to levels in control monocytes (Figure 3C). In addition, induction of cellular RNR, the enzyme catalyzing the direct reduction of ribonucleotides to their corresponding dNTPs was found to be significantly elevated in the Leishmania-maturated MDMs. Specifically, an approximately 40-fold increase in RNR protein levels was observed in immunoblots of day 13 Leishmania-maturated MDMs versus freshly isolated monocytes using an antibody directed against the R2 subunit of human RNR (Figure 4B). That this induction of RNR R2 is regulated at the transcriptional level is supported by the similarly statistically significant elevation of RNR R2 RNA assayed by qRT-PCR (Figure 4C). These findings are particularly intriguing in that the expression of the R2 subunit is known to be strictly and specifically restricted to the S phase of the cell cycle [47], consistent with the observed induction of cell cycle re-entry in both Leishmania- and GM-CSF-maturated MDMs. The present demonstration that Leishmania infection of human monocytes induces elevated dNTP concentrations also has far-reaching implications for Leishmania pathogenesis. Unlike their mammalian hosts, Leishmania lack the metabolic machinery needed for purine nucleotide synthesis. They must therefore rely on the host cell production of purines and have evolved an obligatory purine salvage pathway for this purpose [46]. The dimeric enzyme ribonucleotide reductase is the major source of dNTPs in mammalian and other cells, forming them from the far more abundant pool of rNTPs by the removal of the 2′ OH on the ribose sugar moiety [75]. Our finding that Leishmania infection of human monocytes induces MDMs upregulation of RNR (Figure 4B) is fully consistent with the elevated dNTP concentrations noted above and represents an elegant evolutionary adaptation by which Leishmania can salvage necessary host purines (and pyrimidines). A more recent consequence of Leishmania-mediated induction of host RNR and elevated dNTP concentrations has been to provide a highly permissive environment for HIV-1 replication in the setting of co-infection. These findings are especially significant in light of data that HIV-1 proviral DNA synthesis in non-dividing cells is slower than in dividing cells [76], and can be accelerated by experimentally elevating the intracellular dNTP concentration [42]. They may also be of particular relevance in the setting of infection with Leishmania, in which rapid proliferative expansion of local splenic and bone marrow monocyte/macrophage progenitor populations has been described [70]. In this setting, elevated dNTP concentration would also be expected with accompanying enhancement of HIV-1 replication in such dividing cells. HIV D3 GFP transduction, a model for HIV-1 infection, is also markedly enhanced in these matured cells (Figures 5A, 5B, and Table 1). Both the fluorescent microscopic and flow cytometry results demonstrated substantially increased numbers of HIV-1 D3 GFP+ transduced cells in the setting of Leishmania infection. These findings were further confirmed by a statistically significant elevation of the 2LTR circle copy number ratio in Leishmania infected MDM compared to control monocytes by qPCR (Figure 6). Our results for MDMs maturated by GM-CSF treatment or infection with Leishmania conform well to the majority of studies showing enhanced HIV-1 replication, most likely due to monocytes maturating into macrophages. This is a critical finding in that we have recently reported that HIV replication efficiencies in a wide variety of relevant cell types, including monocytes and macrophages, is directly related to the relative intra-cellular dNTP concentrations [31], [32]. Thus, the finding of elevated dNTP levels in both GM-CSF- and Leishmania-maturated human MDMs, as compared to both freshly isolated monocytes and untreated control cells, offers a novel mechanism to explain both the present results as well as prior in vitro and in vivo studies that demonstrate accelerated HIV-1 replication in both GM-CSF-treated and Leishmania co-infected patients [55], [67], [77], [78]. These results are consistent with the 200–1500 times decrease in replication competence of wild-type HIV-1 in monocytes as compared to the corresponding differentiated MDMs [33]. The present study represents the first demonstration that Leishmania promotes both maturation and proliferation phenotypes in primary human monocytes. During this process we detected elevated intracellular dNTP pools in Leishmania-infected cells, which allows more efficient replication of intracellular co-infected HIV-1. This observation of enhanced pathogen expression in co-infected target cells may be a more generalized phenomenon. For example, the course of HIV-1 related immunodeficiency is also known to be accelerated by active infection with Mycobacterium tuberculosis (MTB) [79], and in vitro studies have demonstrated that MTB-infection of MDMs subsequently infected with HIV-1 produce increased levels of virus as compared to MDMs uninfected with MTB [80]. In matched CD4+ T cell cohorts, both HIV-1 viral load and heterogeneity are increased by MTB infection. In addition, infection of monocytes/macrophages with two other clinically relevant Mycobacterium was found to enhance HIV-1 replication both in vitro and in situ [81]–[83]. Conversely, patients co-infected with HIV-1 and MTB have altered granulomas within the lung [84]. Also higher bacterial burden was detected for HIV-1 and MTB co-infection of MDMs in vitro [85]. Our data suggests that we are just beginning to understand the synergy between virus and parasite co-infections of human cells. These experiments used primary human primary monocytes obtained from human buffy coats (New York Blood Services, Long Island, NY). These are pre-existing materials that are publicly available, and there is no subject-identifying information associated with the cells. As such, the use of these samples does not represent human subjects research because: 1) materials were not collected specifically for this study, and 2) we are not able to identify the subjects. Primary human monocytes were isolated from the peripheral blood buffy coats by positive selection using MACS CD14+ beads as previously described [32]. Three culture condition were used: 1) RPMI 1640 containing 10% FCS and Penicillin/Streptomycin antibiotics without further supplements indicating “control” monocytes, 2) RPMI containing 10% FCS, Pen/Strep antibiotics and 5 ng/ml human recombinant GM-CSF (R&D Systems) indicating “GM-CSF-treated” monocytes, or 3) RPMI 1640 containing 10% FCS, Penicillin/Streptomycin antibiotics and Leishmania major (MOI = 7) indicating “Leishmania-infected” monocytes. Leishmania major promastigotes (strain WHOM/IR/–/173) were grown to stationary phase culture and infectious metacyclic promastigotes were isolated by negative selection using peanut agglutinin [86]. L. major were labeled with 2 µM PKH26 fluorescent cell dye (Sigma) as per manufacturer' s protocol. HIV-1 D3 GFP vector generation: HIV-1 D3 GFP vector encodes the HIV-1 NL4-3 genome with the eGFP gene in place of the HIV-1 nef gene and has a deleted envelope [32]. To generate virus, 293T cells in T225 flasks were transfected with 60 µg pD3-HIV and 10 µg pVSV-g plasmids using 140 µl polyethyenimine (1 mg/ml) in 37 ml DMEM media/flask. At day 1 of HIV-1 production, media was discarded and replaced with fresh DMEM media. At day 2, media was harvested and replaced with fresh DMEM media. The media was centrifuged at 2500 RPM for 7 minutes to remove cellular debris, and then stored at 4°C in T75 flask. Day 3 media was harvested and processed as described for day 2. HIV-1 D3 GFP was concentrated using ultracentrifugation (22K RPM for 2 h in a SW28 rotor). Viral pellets were DNase I digested for 1 h at 37°C. Afterwards, debris was removed by centrifugation (14K for 5 minutes). Sample aliquots were frozen at −80°C until used. Different groups were transduced with HIV-1 D3 GFP and then the samples were analyzed using Accuri C6 flow cytometer monitoring GFP expression at 24 h or 48 h after transduction. Data files were analyzed using FlowJo software (TreeStar). Nucleotide incorporation assay employs a 19-mer DNA template (3′-CAGGGAGAAGCCCGCGGTN-5′). The N indicates the change in template for detecting a specific dNTP within the cellular extract. The template is annealed to a 5′ end 32P-labeled 18-mer DNA primer (5′-GTCCCTGTTCGGGCGCCA-3′). HIV-1 RT is used for this reaction [87]. 1×106 cells for control monocytes, GM-CSF-treated MDMs, and Leishmania-infected MDMs were collected and lysed with 60% cold methanol. Cellular debris was cleared by 14K centrifugation. Supernatant was dried. Pellet was resuspended in 20 µl reaction buffer (50 mM Tris-HCl, pH 8 and 10 mM MgCl2). Two microliters were used in the primer extension assay. Forty-eight hours before harvesting, cells were pulsed with 300 µM BrdU. For microscope analysis, media was removed and the 6-well plate was washed once with PBS. Cells were fixed for using 4% paraformaldehyde for 20 minutes and then washed with PBS. Two milliliters of Target Retrieval Solution (Dako) was added and plates were heated in a rice cooker for 15 minutes at 95°C. Afterwards the plates were removed and allowed to cool. Cells were stained with rat anti-BrdU-FITC antibody (AbD Serotec) for 20 minutes at 4°C. Images were captured using a Zeiss microscope. For FACS analysis, on the day of harvest, the free cells were collected while the adherent cells were Trypsin treated for 30 minutes before scraping the 6-well plate. Both free and adherent cell populations were pooled, centrifuged at 1200 RPM for 5 minutes. Supernatant was removed and the cells were fixed using 4% paraformaldehyde for 20 minutes. After fixing, the cells were washed once with PBS. The cells were stored at 4°C until processing for BrdU staining. For BrdU staining, cells were transferred to a 6-well plate containing 2 ml of Target Retrieval Solution and heated in a rice cooker for 15 minutes at 95°C. Afterwards the plates were removed and allowed to cool. Cells were transferred to tubes and cells washed once with PBS. Next the cells were stained with rat anti-BrdU-FITC antibody for 20 minutes at 4°C. The sample data were collected using an Accuri C6 flow cytometer. Samples were processed in RIPA buffer containing 1 µM DTT, 10 µM PMSF, 10 µl/ml phosphatase inhibitor (Sigma) and 10 µl/ml protease inhibitor (Sigma). The cells were sonicated with 3X, 5 second pulses, to ensure complete lysis. Cellular debris was removed by 15K RPM centrifugation for 10 minutes. Supernatants were stored at −80°C before use. Cell lysates (25 µg) were resolved on an 8% SDS-PAGE gel. Proteins were transferred to a nitrocellulose membrane. The membrane was blocked with 2% non-fat milk in TBST for 1 h, followed by the addition of primary goat anti-R2 antibody (Santa Cruz Biotechnology) and incubation overnight at 4°C. The next day, the membrane was washed (3X, 20 minutes with TBST) followed by staining with donkey anti-goat HRP for 1 h at room temperature. The membrane was washed 3× with TBST and developed using SuperSignal West Femto Kit (Thermo Scientific). The immunoblot was then stripped and re-probed for actin. Images were captured using a BioRad ChemiDoc Imager. 4×106 cells were lysed and RNA prepared using the RNeasy Mini Protocol as per the manufacturers' instructions (Qiagen, Valencia, CA). Pre-mixed Taqman primer/probe sets for RNR R2 and p53R2 were obtained from Life technologies (Cat numbers Hs01072069_gi and Hs00968432_m1, respectively). Template RNA was diluted to 80 ng/µl and 4 µl from each sample, mixed with Express One-Step SuperScript qRT-PCR reagents, was ran in triplicate using an Applied Biosystems 7300 Real Time thermocycler. Data were normalized to GAPDH mRNA. Genomic extracts were prepared using QuickGene-810 Nucleic Acid Isolation System (FujiFilm Global). The DNA was assayed for 2LTR circles by real time PCR using the following primers: 5′-LTR region — 5′-GTGCCCGTCTGTTGTGTGACT-3′ and 3′LTR region — 5′-CTTGTCTTCTTTGGGAGTGAATTAGC-3′, and the probe 5′-6-carboxylfluorsecein-TCCACACTGACTAAAAGGGTCTGAGGGATCTCT-carboxytetramethylrhodamine-3′ (IDT). All samples were normalized to total DNA. The control samples for each donor were set to 1. 0 and 2LTR circle copy number ratio was plotted. Prism software was used for plotting the data. All the data sets were compared for significant difference using ANOVA analysis (Friedman test).

Leishmaniasis is a parasitic disease that infects several human host immune cells, including neutrophils, monocytes, and macrophages. Moreover, while HIV-1 infects monocytes and macrophages, only the infected macrophages productively release viral progenies. Importantly, patients coinfected with both pathogens progress more rapidly to AIDS. In this study, we examine how Leishmania major changes the cellular environment of monocytes in vitro. We found that Leishmania-infected monocytes actively mature into macrophages in the absence of GM-CSF, and that these cells up-regulate the expression of ribonucleotide reductase, an enzyme that catalyzes the formation of deoxynucleoside triphosphates (dNTPs). We confirmed the elevation of dNTP concentrations using a very sensitive dNTP assay for monocytes and monocyte-maturated macrophages. Collectively, these data support a model in which infection of monocytes with Leishmania elevates the intracellular dNTP pools, which is one of the natural anti-viral blocks to HIV-1 infection in monocytes and macrophages in patients.

Abstract Introduction Results Discussion Materials and Methods

medicine infectious diseases biology microbiology molecular cell biology

2012

Leishmania Induces Survival, Proliferation and Elevated Cellular dNTP Levels in Human Monocytes Promoting Acceleration of HIV Co-Infection

Medulloblastoma is the most common malignant brain tumor in children. A subset of medulloblastoma originates from granule cell precursors (GCPs) of the developing cerebellum and demonstrates aberrant hedgehog signaling, typically due to inactivating mutations in the receptor PTCH1, a pathomechanism recapitulated in Ptch1+/− mice. As nitric oxide may regulate GCP proliferation and differentiation, we crossed Ptch1+/− mice with mice lacking inducible nitric oxide synthase (Nos2) to investigate a possible influence on tumorigenesis. We observed a two-fold higher medulloblastoma rate in Ptch1+/− Nos2−/− mice compared to Ptch1+/− Nos2+/+ mice. To identify the molecular mechanisms underlying this finding, we performed gene expression profiling of medulloblastomas from both genotypes, as well as normal cerebellar tissue samples of different developmental stages and genotypes. Downregulation of hedgehog target genes was observed in postnatal cerebellum from Ptch1+/+ Nos2−/− mice but not from Ptch1+/− Nos2−/− mice. The most consistent effect of Nos2 deficiency was downregulation of growth-associated protein 43 (Gap43). Functional studies in neuronal progenitor cells demonstrated nitric oxide dependence of Gap43 expression and impaired migration upon Gap43 knock-down. Both effects were confirmed in situ by immunofluorescence analyses on tissue sections of the developing cerebellum. Finally, the number of proliferating GCPs at the cerebellar periphery was decreased in Ptch1+/+ Nos2−/− mice but increased in Ptch1+/− Nos2−/− mice relative to Ptch1+/− Nos2+/+ mice. Taken together, these results indicate that Nos2 deficiency promotes medulloblastoma development in Ptch1+/− mice through retention of proliferating GCPs in the external granular layer due to reduced Gap43 expression. This study illustrates a new role of nitric oxide signaling in cerebellar development and demonstrates that the localization of pre-neoplastic cells during morphogenesis is crucial for their malignant progression. Medulloblastoma (MB) is a highly malignant tumor of the cerebellum that preferentially develops in children and adolescents. Although the survival rate for standard risk MB is around 70% [1] surviving patients often suffer from neurodevelopmental and cognitive side effects of the aggressive therapy [2]. Therefore, improved understanding of the molecular pathomechanisms driving MB growth is necessary to develop less toxic and more effective treatments. Recent molecular profiling studies suggested at least four MB subtypes that are associated with distinct expression profiles, genomic aberrations and clinical features [3], [4]. One of these MB subtypes is characterized by aberrant activation of the hedgehog (Hh) pathway and typically corresponds to the desmoplastic (nodular) MB variant. This subtype is supposed to develop from granule cell precursors (GCPs) of the external granular layer (EGL) [5]. The EGL is a transient germinal zone at the subpial cerebellar surface consisting of rhombic lip-derived progenitor cells that have migrated tangentially to the emerging cerebellar cortex at late stages of embryonal brain development [6]. During the early postnatal period in mouse, the morphogenic factor sonic hedgehog (Shh) is secreted by subjacent Purkinje cells and binds to patched receptors (Ptch1 and Ptch2) expressed on the GCP surface [7]. Ligand binding to Ptch1 then leads to functional de-repression of Smoh (Drosophila smoothened homolog) and subsequent activation of Gli (Glioma-associated oncogene family zinc finger) transcription factors [8]. This launches a temporally concerted gene expression pattern causing a proliferation burst and massive expansion of the GCP population during the first two postnatal weeks [7]. In particular, the direct Gli-target N-myc [9], [10] and D type cyclins [11] were shown to be crucial for the growth and neoplastic transformation of GCPs [12]. In addition, the set of genes targeted by activated Gli transcription factors also include components of the canonical Hh pathway for feedback-loop regulation, such as the receptors Ptch1 and Ptch2 as well as the hedgehog-interacting protein (Hip) [10], [13]. After several rounds of cell division, GCPs normally exit cell cycle and accumulate at the inner site of the EGL [14], where they start to migrate through the molecular layer (ML) and the Purkinje cell layer to form the internal granular layer (IGL) [15]. The mechanisms underlying the attenuation of the mitotic response and eventually the stop of GCP proliferation are not well understood. The most evident concepts describe extrinsic cues in gradient-based models to trigger GCP differentiation with increasing distance to the region of the outer EGL [16]. Finally, the EGL disappears at about three weeks after birth in mice. PTCH1 was identified as a frequent target of inactivating mutations or genomic loss in sporadic MBs [17]–[19] that belong to the molecular subtype hallmarked by an aberrant activity of hedgehog signaling. The monoallelic inactivation of the Ptch1 gene in mice and thus downstream activation of the Hh pathway leads to MB development at a frequency of about 10–15% [20]. This mouse model has provided substantial insights into the pathogenesis of Hh-dependent MBs and has been used in different cross-breeding experiments to investigate tumor suppressor gene functions in this particular context [21], [22]. Nitric oxide (NO) is a highly reactive gaseous molecule involved in various physiological processes ranging from vasculature modulation to neurotransmission [23], [24]. NO is produced by three distinct enzyme isoforms: i) neuronal nitric oxide synthase (nNos/Nos1), ii) inducible nitric oxide synthase (iNos/Nos2), and iii) endothelial nitric oxide synthase (eNos/Nos3). Though being constitutively expressed in their respective tissue, nNos and eNos activity strongly depends on calcium [25], whereas calcium-independent iNos is primarily regulated by transcriptional induction, e. g. by inflammatory cytokines and endotoxins [26], which permits higher quantities of NO generation. The role of NO in cancer initiation and progression is heterogeneous with opposing effects in different malignancies [27]. Considering effects of tumor stroma, increased angiogenesis was reported to be associated with elevated Nos activity [28] and some immune-related processes were found to be mediated by NO [29], including cytotoxicity of activated microglia [30]. Finally, NO released by vascular endothelial cells was reported to build a niche-like microenvironment for maintenance of glioma stem cells [31]. In the context of cerebellar development, Nos2 (inducible Nos) is initially expressed in early GCPs, whereas Nos1 (neuronal Nos) is hardly present before postnatal day 7 (Cerebellar Development Transcriptome Database [32]). Successively, Nos1 expression increases along with granule cell differentiation [33] and predominantly contributes to the common NO signaling that becomes apparent in the IGL as development proceeds [34]. Evidence has been provided that NO negatively acts on proliferation of neuronal precursors during adult neurogenesis [35]. Similarly, Ciani and colleagues demonstrated enhanced proliferation of cerebellar precursor cells upon inhibition of NO synthases [36]. Here, we report on the generation of Ptch1+/− Nos2−/− mice to investigate the impact of Nos2 on tumor development in Ptch1 hemizygous mutant mice. Interestingly, we observed an approximately two-fold increase in the incidence of spontaneous MB in Ptch1+/− Nos2−/− mice in comparison to Ptch1+/− Nos2+/+ mice. To characterize the molecular pathomechanism underlying the tumor-promoting effect of Nos2 deficiency in Ptch1+/− mice, we performed comprehensive expression and DNA copy number profiling of MB tumors (Ptch1+/− Nos2+/+ versus Ptch1+/− Nos2−/−) as well as expression profiling of normal cerebellar tissue samples from different developmental stages and various genotypes (Ptch1+/− Nos2+/+, Ptch1+/− Nos2−/−, Ptch1+/+ Nos2−/− and wild-type mice). Downregulation of the growth-associated protein 43 (Gap43) was the most striking feature in the cerebellum of Nos2-deficient mice when compared to Ptch1+/− Nos2+/+ and wild-type mice. Subsequent functional analyses and results from in situ studies of GCPs in postnatal cerebellum allowed us to formulate a model for the tumor promoting role of Nos2 deficiency in Ptch1 mutant mice via deregulation of Gap43-dependent migration of GCPs. Survival analyses of 315 wild-type mice, 412 Ptch1+/+ Nos2−/− mice, 215 Ptch1+/− Nos2+/+ mice and 221 Ptch1+/− Nos2−/− mice demonstrated a significantly higher MB incidence in the group of Ptch1+/− Nos2−/− mice relative to the group of Ptch1+/− Nos2+/+ mice (p = 0. 0007, Logrank test, Figure 1A). In total, 11% of the Ptch1+/− Nos2+/+ mice (24/215) and 21% of the Ptch1+/− Nos2−/− mice (47/221) were sacrificed due to the development of cerebellar MB. None of the 315 wild-type and the 412 Ptch1+/+ Nos2−/− mice developed MBs. These observations indicate a MB-promoting role of Nos2 deficiency in Ptch1+/− mice. In humans, Hh-dependent MBs typically correspond to the desmoplastic subtype. MBs in Ptch1+/− mice, however, microscopically resemble the classic MB subtype [20]. Histological analysis of MBs in Ptch1+/− Nos2+/+ and Ptch1+/− Nos2−/− mice demonstrated similar morphological features (Figure 1B–1E). The tumors were composed of densely packed sheets of cells with hyperchromatic carrot-shaped nuclei and scant cytoplasm. There were no obvious histopathological differences between MBs of the two genotypes. For an initial assessment of the molecular tumor characteristics, gene expression of hedgehog signaling pathway components were measured in 21 MBs and 24 normal (adult) cerebellar tissue samples from both Ptch1+/− Nos2+/+ and Ptch1+/− Nos2−/− mice. Using quantitative real-time PCR (qRT-PCR), significant downregulation of the wild-type Ptch1 transcript and upregulation of the Shh target genes Gli1 and N-myc were generally observed in the tumor samples (Figure S1), indicating all examined MBs to be of the same Hh-dependent molecular subtype. However, there were no significant differences for these genes between MBs of the two genotypes. Furthermore, targeted genetic analyses showed a loss of the wild-type Ptch1 allele in 10 of the 21 MBs investigated, while none of the tumors demonstrated a Tp53 mutation or N-myc amplification. The Cdkn2a/p16INK4a locus was retained in all tumors while a single MB demonstrated a homozygous p19ARF deletion (see Table S1 and Text S1 for details). In order to identify the molecular pathomechanism contributing to the increased MB rate in Nos2-deficient Ptch1 mutant mice, we performed array-based gene expression profiling of three Ptch1+/− Nos2+/+ versus six Ptch1+/− Nos2−/− and comparative genomic hybridization (array-CGH) analyses of five Ptch1+/− Nos2+/+ versus seven Ptch1+/− Nos2−/− MB tissue samples. All specimens investigated had tumor cell contents between 70% and 90% as determined on corresponding formalin-fixed and paraffin-embedded (FFPE) reference sections. Differential expression of selected candidate genes was validated by qRT-PCR on an expanded, partially overlapping tumor set of seven Ptch1+/− Nos2+/+ versus seven Ptch1+/− Nos2−/− MB samples. The expression profiling analysis revealed a total of 87 differentially regulated genes between tumors of the two genotypes (Table S2) with the vast majority (87%) showing lower transcript levels in Ptch1+/− Nos2−/− when compared to Ptch1+/− Nos2+/+ mice. As expected from the initial targeted qRT-PCR measurements, there was no difference detectable concerning the activation of Hh pathway genes. Due to the important role of Nos2 during angiogenesis and cancer-associated immune response, including microglia, stromal effects need to be particularly considered in a systemic Nos2 knockout model. However, neither the set of significantly deregulated genes nor selective determination of marker expression for pericytes, vascular endothelial cells or microglia suggested any differences in the tumor stroma between the two genotypes (see Table S3 and Text S1 for details). According to the findings of Ciani and co-workers [37], reduction of NO enhances GCP proliferation through an increased expression of the proto-oncogene N-myc. Therefore, protein levels were particularly examined for differences between tumor samples from Ptch1+/− Nos2+/+ and Ptch1+/− Nos2−/− mice. The results shown in Figure S2, however, revealed similar expression of N-myc in all MBs. Analyses of genomic copy number alterations revealed a trisomy of chromosome 6 in the majority of MBs from both groups (11/12, Figure 2A and 2B). Moreover, a small region on chromosome 13, approximately 1. 5 Mb upstream of the Ptch1 gene, showed a hemizygous deletion in healthy cerebella of Ptch1-mutant mice (data not shown) but a homozygous deletion in most tumors (10/12). Similarly, a second small region 3. 8 Mb downstream of the last Ptch1 exon exhibited a loss in 9 of 12 MBs. This suggests structural changes flanking the Ptch1 locus that likely contribute to inactivation of the wild-type allele. The frequencies of genomic aberrations showed a more heterogeneous karyotype with gross structural changes in Ptch1+/− Nos2+/+ MBs when compared to Ptch1+/− Nos2−/− MBs (see Figure 2A, 2B and Text S1 for details). However, a general difference in chromosomal instability was not obvious between both genotypes. Only a small region (205. 6 kb) on chromosome 14 containing the Entpd4 (ectonucleoside triphosphate diphosphohydrolase 4) gene was more frequently gained in Ptch1+/− Nos2−/− MBs (7/7) than in Ptch1+/− Nos2+/+ MBs (1/5, Figure 2A). Accordingly, Entpd4 expression appeared to be specifically upregulated in expression profiles of Ptch1+/− Nos2−/− tumors. QRT-PCR validation confirmed an elevated mean expression in Ptch1+/− Nos2−/− compared to Ptch1+/− Nos2+/+ MBs in those samples that overlapped with the array-CGH analysis but revealed no significant difference across the expanded tumor set (Figure 2C). As GCPs are considered the cells of origin for the Hh-dependent MB subtype, we examined the effect of Nos2 ablation in the context of cerebellar development. Therefore, gene expression profiles of normal cerebellar tissue samples from postnatal day 9 (P9), 6 weeks after birth (6W), and 1 year of age (1Y) were generated from wild-type, Ptch1+/− Nos2+/+, Ptch1+/− Nos2−/−, and Ptch1+/+ Nos2−/− animals. While specimens of mature cerebellum (6W and 1Y) were investigated separately in 3 biological replicates per genotype and developmental stage, samples of postnatal cerebellum consisted of pooled RNA from 4–5 individuals processed in technical replicates due to limited tissue amounts. An unsupervised hierarchical cluster analysis of transcriptome data clearly separated developing cerebellum (P9) of wild-type mice and the two Ptch1-mutated genotypes from mature cerebellum. Interestingly, P9 cerebellum of Ptch1+/+ Nos2−/− mice displayed different properties highly similar to mature cerebellum (Figure 3A). Expression profiles of MBs formed a distinct cluster clearly separated from all healthy tissue samples. A direct comparison between gene expression profiles from Ptch1+/+ Nos2−/− and wild-type P9 cerebellar tissue samples resulted in a total of 984 deregulated genes with 755 genes (76. 7%) showing a decreased expression in Ptch1+/+ Nos2−/− mice (Table S4). P9 cerebellum from Ptch1+/− Nos2+/+ and Ptch1+/− Nos2−/− mice revealed only 5 and 32 deregulated genes relative to wild-type, respectively (Table S5 and Table S6). This large deviation of postnatal gene expression in the Ptch1+/+ Nos2−/− genotype included a set of downregulated genes that are essential for proliferation of GCPs (e. g. cyclin D1, cyclin D2 and N-myc, Figure 3B). As hedgehog signaling constitutes the main regulatory pathway for neonatal cell proliferation in GCPs of the EGL, the 984 deregulated genes were analyzed for enrichment of Gli transcription factor targets. Matching this list to a set of recently identified Gli-targets in GCPs [10] yielded a significant overrepresentation of Gli1-regulated genes (p = 0. 005, chi-square test). Hence, the reduced transcript levels of these target genes suggests an attenuated hedgehog signaling in postnatal Ptch1+/+ Nos2−/− cerebellum compared to wild-type (or any other genotype). Notably, the decreased expression of Gli1-targets and proliferation-associated genes observed in Nos2-deficient cerebellar tissue was abolished upon additional inactivation of the hedgehog receptor Ptch1 (in Ptch1+/− Nos2−/− mice). Therefore, we examined the transcript levels of patched receptors themselves in more detail. While neither Ptch1 nor Ptch2 expression was changed between wild-type and Ptch1+/− Nos2+/+ P9 cerebellum, a significant increase of Ptch1 and a minor increase of Ptch2 expression were observed in Ptch1+/+ Nos2−/− mice relative to wild-type mice (Figure 3C). Notably, in Ptch1+/− Nos2−/− cerebellar tissue samples, Ptch2 expression was more elevated than Ptch1. However, since Ptch2 is not capable of inhibiting smoothened (Smoh), it probably failed to take over the attenuating effect on Gli activity [38]. MB specimens from Ptch1+/− Nos2+/+ versus Ptch1+/− Nos2−/− mice showed no significant difference in expression levels of either patched receptor, with Ptch2 being substantially increased over Ptch1 in both groups (Figure 3C). These findings indicate that Nos2 deficiency leads to an upregulation of Ptch1 in GCPs, which results in a downregulation of mitotic genes and Gli-targets only in a Ptch1-wild-type background. So far, Nos2 inactivation was shown to counteract proliferation and antagonize hedgehog signaling in developing cerebella. To identify those Nos2-dependent effects promoting MB induction, we determined the features that were common to Ptch1+/+ Nos2−/− and Ptch1+/− Nos2−/− genotypes and persisted in the tumor tissues. Accordingly, the overlap of differential gene expression from three comparisons was built: i) Ptch1+/+ Nos2−/− versus wild-type P9 cerebellum, ii) Ptch1+/− Nos2−/− versus Ptch1+/− Nos2+/+ P9 cerebellum; and iii) Ptch1+/− Nos2−/− versus Ptch1+/− Nos2+/+ MB. As a result, only 2 genes were observed to be deregulated in a Nos2-dependent manner during cerebellar development and in MBs (Figure 4A). Although Stmn1 (stathmin 1) appeared to be upregulated in Ptch1+/− Nos2−/− MBs relative to Ptch1+/− Nos2+/+ MBs, this could not be confirmed by qRT-PCR (Figure S3). Gene expression of Gap43 was consistently reduced in Nos2-deficient cerebellar tissue samples and downregulation in Ptch1+/− Nos2−/− tumors relative to Ptch1+/− Nos2+/+ tumors was also significant in the expanded validation set (Figure 4C and 4D). To further assess the immediacy of Nos2 inactivation and Gap43 deregulation, Gap43 transcript levels were determined in expression profiles of healthy cerebella from all developmental stages (P9,6W and 1Y). Groups for comparison were built according to presence or absence of Nos2, irrespective of the Ptch1 status. The results clearly demonstrated a close association of altered Gap43 transcript levels and Nos2 status (Figure 4B), and indicated downregulation of Gap43 to be the most common effect of Nos2 deficiency in the cerebellum. To investigate differences in Gap43 expression on protein level in situ we performed immunofluorescent double stainings of Gap43 and the proliferation marker Ki-67 on FFPE sections of P9 cerebella from wild-type, Ptch1+/− Nos2+/+, Ptch+/+ Nos2−/− and Ptch1+/− Nos2−/− mice. As illustrated in Figure 4E, Gap43 immunofluorescence was particularly prominent in the outer region of the molecular layer (ML) that is connected to and partially comprised of radial GCP process extensions. Image quantification further indicate a quantitative difference of Gap43 expression in this region with sections from wild-type and Ptch1+/− Nos2+/+ mice showing a more intense staining than sections from Ptch1+/− Nos2−/− and Ptch1+/+ Nos2−/− mice (Figure 4F). The association of Nos2 inactivation and decreased Gap43 expression suggests a gene-regulatory function of NO signaling. In order to investigate this possible link in vitro, we used the murine cerebellar precursor cell line c17. 2 and the human MB cell line D458 (see Text S2 for details). Both cell lines were treated either with the Nos inhibitor L-NAME (Nω-nitro-L-arginine methyl ester) to reduce NO levels or solvent control (Figure S4). Relative expression of Gap43 was assessed every 24 hours by qRT-PCR. In c17. 2 cells, Gap43 transcript abundance was generally low and increased with culture duration. We observed a slightly decreased expression of Gap43 upon L-NAME treatment that reached significance (p = 0. 023) after 120 hours (Figure 5A). NOS inhibition in D458 human MB cells resulted in a significant reduction of Gap43 transcript levels starting already after 72 hours with further decrease after 96 hours and 120 hours (Figure 5B). FACS analyses of apoptosis and cell cycle excluded these observations to be attributed to secondary effects of changing cell conditions (Figure S5). These results suggest Gap43 downregulation as a direct consequence of reduced NO levels in murine neuronal precursors and human MB cells. The dependency of Gap43 expression on NO signaling suggests this gene as key mediator of the effects observed in Nos2-deficient P9 cerebellum and Ptch1+/− Nos2−/− MB, in particular, the upregulation of functional Ptch1 in Ptch1+/+ Nos2−/− mice. Mishra et al. recently reported a central role of Gap43 in the polarization of developing GCPs by regulating centrosome positioning and thus defining correct orientation towards the IGL [39]. Since this is a prerequisite for directed migration, reduced levels of Gap43 in P9 cerebellar tissue may lead to retention of GCPs in the EGL. To test these hypotheses, shRNA-mediated knockdown of Gap43 was performed in c17. 2 cells (see Text S1 for details). Upon knockdown of Gap43 we observed a strong inverse behavior of Ptch1 and Gap43 transcript levels (Figure 5C). Changes in migration characteristics were assayed in a Boyden chamber using recombinant SDF-1α (CXCL12) as chemoattractant, which was reported to participate in guiding migration of embryonal GCPs in vivo [40]. Downregulation of Gap43 yielded a significant decrease in cell migration between 14% (p = 0. 013) and 20% (p = 0. 007) (Figure 5D; Figure S6C). A pseudo-effect of the knockdown due to altered proliferation of the v-myc-immortalized c17. 2 cells was excluded by FACS-based cell cycle analysis (Figure S7). Transcriptome and functional analyses suggest that a decreased Gap43 expression accounts for Ptch1 upregulation and impairment of directed neuronal precursor migration in vitro. Accordingly, Ptch1+/+ Nos2−/− P9 cerebella are supposed to increasingly retain GCPs with reduced mitotic activity in the EGL compared to wild-type and Ptch1+/− Nos2+/+ mice. Moreover, the Ptch1+/− Nos2−/− genotype is also expected to exhibit retention of GCPs, but not to show any cell cycle arrest. To further verify this hypothesis in situ we performed immunofluorescent double staining of proliferating (Ki-67+) and post-mitotic GCPs on FFPE sections of postnatal cerebellum (Figure 6A). Here, post-mitotic cells were delineated by the neuronal marker NeuN (neuronal nuclear antigen A60) [41]. At least three different regions of each mouse cerebellum were analyzed in three to four animals per genotype using confocal laser scanning microscopy. In accordance with the microarray data, averaged cell counts of wild-type and Ptch1+/− Nos2+/+ mice did not show significant difference. In contrast, an increase of post-mitotic GCPs (NeuN+, Ki-67−) was detectable in the EGL of Ptch1+/− Nos2−/− and Ptch1+/+ Nos2−/− mice (Figure 6C). Concurrently, the ratio of dividing to non-dividing GCPs was similar in Ptch1+/− Nos2−/−, wild-type and Ptch1+/− Nos2+/+ P9 cerebella but markedly decreased in Ptch1+/+ Nos2−/− mice. This recapitulated the downregulation of mitotic genes observed in the expression profiles. However, the total amount of proliferating GCPs per EGL section was significantly higher in Ptch1+/− Nos2−/− mice compared to any other genotype (Figure 6C). These results demonstrate a tissue phenotype that corresponds to the effects of reduced Gap43 in developing cerebellar neuronal precursors (in vitro). The increased accumulation of proliferating GCPs in the EGL observed in the Ptch1+/− Nos2−/− genotype supposedly leads to a larger pool of cells susceptible to neoplastic transformation and is therefore likely to promote medulloblastoma development. The Ptch1+/− MB mouse model has been intensively studied and has greatly contributed to our understanding of Hh-dependent MB tumorigenesis in the context of cerebellar development. The data presented here indicate a role of Nos2 and hence NO signaling in Hh-dependent MB by demonstrating a significantly increased MB rate in Ptch1+/− Nos2−/− mice compared to Ptch1+/− Nos2+/+ mice. The global genome-wide screens performed in the present study did not reveal obvious molecular differences between MBs in Ptch1+/− Nos2+/+ versus Ptch1+/− Nos2−/− animals. Assessment of genomic alterations using array-CGH identified trisomy of chromosome 6 as a recurrent feature in tumors of both genotypes. This corresponds to a recent report on MBs of the same molecular subtype with inactivated double-strand break repair proteins targeted to neuronal progenitors of p53−/− mice [42]. The most common loss identified in our analyses affected two small regions on chromosome 13 encompassing the Ptch1 gene and possibly indicate acquired homozygosity for the mutant allele or somatic rearrangements rather than a broad deletion of the locus. Targeted duplex PCR further confirmed loss of the functional wild-type allele to be a frequent event in these MBs. Notably, tumors of the Ptch1+/− Nos2−/− genotype showed a higher frequency of a small gain on chromosome 14. The affected Entpd4 gene encodes for an apyrase located at the internal membrane of lysosomal vacuoles and the Golgi apparatus. It preferentially catalyzes the hydrolysis of UDP to UMP [43] and thereby facilitates the inverse directed import of UDP-GlcNAc [44]. This in turn was reported to increase glycosylation of surface receptors (e. g. EGFR and PDGFR) and foster cell growth [45]. According to the microarray and qRT-PCR expression data, Entpd4 transcript levels were indeed increased in tumors with this chromosomal gain. However, this effect did not turn out to be Nos2-dependent in an expanded sample set. Consequently, Entpd4 likely plays a role in MB pathogenesis but is not directly linked to loss of Nos2. The examination of tumor-relevant changes in developing cerebellum as a consequence of impaired Nos2 activity and hence NO signaling surprisingly revealed a decreased proliferation of GCPs in the cerebellum of Ptch1+/+ Nos2−/− mice. The concurrent upregulation of Ptch1 and the significant enrichment of downregulated Gli1-target genes strongly suggest that this effect is a consequence of reduced hedgehog signaling. Moreover, this phenotype was completely abrogated by a concomitant Ptch1 mutation. The slight increase of Ptch2 in Ptch1+/− Nos2−/− cells points to a compensatory effect and further supports the notion of an inhibitory function of Nos2 loss on the hedgehog pathway in postnatal cerebellum. Since neither a Smoh-regulating domain [38] nor a function for cell cycle arrest through seizing cyclin B1 [46] were reported for Ptch2, its upregulation may be insufficient for preventing MB induction. In contrast to these observations, Ciani et al. demonstrated that proliferation of cultured GCPs increased upon withdrawal of NO and that this effect was mediated by augmented N-myc levels [37]. However, N-myc was not differentially expressed between Ptch1+/− Nos2−/− and Ptch1+/− Nos2+/+ MBs of our series. A possible explanation for this discrepancy might be an unrecognized heterogeneity in the isolated cerebellar cell population used in the Ciani study. Since eNos and nNos are known to attenuate the mitotic activity of subventricular neuronal stem cells [47], [48] Nos inhibitor treatment possibly resulted in a selective growth advantage over GCPs. Downregulation of Gap43 was the only feature observed in Nos2-deficient versus Nos2-proficient postnatal cerebella irrespective of the Ptch1 status. This difference was also conserved between Ptch1+/− Nos2−/− and Ptch1+/− Nos2+/+ MBs, and particularly visible in outer regions of the molecular layer, where maturating GCPs of the EGL develop contact forming projections prior to radial migration. Other studies already suggested a link between Gap43 mRNA levels and NO signaling due to co-induction of nNos and Gap43 during axon regeneration and reactive synaptogenesis following injury of spinal motoneurons [49] and sensory neurons [50]. Furthermore, a downregulation of Gap43 was reported after silencing of soluble guanylate cyclase subunits, the central elements of cGMP-mediated NO signaling [51]. Finally, the present study demonstrates Gap43 downregulation to be a consequence of NO withdrawal in neuronal progenitors and MB cells. A possible mechanism for this regulation refers to decreased protein levels of the poly (U) -binding and degradation factor AUF1 upon NO-dependent cGMP production [52]. AUF-proteins generally bind to AU-rich elements of the 3′UTR (untranslated region) of coding transcripts and associate with proteins of the ELAV-like family to control gene expression via mRNA decay [53]. Tsai et al. demonstrated that Gap43 mRNA levels are post-transcriptionally regulated during neuronal differentiation and that elements of the 3′UTR confer transcript instability, which is abolished upon TPA treatment (inter alia inducing NOS2) [54]. At the same time, Chung et al. demonstrated that indeed ELAV-like family member HuD was binding to 3′UTR regions of GAP43 [55]. Taken together, NO accumulation possibly decreases cellular levels of mRNA-destabilizing AUF1 protein and thus might contribute to a high transcript abundance of Gap43. Gap43 is a membrane-anchored protein at the cytoplasmic side of neuronal cell projections and found to be highly expressed during development of the CNS [56]. It is particularly localized in axonal growth cones and participates in the coordination of extrinsic stimuli and intrinsic cell remodeling [57] by regulating cytoskeleton dynamics [58]. Granule cell (GC) migration follows a sequence of tangential and radial movements controlled by successive formation of leading projections [59]. As maturating GCPs exit cell cycle, positioning of the centrosome determines the site of axon growth cone emergence and thus neuronal polarity including localization of such projections [60]. This defines the structural orientation of GCPs in terms of directing its dendrite to descend across the molecular and Purkinje cell layers to populate the IGL. However, centrosome positioning and therefore accurate polarization of GCPs require phosphorylated Gap43 to bind to the centrosome-associated microtubule-organizing center [61]. Hence, inaccurate GCP migration was observed in Gap43−/− animals [39], a finding that is in full agreement with our data from the functional Gap43 knockdown assays. Downregulation of Gap43 in Nos2-deficient P9 cerebellum therefore likely mediates the retention of GCPs observed in FFPE sections. Accordingly, NO/cGMP signaling was demonstrated to be crucial for accurate migration of the neuronal precursor cell line NT2 [62]. Furthermore, slice culture experiments of neonatal cerebella (P9) exhibited a substantial reduction of proliferation and migration of maturating granule cells to the IGL upon application of NO synthases inhibitors [63]. The elevation of Ptch1 levels upon Gap43 reduction in vitro fits to the data by Shen et al. who reported an upregulation of Ptch1 gene expression in inner EGL regions of Gap43−/− mice compared to wild-type animals. Moreover, cultured Gap43-deficient GCPs show decreased proliferation in response to administered recombinant Shh protein [64]. A possible regulatory link was recently provided as the activation of the hedgehog signaling component Smoh was found to depend on PI4P (phosphatidylinositol 4-phosphate) levels that immediately increase when Shh binds to Ptch1 or when functional Ptch1 is absent [65]. The authors further showed that imbalanced conversion of the precursor molecule PI into PI4P influences hedgehog pathway activity. Alternatively, the production of PI4P can also result from a specific dephosphorylation of PI (4,5) P2 [66]. In this context, Gap43 protein was recently demonstrated to build oligomeric structures in the plasma membrane which sequester specifically PI (4,5) P2 [67]. A similar finding has been reported earlier showing that GAP43 participates in the accumulation of plasmalemma rafts, which promoted retention of PI (4,5) P2 [68]. The amount of Gap43 associated with the plasma membrane therefore possibly modulates the utilization of PI (4,5) P2, including its conversion into PI4P, which in turn directly affects hedgehog signaling through Smoh activation. However, the effective impact on downstream Gli-targets would still be difficult to conclude regarding the multitude of responses to Shh, including negative feedback regulation [13]. Further studies applying depletion and enrichment of specific phosphatidyl derivatives and selective silencing of hedgehog pathway elements will be necessary to elucidate the molecular nature of this proposed signaling axis. The increased accumulation of mitotic granule cells at the EGL seen in the combined Ptch1+/− Nos2−/− genotype supposedly gives a special clue to MB induction. In contrast to the classical view of neonatal EGL organization, which describes radial migration of granule cells to follow a proliferation stop, more and more evidence arises showing that cell cycle arrest is not a prerequisite for migration but rather occurs during a temporally coordinated interplay of gene expression patterns [69]. This corroborates our data shown in Figure 6 (white arrows), where proliferation is still maintained in migrating granule cells and even in cells of the IGL of wild-type cerebellum. Regulation of such expression patterns is largely dependent on Shh stimuli being most intensive in the EGL [70], as well as gradients of other soluble factors such as Bmps, which were reported to account for a regulatory environment along the transition through the cerebellar layers [71]. Further evidence for a niche-like-concept was provided by Choi et al. in Bdnf−/− mice that displayed a severe retardation of GCP migration [16]. The authors could demonstrate that mitotic activity of maturating GCPs was significantly enhanced when cells were retained in the EGL and declined with increasing distance from outer EGL regions. Therefore, the accumulation of GCPs in the EGL in combination with the insensitivity to Ptch1-mediated cell cycle arrest in Ptch1+/− Nos2−/− mice provide a growth advantage and increase the number of putative transformation targets over Ptch1+/− Nos2+/+ mice (Figure 7C). In conclusion, the following picture emerged from our data: Homozygous deletion of Nos2 leads to a reduction of basic NO levels in immature GCPs of the EGL during postnatal development of the cerebellum. This reduction causes a downregulation of Gap43 expression, which results in an increased expression of Ptch1 and impaired directed migration of maturating GCPs. As a consequence, undifferentiated granule cell precursors exit cell cycle and are retained at the EGL (Figure 7B). In case of an additional heterozygous Ptch1 mutation, upregulation of this receptor does not suffice to exert the anti-proliferative stimulus following Gap43 decrease, which results in an increased fraction of continuously dividing cells in the EGL (Figure 7C). As reduced migration towards the IGL further leads to a withdrawal of growth-limiting signals, expansion of the GCP population is additionally supported. Finally, this advances medulloblastoma development in Ptch1+/− Nos2−/− mice compared to Ptch1+/− Nos2+/+ mice. The mechanism described here illustrates a new tumor-promoting concept in MB showing that the localization of pre-neoplastic cells within the developing cerebellum is important for pathogenesis. Ptch1+/− mice (B6; 129P2-Ptch1tm1Mps/Ptch+; [20]) and Nos2−/− mice (B6; 129P2-Nos2tm1Lau; [72]) were obtained from the Jackson Laboratory (Bar Harbor, Maine, USA) and crossbred to generate double heterozygous mice (Ptch1+/− Nos2+/−). The F1 hybrids were backcrossed with Ptch1+/+ Nos2−/− mice to generate Ptch1+/− Nos2−/− mice. Later on, Ptch1+/− Nos2−/− mice were directly mated. For details on housing and genotyping see Text S2 and Table S9. All animal experiments were approved by the responsible federal authorities (Landesamt für Natur, Umwelt und Verbraucherschutz Nordrhein-Westfalen, Recklinghausen, Germany, Az. 50. 05-230-17/06). Total RNA of tumor specimens and normal cerebellar tissue samples was isolated via CsCl density gradient centrifugation [73] and assessed for integrity using an Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, USA). For gene expression microarrays, linear amplification of mRNA and labeling of cDNA were conducted on samples and Mouse Universal Reference RNA (Stratagene, La Jolla, USA) according to the TAcKLE protocol [74]. Both were combined for two-color hybridizations with each sample being performed as two replicates of inverse dye orientation. Global gene expression profiling was performed on self-printed oligonucleotide microarrays. Further details of microarray production and hybridization are described in Text S2. Genomic DNA of tumor specimens was isolated from the interphase of the CsCl gradient by ethanol precipitation, proteinase K digest, and phenol/chloroform extraction. DNA samples were monitored for purity and adequate fragment size using spectrophotometric measurements and gel electrophoresis. Array-based comparative genomic hybridization (array-CGH, matrix-CGH, [75]) was performed on Mouse Genome CGH 244 k Microarrays (Agilent). Cy5-labeled tumor DNA was combined with corresponding Cy3-labeled reference (wild-type genomic DNA) to receive either sex-matched sample pairs or pairs of different gender for internal negative or positive control. Sample preparation, microarray hybridization, and washing procedures were carried out as described in the manufacturer' s protocol. Microarray data are available in GEO (http: //www-ncbi. nlm. nih. gov/geo), under accession number GSE29201. Total RNA isolated from cell culture samples using the RNeasy Mini Kit (Qiagen, Hilden, Germany) or RNA from tissue specimens was subjected to oligo (dT) -primed reverse transcription. QRT-PCR measurements were conducted in an ABI PRISM 7900HT thermal cycler (Applied Biosystems, Foster City, USA) using the SYBR green reaction and detection system (ABgene, Epsom, UK). For relative quantification mean ratios were calculated between genes of interest and a set of five housekeeping genes (Table S7) according to the Pfaffl method [76]. The expression levels of Ptch1, Gli1, N-myc, and Nos2 were determined by real-time reverse transcription PCR analysis using the ABI PRISM 5700 system (Applied Biosystems) as reported before [73]. For these experiments, the mRNA expression level of mitochondrial ribosomal protein L32 (Mrpl32) served as housekeeping reference. All primer sequences are depicted in Table S8. Inhibition of NO synthases was performed in c17. 2 and D458 cells which were seeded at densities of 2×105 and 4×105 cells per well in 12-well plates, respectively. Cells were daily treated with either 1 mM of the inhibitor L-NAME or 1× PBS as solvent control. For knockdown experiments of Gap43, c17. 2 cells were grown in a 12-well plate to 80% confluency and transfected with 2 µg of pLKO. 1-puro vector that contained either shRNA constructs targeting Gap43, shRNA against GFP, or non-target shRNA as a control (Sigma-Aldrich, St. Louis, USA) using 9 µl FuGene HD reagent (Roche, Basel, Switzerland). Transfection was repeated 2 times each after 8 hours and subjected to selection conditions (1 µg/ml puromycin) for 24 hours. Subsequently, cells were trypsinized, adjusted to 4×105 cells/ml and seeded into the inserts of a Costar Polycarbonate Membrane Transwell plate (8 µm pores, Corning, USA). After 24 hours cells were either harvested for gene expression and protein analyses or 0. 1 µg/µl recombinant SDF-1α was applied to the lower compartment for migration assays. Following 12 hours of incubation, cells at the bottom of the insert membrane were methanol-fixed, hematoxylin-stained, and counted. FFPE sections of postnatal cerebella were pre-processed as described in Text S2. For immunofluorescence co-staining, Gap43 (Sigma-Aldrich, clone GAP-7b10) or NeuN (Millipore, clone A60) first primary antibodies were diluted 1∶1000 or 1∶200, respectively and applied using the Dako REAL Detection System (Dako, Glostrup, Denmark). Following over night incubation at 4°C, washing in TBS, and blocking of residual biotin/streptavidin, sections were subsequently incubated with biotinylated anti-mouse secondary antibody (Dako) and stained with 20 ng/µl FITC-conjugated streptavidin (Invitrogen, Carlsbad, USA). The second primary antibody against Ki-67 (Novocastra, Wetzlar, Germany) was diluted 1∶1000 and accordingly applied using biotinylated anti-rabbit secondary antibody (Dako) and 20 ng/µl Cy5-conjugated streptavidin (Invitrogen). Co-stained sections were then covered with DAPI-containing VECTASHIELD Mounting Medium (Vector, Burlingame, USA) and subjected to confocal laser scanning microscopy. Quantification of Gap43 staining was performed for areas of interest using Image J software (NIH). Numbers of dividing and non-dividing cells in the EGL of postnatal cerebellar tissue sections were counted manually and normalized to the corresponding length of the EGL edge. Cell counts for each region were averaged across three sections, each with 10–20 µm distance in z-axis, and per individual. Kaplan-Meier survival plots were calculated for a total of 1167 mice, including 315 wild-type mice, 412 Ptch1+/+ Nos2−/− mice, 215 Ptch1+/− Nos2+/+ mice and 221 Ptch1+/− Nos2−/− mice. MB-free survival was plotted using the GraphPad Prism 5 software (GraphPad, La Jolla, USA). The logrank test was applied to compare survival (tumor occurrence) of the different genotypes. For comparisons of differences of means between two groups of replicates, p-value calculations were performed using an unpaired, two-tailed t-test, unless indicated otherwise. Calculated error bars represent the standard error of the mean (SEM). For details on microarray statistics please refer to Text S2.

Medulloblastoma is a common pediatric brain tumor, a subtype of which is driven by aberrant hedgehog pathway activation in cerebellar granule cell precursors. Although this tumor etiology has been intensively investigated in the well-established Ptch1+/− mouse model, knowledge is still lacking about the molecular interactions between neoplastic transformation and other developmental processes. Nitric oxide (NO) has been reported to be involved in controlling proliferation and differentiation of these cells. Therefore, inactivation of the NO–producing enzyme Nos2 in combination with the mutated Ptch1 gene should provide insight into how developmental regulation influences pathogenesis. Here, we describe a new role for NO in developing neuronal precursors of the cerebellum facilitating physiologically accurate migration via regulation of Gap43. We further demonstrate that disturbance of these processes leads to retention of granule precursor cells to the cerebellar periphery. Together with the sustained proliferation of these cells in combined Ptch1+/− Nos2−/− mice, this effect results in an increased medulloblastoma incidence relative to Ptch1+/− mice and demonstrates a new disease-promoting mechanism in this tumor entity.

Abstract Introduction Results Discussion Materials and Methods

animal models medicine oncology developmental biology model organisms genetics biology genomics morphogenesis genetics and genomics

2012

Nos2 Inactivation Promotes the Development of Medulloblastoma in Ptch1+/− Mice by Deregulation of Gap43–Dependent Granule Cell Precursor Migration

The viruses of the family Flaviviridae possess a positive-strand RNA genome and express a single polyprotein which is processed into functional proteins. Initially, the nonstructural (NS) proteins, which are not part of the virions, form complexes capable of genome replication. Later on, the NS proteins also play a critical role in virion formation. The molecular basis to understand how the same proteins form different complexes required in both processes is so far unknown. For pestiviruses, uncleaved NS2-3 is essential for virion morphogenesis while NS3 is required for RNA replication but is not functional in viral assembly. Recently, we identified two gain of function mutations, located in the C-terminal region of NS2 and in the serine protease domain of NS3 (NS3 residue 132), which allow NS2 and NS3 to substitute for uncleaved NS2-3 in particle assembly. We report here the crystal structure of pestivirus NS3-4A showing that the NS3 residue 132 maps to a surface patch interacting with the C-terminal region of NS4A (NS4A-kink region) suggesting a critical role of this contact in virion morphogenesis. We show that destabilization of this interaction, either by alanine exchanges at this NS3/4A-kink interface, led to a gain of function of the NS3/4A complex in particle formation. In contrast, RNA replication and thus replicase assembly requires a stable association between NS3 and the NS4A-kink region. Thus, we propose that two variants of NS3/4A complexes exist in pestivirus infected cells each representing a basic building block required for either RNA replication or virion morphogenesis. This could be further corroborated by trans-complementation studies with a replication-defective NS3/4A double mutant that was still functional in viral assembly. Our observations illustrate the presence of alternative overlapping surfaces providing different contacts between the same proteins, allowing the switch from RNA replication to virion formation. The Flaviviridae family comprises positive-strand RNA viruses and consists of four genera, Flavivirus, Pegivirus, Hepacivirus, and Pestivirus, with the latter three genera showing a significantly higher degree of similarity [1–3]. Pestiviruses, like bovine viral diarrhea virus (BVDV-1 and -2) and classical swine fever virus (CSFV) are important animal pathogens which cause significant economic damage in livestock industries [3]. The RNA genomes of the Flaviviridae encompass one single open reading frame (ORF), which is flanked by the 5’ and 3’ untranslated regions (UTR) [4]. Upon infection of the host cell the viral RNA genome is translated into a polyprotein that is processed by cellular and viral proteases into the mature structural (SP) and nonstructural (NS) proteins. For members of the genus Pestivirus the array in the polyprotein is the following: NH2-Npro (N-terminal autoprotease), C (capsid protein, core), Erns (envelope protein RNase secreted), E1, E2, p7, NS2-3 (NS2 and NS3), NS4A, NS4B, NS5A, NS5B-COOH [4]. The N-terminal autoprotease Npro generates its own C terminus and thereby the N terminus of the capsid protein core (C). Further cleavages releasing the structural proteins C, Erns, E1 and E2 as well as p7 are mediated by proteases residing in the endoplasmatic reticulum (ER) [4,5]. The cleavage between NS2 and NS3 is catalyzed by an autoprotease in NS2 [6]. The activity of the NS2 protease is temporally regulated by a cellular cofactor leading to significant amounts of uncleaved NS2-3 in pestivirus infected cells (see below). The cleavages downstream of NS3, NS4A, NS4B and NS5A are catalyzed by the serine protease domain of NS3 which requires NS4A as cofactor for full proteolytic activity and is termed NS3-4A protease [7–10]. Co-immunoprecipitation experiments have shown that NS4A forms a stable complex with NS3 (NS3/4A complex). The N-terminal part of NS4A is highly hydrophobic and in analogy to HCV NS4A most likely comprises a transmembrane domain. The most C-terminal region is not required for the protease cofactor function of NS4A [7]. For HCV NS4A it was observed that the central part of NS4A stably intercalates into the N-terminal beta-barrel domain of the NS3 serine protease domain, a process required to gain full NS3 protease activity [11,12]. The downstream part of HCV NS4A is subdivided into the kink region followed by the acidic domain [13,14]. This nomenclature will be also used for the pestiviral NS4A. In addition to its protease function, NS3 of HCV and pestiviruses comprises helicase and NTPase activity [15–17]. Only when NS2 is cleaved off, NS3 is capable of forming the functional viral RNA-replicase together with the viral proteins NS4A, NS4B, NS5A, NS5B and an unknown number of cellular factors [4,6, 18–20]. For the members of the Flaviviridae the processes involved in virion morphogenesis are not completely understood, it is however known that an interplay between the structural and nonstructural proteins is crucial [21]. For HCV all NS proteins are involved in this process with NS2 being of special importance. It interacts with both structural and nonstructural proteins and thus is proposed to represent a platform for virion assembly [22–27]. For pestiviruses, all NS proteins have been shown to be involved in virion formation, with the exception of Npro and NS4B [19,28–31]. Since NS4B is critically involved in virion morphogenesis of HCV [32,33], it seems likely that its pestiviral counterpart is also required for this process. A special feature of pestiviruses is the existence of significant amounts of uncleaved NS2-3 in the infected cell and its essential role in virion formation. A peculiarity of the pestiviral life cycle is the temporal restriction of NS2-3 processing by the NS2 autoprotease [34,35]. This regulation is based on the fact that the activity of the pestiviral NS2 protease depends on the cellular cofactor Jiv (J-domain protein interacting with viral protein, also termed DNAJC14) [6,29,34,36]. Due to the limited amounts of Jiv in the infected host cell, NS2-3 cleavage is mostly restricted to the early phase of infection. NS2-3 translated at later time points is only inefficiently processed leading to the accumulation of uncleaved NS2-3 which temporally correlates with the onset of virion morphogenesis. A second important aspect connected to the downregulation of NS2-3 processing is its crucial role for the non-cytopathogenic (ncp) biotype of pestiviruses in cell culture. Upregulation of the levels of free NS3, e. g. by the insertion of an ubiquitin (Ubi) moiety between NS2 and NS3, results in a cp biotype and interferes with the capability of BVDV to establish persistent infections in cattle [3,37–40]. With regard to virion morphogenesis, previous studies on BVDV and CSFV have shown that the introduction of an internal ribosomal entry site (IRES) or an ubiqutin coding sequence between the sequences coding for NS2 and NS3 result in a complete loss of infectious particle formation [28,29]. Based on these and other studies [6] it became clear that NS3/4A or uncleaved NS2-3 in complex with NS4A (NS2-3/4A) represent essential components of complexes which facilitate either RNA replication or virion formation: While NS3/4A is an essential part of the RNA replication complex, only NS2-3/4A can facilitate virus assembly. The essential role of NS2-3/4A in the pestiviral life cycle is remarkable when compared to the closely related HCV. In cell culture, the cleavage of HCV NS2-NS3 is highly efficient and previous studies demonstrated that HCV genomes that encode an IRES insertion between NS2 and NS3 were not severely affected with regard to infectious particle formation [26,41]. This finding inspired earlier work by Lattwein, Klemens and coworkers [42,43] who succeeded in adapting a pestivirus to virion morphogenesis in absence of uncleaved NS2-3, similar to the situation described for HCV. Surprisingly, this adaptation of a variant of BVDV-1 strain NCP7 that encodes an ubiquitin monomer between NS2 and NS3 (NS2-Ubi-NS3) required just two amino acid exchanges to allow for efficient particle formation in the absence of uncleaved NS2-3 [43]. One amino acid exchange was located in the C-terminal part of the NS2 protease (2/E440V –originally termed E1576V due to its position in the polyprotein), whereas the other exchange was found in the NS3 protease domain (3/V132A –originally termed as V1721A) [43]. In the present study, we addressed the molecular basis of the gain of function observed for NS2 and the NS3/4A complex in virion morphogenesis. In the absence of structural data for NS2 we focused on the mutation of residue 132 of the NS3 protease domain. We report here the crystal structure of a single chain construct encompassing full-length pestivirus NS3 in complex with NS4A residues 21–57. The data revealed that the gain of function mutation in NS3 maps to a hydrophobic surface patch which interacts with the NS4A-kink region. Our structure-guided studies suggest that NS3/4A can adopt two different conformations in the infected cell–a closed form that is used in RNA replication complexes and a more open conformation functional in viral assembly. Furthermore, our data indicate that the NS2-3/4A complex, required for virion assembly of prototype pestiviruses, displays a similar open conformation. Taken together, our study revealed a novel mechanism by which protein complexes adopt alternative conformations to serve in fundamentally different processes such as RNA replication and virion morphogenesis. We believe that our data will stimulate future analyses of other positive-strand RNA viruses, since it seems likely that similar mechanisms will be used to endow viral protein complexes with the capacity to function in genome replication and virion morphogenesis, a task common to so many positive-strand RNA viruses. To express soluble CSFV NS4A37NS3 for crystallization, we used recombinant constructs encoding full-length NS3 (residues 1 to 683) followed by the 8 N-terminal residues of NS4A; the NS3 sequence is preceeded by residues 21 to 57 of the protease cofactor NS4A (Fig 1A). These constructs corresponded to wild-type and single or double mutants, i. e. NS4A37NS3 carrying a mutation in the protease active site (S163A) or in addition, a mutation in the helicase active site (K232A). Wild-type and mutant proteins, previously characterized for helicase activity, express very well in E. coli and were purified as described earlier [44]. By an in vivo trans-cleavage assay described in S1 Fig we confirmed that wild-type NS4A37NS3 but not its derivative containing the active site mutation S163A has proteolytic activity (S1 Fig). We determined the structure of the NS4A37NS3 active site mutant S163A as described in Materials and Methods and refined it to 3. 05 Å resolution (PDB accession code: 5LKL). Table 1 lists the relevant crystallographic statistics, and S1 Table shows the disordered segments. Like in the crystals of the isolated NS3 helicase domain [44], the two molecules of NS4A37NS3 present in the asymmetric unit (AU) are related by a local 2-fold axis around the helicase domains in which D1 and D3 make a head-to-tail crystallographic dimer interaction (Fig 1B, left panel). Within a pair, one molecule adopts an elongated conformation (Zmax = 105 Å) similar to what has been observed for different flavivirus NS3 proteins [45–47] and the other, a more compact conformation (Zmax = 69 Å), similar to HCV NS3 [48] depending on the different orientations of the protease domain relative to the helicase domain (Fig 1C). In the AU, both molecules are displaying no visible electron density at the interdomain region of the protein (residues 195 to 204 in the compact conformation and residues 196 to 201 in the elongated conformation). We had, however, an unambiguous choice for the protease/helicase pairs after measuring the distances between the domains. Also, SDS-polyacrylamide gel electrophoresis analysis confirmed that the protein recovered from our crystals remained intact (S2 Fig). The overall shape of the pestiviral NS4A37NS3 is very similar to its counterparts from the Flavivirus and Hepacivirus genera with two separate globular domains, representing the protease and helicase, linked by a short flexible interdomain region that explains the different relative orientations of the two domains in our structures (Fig 1 and S3 Fig). In the context of full-length NS4A37NS3, the NS3hel domain shows no structural differences and the same flexibility of the helicase domain 2 with respect to the isolated domain [44]. The CSFV NS3pro domain, which comprises 192 residues, shares only 3% of sequence identity among the NS3 proteases within the Flaviviridae family (S4A Fig). It has the canonical trypsin-like fold containing two β-barrels, with the conserved Asp-His-Ser catalytic triad located in a cleft between them (S4B Fig). Despite the low percentage of sequence identity, pestiviral NS3pro superposed well to HCV NS3pro with a Root Square Mean Deviation (RSMD) of 1. 39 Å and to Dengue virus 4 (DENV4) NS3pro with an RMSD of 2. 37 Å, for 129 residues of 178 and 158 Cα positions, respectively, highlighting their close structural similarity (S4B Fig). Like in hepaciviral- and flaviviral proteases, the pestiviral NS3 protease domain seems also to be stabilized by the insertion of a β-sheet from NS4A (residues 21 to 40) (Fig 2A). Most interestingly, the C-terminal region of the pestivirus NS4A, namely residues 41 to 49, hereafter referred to as" kink region" , is interacting with a hydrophobic patch on the surface of NS3pro. For HCV, structural data of the equivalent region of NS4A in complex with NS3 are so far not available. Recently, a critical role in pestivirus morphogenesis was assigned to amino acid exchange 3/V132A at position 132 in the NS3 protease domain of BVDV [43]. Interestingly, residue 132, a methionine in the closely related CSFV (valine in BVDV) and most other pestivirus strains [43] is located on the NS3 protease surface and makes hydrophobic contacts mainly with L45 of the NS4A-kink region (Fig 2B). We thus hypothesized that the gain of function mutation identified in BVDV NS3 favors a conformational state required for virion morphogenesis in which the interaction between NS3 and the NS4A-kink region is destabilized. To challenge this hypothesis, we developed a protease protection assay to determine if mutations at the NS3/NS4A-kink interface indeed disturb this interaction. In order to address the hypothesis that mutation 3/V132A destabilizes the NS3/4A-kink interaction, we established a TEV-protease (TEVpro) protection assay illustrated in Fig 3A and 3B. We assumed that in the wild-type context a tight association between the NS3 protease surface and the NS4A-kink region would protect the C-terminal NS4A region from a proteolytic attack while destabilizing alanine mutations at the NS3/4A-kink interface should result in higher accessibility to proteolytic cleavage (Fig 3A). The basic expression construct of the developed assay, pCITE Ubi-NS3-4A (1–49) –TEV-GST (Fig 3B), encodes for one ubiquitin monomer, full-length NS3, NS4A (aa 1–49) including the putative end of the NS4-kink region, followed by a TEV cleavage site and GST (Fig 3A and 3B). The ubiquitin moiety mediates the generation of the authentic N terminus of NS3 by cellular ubiquitin hydrolases. Processing by the NS3 serine protease in cis is releasing NS4A (1–49) -TEV-GST (Fig 3B). Cleavage at the TEV cleavage site, which liberates GST, is achieved by co-expression of a HA-tagged TEVpro-encoded by pGEM-T HA-TEVpro. In addition to WT pCITE Ubi-NS3-4A (1–49) –TEV-GST derivatives encoding alanine mutations at the NS3/4A-kink interface were generated. Besides the gain of function mutation 3/V132A in the NS3 protease domain, derivatives with mutations in the NS4A-kink region (4A/L45A, 4A/Y47A, and 4A/L45-Y47-AA) were established. Vaccinia virus MVA-T7pol-mediated expression of the described constructs in Huh7-T7 cells resulted in the expected processing pattern for all constructs, i. e. authentic NS3 and NS4A (1–49) -TEV-GST demonstrating the functionality of the NS3-4A protease derivatives (Fig 3C, left lanes). Moreover, free GST was not detected providing evidence that GST is not liberated from NS4A (1–49) -TEV-GST in absence of HA-TEVpro. Upon co-transfection of pCITE Ubi-NS3-4A (1–49) –TEV-GST and pGEM-T HA-TEVpro, free GST could be detected. Importantly, NS4A (1–49) -TEV-GST cleavage did occur to an individual degree (Fig 3C, right lanes). While the lowest level of TEVpro-mediated cleavage was detected for wild-type, single mutations introduced either into NS3 (3/V132A) or into the NS4A-kink region (4A/L45A, and 4A/47A) resulted in an increase of NS4A (1–49) -TEV-GST cleavage which is indicative of a more accessible TEV cleavage site within the mutant NS3/4A complexes supporting our initial hypothesis. Even more efficient cleavage was detected for the derivative encoding the double mutation in the NS4A-kink region (4A/L45-Y47-AA) (Fig 3C). Quantification of the cleavage efficacies revealed for the wild-type construct a cleavage rate of about 8%, while for the single mutants (3/V132A, 4A/L45A and 4A/Y47A) cleavage rates between 15–32% were observed. The highest cleavage rate of about 44% was determined for the NS4A double mutant 4A/L45-Y47-AA (Fig 3D). These results show that the gain of function mutation 3/V132A as well as the mutations in the NS4A-kink region expose the TEV cleavage site suggesting a destabilization of the NS3/4A-kink interaction. An interesting and attractive hypothesis that arises from these results is that the protein-protein interaction at the NS3/4A-kink region interface might dictate the formation of distinct conformational states (more open or more closed) of the NS3/4A complex with the potential to serve different functions in the viral life cycle. Since NS3-4A is the viral main protease, modulation of the interaction at the NS3/4A-kink interface might cause defects in polyprotein processing which finally could result in replication defects. Accordingly, we tested the mutations at the NS3/4A-kink interface first for effects on polyprotein processing and in a second set of experiments for their impact on RNA replication (S5 Fig and Fig 4). Since some mutants might not be replication competent, the replication-independent Vaccinia virus T7 RNA polymerase expression system was used to drive BVDV polyprotein expression in Huh7-T7 cells transfected with WT or mutant BVDV pT7-DI-388 cDNA clones (S5A Fig, [19]). Our analyses revealed authentic polyprotein processing for all NS3-4A mutants (3/V132A, 4A/L45A, 4A/Y47A, and 4A/L45-Y47-AA), as there were no major differences in the accumulation of NS3, NS4A, NS5A or NS5B detectable when compared to wild-type (S5B and S5C Fig). In addition, no differences regarding protein precursors NS3-4A and NS4A-4B were recognized. Since the NS4A mutations interfered with anti-NS4A detection by Western blot, co-radioimmunoprecipitation (Co-RIP) of the respective NS3/4A complex was performed (S5C Fig). No differences regarding polyprotein processing of the individual NS3-4A variants were detected (S5C Fig). In order to test the effect of the individual mutations on RNA replication efficiency the bicistronic replicon Bici RLuc IRES-Ubi-NS3-3’ was applied, which encodes Renilla luciferase in the 5´ open reading frame (ORF) and ubiquitin (Ubi) followed by NS3-5B in the 3´ ORF (Fig 4A). Upon electroporation of the in vitro transcribed wild-type replicon RNA and its derivatives, luciferase activity was determined at 2 h, 24 h, 48 h and 72 h post electroporation (pe) (Fig 4B). The luciferase values obtained at 2 h pe represent input RNA translation and demonstrated equal electroporation efficiencies for all transcripts. For the wild-type replicon 2 x 106 RLUs were measured at 24 h pe, followed by a decline at 48 h and 72 h due to a cytopathic effect [19,43]. The replicon with the 3/V132A mutation showed a slightly reduced RNA replication profile compared to wild-type. In contrast, derivatives encoding single mutations in the NS4A-kink region (4A/L45A and 4A/Y47A) exhibited delays in RNA replication. While the NS4A mutant 4A/Y47A displayed a delay of 24 h compared to wild-type (Fig 4B), mutant 4A/L45A was even more affected: its RNA replication efficacy was 100-fold lower than wild-type 24 h pe with a maximum of 5 x 105 RLUs at 48 h pe (Fig 4B). Interestingly, the introduction of the double mutation 4A/L45-Y47-AA in the NS4A-kink region completely abolished RNA replication (Fig 4B). The observed differences in the RNA replication pattern of the mutants in comparison to the wild-type were further confirmed by statistical analyses (S6 Fig). Taken together, the single or a double mutations introduced at the NS3/4A-kink interface have no obvious negative effects on polyprotein processing. However, those mutations influence RNA replication to different degrees: mutants that affect the NS3/4A-kink interaction in the TEVpro protection assay only moderately (3/V132A, 4A/L45A and 4A/Y47A, Fig 3D) still allow for RNA replication (Fig 4B). The double mutant that was most affected in the NS3/4A-kink interaction was no longer capable of RNA replication and thus not suited for further experiments in the replicative context. A moderate modulation of the NS3/4A-kink interaction could be achieved by either the previously described gain of function mutation in NS3 (3/V132A) or by single amino acid exchanges in the NS4A-kink region (4A/L45A and 4A/Y47A) (Fig 3). These mutations still allowed for RNA replication which enabled us to test whether the NS4A mutations 4A/L45A and 4A/Y47A could functionally substitute the NS3 mutation V132A in NS2-3-independent virion morphogenesis in the context of variants of BVDV strain NCP7 [43]. To this end, NCP7 NS2-Ubi-NS3 (2/E440V + 3/V132A), encoding both previously described gain of function mutations, was compared with NCP7 NS2-Ubi-NS3 (2/E440V + 4A/L45A) and NCP7 NS2-Ubi-NS3 (2/E440V + 4A/Y47A) (Fig 5). Full-length NCP7 NS2-Ubi-NS3 is unable to produce viral progeny and served as negative control for NS2-3-independent virion morphogenesis. In addition, the NCP7 NS2-Ubi-NS3 variant carrying only the 2/E440V mutation was tested to exemplify the importance of the mutation 3/V132A to allow for efficient NS2-3-independent particle formation, as described previously [43]. RNA transcripts of the non-cytopathic BVDV NCP7 wild-type (NCP7) and the replication-defective mutant (NCP7 5B/GAA) were electroporated into MDBK cells and served as positive or negative control, respectively; RNA replication was monitored indirectly via detection of NS3 at 24 h pe by an immunofluorescence assay (Fig 5B, Top panel, RNA replication). MDBK cells electroporated with RNA transcripts of pNCP7 NS2-Ubi-NS3 and its derivatives containing mutations (2/E440V + 3/V132A), (2/E440V + 4A/L45A) or (2/E440V + 4A/Y47A) were positive for NS3 at 24 h pe (Fig 5B, Top panel) demonstrating that the combinations of mutations 2/E440V with either 4A/L45A or with 4A/Y47A allow for RNA replication in the context of a full-length virus. To investigate the formation of infectious progeny, cell culture supernatants of the electroporated cells harvested at 48 h pe were used to inoculate naïve MDBK cells which were analyzed 72 h later for the presence of NS3 to monitor infection (Fig 5B, Bottom panel). The replication-defective NS5B mutant 5B/GAA and BVDV NCP7 wild-type served as negative and positive controls, respectively. As expected, supernatants of cells electroporated with NCP7 NS2-Ubi-NS3 RNA did not contain infectious virus. The NCP7 NS2-Ubi-NS3 variant encoding only for the mutation 2/E440V in NS2 was able to produce infectious virus to a very limited amount, as expected from previous studies [43]. When supernatants from cells electroporated with RNAs NCP7 NS2-Ubi-NS3 (2/E440 + 3/V132A) were used to inoculate naïve MDBK cells a high number of NS3-positive cells and a severe cytopathic effect was observed, corroborating the critical role of these mutations for virion morphogenesis [43]. Next we analyzed virion production of virus variants in which mutation 3/V132A was replaced by a single mutation in NS4A. Most importantly, the transfer of supernatants from cells electroporated with RNAs NCP7 NS2-Ubi-NS3 (2/E440V + 4A/L45A) or NCP7 NS2-Ubi-NS3 (2/E440V + 4A/Y47A) to naïve MDBK cells resulted in NS3-positive cells and a cytopathic effect indicating the presence of infectious virus (Fig 5B, Bottom panel–compare NS3 and DAPI stain). The amounts of infectious viruses within supernatants collected at 48 h pe or 72 h after a defined infection of naïve MDBK cells at MOI 0. 1 were determined in TCID50/ml (Fig 5C). As expected, the NCP7 NS2-Ubi-NS3 virus was not able to generate viral progeny. In accordance to previous studies [43], the virus encoding 2/E440V together with 3/V132A showed high viral titers of 3. 4 x 105 (48 h pe) or 6. 7 x 104 after defined infection at MOI 0. 1. As expected, the derivative of NCP7 NS2-Ubi-NS3 encoding only NS2/E440V produced low amounts of infectious viruses (4. 5 x 102–48 h pe) underlining the importance of the mutation 3/V132A for efficient particle formation in absence of uncleaved NS2-3 [43]. Importantly, viruses encoding mutations in the NS4A-kink region (NCP7 NS2-Ubi-NS3 (2/E440V + 4A/L45A); NCP7 NS2-Ubi-NS3 (2/E440V + 4A/Y47A) ) instead of 3/V132A were also able to reach high titers at 48 h pe (4A/L45A: 4. 0 x 105; 4A/Y47A: 1. 1 x 106) or after defined infection at MOI 0. 1 (4A/L45A: 6. 0 x 105; 4A/Y47A: 2. 0 x 106) (Fig 5C). In sum, these results demonstrate a highly efficient virus production for all BVDV NCP7 NS2-Ubi-NS3 derivatives with single mutations at NS3/NS4A-kink interface when combined with the mutation 2/E440V in NS2. The data obtained so far were in line with a regulatory role of the stability of the NS3/4A-kink interaction for the decision whether a NS3/4A complex is functional in RNA replication and/or virion morphogenesis. In this context the double mutation 4A/L45-Y47-AA which is severely destabilizing the NS3/4A-kink interaction (Fig 3) was of special interest since it leads to a NS3/4A complex that is not able to support viral RNA replication (Fig 4B). However, it was still conceivable that this NS3/4A complex is capable of supporting virion morphogenesis. Since it was not possible to investigate the functionality of this replication-deficient mutant in the context of a recombinant pestiviral genome a trans-complementation assay was established analogous to an approach published by Moulin et al. [29] (Fig 6A). To rescue virion formation of the packaging incompetent NCP7 NS2-Ubi-NS3, MVA-T7pol-mediated plasmid-based expression of E2-4A variants was employed (Fig 6A). SK6-cells were chosen since Vaccinia virus-based gene expression as well as DNA transfection is highly inefficient in MDBK cells. Expression of the wild-type E2-4A polyprotein from pCITE E2-4A (NS2-3) was expected to rescue virion formation and served as positive control; expression of an E2-4A derivative with ubiquitin between NS2 and NS3 (NS2-Ubi-NS3) served as negative control (Fig 6A). Western blot analysis verified the generation of the uncleaved NS2-3 from pCITE E2-4A (NS2-3) and processed NS3 but no uncleaved NS2-Ubi-NS3 precursor protein upon expression of pCITE E2-4A (NS2-Ubi-NS3) (Fig 6B). In addition, correct processing of E2 and NS4A was observed for both constructs (Fig 6B). The mutations at the NS3/4A-kink interface (3/V132A, 4A/L45A, 4A/Y47A or 4A/L45-Y47-AA), each combined with mutation 2/E440V, were introduced into pCITE E2-4A (NS2-Ubi-NS3) and tested for their functionality in viral trans-packaging. To this end, packaging incompetent full-length NCP7 NS2-Ubi-NS3 RNA was electroporated into SK6-cells, followed by Vaccinia MVA-T7pol-based protein expression from the individual pCITE E2-4A (NS2-Ubi-NS3) constructs. At 24 h post transfection, cell culture supernatants were filtered and used to inoculate naïve MDBK cells (Fig 6C). The anti-NS3 IF assay revealed infected cells and therefore successful rescue of virion formation by the positive control E2-4A (NS2-3) but not for E2-4A (NS2-Ubi-NS3) (Fig 6C). Additional trans-complementation experiments with the derivative E2-4A (NS2-Ubi-NS3) encoding only 2/E440V resulted in no viral rescue. In order to analyze for RNA-recombination events that may result in the formation of infectious viruses, we analyzed for infected cells (NS3-positive cells) also at 96 h after inoculation (S7 Fig). Neither single NS3-positive cells nor infected foci were detected (S7B Fig). While the absence of individual NS3-positive cells is most likely due to the cytopathic effect caused by the replication of the NCP7 NS2-Ubi-NS3 RNA the absence of antigen-positive foci excluded the presence of infectious virus generated by RNA recombination. Importantly, all tested E2-4A (NS2-Ubi-NS3) derivatives with mutation 2/E440V in combination with single alanine exchanges either in the NS3 protease domain (3/V132A) or the NS4A-kink region (4A/L45A and 4A/Y47A) were able to rescue virion formation (Fig 6C). Strikingly, also expression of E2-4A (NS2-Ubi-NS3) 2/E440V with the double mutation (4A/L45-Y47-AA) in NS4A allowed virion morphogenesis. Quantitative analysis of supernatants revealed a mean virus titer of 7. 1 x 103 TCID50/ml when E2-4A (NS2-3) was supplied in trans. Importantly, all constructs which encode E2-4A (NS2-Ubi-NS3) variants combining 2/E440V with 3/V132A (1. 3 x 103 TCID50/ml), 4A/L45A (1. 4 x 103 TCID50/ml), 4A/Y47A (5. 9 x 102 TCID50/ml) or 4A/L45-Y47-AA (1. 7 x 103 TCID50/ml) were able to rescue virion morphogenesis of NCP7 NS2-Ubi-NS3 (Fig 6C and 6D). The results of the trans-complementation lead to the intriguing conclusion that the NS4A double mutant 4A/L45-Y47-AA, while interfering with viral RNA replication, is still functional in virion morphogenesis. A destabilization of the NS3/4A-kink interaction appears to be a prerequisite to enable the NS3/4A complex to be active in virion morphogenesis. However, in wild-type pestiviruses NS2-3/4A but not NS3/4A is employed in packaging. Accordingly, the mutations at the NS3/4A-kink interface represent functional surrogates for the presence of NS2 in the NS2-3/4A complex. In consequence, we hypothesized that in the NS2-3/4A complex the NS2 protein moiety should have an effect similar to the one observed for the mutations at the NS3/NS4A interaction surface, namely the destabilization of the NS3/4A-kink interaction. To test our hypothesis we applied the TEVpro protection assay by which we compared pCITE Ubi-NS3-4A (1–49) –TEV-GST with pCITE p7-NS2-3-4A (1–49) –TEV-GST (Fig 7A and 7B). The p7 gene was included in pCITE p7-NS2-3-4A (1–49) –TEV-GST since p7 contains a leader sequence to properly insert NS2 into the ER membrane and to allow the generation of its authentic N terminus by cellular signal peptidase [49]. As described in Fig 3 we determined the levels of TEV cleavage by Western blot and subsequent quantification (Fig 7C and 7D). When analyzed in parallel, a cleavage rate of 10% was observed for pCITE Ubi-NS3-4A (1–49) –TEV-GST while 19% were obtained for pCITE p7-NS2-3-4A (1–49) –TEV-GST (Fig 7D). Thus, the TEVpro protection assay indicates that the NS2-3/4A complex adopts a conformation more accessible to TEVpro cleavage when compared to the NS3/4A complex. Taken together, these results support the hypothesis that the modulation of the NS3/4A-kink interaction either by mutation at the NS3/4A-kink interface or by the presence of NS2 in the NS2-3/4A complex lead to a less compact conformation required for virion morphogenesis. Due to their compact genome size viruses encode a severely limited number of gene products which therefore often exert multiple functions. Along these lines, the nonstructural proteins of the members of the Flaviviridae family have key functions during RNA replication and in the assembly of infectious particles [21]. In addition, like most positive-strand RNA viruses, pestiviruses use regulated polyprotein processing by viral and cellular proteases as means to temporal and spatial regulate protein activity [3]. An excellent example is the observation that in the pestiviral life cycle, RNA replication and virion morphogenesis are highly regulated processes both depending on different protein complexes which derive from the NS2-3-4A region of the polyprotein: while the NS3/4A complex is essential for RNA replication and cannot be functionally replaced by NS2-3/4A [6,18], NS2-3/4A is indispensable for pestiviral particle formation but is not active in RNA replication [28,29]. NS2-3 cleavage by the NS2 autoprotease is temporally restricted to the early time points of infection due to its dependency on a cellular cofactor available only in limiting amounts [6,37]. This leads to a temporal gradient for the formation of NS3/4A vs. NS2-3/4A in pestivirus infected cells. In contrast to pestiviruses, the closely related HCV does not depend on uncleaved NS2-NS3 for infectious virion formation at least in cell culture [26,41] and no major amounts of uncleaved NS2-NS3 can be detected in HCV-infected cells. Thus, HCV and pestiviruses show a major difference with regard to their dependency on uncleaved NS2-3 for the formation of infectious progeny. Interestingly, pestiviruses can be adapted by two amino acid exchanges, one in NS2 and the other one in NS3, to virion morphogenesis without the requirement for uncleaved NS2-3 and thus can carry out virion assembly with strong parallels to HCV [42,43]. The NS3 mutation 3/V132A resulting in a gain of function of the NS3/4A complex in virion morphogenesis was of central importance for this study. In view of the critical role of the NS3/4A protein complex in the switch between RNA replication and virion assembly, the determination of the structure of a single-chain CSFV NS3 serine protease-NS4A complex provided the basis for structure-guided functional analyses. The CSFV full-length NS3 structure in complex with its cofactor NS4A shows that the central region of NS4A forms a β-sheet which intercalates into the N-terminal β-barrel of the NS3 serine protease domain, similar to the results reported for HCV NS3 when co-crystallized with a NS4A cofactor peptide (Fig 1) [11]. However, a question not answered by the available HCV NS3/4A structures was whether its NS4A-kink region undergoes additional interactions with the NS3 surface [11,50]. The structure of the CSFV single-chain NS3/4A complex reported in this study provided insights into this aspect (Fig 2A and 2B). We found that residue M132 maps to the surface of the NS3 protease domain and makes hydrophobic interactions mainly with residues L45 and Y47 of the NS4A-kink region through specific and main chain interactions, respectively (Fig 2B). Our data support the idea that these hydrophobic contacts are critical for a compact conformation of NS3/4A required when serving as a component of the replicase complex. Since an exchange of this residue to Ala (3/V132A) is critical for the gain of function in NS2-3-independent virion formation we speculated that the modulation of this interaction between NS3 and the NS4A-kink region could be the underlying molecular principle. We hypothesized that a destabilization of this interaction (as introduced by e. g. 3/V132A, 4A/L45A or 4A/Y47A) would lead to a more open conformation of the NS3/4A complex required for its function, together with NS2, in NS2-3-independent virion assembly. To follow this hypothesis a method was required to determine the effect of individual mutations on the interaction at the NS3/4A-kink interface. Since the central domain of NS4A stably intercalates into the NS3 protease domain a dissociation of both proteins was not to be expected as a consequence of a weakened NS3/4A-kink interaction and thus could not serve as read out. Therefore, a TEVpro protection assay was established (Fig 3). It was assumed that a more flexible or open conformation of the NS3/4A-kink region will result in a higher accessibility of the TEV cleavage site to TEVpro (Fig 3A and 3B). As proposed, the accessibility at the TEV cleavage site was elevated for all NS3/4A variants with mutations at the NS3/4A-kink interaction surface when compared to wild-type (Fig 3C and 3D). This was in agreement with a more open or more flexible conformation of the mutant NS3/4A protein complexes. Interestingly, the degree of TEVpro cleavage for the individual mutants did inversely correlate with their replication fitness (Fig 4). The previously selected NS3 mutant 3/V132A displayed a moderately increased TEV cleavage rate and RNA replication was slightly reduced when compared to wild-type. Mutant 4A/L45A, which allowed for a more efficient TEVpro cleavage compared to 4A/Y47A showed lower replication fitness than 4A/Y47A. The NS4A double mutant 4A/L45-Y47-AA displayed the highest TEVpro cleavage and was found incapable of RNA replication (compare Figs 3D and 4B). Interestingly, the C-terminal region of HCV NS4A, which is significantly shorter than the one of pestiviruses, was also shown to be critically involved in replicase assembly. In this region of HCV NS4A, two residues (Y45 and F48) were found to inhibit RNA replication in different genotypes when mutated to alanine [13,14]. While structural data about this part of the HCV NS3/4A complex is still missing, these residues could be critically involved in an interaction of the C-terminal NS4A region with NS3, especially since the replication defects of several NS4A mutants could be suppressed by second site mutations on the NS3 surface [13] or in NS4B [14]. Recently, a hydrophobic patch on the surface of the HCV NS3 protease domain was identified as a critical determinant for replicase assembly [51]. Future studies have to reveal whether this hydrophobic area on the HCV NS3 surface is involved in interactions with the C-terminal domain of NS4A. Overall, replication studies within different HCV genotypes indicate that genetic interactions between NS3, NS4A, and NS4B contribute to replicase assembly and genome replication [13,14,33,52,53]. The observation that the replication-deficient BVDV NS4A double mutant retained full NS3-4A protease activity indicates that the respective amino acid exchanges are not interfering with proper folding of the NS3/4A complex (Fig 4 and S5 Fig). This assumption was further corroborated by the fact that the NS4A double mutant was still capable of forming NS3/4A complexes that are active in virion morphogenesis (Fig 6). Moulin et al. had already shown that mature NS4A is required in addition to NS2-3 for virion formation and that NS4A fused to the C-terminus of NS3 is not functional [29]. For HCV a role of the C-terminal part of NS4A in viral assembly has been described [14]. Accordingly, this NS4A region is involved in RNA replication and virion morphogenesis also for HCV and thus might represent a switch between these processes [13]. Taken together, our data indicate that the modulation of the interaction between NS3 and the NS4A-kink region represents a way to control the switch between NS3/4A complexes functional either in RNA replication or virion assembly (Table 2). In this context the role of the gain of function mutation 2/E440V which is also strictly required for virion formation in the absence of uncleaved NS2-3, still remains enigmatic mainly due to the absence of structural information for pestiviral NS2. NS2 residue 440 is located in the C-terminal NS2 protease domain that is residing in the cytoplasm [34]. In previous studies correctly folded NS2 was found critical with respect to NS2-3-independent virion formation, as C-terminal truncations or mutations of the Zn-coordinating residues inhibit particle formation [42]. In the HCV system NS2 also holds a central role in virion assembly by recruiting envelope glycoproteins to the site of assembly and participating in multiple protein-protein interactions with structural and nonstructural proteins that are required for virion morphogenesis [26,27,41,54–56]. Further in depth investigations have to clarify if pestiviral NS2 has a similar role as organizer of the virion assembly pathway. The identification of pestivirus mutants which are capable of NS2-3-independent virion morphogenesis was instrumental for this study. However, in wild-type pestiviruses uncleaved NS2-3 is strictly required for virion morphogenesis. By applying the TEVpro protection assay we observed that NS2 in the context of the NS2-3/4A complex recapitulates the effect of the mutations at the NS3/4A interaction surface, namely a modulation of the NS3/4A-kink interaction (Fig 7). Therefore, we propose that NS2 in the NS2-3/4A complex alters the interaction between the NS4A-kink region and the NS3 surface favoring a packaging-competent conformation. This leads to our current working model (Fig 8) in which a closed and a more open conformation of the NS3/4A-kink region allow the assembly of so far not characterized protein complexes critical either for RNA replication or virion morphogenesis, respectively. This hypothesis is also in line with the fact that the N terminus of NS3 is in close proximity to the NS3/4A-kink interface (Fig 2). In analogy to the pestivirus system, the presence of NS2 at the N terminus of NS3 was also shown to inhibit HCV RNA replication as well as the assembly of a functional RNA-replicase complex indicated by a loss of NS5A hyperphosphorylation [51]. Taken together, these findings support for both virus genera the model that the presence of NS2 in the NS2-3/4A complex leads to a conformation interfering with replicase assembly. However, structural data on uncleaved NS2-3 are absolutely required for further conclusions. The identification of the basic building blocks for RNA replication and virion morphogenesis will facilitate future work aiming at unraveling the respective pathways. An important next step will be the identification of interaction partners of NS3/4A or NS2-3/4A relevant for either replicase assembly or virion morphogenesis. Furthermore, we expect that similar mechanisms are used by other RNA viruses which face the general challenge of how to address a plethora of functions with a very limited set of proteins. To express and purify full-length CSFV NS3 (hereafter named NS4A37NS3) for crystallization, we used the constructs and the expression and purification methods described in [44]. Briefly, the expressed NS4A37NS3 encompasses His-tag, TEV cleavage site (MHHHHHHHENLYFQG), CSFV NS4A amino acids 21 to 57, a GSGS linker, followed by full-length NS3 and the first 8 amino acids of NS4A. Conditions for crystallization of CSFV NS4A37NS3 were found using a Mosquito robot in the format of 96 Greiner plates. Crystals were optimized with a robotized Matrix Maker and Mosquito setups. The best diffracting crystals corresponded to NS4A37NS3 carrying a mutation in the protease catalytic site S163A and were grown in sitting drops at 20°C by vapor diffusion under the condition listed in Table 1. For data collection, the crystals were flash cryo-cooled in liquid nitrogen using 20% glycerol as cryoprotectant. Diffraction data were collected at the synchrotron beam line ID23-1 at the European Synchrotron Radiation Facility (Grenoble, France) indexed and processed using XDS [57] and scaled with SCALA [58]. The structure was determined by molecular replacement with PHASER using as a search models the structure of CSFV isolated helicase domain (NS3hel) [44] and an homology model based in the structure of HCV protease generated with Phyre (http: //www. sbg. bio. ic. ac. uk/phyre2/html/page. cgi? id=index). The model of CSFV full-length NS3 was subsequently manually modified with Coot [59] and refined with BUSTER-TNT [60]. For antigen detection, mouse monoclonal antibody 8. 12. 7 against NS3/NS2-3 [61], anti-E2 (SCR48 6. 6. 11) (kindly provided by H. -J. Thiel, [Justus-Liebig University, Gießen]) [62], anti-NS4A (GH4A1 [4B7]), anti-NS5A (GLBVD5A1 [11C]), anti-NS5B (GLBVD5B1 [9A]) (kindly provided by T. Rümenapf and B. Lamp [University of Veterinary Medicine, Vienna, Austria]) [43], anti-HA (HA. 11 clone 16B12, Covance, New Jersey, USA), anti-GST (New England Biolabs), anti-myc (New England Biolabs) were used. Species specific Cyanogen-3-labeled (Cy3) or peroxidase-coupled (PO) antibodies were obtained from Dianova (Hamburg). Quantitative Western blot analyses were performed using secondary antibody coupled to IRdye-800 obtained from LI-COR Biosciences (Lincoln, Nebraska, USA). Madin-Darby Bovine Kidney (MDBK) cells (American Type Culture Collection—ATCC) were cultivated in minimal essential medium (MEM, Invitrogen) containing 10% horse serum, nonessential amino acids and 1% penicillin/streptomycin (PAA, Pasching, Austria). Swine kidney cells (SK-6) [63] were cultivated in MEM supplemented with 10% fetal bovine serum and 1% penicillin/streptomycin. Huh7-T7 cells [64] were kept in DMEM containing 10% fetal bovine serum gold, 1% penicillin/streptomycin and 125 μg/ml G418 (PAA, Pasching, Austria). All cells were grown at 37°C and 5% CO2. BVDV-1 strain NCP7 was described previously [65,66]. The original Vaccinia virus modified virus Ankara (MVA) -T7pol stock [67] was generously provided by G. Sutter (LMU, Munich, Germany). BVDV-1 genomes NCP7 [66] and NCP7 NS2-Ubi-NS3 [43] have been described. Tables 3 and 4 summarize all constructs which were used in this study. Further details concerning the generation of the constructs and their properties can be found in the supplementary material. The applied procedures have been described [19,43]. Briefly, Huh7-T7 cells, which express T7 RNA-polymerase [64], or SK6 cells were infected with modified vaccinia virus Ankara-T7pol (MVA-T7pol) [67], subsequently transfected with 4–6 μg of plasmid DNA by using 10 μl Metafectene transfection reagent (Biontex). Protein expression was carried out for 18 h and cells were further processed. Cell lysates were separated with SDS-polyacrylamide-Tricine gels (8 or 10% polyacrylamide) [70]. After electrophoresis, proteins were transferred onto a nitrocellulose membrane. Membranes were blocked with 5% skim milk powder (w/v, Roth) in PBS/Tween 20 (0. 05% (vol/vol). Viral proteins were detected with the indicated monoclonal antibodies and visualized with corresponding peroxidase-coupled species-specific secondary antibodies and Western Lightning plus enhanced chemiluminescence reagent (Perkin Elmer, Boston, MA). For the quantification of Western blots, species-specific secondary antibodies coupled to IRdye-800 and an Odyssey SA infrared imaging system were used (LI-COR Biosciences, Lincoln, Nebraska, USA). 2 μg of plasmid DNA were linearized with SmaI (New England Biolabs) and used as template for in vitro transcription with MAXIscript SP6 transcription kit (Ambion, Huntingdon, United Kingdom). The amount of RNA was determined by using Quant-iT RNA assay kit and Qubit-Fluorometer (Invitrogen). RNA quality was verified by agarose gel electrophoresis. 1 μg RNA was used for electroporation of 3 x 106 cells. Electroporation of RNAs in MDBK cells was carried out as described previously [71]. After electroporation, cells were transferred into complete MEM and seeded as required for the assay. Bicistronic reporter constructs, encoding Npro-Renilla luciferase in the 5´ ORF and one monomer of ubiquitin followed by NS3-5B in the 3´ ORF, were used. RNA replication efficiencies were determined as described in [43]. In short, at each time point (2 h, 24 h, 48 h, 72 h) cells were washed with PBS and lysed in 40 μl lysis-juice (PJK-GmbH). 20 μl of cell lysates were mixed with 100 μl of Renilla Glow juice (PJK-GmbH) and measured with a Junior LB 9509 Portable Tube Luminometer (Junior LB9509, Berthold). Cell culture supernatants of MDBK-cells were harvested at indicated time points post electroporation (pe) or post infection (pi) and filtered through a 0. 2-μm cellulose filter (Sartorius). Infection of MDBK-cells was performed as described previously for 1 h at 37°C [34]. Tissue culture infection dose 50 (TCID50/ml) of viral supernatants was determined in three replicates by endpoint titration on MDBK-cells. Viral infection was detected 72 h pi by IF using Mab 8. 12. 7 directed against NS3/NS2-3. Cy3-labeled anti-mouse secondary antibody was used at 1: 1. 000. Titration of viral supernatants which were generated during trans-complementation assays were detected by IF at 24 h pi. BVDV-1 infected or electroporated cells were indirectly analyzed for the presence of BVDV-1 NS3/NS2-3 by indirect immunofluorescence assay (IF) with the monoclonal antibody (MAb) 8. 12. 7 [61]. At indicated time points, cells were washed with 1 x PBS and fixed with 2% paraformaldehyde for 20 min at 4°C. Then, cells were permeabilized by 0. 5% n-octyl-β-d-glycopyranoside for 7 min. Cells were washed with 1 x PBS/Tween 20 (0. 05% (vol/vol) ) and incubated with hybridoma supernatant of anti-NS3 mab at a dilution of 1: 40 in PBS/Tween 20 (0. 05% (vol/vol) ). As secondary antibody, goat anti-mouse IgG conjugated with Cy3 was used at 1: 1. 000. Images were obtained with a Zeiss Axio Observer. Approximately, 5x106 SK6-cells were electroporated with 1 μg of RNA of NCP7 NS2-Ubi-NS3 and seeded into one well of a 6-well plate. 6 h pe cells were infected with MVA-T7pol virus for 1 h at 37°C. Subsequently, cells were transfected with 8 μg of plasmid DNA and incubated for 18 h. Cell culture supernatants were harvested and filtered through a 0. 2-μm cellulose filter (Sartorius). In order to test the supernatants for rescued virus, naïve MDBK-cells were inoculated with the filtered cell culture supernatants and incubated for 24 h. NS3-positive cells were detected by anti-NS3 IF. Viral titers were determined as described above in TCID50/ml. Figs 1B and 2A were prepared using Ribbons and PyMOL (http: //pymol. sourceforge. net) softwares, respectively. Atomic coordinates and structure factors amplitudes of NS4A37NS3 have been deposited at the Protein Data Bank (PDB) under the accession code 5LKL. Initially, a Kruskal-Wallis test was performed to compare all groups for differences. In case p < 0. 05, pairwise differences between WT and all other mutants were tested using Mann-Whitney tests and interpreted as significant if p < 0. 05. For the calculations of the RNA replication assay (S6 Fig), again Kruskal-Wallis tests were performed to test for differences between the groups WT, 3/V132A, 4A/L45A, 4A/Y47A, 4A/L45-Y47-AA and 5B/GAA at each time point separately and regarded to be significant if p < 0. 05. In these cases, pairwise differences between WT and all other mutants were again tested using Mann-Whitney tests. Calculations were performed with GraphPad Prism (La Jolla, CA) and IBM SPSS Statistics for Macintosh, Version 22. 0 (Armonk, NY: IBM Corp).

Many positive-strand RNA viruses replicate without transcribing subgenomic RNAs otherwise often used to temporally coordinate the expression of proteins involved either in genome replication (early) or virion formation (late). Instead, the RNA genomes of the Flaviviridae are translated into a single polyprotein. Their nonstructural proteins (NS), while not present in the virions, are known to be crucially involved in RNA replication and virion formation. The important question how the same proteins form specific complexes required for fundamentally different aspects of the viral replication cycle is not solved yet. For pestiviruses the mature NS3/4A complex is an essential component of the viral RNA-replicase but is incapable of participating in virion morphogenesis which in turn depends on uncleaved NS2-3 in complex with NS4A. However, a gain of function mutation in NS3 enabled the NS3/4A complex to function in virion assembly. Using structure guided mutagenesis in combination with functional studies we identified the interface between NS3 and the C-terminal NS4A region as a module critical for the decision whether a NS3/4A complex serves in RNA replication or as a packaging component. Thus, we propose that subtle changes in local protein interactions represent decisive switches in viral complex formation pathways.

Abstract Introduction Results Discussion Materials and methods

pestivirus medicine and health sciences poxviruses pathology and laboratory medicine enzymes pathogens microbiology enzymology viral structure viruses developmental biology rna viruses dna viruses morphogenesis proteins medical microbiology microbial pathogens vaccinia virus viral replication complex viral replication virions biochemistry flaviviruses virology viral pathogens biology and life sciences proteases organisms

2017

A positive-strand RNA virus uses alternative protein-protein interactions within a viral protease/cofactor complex to switch between RNA replication and virion morphogenesis

Trypanosoma cruzi, the etiological agent of Chagas disease, is a polymorphic species. Evidence suggests that the majority of the T. cruzi populations isolated from afflicted humans, reservoir animals, or vectors are multiclonal. However, the extent and the complexity of multiclonality remain to be established, since aneuploidy cannot be excluded and current conventional cloning methods cannot identify all the representative clones in an infection. To answer this question, we adapted a methodology originally described for analyzing single spermatozoids, to isolate and study single T. cruzi parasites. Accordingly, the cloning apparatus of a Fluorescence-Activated Cell Sorter (FACS) was used to sort single T. cruzi cells directly into 96-wells microplates. Cells were then genotyped using two polymorphic genomic markers and four microsatellite loci. We validated this methodology by testing four T. cruzi populations: one control artificial mixture composed of two monoclonal populations – Silvio X10 cl1 (TcI) and Esmeraldo cl3 (TcII) – and three naturally occurring strains, one isolated from a vector (A316A R7) and two others derived from the first reported human case of Chagas disease. Using this innovative approach, we were able to successfully describe the whole complexity of these natural strains, revealing their multiclonal status. In addition, our results demonstrate that these T. cruzi populations are formed of more clones than originally expected. The method also permitted estimating of the proportion of each subpopulation of the tested strains. The single-cell genotyping approach allowed analysis of intrapopulation diversity at a level of detail not achieved previously, and may thus improve our comprehension of population structure and dynamics of T. cruzi. Finally, this methodology is capable to settle once and for all controversies on the issue of multiclonality. Chagas disease, an American protozoonosis caused by Trypanosoma cruzi, is characterized by various clinical manifestations ranging from asymptomatic to severe cardiac and/or digestive injuries. This complex ailment still afflicts 10 million people across Latin America and another 90 million are at risk of acquiring the disease [1]. A vaccine or specific treatment for large-scale public health interventions is still not available, so the main control strategy relies on prevention of transmission by controlling insect vectors and contamination by food or blood transfusion [2]. T. cruzi is a very polymorphic species, as extensively demonstrated by biological, biochemical, and molecular studies [3], and this certainly contributes to the pleomorphism of the symptoms and to the difficulty in controlling the disease [4]. T. cruzi was recently subdivided into six discrete taxonomic units (DTUs) named T. cruzi I to T. cruzi VI [5], of which at least four are involved with human pathology [6]. Indirect evidence suggests that part of T. cruzi populations is multiclonal. This is in accordance with the theoretical expectation, since patients in endemic areas are infected by multiple contacts with different triatomines and these, in turn, may feed on different infected individuals. Most of the putative multiclonal populations were suggested based on the identification of more than two alleles for different markers [7]–[9]. Whether these reported populations are really multiclonal remains controversial because we cannot exclude the possibility of aneuploidy [7], [8], [10], [11]. Thus, direct evidence for the multiclonality of sylvatic and domestic populations is lacking. Indeed, it is not possible to identify all the clones that constitute a given multiclonal population because current conventional cloning methods (micromanipulation, limiting dilution or cloning in blood-agar plates) may favor individuals over-represented in the original population and/or presenting higher growth rates [12]–[15]. To answer these questions we devised a new strategy for sorting single T. cruzi parasites, adapted from methods originally described for analyzing single spermatozoids [16]. Using this innovative approach the results presented herein may contribute to settling the debate regarding the multiclonality of three T. cruzi representative strains isolated from human and vector hosts. The methodology described here is able to completely dissect the complexity of strains and estimate the relative contribution of each subpopulation, and will therefore have important implications for the study of T. cruzi biology and disease. In this study we used four multiclonal T. cruzi populations: one artificially composed of Esmeraldo cl3 (a standard T. cruzi II strain) and Silvio X10 cl1 (a standard T. cruzi I strain), a naturally occurring putatively multiclonal vector strain called A316A R7, and two populations derived from a naturally occurring putatively multiclonal human strain named Be-78 1B and Be-78 25B. The A316A R7 strain was isolated from a Triatoma sordida circulating in northwestern Paraná state, Brazil [17], [18]. The Be-78 1B and Be-78 25B are two populations derived from a chronic chagasic outbred dog infected with Be-78, originally isolated from Berenice, the first patient of Carlos Chagas [19], harvested after 1 or 25 successive blood passages in mice [20]. About 1. 5 mL (106 cells) of T. cruzi epimastigotes cultured in Liver Infusion Tryptose (LIT) medium were transferred to a siliconed Vacutainer tube (Itupeva, Brazil). The culture was centrifuged at 30×g for 10 minutes at 4°C to separate intact cells from cellular debris. After this centrifugation, the culture was incubated for 10 minutes at room temperature to allow mobile epimastigotes to reach the surface. The collected cells were submitted to three washes with 1× PBS pH 7. 4 at 4°C sterilized using 0. 22 µM Millex GV filter (Millipore, MA, USA) and centrifuged at 1000×g for 10 minutes at 4°C. After the last wash the cells were suspended in 1 mL of 1× PBS pH 7. 4 and fixed for 30 minutes using absolute ethanol maintained at −20°C. Finally, the fixed cells were stored at 4°C overnight, after which they were submitted to sorting on the FACS Vantage apparatus. To calibrate the powerful single-cell sorter FACS Vantage to achieve single T. cruzi specimens, we used fluorescent beads as an indirect control because bead size is similar to the epimastigote form of the parasite [21]. Fluorescent beads (Spherotech Inc. , IL, USA) measuring 10. 0 to 14. 0 µm in diameter were sorted in microscope slides according to the manufacturer' s protocol (Becton Dickinson, CA, USA). We found that the counting of the sorted beads under fluorescence microscope always showed good correspondence with the number of beads programmed for each well. In wells programmed to contain one bead we found an occasional absence of beads, but never more than one bead. To sort T. cruzi strains, small aliquots of the fixed cells were transferred to a Falcon tube (Becton Dickinson, CA, USA) for FACS counting. The solution containing the parasites was diluted in 1× PBS pH 7. 4 until the FACS counting indicated approximately 100 events per second. Samples were gated forward scatter (FSC) versus side scatter (SSC) to select cell populations of interest (Figure S1). The autofluorescence pattern presented by T. cruzi allows cell sorting without staining with fluorescent dyes or reagents. Then, the Clone Cyt apparatus (Becton Dickinson, CA, USA) was used to sort single cells directly into 96-wells plate (Sorensen BioScience, UT, USA) containing 5 µL of 10% Triton X-100. Control wells were programmed to contain 0,2, 5, or 10 sorted parasites from each T. cruzi population. After sorting, a drop of mineral oil was added to each well and the microplates were stored at −20°C until needed. Before the PCR assays, the single sorted T. cruzi was lysed by exposing the microplates to 80°C for 1 hour in a thermocycler. Cells were genotyped by using polymorphic markers to mitochondrial (Cytochrome Oxidase subunit II- COII) and nuclear (24Sα rDNA) genes and four microsatellite loci. To evaluate whether the previously used bead-calibrated method could be successfully used to separate single T. cruzi parasites, we initially sorted an artificial mixture of two monoclonal populations: Silvio X10 cl1 (T. cruzi I) and Esmeraldo cl3 (T. cruzi II). Next, we confirmed separation by genotyping the 24Sα rRNA gene of the single sorted parasites and the controls present in each well (Figure 1). In the control wells A1, A2, A3, and A4 programmed to contain 10 or 5 cells, we detected the expected amplicons of 110 and 125 bp, demonstrating that the two types of cells used in the artificial mixture were present. From the 83 wells programmed to contain the sorted single parasites, 40% (33 wells) showed the amplification of only the 125 bp or the 110 bp amplicon, therefore confirming single cell sorting. This efficiency is approximately the double that obtained with a previously published protocol [25]. Moreover, none of the wells programmed to contain a single cell showed the amplification of both fragments. These results indicate that the FACS Vantage sorting method we modified could be used efficiently to separate single T. cruzi parasites. The A316A R7 strain was isolated from a triatomine bug in Paraná, southern Brazil. Because this strain presented more than two alleles for some of the DNA markers used during its genotyping, it was suggested that it was constituted of a mixed population [17]. However, considerable controversy persisted on whether the multiclonal status of this and other T. cruzi strains is real [7], [8], [26]–[28] or just reflects aneuploidy for the analyzed loci [10], [11]. We therefore used the single T. cruzi sorting and genotyping approach proposed here to address this question. To this end, we sorted single parasites from the A316A R7 strain and genotyped them by using three sets of markers: COII, 24Sα rDNA, and microsatellites (Figures 2,3, 4). To bring further light to the debate regarding the multiclonal status of T. cruzi strains, an innovative approach was successfully adapted from a method originally described to genotype single sorted spermatozoids, to separate single parasites. The evidence described herein reveals that multiclonality holds true for two representative T. cruzi human and vector strains. The approach we developed employed the FACS Vantage sorting method, which allowed us to estimate the relative contribution of each parasite subpopulation within the strains analyzed. The efficiency of the approach was initially established by sorting and genotyping single parasites from an artificially composed multiclonal population and not only all the individuals present in the original mixture could be identified, but also the real proportion of each population could be assigned. The devised method was then used to scrutinize three naturally occurring putative multiclonal populations originally isolated from a vector (A316A R7) and two derived from a patient (Be-78 1B and Be-78 25B). A long-held debate was whether T. cruzi strains presenting more than two alleles for some polymorphic markers were really multiclonal or aneuploid populations [7], [9], [10]. Previous work suggested that the A316A R7 strain was a mixture of distinct populations [18], [29]. Our new results involving the microsatellite analysis of single sorted parasites confirmed the presence of different subpopulations within this strain, ruling out aneuploidy as a possible explanation for the complex genetic patterns presented by this population. Interesting results were observed with the human T. cruzi strain. Great controversy remains on whether or not Berenice, the first patient of Carlos Chagas, has been infected with a multiclonal T. cruzi. This was due to the fact that two different but apparently monoclonal populations (Be-62 and Be-78) were isolated from her [34]. The observation that Be-78 has changed its molecular and biochemical characteristics after passages in dogs and mice strengthened this possibility, suggesting that the passage in different hosts could have grown previously undetectable subpopulations within the original strain. Here we showed that Be-78 1B and Be-78 25B (two of the T. cruzi populations derived from a dog infected with Be-78 after 1 or 25 successive passages in mice) were constituted by three subpopulations: one compatible with Be62, another equivalent to the Be78, and a third one not identified before. Therefore, the complexity of this T. cruzi strain is even greater than originally expected. We also detected alleles not previously reported for the analyzed strains. The probable reason they were not found before is that when alleles of less representative subpopulations are lower than 15% in the DNA artificial mixture, they cannot be detected by conventional PCR (our unpublished data). Another remarkable result was observed regarding the TcAAAT6 profiles of the single sorted parasites. The analysis of the vector strain showed the presence of two alleles of 251 and 275 bp, initially interpreted as a typical heterozygote genotype. However, the analysis of the single parasites using this locus showed the presence of two different subpopulations both presenting homozygote profiles. This finding emphasizes the advantage of genotyping single cells instead of pools of parasites, because conventional PCR cannot distinguish between the presence of a homogeneous population presenting two different alleles, and a heterogeneous population constituted by two different and homozygous individuals. Interestingly, by using the methodology presented here we successfully identified the subpopulations presented in a multiclonal strain as well as estimate the proportion of each subpopulation. A drawback inherent to use of a single cell is failure of amplification of one of the two alleles of some loci, a phenomenon known as allele drop-out (ADO). Due to its characteristics, ADO can be misleading as false homozygosity, as one of the alleles in heterozygote loci is undetectable [35]. Although we observed ADO in some of our analyses, it did not interfere with the single-cell genotyping because we analyzed simultaneously 96 plated cells and a variety of loci. Contrasting to real homozygosis, ADO is a random phenomenon and it does not occur in the same way in all wells and loci. Thus, the more plates and therefore the more single cells analyzed, more reliable the results will be, a possibility easily achievable by the cell sorting method devised. In summary, by using the strategy proposed here for sorting and genotyping T. cruzi single cells we successfully identified the complexity of representative human and vector T. cruzi strains and estimated the relative contribution of each subpopulation. These results have important implications in settling the debate regarding multiclonality versus aneuploidy in these representative strains. Moreover, our results indicate that the complexity of T. cruzi strains may be even greater than initially expected. It is worth noting that the methodology described herein has the potential to close the discussion concerning multiclonality. Knowledge about complexity of T. cruzi strains is essential for determining the aspects involved in differential parasite tissue tropism, clinical manifestations of the disease, and drug resistance. The single-cell genotyping approach allowed the analysis of the intrapopulation diversity to a degree not achieved previously by conventional methods, and may represent a powerful and new tool for the understanding of the population structure and dynamics of T. cruzi.

Trypanosoma cruzi, the cause of Chagas disease, is responsible for high levels of mortality and morbidity in many countries of Latin America. This species is very polymorphic and was recently classified into six discrete typing units (TcI-VI) that are associated with different geographical distribution, transmission cycles, and varying disease symptoms. Natural parasite populations are thought to have a complex structure since in endemic areas the same hosts may be simultaneously infected by different T. cruzi strains due to frequent exposure to insects as well as the variety of reservoirs and vectors in the sylvatic environment and the heterogeneous composition of the parasite populations. Most of the putative multiclonal populations have been suggested based on indirect evidence, or current conventional cloning methods that may favor some individuals' growth. In an effort to elucidate the complexity of T. cruzi isolates we adapted a methodology originally described for analyzing single spermatozoids, to sort single parasites. This approach allowed us to completely dissect the complexity of four strains and estimate the relative contribution of each subpopulation. Knowledge about complexity of T. cruzi strains is essential for determining the aspects involved in differential parasite tissue tropism, clinical manifestations of the disease, and drug resistance.

Abstract Introduction Methods Results Discussion

population genetics ploidy microbiology parastic protozoans dna genetic polymorphism dna amplification biology molecular biology trypanosoma nucleic acids genetics protozoology molecular cell biology genetics and genomics

2012

Unequivocal Identification of Subpopulations in Putative Multiclonal Trypanosoma cruzi Strains by FACs Single Cell Sorting and Genotyping

Lymphatic filariasis and onchocerciasis are two chronic diseases mediated by parasitic filarial worms causing long term disability and massive socioeconomic problems. Filariae are transmitted by blood-feeding mosquitoes that take up the first stage larvae from an infected host and deliver it after maturation into infective stage to a new host. After closure of vector control programs, disease control relies mainly on mass drug administration with drugs that are primarily effective against first stage larvae and require many years of annual/biannual administration. Therefore, there is an urgent need for alternative treatment ways, i. e. other effective drugs or vaccines. Using the Litomosoides sigmodontis murine model of filariasis we demonstrate that immunization with microfilariae together with the adjuvant alum prevents mice from developing high microfilaraemia after challenge infection. Immunization achieved 70% to 100% protection in the peripheral blood and in the pleural space and furthermore strongly reduced the microfilarial load in mice that remained microfilaraemic. Protection was associated with the impairment of intrauterine filarial embryogenesis and with local and systemic microfilarial-specific host IgG, as well as IFN-γ secretion by host cells from the site of infection. Furthermore immunization significantly reduced adult worm burden. Our results present a tool to understand the immunological basis of vaccine induced protection in order to develop a microfilariae-based vaccine that reduces adult worm burden and prevents microfilaraemia, a powerful weapon to stop transmission of filariasis. Infections with filarial nematodes are classified among the “neglected tropical diseases” and cause serious public health problems in the tropics and subtropics with more than 150 million people infected and many more at risk. Lymphatic filariasis (LF) caused by the filarial nematodes Wuchereria bancrofti and Brugia spp. affects 120 million people with one third of them suffering from clinical presentations of the infection, namely lymphedema of the extremities and hydrocele, making LF the second-largest cause of long-term disability [1]. Human filariasis is transmitted by blood feeding vectors that ingest first stage larvae (microfilariae, Mf) from infected patients. Within the vector, Mf undergo two obligatory molts to become infective third stage larvae (L3). After their transmission to a new host infectious L3 molt twice into adult worms, which mate and release thousands of new Mf [2]. Current elimination strategies of the WHO such as the Global Programme to Eliminate LF (GPELF [3]) or the African Programme for Onchocerciasis Control (APOC [4]) are based on the mass drug administration (MDA) of the microfilaricides ivermectin (IVM), diethylcarbamazine and albendazole that have been successful in reducing Mf-burden. However, only IVM and albendazole are used in MDA programs against LF in Africa, because diethylcarbamazine causes rapid death of Mf, thereby increasing chances of adverse reactions, such as ocular damage in onchocerciasis [2]. In addition, doxycycline has been introduced for individual drug administration [5] directed against the obligate endosymbiotic Wolbachia bacteria of the filariae [6], [7]. Doxycycline inhibits filarial embryogenesis, and has been proven to be macrofilaricidal and to halt or reduce pathology [8], [9]. However, doxycycline is contraindicated in children ≤9 years and pregnant woman and improvement of anti-wolbachial chemotherapy to be used in public health control programs is a focus of ongoing research [2]. Despite the success of anti-helmintic drugs used in MDA programs in order to reduce infection and morbidity, certain drawbacks have to be considered. IVM has only limited macrofilaricidal efficacy [2] and repeated treatment for the life of the adult worm (up to eight years) is needed in order to stop transmission. Together with the limited logistics, especially in areas with civil unrest, the occurrence of adverse events after treatment such as scrotal pain or systemic inflammation can substantially corrupt the degree of compliance to therapy [1]. Finally, emerging resistance to drugs [10] reinforces the urgent need of alternative ways of disease control. Hence, besides drug therapy and vector control [11], the development of a vaccine against filarial infections would be a pivotal step towards the elimination of this disease [12]. As filarial nematodes have a high reproductive capacity with a total daily turnover of thousands of Mf in chronically infected human individuals [13], a vaccine achieving substantial clearance of circulating Mf would be a step towards stopping disease transmission. At best, a vaccine would be used in conjunction with MDA after Mf loads in a population were reduced either to prevent re-infection or to prevent circulation of Mf in the blood, particularly in areas of very high transmission. Despite the severity of infection and the vast number of infected people and individuals at risk, there is no vaccine against filarial infections available [12], [14]. Addressing this issue, various animal immunization studies using different approaches have been conducted. For example the group of Odile Bain used Litomosoides sigmodontis L3 to immunize mice and achieved up to 58% protection [15]. The protection was established within a few days after challenge infection and was characterized by L3-specific immunoglobulins, eosinophilia and high levels of IL-5. Lange and colleagues used a similar approach to immunize mice with Onchocerca volvulus L3 and also observed fast protection, which led to reduction of the recovery rates by 54% to 77% between five and 28 days post challenge infection (p. i.), and this was also associated with eosinophils and IL-5 [16]. In the experiments of Dixit el al [17], immunization of Mastomys coucha with a fraction of adult B. malayi extract reduced the recovery rate of adult worms by 85. 7%. Other groups used recombinant peptides instead of complete extracts [18], [19], [20], [21]. The immunization with B. malayi heavy chain myosin for example generated a high level of protection against challenge infection in jirds and M. coucha [21]. Different to all these setups, Anand and colleagues used a cocktail of B. malayi DNA to immunize mice and found high cytotoxcicity against B. malayi Mf in immunized mice, associated with specific Ig and increased IFN-γ responses [22]. However, none of these studies reported complete clearance of adults or Mf. In the 1960s, Wenk and colleagues found that cotton rats immunized with L. sigmodontis Mf had fewer blood-circulating Mf, although adult worms were present [23], [24], [25]. In this study we have taken this approach another step forward using the fully permissive BALB/c mouse model to study filarial infections. In this model with the advantage of greater access to immunological tools, female L. sigmodontis worms release the Mf into the pleural space of the thoracic cavity, the site of infection. From there they migrate into the peripheral blood [26]. Here, we show that immunization with Mf together with the adjuvant alum reduces microfilaraemia by apparently inhibiting embryogenesis. Eight - 12 week old female BALB/c wild type mice (Janvier, La Genest St. Isle, France) were maintained under specific pathogen-free conditions, according to animal welfare guidelines. All animal experiments were approved by and conducted in accordance with guidelines of the appropriate committee (Landesamt für Natur, Umwelt und Verbraucherschutz, Köln, Germany). Mf were purified from the peripheral blood of infected cotton rats on a Percoll gradient as described [27]. In brief, isoosmotic Percoll (Sigma-Aldrich, Munich, Germany) was prepared by mixing 9 parts of Percoll (density, 1. 130 g/ml) with 1 part of 2. 5 M D-sucrose (Sigma-Aldrich). Dilutions of the isoosmotic Percoll in 0. 25 M sucrose were made to obtain 25% and 30% solutions. Three ml of both gradient dilutions were layered and the peripheral blood diluted 1∶2 in phosphate-buffered saline (PBS) (PAA, Cölbe, Germany) was placed on top. After centrifugation at 400×g for 30 min at room temperature (RT) without brakes, Mf (between the 25% and 30% layers) were recovered and washed twice with PBS and 1×105 viable Mf per mouse were used for each immunization. Injection was performed via different administration routes as indicated in the text (see also Figure S1). For Mf attenuation, 1×106 Mf/ml were irradiated 40 min at 140 kV and 25 mA, corresponding to an absorbed dose of 400 Gray (Gy) at the Facility of Experimental Therapy of the University Hospital Bonn. Microscopic analysis of irradiated Mf confirmed their attenuation by monitoring their motility (data not shown). For IVM (Merck, Darmstadt, Germany) treatment after immunization, mice received 800 µg per kg mouse body weight. For immunization with alum (Thermo Scientific, Bremen, Germany), Mf were added slowly to the adjuvant to a final adjuvant concentration of 25% and then mixed on an automatic shaker at 1,000 rpm for 30 min. After this procedure Mf were morphologically intact, however they were amotile and motility was not reconstituted after 72 h at 37°C and 5% CO2, suggesting that the Mf were not viable. Directly before injection, the suspension was intensively vortexed. For the sham injection control mice received alum or PBS. In all experiments, second and third immunization injections were performed two and three weeks after the initial immunization. For challenge infection, infective L3 larvae were transmitted through the bite of the vector mite Ornithonyssus bacoti as described [28]. Natural infection was performed one week after the last immunization. Peripheral blood was taken from the tail vein and directly transferred into 500 µl Hinkelmann solution (0. 5% [wt/vol] eosin Y, 0. 5% [wt/vol] phenol (both Merck) and 0. 185% [vol/vol] formaldehyde (Sigma-Aldrich) in deionized water). After centrifugation (5 min, 250×g) supernatant was discarded and pellet suspended in 20 µl PBS before counting under the microscope (40×). For monitoring Mf in the pleural space, 20 µl from 1 ml pleural space lavage (see section below) were added to 450 µl Hinkelmann solution and treated as described for peripheral Mf. Mice were euthanized with Isofluran (Abbott, Wiesbaden, Germany). Parasites and cells were harvested from the pleural space by lavage with cold PBS. At 15 days p. i. , L4 (L3/L4 molting around day eight in BALB/c mice [29]) and at days 70 and 90 p. i. adults (L4/adult molting around day 22 p. i. [29]) were separated from the cells by 15 min sedimentation. For the embryogram each single female worm was transferred into 80 µl PBS, cut into several pieces and embryonic stages squeezed out of the uterus using a 1. 5 ml plastic tube and plastic pestle. The embryonic stages were stained by adding 20 µl Hinkelmann solution and 10 µl of this preparation (thus 10% of total uterine content of each analyzed female filariae) were analyzed under the microscope. If present, three female worms from each mouse were investigated. Damaged females, empty females or females with only oocytes were excluded from analysis. Ex vivo stimulations were performed at days 22,70 or 90 p. i. . Lysis of red blood cells from the pleural space exudate cells was done by 5 min incubation with trisammoniumchloride (Sigma-Aldrich). Cells were washed twice with PBS, filtered through sterile 41 µm gaze (Bueckmann, Moenchengladbach, Germany), and 2. 5×105 cells per well in RPMI medium (supplemented with 10% fetal calf serum, 1% L-glutamine, 1% penicillin/streptomycin, 1% non-essential amino acids, 1% sodium bicarbonate, and 1% sodium pyruvate (all PAA) ) were stimulated with 5 µg/ml concanavalin A (Sigma-Aldrich) in a 96 well plate (Greiner Bio-One, Frickenhausen, Germany) or the respective worm extracts for 72 h at 37°C and 5% CO2. For preparation of L. sigmodontis extract, freshly isolated adult worms were rinsed in sterile PBS before being mechanically minced. Insoluble material was removed by centrifugation at 300×g for 10 min and 4°C. Protein concentrations of crude extracts were determined using the Advanced Protein Assay (Cytoskeleton, Denver, CO, USA). All procedures were conducted under sterile conditions. L. sigmodontis Mf extract was similarly prepared with sonicated (Bandelin Electronics, Berlin, Germany) freshly isolated Mf. Systemic Mf-specific IgG was measured from plasma of mice directly before immunization injections (days −28, −14 and −7) and in weekly intervals after infection (days 0,7 and 14 p. i.). Blood was taken submandibular from anesthetized (Ketanest, Medistar, Ascheberg, Germany/Rompun, Bayer, Leverkusen, Germany) mice. After centrifugation (5 min at 6,500×g), plasma was taken and stored at −20°C until further usage. Mf-specific Ig of the pleural space were measured from the supernatant of the pleural space lavage at days 22 and 70 p. i. . Polysorb ELISA plates (Nunc, Roskilde, Denmark) were coated overnight at 4°C with 10 µg/ml of Mf crude extract in PBS at pH 9. After blocking 1 h with 1% BSA-PBS (PAA), plates were washed with PBS containing 0. 05% Tween 20 (Sigma-Aldrich) and incubated for 2 h at RT with either 50 µl of pleural space lavage or a 1∶10 (IgE) or 1∶1,000 (IgG) dilution of plasma. After another washing step, biotinylated detection antibody (BD Pharmingen, Heidelberg, Germany) was added as recommended by the manufacturer. After a final wash, alkaline phosphatase-conjugated streptavidin (Roche, Grenzach, Germany) was added and tetramethylene benzidine (Carl Roth, Karlsruhe, Germany) was used as substrate. The reaction was stopped by adding 1 M H2SO4 (Merck) and the absorbance was measured at 450 nm. IFN-γ (eBioscience, Frankfurt, Germany), IL-5 (BD Pharmingen), IL-13, macrophage inflammatory protein 2 (MIP) -2α, chemokine C-C motif ligand 5 (CCL5), granzyme B, eotaxin-1 and eotaxin-2 (R&D Systems, Wiesbaden, Germany) ELISA of the pleural space lavage and the supernatants of restimulated cells were performed according to manufacturer' s instructions. Statistical analyses were performed with GraphPad Prism 5. 0 software (GraphPad Software, La Jolla, CA, USA), using the Student' s unpaired t-test for parametric, the Mann Whitney t-test (u-test) for nonparametric data and Welch' s correction for data sets with different variances. Variances were tested with the D' Agostino & Pearson omnibus normality test. P-values ≤0. 05 were considered significant. Microfilaraemia, Ig kinetics and cytokine responses after ex vivo restimulation were analyzed with regular 2-way ANOVA and Bonferroni post tests. Data were graphed with means ± standard error of mean (SEM). Revisiting some of the known immunization protocols in animal models, we immunized mice in various ways (for detailed information see Figure S1). Initially, because Mf are mainly located in the blood of the infected host, we immunized BALB/c mice three times intravenously (i. v.) with 100,000 living Mf. This injection resulted in a transient presence of Mf in the peripheral blood lasting about two weeks (Figure S2A). After challenge infection, natural Mf levels in the peripheral blood were monitored from the onset of peripheral microfilaraemia at day 50 until the end of patency around day 90 p. i. . This immunization neither delayed the onset of natural microfilaraemia nor changed the Mf levels in the peripheral blood after challenge infection compared to control animals at any time point during patency (Figure 1A). Next, since healthy Mf may modulate immune responses in the immunized host, and in order to enrich the amount of immunogenic material mice were treated after immunization with the microfilaricide IVM, which suppresses the ability of Mf to secrete immunomodulatory proteins [30] and inhibits their neuromuscular control [31]. Accordingly, after IVM injection, Mf disappeared from the peripheral blood within one day after injection (Figure S2B). However, as observed for mice immunized three times with live Mf, microfilaraemia was not reduced in infection-challenged mice (Figure 1B). According to the successful scheme originally used in cotton rats [24], we then immunized mice first subcutaneously (s. c.), followed by an intraperitoneal (i. p.) immunization two weeks later and an i. v. immunization three weeks after primary immunization. As with the two former schemes, this route of immunization failed to protect mice and this was independent of the usage of either healthy (Figure 1C) or irradiation-attenuated Mf (Figure 1D). Finally, to investigate whether a standard adjuvant is able to establish protective immunity, mice were then immunized three times with 100,000 Mf together with the adjuvant alum. Due to the viscosity of alum this immunization was performed s. c. . Mice immunized with Mf in alum had significantly reduced numbers of circulating Mf after challenge infection compared to control animals throughout patency (P<0. 05, Figure 2A). Furthermore, the frequency of mice that became microfilaraemic until the end of observation was significantly reduced in the immunized group (P<0. 05, Figure 2B). Taken together, effective vaccination of 70–100% was only observed in mice after s. c. immunization with Mf in alum, but not after immunization with Mf alone, irrespective of the administration route, irradiation of Mf or IVM treatment of mice after immunization. Consequently, all further experiments were performed with 100,000 s. c. -administrated Mf in alum. To investigate whether immunization inhibits the ability of females to generate and release Mf or just hinders the Mf migration into the blood, the pleural space lavage was analyzed for the presence of Mf on days 70 and 90 p. i. . Figure 2C and D show, that the number of Mf in the pleural space was significantly reduced after immunization compared to the alum-treated control group, the latter showing a wide range of microfilaraemia that is well described for this model [28]. The few immunized mice that were Mf+ at day 70 p. i. had only low Mf levels with a mean of four Mf compared to 226 Mf/20 µl lavage in the alum treated control group (P<0. 005, Figure 2C). Furthermore, at day 90 p. i. 90% of the immunized mice were free of Mf with only one mouse having five Mf/20 µl lavage. In contrast, 90% of control mice still harbored Mf with a mean of 68 Mf/20 µl lavage (P<0. 005, Figure 2D). To rule out that alum itself influences the course of infection, we compared mice injected s. c. with either alum or PBS and did not find significant differences in the course of infection (data not shown, 2-way ANOVA of peripheral microfilaraemia P = 0. 4898, Welch-corrected t-test of pleural space Mf P = 0. 7377). The reduction of Mf levels not only in the blood but also in the pleural space suggested that either Mf were cleared immediately after being released or the Mf output of female worms was reduced. Consequently, the embryogenesis of female worms was analyzed. During the filarial embryogenesis four main developmental stages can be distinguished in the uteri of female worms (Figure 3A–D): oocyte, divided egg (fulfilled first cell division), pretzel and stretched Mf [32], [33]. If present, three female worms from each mouse were investigated. Empty females or females with only oocytes were excluded from analysis. In the embryograms of three independent experiments the percentage of those excluded females was similar between immunized (33,20 and 27%) and non-immunized mice (16,18, and 28%), indicating that immunization did not interfere with insemination. At day 70 p. i. we found all stages to be present in the uteri of female worms of control mice, whereas females of immunized mice contained mainly the first two developmental stages (oocyte and divided egg) but rarely pretzel stages (P<0. 001) and fully developed Mf (P<0. 01, Figure 3E, see also Figure S4A). To confirm this, any remaining worms were checked for the presence or absence of later stages such as fully stretched Mf and pretzel stages. Only two out of 23 female filariae were positive for later stages in the Mf-vaccinated group (see also Figure S4A; 0/6 worms), whereas in the control group 25 of 31 females contained stretched Mf (see also Figure S4A; 14/26 worms). Inhibition of embryogenesis at day 70 p. i. was exemplarily documented by live video analysis of the uteri of freshly isolated healthy females (video S1, and Video S2). Finally, in an additional experiment inhibition of embryogenesis could already be observed at the beginning of patency (day 56 p. i. , see Figure S4B). Taken together, these data suggest that immunization induces the inhibition of larval development. Cross reactive protection with respect to other developmental stages is known for immunization with L. sigmodontis L3. Thus, we asked whether Mf immunization may also affect stages other than the Mf. Analysis of the L4 burden at day 15 p. i. and adult burden at day 56 p. i. showed that immunized mice had similar worm numbers as control animals (Figure 4A, B). However, at day 70 (P<0. 005, Figure 4C) and 90 p. i. (P<0. 0001, Figure 4D) immunized mice contained significantly fewer adult worms and this reduction was associated with decreased numbers of both males and females, as the gender balance was similar in immunized and control mice (Figure 4E). Male and female worms did not differ in length to the corresponding worms of control mice (Figure 4F, G). Taken together, our data show that immunization with 100,000 Mf in alum not only inhibited microfilaraemia, but also reduced adult worm burden at later time points. To investigate whether immunization-induced Mf-specific Ig responses were associated with protection, Mf-specific IgE, IgG1 and IgG2 levels were measured in the plasma and pleural space lavage at different time points throughout immunization and infection. As Figure 5A and B illustrate, the immunization induced an Mf-specific humoral response and both IgG1 and IgG2 antibodies were elevated in the blood. The most prominent increase was observed after the boost immunizations, as indicated by the levels seven days before the challenge infection. A comparison of both immunized groups (infected vs. uninfected) revealed that these humoral responses were not further enhanced by the infection itself. The same picture was found at the site of infection with Mf-specific IgG1 and IgG2 levels being significantly elevated in immunized mice compared to controls on day 22 (P<0. 001, Figure 5C, D) as well as on day 70 p. i. (P<0. 001, Figure 5E, F). Albeit the differences in IgG1 levels remained significantly higher in immunized mice at day 70 p. i. , the IgG1 levels of infected but non-immunized mice increased on days 28 and 42 (Figure S6C, D) compared to day 22 p. i. (Figure S6A, B). This however indicates a Th2 shift induced by the parasite itself and is well-known for primary infected BALB/c mice [34], [35]. The amount of Mf-specific IgE was not increased at day 22 p. i. in the pleural space of immunized mice (mean OD of 0. 047) compared to non-immunized mice (mean OD of 0. 0892). Later during infection, Mf-specific IgE was elevated in the blood of immunized mice with a mean OD of 0. 121 (day 28 p. i.) and 0. 187 (day 42 p. i.) in immunized mice, and a mean OD of 0. 043 and 0. 050 in control mice, respectively. However, these levels of Mf-specific IgE clearly did not reach the IgE levels of chronically infected mice (OD on day 100 p. i. 1. 683; single experiment, data not shown), suggesting that immunization per se does not lead to a strong IgE induction. To classify cellular responses, we analyzed major cell populations in the pleural space lavage by flow cytometry. However, no consistent differences were observed (Table S1). We also measured various cytokines in the pleural space at day 22 p. i. . Since eosinophils are known effector cells in helminth infections [36], we measured molecules involved in eosinophil recruitment or activity, i. e IL-13, MIP-2α, CCL5, granzyme B, eotaxin-1 and eotaxin-2. Results from three independent experiments did not reveal any significant differences between immunized mice and control animals (data not shown). However, analysis of hallmark cytokines IL-5 and IFN-γ of type 1 and 2 immunity showed that immunized mice had significant more IFN-γ in the pleural space of the thoracic cavity (P<0. 001), whereas level of IL-5 were low irrespective of the immunization (Figure 6A). A similar picture was observed when cells recovered from the pleural space were restimulated with worm extracts (Figure 6B, C). Strikingly, 22 days p. i. , a time point when parasites are already present in the pleural space, only cells of immunized mice secreted IFN-γ regardless of whether they were infected or not. This effect was seen after specific restimulation with crude extract of adult worms and Mf, as well as with nonspecific stimulation by concanavalin A (P<0. 001, Figure 6B). Although less pronounced, enhanced IFN-γ responses after restimulation were also present throughout patency (Figure S7A, B). Different to IFN-γ, the IL-5 responses were dependent on the infection itself, as only cells from infected mice secreted IL-5 after restimulation irrespective of immunization (Figure 6C, Figure S7C, D). Current public health control of human filarial infections relies on chemotherapy provided by MDA programs. Antifilarial drug therapy has to be implemented for years with high coverage, incurring high logistical costs and the emergence of drug resistance is a potential threat [12]. Thus, a vaccine that results in the reduction of parasite burden would complement the MDA efforts, as suggested for other neglected tropical diseases [37] and that would be complementary step towards elimination of the diseases. The present study describes a successful immunization protocol against L. sigmodontis Mf in the murine model of filariasis, which additionally resulted in a reduced adult worm burden. Subcutaneous immunization with Mf in alum prevented the onset of microfilaraemia after challenge infection in the majority of mice. Reduced Mf loads were observed in the peripheral blood and at the site of infection in conjunction with intrauterine inhibition of embryogenesis. Protection was further associated with systemic and local Mf-specific IgG and IFN-γ secretion of pleural space exudate cells. In mouse models the adjuvant alum is often administered i. p. and this route has been referred to establish “systemic” responses in contrast to the “local” s. c. route [38]. Interestingly, we found neither the systemic i. v. route nor the combination of the local s. c. and both systemic i. p. and i. v. routes to be able to reduce microfilaraemia (Figure 1). As systemic immunizations have been reported to induce tolerance rather than immunity [39], out of the protocols tested in this study, only the s. c. immunization was able to immunize mice successfully against L. sigmodontis Mf. The s. c. immunization would also be a route, which is best applicable to humans. It is known that immunization with irradiated L3 stages reduces the recovery rate of larvae in the pleural space [40]. However, immunization with Mf did not render mice less susceptible for infection per se, as the worm burden did not differ on day 15 p. i. . The short time between immunization and challenge infection might explain the absence of enhanced immunity against incoming L3. The observation that the adult worm burden remained similar in both groups until the onset of patency at day 56 p. i. , may suggest that the accessibility of Mf in the pleural space of the thoracic cavity could be a critical step in initiating the responses that affect the adult filariae. However, only one experiment was performed on day 56 and a more detailed analysis of the efficacy on adult worm burden is needed to identify the time point when adulticidal immune responses are effective. Interestingly, at later time points, namely on days 70 and 90 p. i. , we could observe a reduced worm burden in immunized mice compared to the controls. This reduction was seen with both male and female worms, suggesting effector mechanisms not only acting against intrauterine Mf in female worms, but against target structures of adult worms. In line with this, different developmental stages of filariae share many molecular structures [41] and cross-reactive immunization effects are documented for filarial immunizations [19], [42]. Importantly, it is unlikely that the lower worm burden observed on days 70 and 90 p. i. is the reason for reduced microfilaraemia in immunized mice, because all immunized mice contained male and female worms and it has been shown that even a few fecund females can establish peripheral microfilaraemia [43]. Inhibition of embryogenesis in immunized mice was further indicated by the reduced number of Mf in the pleural space and, more importantly, the presence of embryonic stages that had not developed beyond several divisions of the fertilized oocytes. This is in contrast to Wenk and colleagues who found fully developed Mf in the uterus of female worms and in the pleural space of infected and Mf-immunized cotton rats [23]. The differences between both animal models in the mode of Mf reduction may be due to different susceptibilities of the hosts to L. sigmodontis. Both, BALB/c mouse and the cotton rat develop patent infection, but BALB/c mice clear the infection after 3–4 months, whereas the cotton rat is the natural host and can harbor filarial parasites for years [29]. This is the general downside of using laboratory mice for L. sigmodontis infection rather than the natural host; however the advantage is that cytokine responses can be measured and associated with protection, and for future studies cytokine deficient mice may answer further questions about the essentiality of key cytokines for vaccination success. In our experiments, immunization induced Mf-specific IgG1 and IgG2 antibodies, present throughout the whole infection including patency, were associated with protection. Because protective responses induced by immunization often rely on protective antibodies [44] and the importance of B cells in promoting immune responses against filarial Mf is well documented for the murine model of filariasis [45], [46], we analyzed the frequency and total number of pleural space B1 and B2 B cells by flow cytometry, but found no consistent differences between immunized and non-immunized mice (Suppl. Table 1), suggesting differences in B cell activation rather than in total B cell numbers. Interestingly, despite a strong antibody production female worms of immunized mice appeared morphologically as intact as those in non-immunized control mice. Furthermore we did not find any host cells within the female worms. These two findings suggest a blockade of embryonic development, rather than a cell-dependent destruction of the embryonic stages. In our experiments, Mf-specific IgGs may have entered the female worm uterus and bound to the developing early developmental stages, thereby hindering their further growth in an Ig-dependent, but cell-independent manner. Indeed, it is known that filarial-infected humans produce filarial-specific Ig that is able to bind early intrauterine filarial stages, as shown by using sera of chronic LF patients against isolated intrauterine Mf stages from the filariae Setaria digitata [47]. Another possibility for cell-independent, Ig-mediated responses is the activation of the complement cascade resulting in the formation of the membrane attack complex (MAC), due to insertion of complement proteins into a phospholipid bilayer [48]. Although there is as yet no evidence for MAC formation in the sheath of adult nematodes, earlier developmental stages may be more sensitive to MAC formation. The next step to clarify the role of antibodies in the establishment of immunization-induced protection would be the verification of embryonic stage-bound Mf-specific Ig, e. g. by immunohistochemistry. Furthermore, experiments with mice having a defect in immunoglobulin production would give important insights into the relevance of Ig for impairment of embryogenesis. Also, a possible IgE reaction to the immunization has to be elucidated. We did not observe a strong IgE response after immunization and infection in the blood nor at the site of infection. However, future experiments should clarify how chronically infected mice that already have Mf-specific IgE respond to immunization. In humans, a vaccination may be favourable one month after IVM treatment, when individuals have no skin or blood Mf, as the risk of urticaria due to immune attack on remaining Mf would be at its minimum. IFN-γ and IL-5 are well known players in innate, adaptive and vaccine-induced immunity against helminths [49], with the adaptive type 2 response referred to as “typical” for helminth infections [50]. We found that immunization was associated with strong IFN-γ responses, mirrored by increased levels in the pleural space and after restimulation of pleural exudate cells. The importance of IFN-γ production in immune responses against Mf in permissive BALB/c mice is underlined by several findings. IFN-γ−/− mice have increased numbers of circulating L. sigmodontis Mf compared to the wild type littermates [51]. In addition it has been shown that IFN-γ RNA levels of restimulated splenocytes obtained from L. sigmodontis-infected BALB/c mice are strongly increased within days after the beginning of patency [52]. These observations may reflect the moderate increase in IFN-γ production also in non-immunized mice upon natural infection and the less pronounced but still significant differences in IFN-γ between immunized and non-immunized mice during patency (Figure S7A, B). Furthermore, it is known that injected B. malayi Mf, but not implanted adult stages induce IFN-γ and Th-1-assiociated IgG2a in BALB/c mice [53]. IFN-γ is an inducer of IgG2a [54] therefore it is most likely that in our experiments, Mf-induced IFN-γ has promoted the secretion of IgG2. Importantly, induction of IFN-γ is not in conflict with the use of the adjuvant alum, which is generally referred as Th2-promoting, because recent findings have shown that alum also can influence proliferation and IFN-γ production of CD8+ T cells [55]. Furthermore, Toll-like receptor agonists have been found to be able to bias alum towards a mixed Th1/Th2 response [56]. L. sigmodontis, like many other filariae contains endosymbiotic Wolbachia bacteria that are recognized by Toll-like receptors [57]. Although IL-5 responses did not differ between immunized and control mice, this cytokine may also play a role for the overall effect of immunization in the infected mice. It may even be possible that the effect of immunization, although predominated by IFN-γ, may be dependent on at least baseline levels of IL-5, since this cytokine has been shown to be important for both adult worm and Mf containment in L. sigmodontis infection in our earlier reports [58], [59]. Future immunization experiments with BALB/c mice defective for IFN-γ or IL-5 responses will shed more light on the importance of both key cytokines for the inhibition of embryogenesis. Taken together, the immunization scheme presented in this study demonstrates the feasibility of an immunization that is directed against the Mf stage, leading to protection against peripheral microfilaraemia with an efficacy of up to 100%. The IFN-γ that has been induced by the immunization suggests a shift towards a Th1-like milieu in the host that may furthermore promote direct or indirect responses against the Mf during patency, possibly through IFN-γ-promoted IgG2a. It is known for human LF that there is a threshold for Mf density in the peripheral blood to achieve transmission and a high number of infective bites is needed to produce a patent infection [60]. We hypothesize that the reduction of circulating peripheral Mf at the level we observed might prevent transmission. The study presented here contributes to the understanding of the immune mechanisms needed to develop a vaccine against filarial parasites. Whereas the use of Mf recovered from infected humans would be costly and the number of Mf limited, even disregarding the potential transmission of other infections, our data may serve for a better understanding of the nature of protective Mf vaccination. Future assessments should address the characterization of microfilarial molecular subunits that account for this protection, as the growing fields of helminth genomics [61] and systems biology [62] may predict such potential Mf-related vaccine candidates. Administration of only a subunit vaccine may also avoid vaccination with tolerogenic molecules contained within the Mf and lead to better efficacy of protection.

Lymphatic filariasis is caused by parasitic filarial worms that are transmitted by mosquitoes, requiring uptake of larvae and distribution into the blood of the host. More than 120 million people are infected and about 30% of these individuals suffer from clinical symptoms. Reduction in transmission currently depends on mass drug administration, which has significantly reduced transmission rates over the past years. However, despite repetitive rounds of administration, transmission has not been eliminated completely from endemic areas. In some infected individuals the immune system can partially control the parasite, such that a proportion of infected individuals remain microfilaria-negative, despite the presence of adult worms. Therefore mechanisms must exist that are able to combat microfilaraemia. Identifying such mechanisms would help to design vaccines against disease transmitting microfilarial stages. Using the Litomosoides sigmodontis murine model of filariasis research we show a successful immunization against the blood-circulating larval stage that is responsible for arthropod-dependent transmission of the disease. Reduced microfilaraemia was associated with impairment of worm embryogenesis, with systemic and local microfilarial-specific host IgG and with IFN-γ secretion by host cells at the site of infection. These results raise hope for developing a microfilariae-based vaccine, being a pivotal step towards eradicating filariasis.

Abstract Introduction Materials and Methods Results Discussion

medicine infectious diseases neglected tropical diseases immunology biology zoology parasitic diseases immune system

2012

Immunization with L. sigmodontis Microfilariae Reduces Peripheral Microfilaraemia after Challenge Infection by Inhibition of Filarial Embryogenesis

The locomotor gait in limbed animals is defined by the left-right leg coordination and locomotor speed. Coordination between left and right neural activities in the spinal cord controlling left and right legs is provided by commissural interneurons (CINs). Several CIN types have been genetically identified, including the excitatory V3 and excitatory and inhibitory V0 types. Recent studies demonstrated that genetic elimination of all V0 CINs caused switching from a normal left-right alternating activity to a left-right synchronized “hopping” pattern. Furthermore, ablation of only the inhibitory V0 CINs (V0D subtype) resulted in a lack of left-right alternation at low locomotor frequencies and retaining this alternation at high frequencies, whereas selective ablation of the excitatory V0 neurons (V0V subtype) maintained the left–right alternation at low frequencies and switched to a hopping pattern at high frequencies. To analyze these findings, we developed a simplified mathematical model of neural circuits consisting of four pacemaker neurons representing left and right, flexor and extensor rhythm-generating centers interacting via commissural pathways representing V3, V0D, and V0V CINs. The locomotor frequency was controlled by a parameter defining the excitation of neurons and commissural pathways mimicking the effects of N-methyl-D-aspartate on locomotor frequency in isolated rodent spinal cord preparations. The model demonstrated a typical left-right alternating pattern under control conditions, switching to a hopping activity at any frequency after removing both V0 connections, a synchronized pattern at low frequencies with alternation at high frequencies after removing only V0D connections, and an alternating pattern at low frequencies with hopping at high frequencies after removing only V0V connections. We used bifurcation theory and fast-slow decomposition methods to analyze network behavior in the above regimes and transitions between them. The model reproduced, and suggested explanation for, a series of experimental phenomena and generated predictions available for experimental testing. Central Pattern Generators (CPGs) are neural networks that can produce organized rhythmic motor activities in the absence of rhythmic inputs and feedbacks from other parts of the nervous system. The CPGs that generate basic rhythmic locomotor activities and control locomotion in vertebrates are located in the spinal cord [1–11]. It appears that each limb in mammals is controlled by a separate CPG because cats with chronic thoracic spinal lesions were shown to step on a split treadmill with different speed for the left and right hind limbs [12]. The gait of locomotion is defined by the coordinated limb movements and hence by the phase relationships between rhythmic patterns generated by CPGs controlling each limb. In turn, these relationships are defined by neural circuits within the spinal cord providing direct or indirect interactions between the CPGs. Several computational models of neural circuits were proposed to reproduce and explain the left-right and segmental coordination during locomotion in lamprey [13–16] and left-right hindlimb and quadrupedal limb coordination in the salamander [17–19] and mammals [20–22]. Some computational studies focused on gait transitions analyzing them as bifurcation phenomena [23–25]. Spinal circuits in some models were simulated and analyzed as coupled nonlinear oscillators [26–28]. In these models, each oscillator represented an independent rhythm-generating center (simulated as a single neuron or a neural population), and the main goal of these models was to study how inter-oscillator couplings could affect the synchronization and the phase lags between rhythm-generating centers [18]. It is considered that the locomotor gait and pattern in these models mostly depend on the topology of interactions and properties of neurons involved, rather than on the local rhythm-generation mechanisms [18]. However, even in lower vertebrates such as the lamprey, it is extremely difficult to identify and characterize all neurons and neural circuits that are involved in such coupling/interactions during locomotion. It is generally considered that phase relationships between the activities of corresponding neurons on the left and right sides of the spinal cord are provided by so-called commissural interneurons (CINs), whose axons cross the midline and innervate neurons on the contralateral side of the cord [6,29–32]. There are different types of inhibitory and excitatory CINs, which are involved in the left-right alternating (e. g. , during normal walking) or left-right synchronized (e. g. , during hopping or galloping) locomotor activities [6–8,29–32]. Significant progress in understanding the functional and structural organization of spinal circuits has been achieved due to recently developed combinations of genetic, molecular, and developmental techniques. Several populations of neurons were found to be derived from genetically distinct populations of embryonic neurons in the spinal cord of adult mice [33–36]. Some of these genetically characterized neurons were identified as CINs, including the V0 and V3 interneurons [6,30,33,37,38]. The V0 population contains both excitatory and inhibitory CINs that both contribute to alternating left and right neuronal activities [6,30,37]. The V3 neurons are excitatory and mediate interactions promoting left-right synchronized hopping behavior [38]. The V0 population contains genetically distinct subpopulations of CINs: V0V and V0D. The inhibitory V0D neurons constitute about two-thirds of the V0 population and the excitatory V0V neurons constitute about one-third [6,30,39]. The specific functional roles of V0V and V0D subpopulations in coordinating left and right locomotor activities have been recently studied using isolated spinal cord preparations from genetically transformed mice with either V0V, or V0D, or both V0 subpopulations knocked out [37]. The spinal cord was isolated in vitro to ensure that supraspinal structures were not participating in rhythm generation. The specific contribution of each subpopulation to the left-right coordination has been found to depend on the frequency of locomotor oscillations. Locomotor oscillations were induced by application of a mixture of N-methyl-D-aspartate (NMDA) and 5-hydroxytryptamine (5-HT). These oscillations were characterized by alternating activities recorded from left and right flexor-related (LL2 and RL2) and left and right extensor-related (LL5 and RL5) lumbar roots. The speed of locomotion (oscillation frequency) was regulated by the NMDA concentration [37,40]. Measuring the phase differences between rhythmic activities recorded from LL2 and RL2 roots and from LL5 and RL5 roots allowed for identification of an alternating or synchronized pattern. The results of these studies can be summarized as follows [37]: (i) spinal cord preparations from control mice showed left-right alternating behavior at all locomotor frequencies; (ii) preparations from the mice with genetically ablated V0 populations (both V0V and V0D types) showed synchronized (hopping) activity at all locomotor frequencies; (iii) preparations from the mice with only V0V populations ablated maintained left-right alternation at low frequencies but switched to hopping gait at high frequencies; (iv) preparations from the mice with selectively ablated V0D populations showed hopping at low frequencies and switched to alternation at high frequencies. This model represents a simplified version of the large-scale model [41], which was used for computer simulation of spinal circuits involved in rhythm generation and frequency-dependent left-right CIN-mediated interactions, but was too complicated to use methods of qualitative theory of differential equations and dynamical systems. Therefore, in this study, we formulated and analyzed a simplified mathematical model. The model consisted of four pacemaker neurons representing left and right, flexor and extensor rhythm-generating centers interacting via commissural pathways representing V3, V0D, and V0V CINs. We used this model to investigate different regimes of behavior of the intact circuitry, and following a removal of commissural interactions mediated by V0D, or V0V, or both V0 CIN pathways to mimic recent experimental studies of Talpalar et al. [37]. Then, using methods of bifurcation theory and fast-slow decomposition, we analyzed the essential properties of this system under different conditions and propose plausible explanations for the experimentally observed behavioral transformations. Most of the recent models of mammalian spinal circuits and locomotor CPGs were based on interacting populations of tens or hundreds of spiking neurons modeled in the Hodgkin-Huxley style and described by 5–10 differential equations per neuron [41–46]. Such large networks of spiking neurons are too complex to be analyzed by traditional mathematical approaches, and hence certain simplifications need to be applied. Such a simplification has been previously proposed in the analysis of the model of the mammalian respiratory CPG [47,48]. In these studies, each population of spiking neurons was described by a single, non-spiking neuron model, in which the voltage variable represented an average voltage for the population and the output activity was described as a nonlinear function of voltage, f (V) [47–49]. It was established that such a simplification can provide a reasonably accurate description of population activity, including transitions between quiescent (silent) and active states in population dynamics [47]. This simplification was used in our present model. The other simplification of the current study was that, in contrast to the previous multi-scale models [42–45], we focused only on rhythm-generating circuits and did not consider motoneurons and interneurons (such as Ia and Renshaw cells) that are not critically involved in rhythmogenesis and left-right coordination. Therefore, the reduced neural circuit that we analyzed included two (left and right) rhythm generators (RGs) bilaterally interconnected via several commissural pathways. Each RG consisted of one flexor (left, LF, or right, RF) and one extensor (left, LE, or right, RE) centers reciprocally inhibiting each other via inhibitory interneurons (Ini1 and Ini2, respectively) (Fig 1A). Three major commissural pathways were considered: (1) the excitatory V3 CINs mediate mutual excitation between the left and right flexor centers (LF and RF); (2) the inhibitory V0D CINs mediate mutual inhibition between these centers; and (3) the excitatory V0V CINs that were also shown to contribute to mutual inhibition between these centers and promote left-right alternation. To provide mutual inhibition, the excitatory V0V CINs should either receive inputs from the ipsilateral flexor centers and affect the contralateral flexor centers via inhibitory interneurons (as suggested in [37,41]) or mediate a crisscross excitation from each extensor center to the contralateral flexor center as shown in Fig 1A (see also [41]). Because no recordings from V0V neurons have been made so far, it is difficult to determine which of the two V0V pathways suggested is more realistic or whether both these pathways are present in the spinal cord (see discussion in [41]). Preliminary large-scale simulations have shown that both pathways lead to qualitatively similar behavior [41]. Therefore, in this study, we only focused on the V0V CIN pathways providing the direct crisscross excitation from each extensor center to the contralateral flexor center (Fig 1A). The final simplification concerned all interneurons (i. e. CINs, Ini1, and Ini2), which, for simplicity, were removed and replaced by the corresponding direct synaptic connections as shown in Fig 1B. Therefore, the final simplified model under investigation consisted of four centers (LF, LE, RF, and RE) with the unilateral inhibitory interactions between flexor and extensor centers on each side via Ini1 and Ini2 pathways and the bilateral V3, V0D, and V0V commissural interactions between the left and right centers as shown in Fig 1B. Each center is described using a reduced activity-based non-spiking neuron model with only two dynamical variables. This model was adapted from a reduced model of the population of intrinsically bursting neurons in the pre-Bötzinger complex developed by Rubin et al. [47,48]. Endogenous bursting in these neurons was suggested to involve the persistent (slowly inactivating) sodium current, INaP, first described by Butera et al. [50] (Model 1), with burst termination based on the slow inactivation of this current. This model was shown able to generate busting activity with a wide range of a parameters defining neuronal excitation and burst frequencies, and exhibited the correct change (reduction) of burst amplitude when the frequency increased [47,48]. This model has been successfully implemented in several previous models of the locomotor CPG [41–46,51,52]. The persistent sodium current was indeed found in spinal interneurons [52–56] and its blockade terminated rhythm generation in the rat spinal cord [53]. Similar to the Rubin et al. models [47,48], our formulation for each center includes an explicit representation of INaP: C⋅V˙=−INaP−IL, (1) where C is the membrane capacitance, INaP represents the persistent sodium current, and IL is the leak current. The currents in Eq (1) are described as follows: INaP=g¯NaP⋅mNaP∞ (V) ⋅hNaP⋅ (V−ENa); (2) IL=g¯L⋅ (V−EL), (3) where g¯NaP and g¯L are the maximal conductances of the persistent sodium and leak channels, and ENa and EL are the reversal potentials for sodium and leak currents, respectively. hNaP is the INaP inactivation gating variable, and mNaP∞ (V) represents voltage-dependent steady state of INaP activation: The INaP activation, mNaP∞ (V), is considered instantaneous; the INaP inactivation is slow and hNaP represents the “slow” dynamical variable in this model described as follows: τhNaP (V) ⋅h˙NaP=hNaP∞ (V) −hNaP, (5) where hNaP∞ (V) and τhNaP (V) represent the voltage-dependent steady state and time constant for inactivation, respectively: In Eqs (4), (6) and (7), VxhNaP and kxNaP for x ∈ {m, h, τ) represent the gating variable’s half-activation voltage and slope, respectively. The values of all parameters used are provided in Table 1. The output of each neuron (center) describing its activity (normalized firing rate) is represented by a piecewise linear function f (V), changing between 0 and 1 such that: f (V) ={0, ifV<Vmin (V−Vmin) / (Vmax−Vmin), ifVmin≤V<Vmax1, ifV≥Vmax (8) where Vmin and Vmax define the voltages at which threshold and saturation are reached, respectively. The persistent (slowly inactivating) sodium current, INaP, provides a neuron with intrinsic rhythmicity. The behavior of this neuron can be considered using a (V, hNaP) phase plane (Fig 2A). The large time constant of INaP inactivation makes the time scales of the two dynamical variables, V and hNaP, substantially different which allows us to apply the fast-slow decomposition technique for qualitative analysis of neuronal behavior. In the phase plane shown in Fig 2A, V- and hNaP-nullclines (red and black, respectively) are calculated by setting the right hand side of Eqs (1) and (6) equal to zero. EL was increased to mimic higher levels of excitation and this was reflected by a downward movement of the V-nullclines on the phase plane. The band of nullclines where intrinsic bursting activity occurs is shaded gray. Within this area the hNaP-nullcline intersects with the V-nullcline at a point between the V-nullcline knees. This corresponds to an unstable critical point and allows for the emergence of a stable limit cycle. This limit cycle is shown in the (V, hNaP) phase plane by a neuron’s trajectory with the direction indicated by arrows. The neuron’s image point moves slowly along the left and right segments of the V-nullcline (slow timescale, silent and active phases) and quickly jumps between them (fast timescale, switching between phases) showing the behavior typical for a relaxation oscillator. When a neuron is silent its image point travels up the left branch of the V-nullcline until reaching the left knee causing it to jump toward the right branch of the V-nullcline in a movement representing activation of a neuron. Once the neuron is active, the corresponding trajectory will travel down the right branch of the V-nullcline until it reaches the right knee where it will then jump back to the left branch and the neuron will once again be silent. Based on the above description, each isolated neuron can be in one of three regimes: silent, bursting (oscillatory), and tonic (constant activity) (Fig 2A) which depends on the level of neuronal excitation that, in the case of an isolated neuron, is defined by the leak reversal potential, EL. As shown in Fig 2B, bursting in the isolated neuron occurs when EL ∈ (−62. 7, −54. 2) mV (gray area). When EL is less than -62. 7 mV, the hNaP-nullcline in Fig 2A intersects with the V-nullcline at a position to the left of its left knee (all nullclines that exist in the area marked" silence" ). This creates a stable critical point on the inactive (left) branch of the V-nullcline where the neuron' s position represents an inactive (silent) state. When EL exceeds -54. 2 mV, the V-nullcline in Fig 2A loses its cubic-like shape. This creates a stable critical point on the V-nullcline (all nullclines that exist in the area marked" tonic" in Fig 2A) and the neuron therefore demonstrates a constant level of activity, which is referred to as tonically active. Therefore, with an increase of excitation defined by EL the neuron can potentially go through silent to bursting and then to tonically active regimes. The model considered here consists of two (left, L, and right, R) bilaterally connected rhythm generators, each of which represents one flexor (F) and extensor (E) center (Fig 1). As described above, each center can be in silent, bursting, or tonic regimes, depending on the level of its excitation. However, we assume that, at least in the conditions of rhythmic bursting (fictive locomotion) evoked in the isolated rodent spinal cord by neuroactive drugs, the rhythm generators are asymmetric so that only the flexor centers (left, LF, and right, RF) operate in the bursting mode, whereas the extensor centers (left, LE, and right, RE), if isolated, operate in the regime of tonic activity (this issue is specifically addressed in the Discussion). The excitabilities of the flexor and extensor centers (defined by their leak reversal potentials, ELF and ELE, respectively) in the model are regulated by a parameter α. By introducing this parameter we did not intend to perform a biologically realistic simulation of the cellular mechanisms responsible for NMDA-induced bursting in spinal interneurons, which are not well understood. Instead, this parameter was used to simulate the general effect of NMDA, whose application in the isolated spinal cord preparations leads to an increase in neuronal activity and locomotor frequency [37,40]. To simulate the effect of the drug on the excitabilities of both centers, we suggest that: ELF=ELFO⋅ (1−βE⋅α); (9) ELE=ELEO⋅ (1−βE⋅α), (10) where ELFO and ELEO define the basal values of the leak reversal potentials of the corresponding centers (at α = 0), and βE defines the rate of their increase with α. To introduce the asymmetry suggested above, we set ELFO close to the border between silence and bursting and ELEO slightly above the bursting area (Fig 2B) so that with increase of α (in the uncoupled case), the flexor centers would always operate in the bursting mode, whereas the extensor center would be in the tonically active regime. The bilaterally symmetrical network of four interacting centers with their interconnections (Fig 1B) can be described as follows: C⋅V˙i=−INaPi−ILi−ISynEi−ISynIi; (11) INaP=g¯NaP⋅mNaP∞ (V) ⋅hNaP⋅ (V−ENa); (12) ILi=g¯L⋅ (Vi−ELi), (13) where i (i = 1,2, 3,4) is the index corresponding to the center’s number as shown in Fig 1B. Currents ISynEi and ISynIi define, respectively, the excitatory and inhibitory synaptic interactions between the centers: ISynE1= (a3⋅f (V3) +aV⋅f (V4) ) ⋅g¯SynE⋅ (V1−ESynE); (14) ISynE3= (a3⋅f (V1) +aV⋅f (V2) ) ⋅g¯SynE⋅ (V3−ESynE); (15) ISynI1= (aD⋅f (V3) +b2⋅f (V2) ) ⋅g¯SynI⋅ (V1−ESynI); (16) ISynI2=b1⋅f (V1) ⋅g¯SynI⋅ (V2−ESynI); (17) ISynI3= (aD⋅f (V1) +b2⋅f (V4) ) ⋅g¯SynI⋅ (V3−ESynI); (18) ISynI4=b1⋅f (V3) ⋅g¯SynI⋅ (V4−ESynI), (19) where g¯SynE and g¯SynI are the maximal conductances of excitatory and inhibitory synaptic channels, ESynE and ESynI are the reversal potentials of these channels, and aD, aV, a3, b1, and b2 define the synaptic weights of, respectively, V0D, V0V, V3, In1, and In2 inputs to the centers in accordance with their interactions shown in Fig 2B. An increase in NMDA in the spinal cord preparation affects not only the level of excitation of neurons representing the centers, but also the level of excitation and recruitment of CINs. To take this into account in our simplified model, in which CINs are not explicitly included and the pathways they mediate are replaced by direct synaptic connections (see Fig 1B vs. 1A), we need to make synaptic weights of these connections to increase with the NMDA concentration represented by the parameter α. This especially concerns the excitatory commissural pathways that are amplified by neuron recruitment via excitatory interactions within each population and may also involve other excitatory neurons, such as the V2a, which are also recruited and activated with an increase in the NMDA concentration and locomotor speed and mediate inputs to the excitatory V0v CINs [37,57,58]. In other words, the increase of synaptic weights of CIN pathways was included in our simplified model to compensate for removal of CIN neurons whose excitability should also be affected by the drug. Therefore, we suggested that: a3=a3O⋅ (1+βa⋅α); (20) aV=aVO⋅ (1+βa⋅α), (21) where aVO and a3O define the basal weights of the corresponding excitatory connections and βa defines the rate of their increase with α. All model parameters are specified in Table 1. Synaptic weights were first set by manual adjustment to match experimental recordings. Weights were then optimized by running several iterations of one-dimensional bifurcation diagrams (see next section), where the robustness of a given regime was assessed by the width of the parameter range across which it displayed proper behavior, i. e. matching experimental phase diagrams and flexor-flexor phase relationships. Simulations were performed using custom C++ programs and visualized using gnuplot. To inspect changes in the model behavior with changing neuronal excitation (defined by the parameter α), phase differences between the left (LF) and right (RF) flexor activities or between ipsilateral flexor (LF) and extensor (LE) activities were calculated as the time differences between onsets of the corresponding bursts divided by the oscillation period measured in the activity of the flexor center. In the chosen range, α ∈ [0,1. 2] was progressively increased or decreased in a step-wise manner with a step size comprising 1/1000 of the range. At each point, the simulation was run and the phase differences were computed. Moreover, at every time step the initial conditions were chosen as a final state of the system from the previous time step to minimize a transient period. After convergence of the phase difference to a steady state with preset accuracy its value was plotted versus the parameter value. For each bifurcation diagram two series were generated: one with α increasing from 0 to 1. 2 in the described manner and another with α decreasing from 1. 2 to 0. The qualitative changes in system behavior (bifurcations) can be seen on such diagrams as jumps, corners, branching and other types of discontinuities. In addition, the existence of non-overlapping branches obtained by changing the parameter in different directions serves as an unequivocal indication of bi- or multi-stability in the system in the corresponding parameter ranges. In addition to the phase differences, the values of amplitude and instantaneous frequency (reciprocal to the oscillation period) were also calculated and the corresponding plots were built. Furthermore, the frequency of flexor activity was used to construct diagrams where phase differences between left (LF) and right (RF) flexors were plotted against the frequency. The experimental data included for comparison with modeling results had been obtained in earlier experimental studies published in Nature [37]. No new animal data were collected. All experiments were approved by the local ethical committee and performed in accordance with European guidelines for the care and use of laboratory animals. Briefly, the experiments were performed in the isolated spinal cords of wild-type (control) and transgenic mice with ablated V0D or V0V or both V0 neuron types (the detailed description of the transgenic lines of mice used can be found in [37]. Mice aged E18. 5 (with genetically deleted V0D neurons) or newborn mice aged 1–2 days (for all other studies) were used. The isolated spinal cords in a chamber were perfused with normal Ringer’s solution. Locomotor-like activity was induced by the exogenous application of mixtures of serotonin (or 5-hydroxytryptamine, 5-HT) and NMDA (N-methyl-D-aspartate). The locomotor frequency was controlled by concentrations of locomotor-inducing drugs (mostly NMDA) [37,40]. All recordings were performed at room temperature (22–24°C). Locomotor activity was recorded with suction electrodes attached to the L2 and L5 lumbar roots on both sides of the cord. The raw activity was band-pass filtered at 100 Hz to 1 kHz. Data points for analyzing cycle periods and phases were taken after the locomotor activity had stabilized 10–15 min after the initial burst of activity. All details of recordings and data processing can be found in [37]. The performance of the intact network is shown in Fig 3. Panel A of the figure shows the changes in the output activity, f (V), of all four centers with α changing from 0 to 1. 2. The vertical dashed lines in this panel indicate that activities of left (LF) and right (RF) flexor centers alternate at all values of α. Panel B shows how the frequency of oscillations (top diagram), the amplitude of flexor activity (second diagram), and the phase differences between the activities of left and right flexor centers (LF-RF) and left flexor and left extensor centers (F-E) (two bottom diagrams) changed with α. Note that in contrast to panel A, panel B shows α being changed in both directions, first forward from 0 to 1. 2 (red) and then backward from 1. 2 to 0 (blue). However, in all diagrams in Fig 3B, the graphs for forward and backward changes of α fully overlapped. Fig 3A and two top diagrams in Fig 3B show that with increasing α (simulating an increase of neuronal excitation in the experimental preparations by the administration of NMDA) the amplitude of flexor center activity in the model monotonically decreases and the locomotor frequency monotonically increases, which generally fits experimental data [40]. The bifurcation diagram depicting the phase difference between flexor centers in the intact model (Fig 3B, Phase diff. LF-RF) shows two intervals where the model exhibits qualitatively different behavior. On the right (α > 0. 55) flexor centers demonstrate perfect anti-phase synchronization (Δφ = 0. 5). This represents a symmetric alternation between the activities of left and right flexor centers that is in contrast to the behavior seen at lower values of α (α > 0. 55) where the phase difference between the alternating flexor center activities is not 0. 5 and a pitchfork bifurcation occurs at α ≈ 0. 55. The branches of the pitchfork reflect direct inhibitory interactions between the flexor centers (rebound) and differ depending on which left or right flexor center activates first. Moreover, the supercritical pitchfork bifurcation results in instability of the symmetric anti-phase regime with a decrease in α. The existence of two symmetric branches both emerging from Δφ = 0. 5, one with Δφ > 0. 5 and the other with Δφ < 0. 5, is concerned with the symmetry of the underlying network. It is worth mentioning that a slightly non-symmetric network would favor one of these two branches and weaken the other. In this case, the pitchfork bifurcation would be replaced with a saddle-node bifurcation resulting in the emergence of the “weak” branch, while the “strong” branch would replace the perfect anti-phase regime. Accordingly, the major qualitative change in the system dynamics as α is decreased below the bifurcation value of α ≈ 0. 55 is concerned with the appearance of two distinct stable regimes with phase shift between flexor center activities greater and less than 0. 5. It is important to note that at high values of α the periods of flexor and extensor activity have the same duration, while at low values of α the flexor phases can be significantly shorter than the extensor ones (see left parts of the graphs in Fig 3A). However, although the activity of left and right flexor centers in this case is not symmetric (does not alternate in perfect anti-phase), the left and right flexor centers are never active at the same time. Therefore, at any value of α (and at any locomotor frequency, see the top diagram in Fig 3B), the intact network exhibits left-right alternation of flexor activity. After both V0 commissural connections are eliminated, the interactions between left and right centers are provided exclusively by the excitatory V3 pathways (see Fig 1). As shown in Fig 4 (panel A and two top diagrams in panel B), similar to the intact model the amplitude of flexor activity monotonically decreases with increasing α, whereas the locomotor frequency monotonically increases. In contrast to the intact case, the activities of left and right centers are fully synchronized at all values of α. This is seen in Fig 4, panel A (indicated by vertical dashed lines) and panel B (the phase difference between the activities of left and right flexor centers remains 0 or 1 for all values of α - see “Phase dif. LF-RF” diagram). This left-right synchronous activity corresponds to hopping behavior. Selective deletion of the V0V subtype of V0 CINs corresponds, in our model, to elimination of V0V excitatory connections from extensor centers to the contralateral flexor centers (see Fig 1). In this case, the left and right sub-networks interact only through the excitatory (V3) and inhibitory (V0D) pathways between the flexor centers. Independent of commissural connections and similar to the intact case, the amplitude of flexor activity monotonically decreases with increasing α, whereas the locomotor frequency monotonically increases (see Fig 5, panel A and two top diagrams in panel B). Depending on the balance between the excitatory V3 and inhibitory V0D pathways, the net interactions can be excitatory (if the excitatory connections prevail) or inhibitory (if mutual inhibition is stronger than excitation). In our model, a pharmacologically-induced increase in excitation is implemented as an increase in the leak reversal potentials in the centers as well as an increase in the weights of excitatory pathways (to account for drug effects on neurons involved in these pathways). Based on our suggestions, the model parameters are set in such a way that at low levels of excitation (at low values of α) the inhibitory connections between flexor centers dominate, while at high excitation (at high values of α) the net interaction becomes excitatory. As a result, the model demonstrates left-right alternation of flexor activity at low values of α and left-right synchronization at high values of α (both indicated by vertical dashed lines in Fig 5A). The bifurcation diagram in Fig 5B (see “Phase dif. LF-RF” diagram) shows that at extremely low levels of excitation (α < X1) the activities of flexor centers alternate similarly to the intact case (compare the two branches left of X1 in “Phase dif. LF-RF” diagram in Fig 5B with the corresponding diagram in Fig 3B). At very high excitation levels (α > X5) left-right synchronous activity occurs similarly to the case when all V0 pathways are removed (compare two blue branches in the right part of “Phase dif. LF-RF” diagram in Fig 5B with the corresponding panel in Fig 4B). For intermediate excitation levels (X1 < α < X5) the system demonstrates a complex transition scenario from alternating to synchronous activity that is explained below using the technique of fast-slow decomposition. As seen in Fig 2A, the left knee on the V-nullcline is affected by changes in excitation to a much greater extent than the right knee (compare, for example, the width of the shaded area in terms of hNaP values at V = –55mV and at V = –30mV). Therefore, the activity of each flexor center can be affected by a synaptic input during a period of its inactivity, i. e. during the ipsilateral extensor phase, more efficiently than during its active phase. Due to intrinsic bursting properties each flexor center is capable of endogenous escape and inactivation. Accordingly, two types of strong synchronizing events are possible: a release of one flexor center on an escape of the other (release-on-escape) and a release of one flexor center when the other deactivates (release-on-shutdown). The former mechanism creates (in-phase) synchronization between flexor centers (“hopping”), while the latter underlies their alternation (anti-phase synchronization). Fig 6A–6C illustrates and allows for a more comprehensive understanding of system behavior after V0V pathways are deleted. Removal of V0D corresponds to termination of reciprocal inhibition between the flexor centers (see Fig 1). Since V0V-dependent connections remain intact, the network has two competing mechanisms of left-right coordination. One of them concerns reciprocal excitation between the flexor centers provided by V3 connections, which tends to synchronize their activity. The other is characterized by excitation from contralateral extensor centers to each flexor center, which contributes to alternation of flexor center activities. The results of simulations in this case are shown in Fig 7A and 7B. Again, similar to the intact case, the amplitude of flexor activity monotonically decreases with increasing α, whereas the locomotor frequency monotonically increases (see panel A and two top diagrams in panel B). In contrast to the previous case, when V0V pathways were removed, the model now demonstrates left-right synchronization of flexor center activity at low values of α and left-right alternation at high values of α (indicated by vertical dashed lines in Fig 7A). The transition scenario from synchronization to alternation is best seen in “Phase dif. LF-RF” diagram in Fig 7B. The above analysis of our model is centered on two major regimes characterized by left-right alternation and left-right synchronization (hopping). We have shown that the operating regime depends on the network integrity (presence or lack of particular commissural interactions), level of excitation in the network, and initial conditions. The general conclusion from this analysis is that is in the intact system, the left-right alternation is secured by the V0D CIN pathways at low excitabilities/frequencies and by V0V CIN pathways at high excitation/frequencies. The regimes of operation and patterns exhibited are summarized in Table 2 below: In the following section we provide a more general description of the network to elucidate the critical elements of the model that lead to the synchronization properties described above. In the model proposed the connections between neurons are relatively weak. In this case, the dynamics of the system may be described in terms of phase synchronization. This approximation implies that the trajectory of each oscillator in its own phase subspace is perturbed negligibly by the interactions between the oscillators, and hence the state of each oscillator can be described by a single variable, phase. Accordingly, taking into account the left-right symmetry of the system, the full set of differential equations describing the system can be reduced as follows in the approximation of asymptotically small connections (see, for example, [59]): ϕ˙1=Ω+F (ϕ1, ϕ2); ϕ˙2=Ω+F (ϕ2, ϕ1), (22) where ϕ1 and ϕ2 are the phases of the oscillators, Ω is an oscillation frequency and the function F (. , .) describes interactions between oscillators. For the phase difference ϕ = ϕ2 − ϕ1 we will have Hypothesizing that the right hand side of this equation depends only on the phase difference, i. e. F (ϕ2, ϕ1) − F (ϕ1, ϕ2) = G (ϕ), allows Eq (23) to be rewritten in a simpler form: The synchronized regimes of this system are the fixed (equilibrium) points of this first order differential equation (i. e. solutions of the equation G (ϕ) = 0) and their stability is defined by the sign of the first derivative dG (ϕ) / dϕ at the equilibrium. Specifically, the point is stable if the first derivative dG (ϕ) / dϕ is negative and the point is unstable if the derivative is positive. If we introduce a potential function as P (ϕ) = −∫G (ϕ) dϕ, then stable equilibrium points in Eq (24) will be local minima (valleys) of the function P (ϕ), and unstable points will be local maxima (hills) of P (ϕ). Because G (ϕ) is a 2π-periodic function that is odd, i. e. maintaining symmetry about the origin such that –G (φ) = G (−φ), its Fourier series expansion may be represented as: Restricting the expansion to the first two terms gives the following equation: This equation always has at least two fixed points at ϕ = 0 and ϕ = π representing synchronization and alternation regimes, respectively. The former is stable whenever A – 2B < 0, and the latter is stable if A + 2B > 0. Accordingly, for A < −2B only the synchronization regime ϕ = 0 is stable, for A > 2B only the alternation regime ϕ = π is stable, and between these values for −2B < A < 2B both regimes are stable. This inequality has solutions only if B > 0, so the second term in the expansion (Eq (26) ) with positive coefficient B is responsible for the existence of bistability in the system. Another qualitative conclusion from these speculations is that greater values of the parameter A correspond to “less stable” synchronization and “more stable” alternation in a manner referring to these behavior' s basins of attraction. Conversely, smaller values of A make synchronization “more stable” and alternation “less stable”. It is straightforward to assume that excitatory connections between oscillators contribute to a decrease in A, while inhibitory connections increase A. In the system under consideration, V3 and V0D commissural pathways mediate the direct excitatory and inhibitory connections between the flexor centers, respectively, and V0V pathways represent excitatory connections from the contralateral extensor to each flexor center, and hence can be considered as effectively inhibitory pathways between the flexor centers. Accordingly, the parameter A can be constructed from synaptic weights of these connections as A = −V3 + V0V +V0D, where V3, V0V, and V0D represent the strengths of the corresponding pathways. Assuming that all considered pathways include interneurons whose excitability is affected by NMDA, we can suggest that the strength of these pathways linearly increases with α: Accordingly, we can rewrite an expression for A as: In these terms the bifurcation scenarios described above allow for an elegant qualitative interpretation (see Fig 9A–9D). In the case when both subtypes of V0 pathways are deleted V0V0=V0D0=0, k0V = k0D = 0, and the parameter A=−V30−α⋅k3 is always negative and becomes even more negative with increasing α due to the negative slope. Accordingly, the synchronization regime remains stable for all values of α and the alternation regime is always unstable (see Fig 9D). Let’s now consider the case when only V0V pathways are deleted. In this case V0V0=0 and k0V = 0. Therefore, A=−V30+V0D0+α⋅ (−k3+k0D). If we assume that the connections mediated by inhibitory V0D pathways have a much stronger basal component (V0D0>V30), but weaker dependence on α than the excitatory V3 pathways (k0D < k3), then the baseline activity of the system (with small α) will be the opposite of the case when both V0 pathways are deleted since A can now be positive and large enough. With an increase in α the α-dependent term will eventually prevail and we will get the same situation as when all V0 pathways are deleted. In Fig 9B this scenario can be seen on the bifurcation diagram shown at the bottom on the (α, ϕ) -plane. We start, when moving from smaller to larger values of α, with the synchronization regime unstable and alternation regime stable. Then the fixed points ϕ = 0,2π become stable through the supercritical pitchfork bifurcation and new unstable fixed points appear which now separate the attraction basins of synchronization and alternation regimes (shown by dashed blue line). As α further increases these points move towards, and eventually merge to, ϕ = π making it unstable. On the surface described by the potential function, P (ϕ), this scenario manifests itself by the valley at ϕ = π (Δφ = 0. 5) where the surface becomes progressively more shallow and finally becomes a hill (see α close to 5). With V0D deleted, the expression (28) takes the form: A=−V30+VV0+α⋅ (−k3+k0V). Here we make an opposite assumption, i. e. V0V interactions have a smaller basal component than V3 pathways, i. e. V0V0<V30, but stronger α-dependent component, i. e. k0V > k3. Accordingly, the baseline in this case coincides with the case when both V0 pathways are deleted, i. e. we see stable synchronous and unstable alternation regimes at low values of α (see Fig 9C). As α increases the parameter A decreases due to its negative slope. This ultimately results in switching to the stable synchronization regime. Qualitatively this scenario is similar to the previous one but reversed (see the bifurcation diagrams on the (α, ϕ) -plane in Fig 9C). Finally, in the intact system both the intercept and the slope in Eq (28) are positive. The former is due to the fact that V0D0>V30, and the latter is because k0V > k3. Accordingly, A is always positive and only the alternating (out-of-phase synchronization) regime when ϕ = π (Δφ = 0. 5) is stable (see Fig 9A). In summary, this simplified model offers an explanation for why elimination of functionally similar interactions provided by V0V and V0D commissural pathways has such dramatically different effects. Based on our analysis, this may happen if the excitatory interactions between left and right rhythm generators are more dependent on (and change with) general neuronal excitation in the network than the inhibitory interactions. One of the most important characteristics of locomotion is the relative change in the flexor and extensor phase durations with changes in the locomotor speed or step cycle period [60,61]. Fig 10A shows how the durations of both phases in our intact model change with an increase in the locomotor period (slowing down locomotor oscillations). These data demonstrate clearly asymmetric changes in phase durations with variations of the step cycle, so that changes in the duration of the flexor phase are significantly less than changes in the duration of the extensor phase. This means that an increase in locomotor frequency mainly occurs due to shortening the extensor phase. Fig 10B shows a similar diagram representing the changes in flexor and extensor phase durations vs. step cycle period built using the recordings from wild-type mouse spinal cord preparations. One can see that the asymmetric changes in phase durations in experimental studies are qualitatively similar to those in our simulations. The main objective of this theoretical study was to formulate and analyze a simplified model of the spinal circuitry in order to reproduce and explain recent experimental data concerning frequency-dependent differential contributions of two distinct commissural pathways to the left-right alternation of neural activity in the isolated spinal cord [37]. We used a simplified connectivity scheme adapted from both the experimental data [37] and the previous large-scale computer models [41]. Bifurcation diagrams were constructed for each experimental case to observe the system’s behavior over a large range of excitability (Figs 3–5 and 7). These diagrams allowed us to determine the values of parameters leading to the observed behaviors and provide a strong mathematical explanation of the critical changes seen in left-right phase relationships leading to a normal alternating pattern and to switching to synchronous (hopping) behavior. Moreover, the bifurcation diagrams revealed regions of bistability, where two stable states co-existed for a given set of parameters. To complete our understanding of the dynamics underpinning the transitions, we used fast-slow decomposition methods and generated phase plane diagrams that corresponded to removal of V0V or V0D commissural pathways (Figs 6 and 8). Our analysis showed that left-right alternation and synchronization occurred in both cases when one of the pathways (V0 or V0D) was removed, because of the release-on-shutdown and release-on-escape mechanisms, respectively. These findings allowed the interpretation of experimental results in terms of qualitative theory of dynamical systems. In addition, at the intermediate locomotor frequencies, our model exhibited bistable behaviors and predicted a coexistence of both alternating and synchronized regimes, leading to an intermittent alternating/synchronous activity observed in experimental studies (Fig 15). The exact intrinsic cellular mechanisms involved in rhythmic bursting in the spinal cord remain unknown. Previous modeling studies suggested that these mechanisms may involve the persistent (or slowly inactivating) sodium current, INaP, [9,41–46,51,52] and this suggestion has been supported by several experimental studies [52–56]. Although we used the INaP-dependent mechanism in this model, the results of this study concerning network behavior can be considered independent of the exact cellular mechanism employed. At the same time, it should be noted that some particular features of the model fit to known experimental data specifically due to the INaP—dependent mechanism incorporated (although they can be provided by other cellular/network mechanisms). These features include (a) a monotonic increase of burst frequency and (b) monotonic decrease of (flexor) burst amplitude with an increase of neuronal excitation (by increasing NMDA or glutamate concentration, see [37,40] and Figs 3–5 and 7), as well as (c) small changes in the flexor phase duration relative to the extensor phase duration at low frequencies (the same figures). These observations provide additional indirect support for the idea that INaP—dependent mechanisms are involved in the generation of bursting activity in the spinal cord. A recent study using optogenetics [62] has demonstrated that locomotor-like rhythmic bursting can be induced independently in flexor and extensor networks. This means that both flexor and extensor populations can potentially, under certain conditions, generate endogenous locomotor-like, oscillatory activity. However these findings do not necessary imply that both these populations are in the bursting mode under the conditions of isolated spinal cord preparation with locomotor oscillations induced by drugs or other methods. Moreover, analysis of non-resetting deletions (missing bursts) in this preparation showed that missing flexor bursts were always accompanied by a sustained (tonic) extensor activity, whereas missing extensor bursts occurred without an effect on flexor bursting [45]. Finally, the duration of extensor activity between flexor bursts at low burst frequency can be pretty long, exceeding 5–6 seconds when flexor burst duration is usually less than 2. 5 seconds [37], which also implies that intrinsic bursting is unlikely to occur in rodent extensors. Based on the observations described above, we suggest that each flexor or extensor center can potentially, in certain conditions, intrinsically generate bursting activity. However, we think that under the conditions considered here (i. e. , the isolated spinal cord with 5-HT and NMDA-induced locomotor-like oscillations) only the flexor centers (left and right) operate in a bursting mode whereas the extensor centers (left and right) operate in a state of tonic activity (if isolated) and exhibit rhythmic bursting because of rhythmic inhibitory inputs from the corresponding ipsilateral flexor center (see the section “Rhythm-generating Centers and Connections”). This may occur because the mixture of 5-HT and NMDA used to evoke locomotor activity in the isolated spinal cord, excites the extensor centers much stronger than the flexor centers (and hence ELFO < ELEO, see Fig 2B). The above suggestion leads to the concept that each (left and right) CPG is asymmetric, so that the flexor center on each side differs from the extensor center not only by a level of excitation described above, but also by the organization of left-right and flexor-extensor interactions (see Fig 1B and weights of interactions in Table 1). As a result of this flexor-extensor asymmetry, the burst frequency and the flexor phase duration are mostly defined by the flexor centers, and hence at low frequencies the duration of the flexor phase may be significantly less than the duration of the extensor phase. Fig 10A shows that when frequency goes down (with decreasing α) the increase of step cycle period mainly results from an increase of the extensor phase duration with a very small increase of the flexor phase duration. Interestingly, this phase durations vs. period plot built using simulation data fits well to the plot built using data from experimental studies [37] (Fig 10B). This provides additional support for the idea of an asymmetric CPG employed in our study. The concept of an asymmetric CPG organization with a dominant role of flexor centers in rhythm generation has been previously suggested by Duysens and Pearson [63–66] and implemented in recent computational models [41,45]. Such asymmetric flexor-extensor phase relationships are consistent with the previous experimental studies in cats [60] and support the idea that this flexor-extensor asymmetry is an inherent property of the CPG [61] rather than a result of afferent influences on a quasi-symmetric CPG [67]. Our simulations support the earlier suggestion that left-right alternation is provided by two commissural pathways involving the inhibitory V0D and excitatory V0V CINs [37]. Similar to the experimental [37] and previous modelling [41] studies, our simulations demonstrate that the contribution to left-right alternation of the V0D pathway is dominant at low frequencies and reduces as locomotor frequency increases, whereas the contribution of the excitatory V0V pathway is weak at low frequencies but becomes stronger as frequency increases. The analysis of our models suggests that the different frequency-dependent roles of these pathways are based on the following: (1) a relatively weak dependence of V0D CIN pathways on the neuronal excitability (neuroactive drug concentration), which leads to a net reduction of V0D activity due to the reduction of amplitude of rhythm-generating activity when the oscillation frequency increases; (2) a strong dependence of V0V CIN pathways on the neuronal excitability, leading to the increase of their activity with oscillation frequency. This may be based on the existing data that inputs to V0V CINs are often mediated by ipsilaterally-projecting excitatory V2a neurons [37,68,69] These were shown to increase their activity and recruitment with an increase in neuroactive drug concentration [58,70] and their selective ablation mimics the V0V ablation [70]. Our study support the concept that the resultant locomotor pattern and locomotor gait depend on the balance between different commissural interactions, which, in turn, depends on the level of neuronal excitation and locomotor speed dictated by the locomotor task. This conclusion is consistent with a general concept that the expression of a particular left-right coordinated pattern, such as the left-right alternation of activity typical for regular walking or the left-right synchronized activity characteristic for hopping or galloping, depends on the balance between the commissural pathways working for or against synchronization of left and right spinal CPGs. Therefore, under normal conditions this pattern and the corresponding gait can be changed by additional speed-dependent or speed-independent descending or afferent signals to different commissural interneuron populations involved in these pathways.

Movements of left and right limbs in mammals during locomotion are controlled by distinct rhythm-generating neuronal circuits in the spinal cord. Complex interactions between these circuits provide flexible coordination of limb movements in different gaits. It was shown that interactions between left and right spinal circuits are mediated by commissural interneurons. Genetic ablation of a particular type of these interneurons, called V0, leads to switching from a regular, left-right alternating “walking” activity to a left-right synchronous “hopping” pattern. Moreover, the V0 commissural interneurons have excitatory and inhibitory subtypes that appear to play different roles in the left-right coordination depending on locomotor speed. In this theoretical study, we build a simplified mathematical model of spinal circuits that describes left and right rhythm generators interacting bilaterally via several types of commissural connections. Using this model, we simulate different experimental manipulations, analyze the resultant alternating and synchronous regimes of activity, and propose explanations for the results of experimental studies. We show that although both excitatory and inhibitory V0 commissural pathways support left-right alternation, the resultant locomotor pattern and gait depend on the balance between different commissural interactions, which in turn may depend on the level of neuronal excitation and locomotor speed.

Abstract Introduction Methods Results Discussion

2015

Mechanisms of Left-Right Coordination in Mammalian Locomotor Pattern Generation Circuits: A Mathematical Modeling View

Buruli Ulcer is a tropical skin disease caused by Mycobacterium ulcerans, which, due to scarring and contractures can lead to stigma and functional limitations. However, recent advances in treatment, combined with increased public health efforts have the potential to significantly improve disease outcome. To study the Quality of Life (QoL) of former Buruli Ulcer patients who, in the context of a randomized controlled trial, reported early with small lesions (cross-sectional diameter <10 cm), and received a full course of antibiotic treatment. 127 Participants of the BURULICO drug trial in Ghana were revisited. All former patients aged 16 or older completed the Dermatology Life Quality Index (DLQI) and the abbreviated World Health Organization Quality of Life scale (WHOQOL-BREF). The WHOQOL-BREF was also administered to 82 matched healthy controls. Those younger than 16 completed the Childrens' Dermatology Life Quality Index (CDLQI) only. The median (Inter Quartile Range) score on the DLQI was 0 (0–4), indicating good QoL. 85% of former patients indicated no effect, or only a small effect of the disease on their current life. Former patients also indicated good QoL on the physical and psychological domains of the WHOQOL-BREF, and scored significantly higher than healthy controls on these domains. There was a weak correlation between the DLQI and scar size (ρ = 0. 32; p<0. 001). BU patients who report early with small lesions and receive 8 weeks of antimicrobial therapy have a good QoL at long-term follow-up. These findings contrast with the debilitating sequelae often reported in BU, and highlight the importance of early case detection. Buruli ulcer (BU) is listed by the World Health Organization (WHO) as a neglected tropical disease, caused by infection with Mycobacterium ulcerans. Although the disease has been reported from as many as 30 countries around the world, it is currently most common in West and Central Africa, and it predominantly affects the rural poor. Typically, the disease starts with a small, painless nodule that progresses into a large necrotizing ulcer over the course of several weeks. After treatment, although the ulcer usually heals, there is a high risk of significant scarring, contractures and functional limitations [1]–[3]. In the socio-economic context of rural Africa, functional limitations and stigmatizing scars can have severe consequences. In a study of 638 former BU patients in Ghana and Benin, 57% appeared to have a functional limitation, and school dropout, financial difficulties and job loss were frequent consequences of the disease [2]. People in endemic communities sometimes perceive the disease to be caused by a curse or witchcraft, and the resulting stigma can cause social isolation and problems with finding work or a spouse [4]–[6]. Although the treatment for BU is free of charge, the costs associated with hospitalization can be devastating for the household economy and frequently cause family members to cease providing financial and social support to patients [7]. However, over the past decade, the main mode of treatment has shifted from surgery to antibiotics, with high rates of cure [8]. In addition, significant public health efforts have been directed at detecting BU at an early stage and educating affected communities about the disease. Conceptually, both factors combined should reduce scarring, contractures, and stigma, and improve the subsequent quality of life (QoL) of former patients. To our knowledge, the QoL of former BU patients has not yet been studied. In leprosy and podoconiosis, similarly deforming and stigmatizing skin conditions occurring in the tropics, several studies on QoL have been conducted in Bangladesh, Brazil, China, Ethiopia, Ghana, and India [9]–[15]. In general, these studies show that patients report a substantially lower QoL than controls [9]–[11]. There appears to be some relationship between QoL and the severity of the disease in terms of number of lesions, functional limitations, stigma, and deformities [11]–[15]. Studies from Ethiopia in podoconiosis patients, a disabling and stigmatizing geochemical elephantiasis of the foot, found that a dermatology-specific QoL instrument was valid and sensitive to therapeutic change [16], [17]. These studies demonstrate that measuring QoL with standardized questionnaires in rural Africans suffering from skin diseases is feasible. Moreover, as both leprosy and podoconiosis were shown to have a significant impact on QoL they show that measuring QoL in these populations is warranted. In the current study, we report on the disease-specific and general QoL of former BU patients who previously presented with small, early ulcers and who were treated with a full course of 8 weeks of antibiotics. Our study subjects were former BU patients that had earlier participated in the BURULICO trial, conducted between 2006 and 2009 in Ghana, registered with number NCT00321178 at clinicaltrials. gov. For that trial, patients aged 5 years or older, clinically diagnosed with early (duration <6 months), limited (cross-sectional diameter of induration <10 cm, including plaques and oedemas) M. ulcerans infection were included, and randomized to receive either 8 weeks of streptomycin at 15 mg/kg daily (max 1000 mg daily) and rifampicin at 10 mg/kg daily (max 600 mg daily), or 4 weeks of streptomycin and rifampicin, followed by 4 weeks of rifampicin and clarithromycin at 7. 5 mg/kg daily. The rate of healing did not differ between both arms. Patients had a median age of 12 and 30% were male [8]. For the present follow-up study, participants were traced between June and November 2012 by visiting their last known village or through telephone contact if available. If the former patient was no longer living at the last known village, neighbors, relatives, and community leaders were asked for additional information. When a former patient was located, he or she was informed about the study, given time to consider participation, and asked for consent. The Cardiff Dermatologic Life Quality Index (DLQI) and its pediatric adaptation the Childrens Dermatologic Life Quality Index (CDLQI) are dermatology-specific QoL instruments [18], [19]. Both contain 10 questions with scores on a question ranging from 0 to 3. The total score is calculated by summing the score of each question resulting in a maximum of 30 and a minimum of 0. The higher the score, the more QoL is impaired. In addition, to facilitate interpretation, banding scores are available for both the DLQI and CDLQI, with a score of 0 or 1 indicating no effect, a score between 2 and 5 a small effect, a score between 6 and 10 a moderate effect, a score between 11 and 20 a very large effect, and a score between 21 and 30 an extremely large effect on the patients' life. The DLQI was designed for patients aged 16 or above, and the CDLQI for those aged between 4 and 16. Both the DLQI and CDLQI have been extensively validated, but only the DLQI has been used in low and middle Income countries, including sub-Sahara Africa [9], [16], [20], [21]. Both the DLQI and CDLQI were translated into the local language, Twii, according to the instructions of the authors of the original questionnaires. Two independent translators separately translated the questionnaire from English into Twii, and discussed their translations to arrive at a single Twii translation. A third and fourth translator then independently translated the questionnaires back into English. Next, the back-translations were reviewed by the original authors of the questionnaire. After initial comments and a subsequent cycle of translation and back-translation, the questionnaire authors approved this back-translation for further use. Finally the agreed translation was pretested in a group of 8 former BU patients that did not participate in the BURULICO trial, asking them about the clarity, understandability and wording of the questions. In this pretest no further issues arose. The WHO Quality of Life-BREF (WHOQOL-BREF), is an international cross-culturally comparable generic QoL assessment instrument [22]. It assesses the individuals' perceptions in the context of their culture and value systems, and their personal goals, standards and concerns. It comprises 26 items, which measure 4 domains: physical health, psychological health, social relationships, and environment (satisfaction with one' s living conditions). In addition, two questions measure general health and general QoL. The score on each question ranges between 1 and 5, and for each domain a total score is computed that ranges between 20 and 100, with a higher score indicating a better QoL. No specific age range is given for the WHOQOL-BREF, but it was designed for adults. The WHOQOL-BREF was also translated into Twii according to the procedure outlined above for the DLQI and CDLQI, with the exception that the back translations were not reviewed by the authors of the questionnaire but by the study team. Effect sizes of the difference in WHOQOL-BREF domains between former patients and controls were calculated using the formula z/√N, where z is the z-score of the U statistic, and N is the total sample size. The Buruli Ulcer Functional Limitation Score (BUFLS) is a questionnaire that consists of questions related to 19 common daily activities of people living in endemic areas [2], [23]. Each item is scored between 0 and 2, with 0 indicating no difficulties in performing the activity compared to age- and sex matched community members, 1 indicating difficulties performing the activity, and 2 indicating that the former patient is unable to carry out the activity at all. In the calculation of the individual functional limitation score, the sum is divided by the maximum possible score for that individual, multiplied by 100%. A higher score therefore indicates more functional limitations, with a range between 0% and 100%. A score cannot be calculated if more than 6 items are not applicable. Former patients currently aged 16 or above were administered the WHOQOL-BREF, DLQI and BUFLS questionnaires, and former patients currently aged below 16 were administered the CDLQI and BUFLS only. The WHOQOL-BREF was also administered to 82 age, sex, and occupation (farming, schooling or other) matched healthy controls that lived in the same villages as the former BU patients on separate visits. All potential controls were asked whether they were currently sick or injured, and whether they were on any medication; they could only participate if they answered both questions with no. Due to high rates of illiteracy among the participants, all questionnaires were administered orally in a quiet private place by three trained local hospital staff members. In addition all patients were asked about the presence of pain (Yes/No) and itch (Yes/No), marital status, level of education and employment. Also, for all former patients the scar size was determined by tracing it on a transparent sheet, scanning it, and measuring the surface with ImageJ software. The study protocol was approved by the Committee on Human Research, Publication, and Ethics of the Kwame Nkrumah University of Science and Technology and the Komfo Anokye Teaching Hospital, Kumasi (reference number CHRPE/AP/133/12). Written and verbal informed consent or assent was obtained from all participants aged ≥12 years, and consent from parents, or legal representatives of participants aged ≤18 years. 127 individuals (84%) of the 151 former participants of the BURULICO trial were located for follow up, and none declined to participate. The median duration between drug treatment and enrolment in the current study was 5 years. Although the trial had taken place in the Ashanti region of Ghana, many former patients had moved away from the study site, and patients were retrieved in 9 of 10 Ghanaian regions, including the three Northern regions more than 700 kilometers and approximately 10 hours by road from the original study site. 68% of the retrieved former patients were female, and the median age at follow-up was 18 years. 71 former patients were age-eligible, and completed the DLQI, but due to an error the WHOQOL-BREF was not administered to 4 former patients, so data on this questionnaire were only available for 67 former patients. The remaining 56 former patients were administered the CDLQI only. There was no missing data on the DLQI and CDLQI, and a maximum of 2 missing answers per QoL domain on the WHOQOL-BREF, meaning that domain scores could be calculated for every subject. Item 21 of the WHOQOL-BREF, “how satisfied are you with your sex life? ”, caused considerable confusion among many former patients who were not married. Even after repeated explanation that one could still have an opinion about one' s sex life if you are not married (i. e. be dissatisfied with it), it was poorly understood and hence left open on 20 of the 67 questionnaires. Due to a printing error, the BUFLS was not administered to one former patient, but meaningful scores (i. e. less than 6 items not applicable) could be computed for all other patients. Of the 24 former patients not retrieved, 4 were already lost to follow-up during the BURULICO trial, 3 had moved abroad, 2 had deceased, and the fate of the remaining 15 was unknown. The former patients that were lost to follow-up did not differ significantly from those that were retrieved in terms of age, gender or treatment arm. Scores, range, and reliability for the DLQI, CDLQI, and WHOQOL-BREF subscales are shown in Table 1. The frequency distributions of the QoL scales are shown in Figure 1. Upon inspection, it appears that the psychological and environmental subscales of the WHOQOL-BREF were normally distributed. The distribution of the DLQI and CDLQI is skewed to the left. The banding scores for the DLQI and CDLQI are shown in Table 2. Former patients had a median (IQR) scar size of 4. 6 (1. 4–13. 5) cm2, and a median (IQR) BUFLS score of 2. 6% (0%–13. 3%). Only 4 (3%) former patients were unemployed, all others were either studying or working. Only 52% of former patients were married, but those not married were significantly younger (t = −2. 31; p = 0. 024). The scar was itchy in 26% and painful in 9% of former patients. Spearman' s Rho correlations between the DLQI, CDLQI and WHOQOL-BREF subscales and other continuous variables are shown in Table 3. Age was not significantly related to any of the QoL scales. Scar size correlated significantly with the DLQI (ρ = 0. 32; p<0. 01). The BUFLS correlated significantly with the environmental subscale of the WHOQOL-BREF (ρ = −0. 24; p<0. 05). Relationships between the QoL instruments and discrete variables were tested through either student T-tests or Mann-Whitney U tests depending on whether the scales were normally distributed. The QoL of women did not differ from men on any of the scales or subscales. Farmers, scored significantly higher on the social domain of QoL compared to other professions (U = 435; p = 0. 050). Those who were married also scored significantly higher on the social domain than those who were single (U = 375; p = 0. 019). Those who were ethnically Akan scored significantly higher on the DLQI (U = 332; p = 0. 026), indicating lower QoL, and significantly lower on the physical domain of QoL (U = 292; p = 0. 003), again indicating lower QoL than those from the northern tribes. None of the QoL scores differed between those that did or did not report painful or itchy scars. The DLQI correlated significantly with the physical (ρ = −0. 54; p<0. 001) and psychological (ρ = −0. 34; p<0. 001) subscales of the WHOQOL-BREF, but did not correlate at all with the social (ρ = 0. 00; p = 0. 99) and environmental (ρ = −0. 01; p = 0. 98) subscales. Scores on the WHOQOL-BREF domains of patients and healthy controls are shown in Figure 2. Former BU patients scored significantly higher than controls on the physical and environmental domains of the WHOQOL-BREF by Mann-Whitney U tests. The effect sizes for the four domains were as follows: physical 0. 29, environmental 0. 41, psychological 0. 10, and social 0. 14. This is the first study to address QoL in former BU patients. The former BU patients with small lesions that we studied appeared to have a good QoL. When measured with the DLQI, a disease specific questionnaire, 85% of patients indicated no or a small effect of the disease on their current life. The median score found was 0 compared to 13 in active and 3 in healed podoconiosis patients in Ethiopia [16], and median scores between 2 and 12 in patients with various active skin diseases in South Africa [21]. In the current study, scores on a measure of general QoL were high, especially in the physical domain of QoL, but – likely owing to the poor rural setting – lower on the environmental domain, which relates to issues such as access to healthcare and transportation, and financial resources. In addition, former patients indicated an equal or better QoL compared to healthy controls in all four domains measured. As their scars and functional limitations were limited, it could be expected that former patients would not differ much from controls in QoL. However, it is surprising that they would report a higher physical and environmental QoL, with small to medium effect sizes. As controls were matched to former patients on age, gender, village and occupation, it is difficult to explain these findings on the basis of social or economic factors. Perhaps having been confronted with BU and the prospect of scarring and disability, made patients adjust their internal standards of what constitutes good physical functioning. This, in turn, would cause them to appreciate the preserved physical functioning they have today, a phenomenon known as response shift [24]. One possible explanation for indicating a better environmental QoL could be their positive experience with the healthcare system for the treatment of their BU, which is free of charge and relatively well-organized in the study regions. However, it is also possible that the interviewers were viewed by former patients as representatives of the hospitals where they were treated, causing them to answer in a more socially desirable way than controls. Another possible explanation is that we followed-up a group who presented early, in contrast with most BU patients, and that this was a consequence of this group having more access to healthcare and other resources, i. e. through being more educated or living less isolated, and hence a better baseline QoL than the average community member. The good QoL of former patients with small lesions found in this study corresponded with the more physical measures of BU sequelae. Patients had a relatively small median scar size. In addition their level of functional limitations as measured by the BUFLS was low, with a median score of 2. 6%, compared to earlier scores of 5. 3% in a large cohort of BU patients with varying lesion sizes that was treated outside of any study context [2], and 16% in patients who had only received surgical or traditional treatment [25]. The DLQI had a Cronbach' s alpha of 0. 78 indicating satisfactory internal consistency [26]. In addition, it was significantly related to scar size, which can be seen as a proxy measure of disease severity, and to both the psychological and physical QoL domains of the WHOQOL-BREF in the expected directions. Similar to two previous studies with the DLQI in Sub-Saharan Africa, we did not find DLQI scores to be related to gender [16], [21]. Together, these findings suggest that the DLQI is a valid instrument for measuring disease-specific QoL in BU patients aged 16 and above, although further research in larger samples, including larger lesions is needed. In children, the CDLQI had a low Cronbach' s alpha, and was not related to any of the background variables, which makes it difficult to ascertain its validity. The Cronbach' s alphas for the WHOQOL-BREF subscales were all below 0. 7, indicating questionable internal consistency. The social domain of QoL was related to being married and working as a farmer, and the physical and psychological domains were related to disease specific QoL in the expected directions. Former patients indicated a higher QoL on all 4 domains of the WHOQOL-BREF than controls, which is rather unexpected. In addition, item 21, satisfaction with ones sex life, was poorly understood by those who were not married, and was left open by a considerable proportion/number of former patients. Overall, these data do not clearly establish the validity of the WHOQOL-BREF for measuring general QoL in former BU patients. The present study suffered from several limitations. First the sample size was relatively small, with only 67 adults completing both adult QoL instruments. As the cohort under study was predetermined by the BURULICO trial, we were not able to increase the sample size. This could have left the study underpowered to pick up associations between the questionnaires and background variables such as gender. However, several significant associations were indeed found. In addition, we were not able to recruit healthy controls for the DLQI and CDLQI as these are disease specific questionnaires, and thus had to rely on the earlier proposed norm scores, but the very low scores on both instruments are likely to reflect good QoL. Finally, we did not study QoL in former BU patients that were treated under normal service conditions. These patients would have likely had larger lesions and might have reported a lower QoL, and data on these patients could have helped to establish the validity of the questionnaires, although in our sample the DLQI already correlated significantly with scar size. In this study, we show that in BU patients who reported early with small lesions and received either 8 weeks of streptomycin and rifampicin or 4 weeks of streptomycin and rifampicin followed by 4 weeks of clarithromycin and rifampicin, scars were small, functional limitations were uncommon and long term QoL was preserved. These findings contrast the debilitating sequelae often reported in BU, highlighting the importance of active case finding and antimicrobial treatment of BU lesions. The Twii version of the DLQI appeared to be a valid instrument to measure disease specific QoL in BU patients aged 16 or older, and was quick and easy to administer under field conditions. For children, the Twii version of the CDLQI appeared to be a less valid instrument in this population, though alternatives are lacking. The validity of the WHOQOL-BREF for measuring general QoL in former BU patients is questionable. Future studies on QoL in BU should attempt to include patients with active BU, and should include patients with larger lesions. In addition they should include alternative questionnaires for measuring general QoL in BU, such as the SF-36, or limit themselves to disease-specific QoL using the DLQI.

Buruli ulcer is an infectious skin disease, mainly occurring in West Africa. It usually starts with a small nodule that over the course of weeks progresses into an ulcer. Effective treatment with antibiotics is available, but patients often report to the hospital late, and there is a serious risk of scarring, contractures and functional limitations of the affected body parts. Although it is often reported that Buruli ulcers can have serious sequelae, the quality of life of former Buruli ulcer patients has not been studied before. In this study we assessed the quality of life of healed Buruli ulcer patients that took part in a drug trial between 2006 and 2009, which only included small and early lesions. On follow-up, we found that most scars were small and functional limitations were rare, and that both general and skin specific Quality of Life was good. Our results demonstrate the potential of the combination of early detection and proper antibiotic treatment of Buruli ulcer and hence stress the importance of public health efforts aimed at diagnosing the disease in its early stage, and providing standardized treatment in endemic areas.

Abstract Introduction Methods Results Discussion

public and occupational health bacterial diseases infectious diseases buruli ulcer medicine and health sciences psychology disabilities behavioral and social aspects of health global health neglected tropical diseases biology and life sciences tropical diseases mental health and psychiatry psychological stress

2014

Good Quality of Life in Former Buruli Ulcer Patients with Small Lesions: Long-Term Follow-up of the BURULICO Trial

Lymphatic filariasis (LF) is a chronic nematode infection transmitted by mosquitoes and in sub-Saharan Africa it is caused by Wuchereria bancrofti. The disease was targeted for global elimination by 2020 using repeated community-wide mass drug administration (MDA) distributed in endemic areas. However, recently, there has been a growing recognition of the potential role of including vector control as a supplement to MDA to achieve elimination goal. This study was carried out to determine mosquito abundance and transmission of bancroftian filariasis on Mafia Islands in Tanzania as a prerequisite for a search for appropriate vector control methods to complement the ongoing MDA campaign. Mosquitoes were collected indoor and outdoor using Centre for Disease Control (CDC) light and gravid traps, respectively. Collected mosquitoes were identified based on their differential morphological features and Anopheles gambiae complex and An. funestus group were further identified to their respective sibling species by polymerase chain reaction (PCR). Filarial mosquito vectors were then examined for infection with Wuchereria bancrofti by microscopy and PCR technique. Overall, a total of 35,534 filarial mosquito vectors were collected, of which Anopheles gambiae complex, An. funestus group and Culex quinquefasciatus Say accounted for 1. 3,0. 5 and 98. 2%, respectively. Based on PCR identification, An. gambiae sensu stricto (s. s) and An. funestus s. s sibling species accounted for 88. 3% and 99. 1% of the identified members of the An. gambiae complex and An. funestus group, respectively. A total of 7,936 mosquitoes were examined for infection with W. bancrofti by microscopy. The infection and infectivity rates were 0. 25% and 0. 08%, respectively. Using pool screen PCR technique, analysis of 324 mosquito pools (each with 25 mosquitoes) resulted to an estimated infection rate of 1. 7%. The study has shown that Cx. quinquefasciatus is the dominant mosquito on Mafia Islands. By using mosquito infectivity as proxy to human infection, the study indicates that W. bancrofti transmission is still ongoing on Mafia Islands after more than a decade of control activities based on MDA. Lymphatic filariasis (LF) is a chronic infection with serious physical, mental and socio- economic consequences to the affected individuals, and ranked as one the leading causes of long-term disability in the world [1,2]. In Sub-Saharan Africa, LF is caused by the filarial nematode Wuchereria bancrofti and transmitted mainly by Anopheles and Culex mosquitoes [3]. Globally, it has been estimated that more than one billion people live in endemic areas and are at risk of infection, and more than one third of these are in Sub-Saharan Africa [4]. In Tanzania, it has been estimated that 34 million people are at risk of LF infection and about 6 million live with debilitating manifestations of the disease [5]. LF was considered eradicable and the World Health Organization (WHO) launched a Global Programme to Eliminate Lymphatic Filariasis (GPELF) by year 2020 [6]. The principal elimination strategy in endemic countries is based on yearly community-wide mass drug administration (MDA) with ivermectin or diethyl-carbamazine in combination with albendazole [6,7]. The drugs mainly kill microfilariae and it is assumed that the reduction of microfilarial load in endemic communities will lead to reduction or even elimination of transmission [8]. Since the inception of GPELF, countries have initiated their local control programmes and encouraging reduction in disease prevalence as a result of MDA have been reported elsewhere [9,10]. In Tanzania, MDA intervention was launched on Mafia Islands in the year 2000 and geographical coverage has been expanded in most of the endemic districts [5,11]. Recently, there is growing recognition of the potential role of inclusion of vector control to achieve interruption of LF transmission in different epidemiological settings [3,12]. In line with this assumption, studies have indicated that use of insecticide treated bed nets (ITNs) resulted in reduction in prevalence and transmission of LF [13–17]. However, insecticide based mosquito vector control interventions are threatened by development of insecticide resistance [18] and change in behaviour or shift of mosquito vectors species [19,20]. Other studies have shown that Cx. quinquefasciatus, an important filarial vector is relatively tolerant to insecticides used for ITNs and IRS interventions [13,21]. Thus, to expedite LF elimination efforts, novel control methods are needed to tackle the growing population of insecticide tolerant Cx. quinquefasciatus which is responsible for most of LF transmission in Tanzania as previously reported [19]. This study was carried out to determine mosquito abundance and transmission of bancroftian filariasis on Mafia Islands in Tanzania as a prerequisite for a search for appropriate vector control method to complement the ongoing MDA campaign. Xenomonitoring in filarial vectors has been considered as an integral component of monitoring the impact of MDA and has been reported to provide real time information on LF transmission [22,23]. For mosquito surveys, both Centre for Disease Control (CDC) light and gravid traps have been found to be useful tools for collection of filarial mosquito vectors [24,25]. Dissections of the vectors and molecular tests based on polymerase chain reaction (PCR) have proved useful in detection of W. bancrofti in mosquitoes [19,26,27]. The potential of PCR to screen large number of mosquitoes relatively quickly with high precision are requirements when infection rates in mosquitoes decrease after repeated MDA cycles. By using xenomonitoring as a proxy to human infection, this study reports W. bancrofti transmission on Mafia Islands 15 years after the launching of MDA campaign by the Tanzanian National Lymphatic Filariasis Elimination Programme. The study was conducted on Mafia Islands (07°91554’S, 39°65529’E) in the Indian Ocean, off shore of the Pwani Region at about 195 km south-east of Dar es Salaam, Tanzania. A distance of about 40 km separates the islands from the mainland Tanzania (Fig 1). Mafia is an archipelago of islands, with the main island surrounded by seven small islets. Of these, five islets namely, Mafia (main islet), Bwejuu, Jibondo, Juani and Chole are inhabited. Mafia Islands have an estimated population of 50,167 people [28], of which 92. 6% live in the main island of Mafia. The inhabitants of Mafia islands are subsistence farmers of coconuts and rice, and some are fishermen. The islets receive two rain seasons, long rains in March to June and short rains in October to December. As in many coastal areas of Tanzania, LF is an important mosquito borne diseases on the Mafia Islands. Before the start of MDA campaign in 2000, the baseline prevalence of W. bancrofti circulating filarial antigens (CFA) on the Mafia Islands was 49% and declined to 4% in 2006 [29]. This drop of prevalence indicates that the MDA rounds had marked impacts on the prevalence of W. bancrofti [30], but not yet reached the elimination threshold. Three villages, Kilindoni and Kiegeani (from the main Mafia Island) and Chole islet were purposely selected for the study. Other islets were excluded due to transport-related challenges and very low filarial vectors collected during preliminary surveys. All hamlets in Kiegeani and Chole villages (3 hamlets each) and an equal number of hamlets were selected from Kilindoni village in a non-random fashion to increase the odds of catching mosquitoes. All households in the selected hamlets were mapped using hand held Global Positioning System (GPS) device (Garmin etrex Legend H, Garmin ltd, USA). Three households in each hamlet were randomly selected for indoor mosquito collections using Centers for Disease Control (CDC) light traps (John W Hock Co, Gainesville, FL, USA). Light trapping was conducted as previously described [31] and mosquitoes were collected in each of the selected households every other day from 22nd January to 10th March 2014, resulting in a total of 22 light trap catch nights. In brief light traps were set in the evening between 17. 00 to 18. 00 hours and retrieved from 06. 00 to 8: 00 hours the following morning. Caught mosquitoes were transferred from the traps to labelled paper cups covered with netting material and transported to the field laboratory for identification and processing. Moreover, two households were selected from each of the 9 study hamlets for outdoor mosquito collection using CDC gravid traps (John W. Hock Co. , Gainesville FL). Gravid traps were set in peri-domestic areas and trapping was conducted as described previously [25,32]. Traps were set in the evening between 17: 00 and 18: 00 hours, and retrieved the following morning between 06: 00 and 08: 00 hours in alternating days from 26th January to 7th March 2014. At each household the traps were ran for 18 nights. Collected mosquitoes were treated as described for light trap catch. Upon arrival in the field laboratory live mosquitoes were knocked down with chloroform and both (live and dead) were identified using morphological criteria [33,34]. In the field laboratory, freshly killed Cx. quinquefasciatus, Anopheles gambiae complex and An. funestus group were processed for W. bancrofti detection by microscopy. The rest were stored in Eppendorf tubes containing silica gel desiccants for later identification of sibling species of An. gambiae complex, An. funestus group and detection of W. bancrofti by PCR technique. Members of the An. gambiae complex were identified by PCR based on method previously described to identify An. gambiae sensu stricto (s. s), An. arabiensis, An. quadriannulatus, An. melas and An. merus [35,36]. In brief, DNA was extracted by using Bender buffer method [36,37] that involved homogenizing individual mosquito and precipitating extracted DNA using potassium acetate and ethanol. PCR reactions were conducted in a final volume of 20μl consisting of 0. 25μM of each of the five primers, 1: 1 TEMPase Hot Start polymerase master mix (Ampliqon III, VWR-Bie Berntsen, Denmark) and 2μl of DNA extract. The samples were amplified in GeneAmp PCR Systems 9700 (Applied Biosystems, USA) and cycling conditions were 95°C for 15 minutes followed by 35 cycles of denaturation at 94°C for 30 seconds, annealing at 50°C for 30 seconds, extension at 72°C for 30 seconds and final extension at 72°C for ten minutes. On the other hand, sibling species of the An. funestus group were identified based on species-specific primers in the ITS2 region on the rDNA genes, a method previously described to identify An. funestus, An. vaneedeni, An. rivulorum, An. leesoni and An. parensis [38,39]. DNA was extracted as described previously for sibling species of the An. gambiae complex. Each PCR run was conducted in a final volume of 25 μl consisting of 0. 5 μM of each of the six primers, 1: 1 TEMPase Hot Start polymerase master mix and 3 μl of DNA extract. The samples were amplified in GeneAmp PCR Systems 9700 and cycling conditions were 94°C for 15 minutes followed by 45 cycles of denaturation at 94°C for 30 seconds, annealing at 50°C for 30 seconds, extension at 72°C for 40 seconds and final extension at 72°C for ten minutes. Freshly killed An. gambiae s. l. , An. funestus group and Cx. quinquefasciatus from both light and gravid traps were dissected and examined under microscopy for the first, second and human infective third stage larvae of W. bancrofti as previously described [40]. The required sample size for filarial vectors (mainly Cx. quinquefasciatus) examined by microscopy was estimated based on thresholds criteria outlined by the World Health Organization [41]. Moreover, using PCR technique, a separate sample (proportionally equal to dissected specimens) of randomly selected filarial mosquito vectors were pooled (25 mosquitoes in each reaction tube) and examined for presence of W. bancrofti infection as previously described [39,42]. DNA was extracted from the pooled mosquitoes in the same way as explained for identification of sibling species of An. gambiae s. l. and An. funestus group. Extracted DNA was examined for presence of W. bancrofti by PCR targeting a highly repeated DNA sequences (the SspI repeat) found in W. bancrofti. In the reaction mixture, each of the 20 μl of PCR consisted of 0. 25μM of each of the two primers (NV1&NV2), 1: 1 Hot-Start TEMPase polymerase master mix and 2 μl of DNA extract. PCR thermal cycling conditions were 95°C for 15 minutes followed by 54°C for 5 minutes: then 35 cycles of denaturation at 94°C for 20 seconds, annealing at 54°C for 30 seconds, extension at 72°C for 30 seconds and final extension at 72°C for 5 minutes. The amplified DNA for both sibling species and W. bancrofti specimens were separated based on their fragment size by gel electrophoresis and visualized under ultra violet light as previously described [35,38]. Data were entered in Excel and later transferred to STATA 12 (Stata Corp, College Station, Tx, USA) for analysis. The" infectivity rate" of the dissected mosquitoes was calculated as the percent of mosquitoes infected with infective larvae (L3) and the" infection rate" as the percent of mosquitoes infected with any stage of the parasite (L1, L2 and/or L3). For the PCR technique used for pooled mosquitoes, the probability that any one mosquito is infected with any stage of the W. bancrofti parasite were calculated using Poolscreen 2. 02 software, providing maximum likelihood estimates for the rate of infection [43]. The 324 mosquito pools screened for W. bancrofti by PCR were randomly selected from a total of 563 pools made using random number generator programme in Microsoft Excel 2007. Mosquito infection and infectivity rates were compared using two sample test of proportions and p-value ≤ 0. 05 was considered statistically significant. The study received ethical approval from the Medical Research Coordinating Committee of the National Institute for Medical Research, Tanzania (Ref: NIMR/HQ/R. 8a/VOL. 9/1616). Before data collection, meetings were held with the district and respective village leaders to inform them about the study and to obtain their cooperation. Written informed consent was obtained from the heads of households before commencing mosquito collection in their respective houses or peri-domestic areas. A total of 38,505 mosquitoes were collected in the three villages of Chole, Kiegeani and Kilindoni during the study period. CDC light and gravid traps collected 17,831 (46. 3%) and 20,674 (53. 7%) mosquitoes, respectively. Out of the collected mosquitoes, 35,534 (92. 3%) were filarial vectors belonging to members of the An. gambiae complex (1. 3%), An. funestus group (0. 5%) and Cx. quinquefasciatus (98. 2%). All members of the An. funestus group and 99. 8% of the members of An. gambiae complex were collected with light trap method. On the other hand, of 34,899 collected Cx. quinquefasciatus, 57. 8% were collected using gravid traps. Majority (72. 8%) of the filarial mosquito vectors were collected in Kiegeani village (Table 1). Of the collected Anopheles, 270 members An. gambiae complex and 114 An. funestus group were processed for sibling species identity using PCR technique. An. gambiae sensu stricto (s. s) sibling species accounted for 88. 3% of the analysed members of the An. gambiae complex. Other members of the An. gambiae complex identified were An. arabiensis, An. quadriannulatus and An. merus. On the other hand, An. funestus s. s was the majority (99. 1%) of the identified sibling species in the An. funestus group (Table 2). A total of 3,866 filarial mosquito vectors collected with CDC light traps were dissected and examined for infection and infectivity with W. bancrofti. Nine (0. 23%) Cx. quinquefasciatus were found to be infected with any of the three larval stages (L1, L2 and /or L3) of W. bancrofti and three mosquitoes (0. 08%) were infective. None of the dissected members of the An. gambiae s. l. and An. funestus were found to carry W. bancrofti larvae of any stage. On the other hand, a total of 4,070 Cx. quinquefasciatus mosquitoes collected with CDC gravid trap were dissected and examined for infection and infectivity with W. bancrofti. Eleven (0. 27%) Cx. quinquefasciatus were found to be infected with any of the three larval stages (L1, L2 and /or L3) of W. bancrofti and three (0. 07%) were infective. Mosquito infection and infectivity rates between the two trap types were not significantly different (Table 3). Using PCR technique, of 324 mosquito pools (each with 25 mosquitoes) tested, 115 were found to be infected with at least a larval stage of W. bancrofti. Analysis by trap type revealed that of 163 gravid trap mosquito pools processed, 70 were infected whilst out of 161 light trap pools processed, 45 pools were infected. The infection rates between the two trapping methods were not significantly different (two sample test of proportions, p>0. 05). On the other hand, of 6 Anopheles pools processed, only one (belonging to An. funestus group) was infected. For both trap types and species, the probability that any one mosquito in the pool was infected with any stage of the W. bancrofti parasite was estimated at 1. 7%. Comparison of mosquito infection rates as measured by the two xenomonitoring methods have shown that PCR estimate seven-fold higher infection rate than dissection (Table 3). The study has shown that Cx. quinquefasciatus was the dominant man-biting mosquito on Mafia Islands and W. bancrofti infection is confined to this vector group. Both CDC light and gravid traps were found useful for mosquito vector surveillance. Moreover, it was found out that molecular method based on PCR was seven fold more sensitive than dissection in detecting W. bancrofti infection in mosquitoes. By using xenomonitoring as proxy to human infection, the study indicated that W. bancrofti transmission was still ongoing on Mafia Islands after more than a decade of control activities based on MDA. Our findings suggest that inclusion of mosquito control method that target Cx. quinquefasciatus will accelerate LF elimination on Mafia Islands and other coastal areas of Tanzania.

Lymphatic filariasis is a chronic human disease caused by parasitic worms and transmitted by mosquitoes. The disease is targeted for elimination by 2020 through the treatment of the entire population at risk in endemic areas using a mass drug administration (MDA) strategy. After several years of MDA, there is now growing interest in including vector control as a supplement to MDA to achieve elimination goal. This study was carried out to determine mosquito abundance and transmission of lymphatic filariasis on Mafia Islands in Tanzania after nine rounds of MDA. Mosquitoes were collected indoor and outdoor using Centre for Disease Control (CDC) light and gravid traps, respectively. Filarial mosquito vectors were examined for infection with Wuchereria bancrofti by microscopy and PCR technique. A total of 35,534 filarial mosquito vectors were collected, of which Anopheles gambiae complex, An. funestus group and Culex quinquefasciatus Say accounted for 1. 3,0. 5 and 98. 2%, respectively. Using PCR, An. gambiae sensu stricto (s. s) and An. funestus s. s sibling species accounted for 88. 3% and 99. 1% of the identified members of the An. gambiae complex and An. funestus group, respectively. A total of 7,936 mosquitoes were examined for infection with W. bancrofti by microscopy. The infection and infectivity rates were 0. 25% and 0. 08%, respectively. Using PCR technique, of 324 mosquito pools (each with 25 mosquitoes) tested, 115 were found to be infected with at least a larval stage of W. bancrofti. The study concludes that Cx. quinquefasciatus is the dominant mosquito on Mafia Islands and that W. bancrofti transmission is still ongoing on Mafia Islands after a decade of control activities based on MDA.

Abstract Introduction Methods Results Discussion

geomorphology invertebrates medicine and health sciences landforms topography vector-borne diseases geographical locations animals tanzania molecular biology techniques infectious disease control insect vectors africa islands research and analysis methods infectious diseases wuchereria bancrofti artificial gene amplification and extension anopheles gambiae wuchereria molecular biology disease vectors insects arthropoda people and places mosquitoes eukaryota polymerase chain reaction earth sciences nematoda biology and life sciences species interactions organisms

2017

Lymphatic filariasis transmission on Mafia Islands, Tanzania: Evidence from xenomonitoring in mosquito vectors

Tuberculosis is still a major health problem worldwide. Currently it is not known what kind of immune responses lead to successful control and clearance of Mycobacterium tuberculosis. This gap in knowledge is reflected by the inability to develop sufficient diagnostic and therapeutic tools to fight tuberculosis. We have used the Mycobacterium marinum infection model in the adult zebrafish and taken advantage of heterogeneity of zebrafish population to dissect the characteristics of adaptive immune responses, some of which are associated with well-controlled latency or bacterial clearance while others with progressive infection. Differences in T cell responses between subpopulations were measured at the transcriptional level. It was discovered that a high total T cell level was usually associated with lower bacterial loads alongside with a T helper 2 (Th2) -type gene expression signature. At late time points, spontaneous reactivation with apparent symptoms was characterized by a low Th2/Th1 marker ratio and a substantial induction of foxp3 reflecting the level of regulatory T cells. Characteristic gata3/tbx21 has potential as a biomarker for the status of mycobacterial disease. Tuberculosis (TB) is a pulmonary disease spread worldwide. It is caused by an infection with Mycobacterium tuberculosis. Only 5–10% of infected individuals develop a primary active disease while the most common outcome of infection is a latent or subclinical disease with no evident symptoms. This latent disease has the inherent ability to reactivate after years or even decades of latency and is therefore a major global threat. The existing vaccine, the Bacille Calmette-Guérin (BCG), is not entirely safe and does not confer protection against latent or reactivated TB. Current antibiotic regimens have started losing their efficacy due to the spread of antibiotic resistance genes [1]. In total, the primary active infections and reactivated infections cause 1–2 million deaths yearly, which makes M. tuberculosis the deadliest bacterium for humans [2]. The dichotomy to a latent and active tuberculosis is an over-simplification, as the infection can actually lead to a wide spectrum of disease states ranging from a well-controlled (or even cleared) latent disease to fulminant, severe forms of TB. Within the latent population, a “sub-spectrum” exists leading to differences in the risk of reactivation [3]. The pathogenesis of tuberculosis has been widely studied for decades, but as it seems that TB is not a single disease but a spectrum of different outcomes, it remains poorly understood. Better understanding on the factors that contribute to the type of TB disease is crucial for the development future treatment strategies. The TB spectrum is likely to arise from genetic variation both in the host and in different pathogen strains as well as from environmental factors. It is known that adaptive immunity and especially T helper (Th) cells are required for controlling the disease. HIV-infected individuals are more susceptible to active and reactivated tuberculosis due to the defective T lymphocyte response [4]. Mice lacking T helper responses are hypersusceptible to TB [5]. Based on the observations that IL-12 or IFN-γ deficient mice are unable to restrict mycobacterial infection, it was initially concluded that Th1 cells are the predominant mediators of protective immunity to M. tuberculosis [6]–[8]. In mice, observations of an early Th1 response (2–3 weeks post infection, wpi) followed by a Th2 response simultaneously with the onset of a chronic phase, have led to a presumption that Th2 response is detrimental to the host by leading to a failure of Th1 response to clear the infection [9], [10]. Subsequently, it has also been shown that the lack of Th2 responses in IL-4, IL-13 or Stat-6 deficient mice does not lead to better resistance to M. tuberculosis infection but, instead, to increased bacterial burdens at later stages of infection [11]. Although the role of humoral immunity in response to mycobacterial infection is still unclear, there is evidence that Th2 responses are needed as well for optimal protective immunity [1], [12]. Despite the emerging understanding of the roles of different Th subtypes in TB immunity, it is still not known what type of Th profiles are needed at different phases of infection to provide optimal protection. In part, this is due to the lack of suitable animal models for studying the full spectrum of disease outcomes, including latency and reactivation. Several animal models have been developed with the aim of understanding the complex pathogenesis of tuberculosis. The murine model of pulmonary TB is well standardized and has made many valuable contributions to the knowledge of the disease pathomechanisms, especially on the role of T cells as mediators of protective immunity [13]. A major constraint of the model is that mice do not develop spontaneous latency although they can restrict the bacterial growth to chronic progressive infection. One of the rare animal models developing true latency is the Cynomolgus macaque. In the macaque, a low-dose M. tuberculosis infection leads to active primary disease in 50% and latent disease in 50% of individuals [14]. In the rabbit model of latent TB, the lung bacterial burdens start declining at 4 wpi following a primary phase with limited bacterial growth. In the rabbit TB model, different outcomes of infection can be induced by using mycobacterial strains with different virulence properties [15], [16]. In addition to the mammalian models of TB, we have previously shown that infection of adult zebrafish with their natural pathogen, Mycobacterium marinum, can be used to model latent TB [17]. M. marinum is a close genetic relative of M. tuberculosis, and typically infects cold-blooded hosts, such as frogs and various freshwater and saltwater fish species [18]. M. marinum infection of zebrafish embryos has been established as an elegant model to dissect the innate mechanisms of protective host responses in active mycobacterial infection [19]–[21]. However, the full spectrum of mycobacterial disease outcomes can be observed only in the adult zebrafish, due to the full maturation of adaptive immune system after the first four weeks post fertilization [21], [22]. In the adult zebrafish model, the injection of a low dose of M. marinum (ATCC 927 type strain) into the abdominal cavity leads to a systemic infection, characterized by an initial 3–4 week phase with rapid bacterial growth, followed in most individuals by a latent phase with stable bacterial burdens. In the latently infected fish, the majority of the mycobacterial population passes into a non-replicative state, dormancy, but can be experimentally reactivated by immunosuppression [17]. The wide disease spectrum typical of mycobacterial disease results from various host- and pathogen-associated factors. It is known that genetic determinants lead to an inherent, stable preference towards either T helper 1 or T helper 2 response that varies between human individuals [23], [24]. As a starting point for our study, we hypothesized that the differences that control the T helper response might be associated with the establishment of the wide spectrum seen in TB patients and that differences in T cell polarity might be related to the progression of the disease. Taking advantage of the heterogeneity of the zebrafish population and the wide spectrum of mycobacterial disease outcomes in the zebrafish model, we set out to look for differences in T helper responses involved in regulating protective response. Finding such differences would 1) allow the use of T helper markers among latently infected individuals to distinguish between those at high or low risk of reactivation and 2) provide understanding on what type of T cell response gives the optimal protection against mycobacterial infection and allow development of novel kinds of therapeutic or preventive approaches. Our previous work [17] provided evidence that functional lymphocyte response is a prerequisite for latency and mycobacterial dormancy in the M. marinum infection of zebrafish. To further demonstrate the significance of lymphocyte responses in the immune defence against mycobacteria in zebrafish, we carried out adoptive transfer experiments on low-dose (21±7 cfu) M. marinum-infected rag1 (−/−) fish. Spleen and kidney marrow cells were transferred from WT or rag1 (−/−) zebrafish immunized with heat-killed M. marinum to rag1 (−/−) recipients at 2 wpi. At 4 wpi, bacterial burdens were significantly lower in the fish that received transplants from immunized WT donors, compared to the fish that received transplants from rag1 (−/−) donors (3. 2×105 vs. 1. 8×106). This indicates that heat-killed M. marinum-induced lymphocytes, rather than NK cells or other innate immune cells, transferred additional immune protection against M. marinum infection to rag1-deficient zebrafish. (Figure 1A). In our current study of ∼150 individuals, a total of 10% of zebrafish were able to clear the bacterial number below the detection limit of M. marinum q-PCR-analysis (∼100 bacteria) (Figure S1B–D). These individuals capable of clearance were not detected in the groups that were collected at 2 wpi, suggesting that the clearance is likely to occur after the activation of adaptive responses. These results attest the significance of adaptive responses in the immune protection against zebrafish mycobacteriosis, and support the view of similarity of human and zebrafish anti-tuberculosis immunity. In the current study, we aimed at further elucidating the details of adaptive immune response leading to a variety of infection outcomes. Unlike many other commonly used laboratory animals, zebrafish populations are genetically heterogeneous. This characteristic causes large variations and standard deviations in most studies utilizing this model, including our studies on mycobacterial infection. On the positive side, the zebrafish population provides starting material for studying the natural differences between individuals. To be able to elucidate the host factors affecting the outcome of infection, the variation in environmental and bacterial factors was first minimized: the bacterial strain, bacterial growth conditions, infection procedure, infection dose and housing of infected fish (water quality and temperature, feeding etc.) were carefully standardized. WT adult zebrafish were infected with a low dose (21±7 cfu) of M. marinum, collected at various time points and divided into three subpopulations, based on their bacterial burdens. The subpopulations were named Low (25% of the population, individuals with the lowest bacterial burdens), Medium (50% of individuals) and High (25% of the population, individuals with the highest bacterial burdens, including the primary-progressive). The Low, Medium and High subpopulations were analyzed at various stages of mycobacterial disease: at 2 wpi (primary active disease), 4 wpi (the onset of latency in the majority of zebrafish) and 5 months post infection (mpi; late stage at which most individuals maintain latency). The reactivation risk of latent mycobacterial infection is thought to increase with increasing bacterial load [25]. The bacterial burdens in the subgroups at different time points are shown in Figure S1. To dissect the differences in the total T lymphocyte numbers between the Low, Medium and High subpopulations, we quantified the cluster of differentiation (cd) 3 levels from internal organs of the zebrafish by q-RT-PCR. Zebrafish cd3 has been shown to be an ortholog of the mammalian T cell marker cd3 [26]. Here, the cd3 expression as a marker for T cell numbers in zebrafish was further validated as described in Figure S2. Induction of T cell expansion was similar in the Low and Medium groups, seen as a 2-fold induction in cd3 expression level already at 2 wpi (Figure 1B) and peaking to 4-fold around 4 wpi (Figure 1C), compared to the cd3 expression levels in non-infected zebrafish. The High group differed from the rest of the population by showing a modest T cell expansion (max. 1. 7±1. 3), which was only seen at 4 wpi (Figure 1C). A similar pattern remained at a late stage of the infection (5 months, Figure 1D). To assess whether the limited T cell expansion is the cause or the consequence of enhanced bacterial growth in the High group, the low-dose-infected fish were compared with a group infected with a high dose of M. marinum (2691±520 cfu). Based on our previous work [17] a high initial dose causes the bacterial load to be significantly higher than with a low initial dose during the first 2 weeks of infection and this difference will even out by 4 wpi. At 4 wpi, cd3 expression levels were significantly lower in the high-dose group (Figure 1E), suggesting that the reduced T cell numbers in the High subpopulation may at least partly be affected by the rapid bacterial growth. Taken together, these results indicate that an early T cell expansion associates with protective response against mycobacterial infection, as the fish with highest cd3 expression levels were always found in the Low and Medium subgroups. However, individuals with modest lymphoproliferative response were found equally in all the three subgroups, suggesting that other factors besides efficient T cell expansion are required for mounting a protective response against mycobacterial infection. As Th cells are potent orchestrators of immune responses during infection, it is reasonable to assume that in addition to total lymphocyte numbers, variation in Th response types may be an important factor underlying the wide spectrum of outcomes in mycobacterial infections. For zebrafish, antibody markers or reporter lines for FACS (fluorescence-activated cell sorting) analysis of different T lymphocyte populations are not available. To assess the Th1/Th2 balance of individuals with different infection outcomes, we measured the levels of master regulator transcription factors for Th1/Th2 lineage development, T-box transcription factor 21 (tbx21) and gata3, from the internal organs of infected zebrafish. Tbx21 is a Th1 cell transcription factor important for Th1 lineage commitment and gata3 is a well-known regulator of Th2 cell differentiation also playing a role in endothelial cell biology [27]. The central T cell transcription factors tbx21, gata3 and foxp3 have been identified in the fish [28]–[30]. The enrichment of tbx21 and gata3 in the zebrafish T cell population was validated as described in Figure S2. During infection, the alterations in the transcript levels of these transcription factors reflect the changes in the numbers of the corresponding T helper cells. The ratio of Th2/Th1 markers was used to assess the balance of T helper cell response. In addition, the induction of a Th2-type cytokine IL-4 (IL4b) [31] and a Th1-type cytokine IFN-γ (ifnγ1-2) was measured and the ratio was calculated. At 2 wpi, the induction of both gata3 and tbx21 was significantly higher in Low and Medium than in the High group (Figure S3A&D). At this time point, there were no significant differences in the gata3/tbx21 ratio between the three groups (Figure 2A). At 4 wpi, gata3 was still significantly more induced in Low and Medium groups compared to High group (Figure S3B). However, the tbx21 levels were similar in all groups (Figure S3E). As determined by the gata3/tbx21 ratio, the Low group had developed a significantly more Th2-biased response than the High group by 4 wpi (Figure 2B), suggesting that insufficiency of Th2 cells is a differentiating factor between the Low and High individuals. The il4/ifnγ ratios generally followed a similar pattern (Figure 2D–E). Also at 2 wpi, the il4/ifnγ ratio was significantly higher in the Low group compared the other two, although there were no significant differences in the gata3/tbx21 ratio at this time point. At 2 weeks, it is likely that the adaptive Th response is in the process of maturation conducted by the cytokines excreted by innate immune cells. Similar patterns were observed at the late time point 5 months post infection (Figure S3C&F, Figure 2C&F). At 4 wpi, the Th2/Th1 balance was also assessed at the protein level by semi-quantitative Western blot analysis of gata3 and CXCR3 (a CXC chemokine receptor preferentially expressed on Th1 cells) from individuals in Low, Medium and High groups (Figure 2J, Figure S4). The results showed a similar trend as seen with q-RT-PCR analyses. To assess the importance of functional, specific lymphocytes for the changes in the levels of the markers used in this study, we also carried out similar infection experiments in rag1 (−/−) mutants. The fish were infected with a low dose (35±18 cfu) and collected 4 wpi. We found that there was some induction of gata3, tbx21, il4 and ifnγ in the infected rag1 (−/−) mutants (Figures S2H&I and 2H&I) showing the proportion of lymphocyte-independent induction of these markers. However, the induction of gata3, tbx21 and il4 was significantly higher in the WT fish (Figures S2H&I and 2H) than in rag1 (−/−) fish at 4 wpi. This clearly demonstrates the major contribution of functional lymphocytes in the changes seen in these markers during mycobacterial infection. The rag1 (−/−) fish were grouped according to bacterial load (Figure S1E) as previously described for the WT fish, and association of the gata3/tbx21 ratio and bacterial load was assessed. In the absence of functional lymphocytes no association was detected, implying that the differences in this ratio relevant to the course of mycobacterial infection seen in WT fish are indeed derived from lymphocytes. However, the expression levels of ifnγ were similar in rag1 (−/−) and WT fish showing that the induction of this Th1-type cytokine in mycobacterial infection might not be as dependent on functional lymphocytes as the other markers used. To assess whether the Th2/Th1 balance is directly influenced by the bacterial burden in the beginning of the infection, low-dose (21±7 cfu) infected fish were compared to fish infected with a high mycobacterial dose (2691±520 cfu) at 4 wpi. Average of gata3/tbx21 ratio was found to be lower in the high-dose infected group (Figure 2K), suggesting that rapid bacterial growth can lead to changes in this ratio. To investigate whether the differences in the disease outcome could result from genetically defined Th1/Th2 preferences, we stimulated healthy WT zebrafish by an i. p. injection of heat-killed M. marinum and 10 days later, analyzed the gata3 and tbx21 transcript levels. The spectrum of individual Th2/Th1 responses was broad, similarly to that seen in humans [23]. The gata3/tbx21 ratio varied from 0. 6 to 12. 8 within a group of 14 zebrafish. 22% of the individuals were substantially Th2-biased (gata3/tbx21>10), whereas 28% had a bias towards Th1 (gata3/tbx21<1) (Figure 2L). This observation of inherent Th1/Th2 phenotypes in healthy fish suggests that genetic Th1/Th2 preferences may in part lead to the development of wide disease spectrum in mycobacterial infections. Based on these results, it seems plausible that both the bacterium and the host can affect the gata3/tbx21 ratio during mycobacterial infection. At any of the time points of the study, the average gata3/tbx21 ratio was never >1. 0 in the progressive High group, whereas Th2 dominant response (average gata3/tbx21 1. 5–2. 7) was seen in the Low group in the primary infection. Altogether, these results show that induction of Th2-type responses during the first four weeks of mycobacterial infection are associated with controlling the bacterial growth. In order to further confirm the reliability of our markers and to characterize the response in the different subgroups, we measured a wider selection of Th2 and Th1 signature genes at 4 wpi: Th2: interleukins il13 and il4, immunoglobulin M (IgM) constant region, V-maf musculoaponeurotic fibrosarcoma oncogene (cmaf), STAT6, St2 (Figure 3A–F) Th1: il12, interferon gamma 1-2 (IFNγ1-2), nitric oxide synthase 2b (Nos2b), tumor necrosis factor alpha (TNFα); (Figure 3G–J). It has been previously shown that the expression levels of the selected Th2 marker genes are upregulated in zebrafish in response to recombinant IL-4 treatment, thus representing a Th2-type response in the zebrafish [31]. In line with gata3 and il4 markers, there was a significantly higher induction of Th2 signature cytokines and other Th2 related genes in the Low group than in the High group (Figure 3A–F). At the same time, Th1 markers showed a somewhat higher induction in the High group (Figure 3G–J). However, the differences in Th2 markers were more indicative of the bacterial burden and the disease state than those seen in Th1 markers. The Low group showed substantial IgM induction at 4 wpi differing significantly from the rest of the population (Figure 3C). Only three antibody classes exist in zebrafish, namely IgM, IgD, and fish-specific IgZ, whereas IgG, IgA and IgE are absent [32]. Secreted tetrameric IgM is the most abundant zebrafish serum immunoglobulin [32], and is induced by Th2-mediated IL-4 signaling [31]. Taken together, these data suggest that Th2-type cytokines participate in the control of bacterial growth during the first four weeks of primary mycobacterial infection. As zebrafish with higher T cell numbers and a Th2-biased response were able to control the infection most efficiently, we next wanted to study whether these protective features were relevant for the ability of maintaining long-term latency. Even though the individuals with a primary progressive disease had already been removed from the experiment, large variations in the bacterial numbers (Figure S1D) and dormancy (Figure 4A) were seen at a late time point, when the population was again divided into subgroups according to their bacterial loads at 5 mpi. The individuals with highest levels of dormancy-associated mycobacterial citrate synthase I (GltA1) expression were found in the Low group, whereas in the High group GltA1 expression levels were significantly lower than in the other groups (Figure 4A). This shows that there is a spectrum even within the “latent” population that has survived the primary infection. The total T cell number still seemed to play a role at this late time point so that the High group had significantly less T cells than the other groups (Figure 1D). However, there was no significant difference in the total T cell number between the Low and Medium group. The protective role of Th2-type responses associated with lower bacterial numbers remained even at this late time point (Figures 2C&F). These results indicate the importance of high T cell numbers and a tendency to Th2-biased responses in determining how well the infection is controlled. It is plausible that the ideal response for inducing stable latency is similar to that required for maintaining long-term latency. As we followed the large population of fish infected with a low initial dose, we saw that from 8 wpi to 5 mpi, there was a 17% overall morbidity in the population (data not shown). Following an eight-week period of asymptomatic infection, until the end of the 5 month follow-up, these spontaneously reactivated individuals were collected as they started showing symptoms of mycobacterial disease. These individuals had a slightly higher bacterial load than the fish in the High group (Figure S1D) but did not differ in terms of dormancy (Figure 4A). Thus, it is likely that the disease is active in the fish in the High group as well, although the fish did not yet show external signs of disease. Based on the gata3/tbx21 ratios, the Th2/Th1 balance was similar in the High and Reactivated groups (Figure 4B), differing significantly from the Low and Medium groups. As direct assessment of the bacterial load within a population of asymptomatic humans is difficult, measuring the Th2/Th1 balance from a blood sample could be used as a clinical biomarker for estimating the activity of the disease. We carried out a ROC-analysis of gata3/tbx21 ratio and found that gata3/tbx21 ratio had a high sensitivity and specificity as a marker in distinguishing the high load individuals (High&Reactivated) from the well-controlling population (Low&Medium) (Figure 4C). We next wanted to study whether there were some differences in the T cell responses between the progressive High and Reactivated groups. In the Reactivated group, half of the individuals showed high induction of cd3. This was surprising, as within all the other groups, high cd3 was generally associated with better disease control. A possibly detrimental role of the induced T cell response in the Reactivated group could be explained by a non-ideal polarization of these cells. The most striking difference between the High and Reactivated group was the significantly higher induction of foxp3 in the Reactivated group (Figure 4E), indicating a pronounced regulatory T helper (Treg) proliferation. When the expression levels of gata3, tbx21 and foxp3 were analyzed relative to cd3 in the Reactivated, only foxp3 was significantly different between the Reactivated and High group (P = 0. 0006,10-fold higher proportion than in the High group, Figure 4E). In fact, the induction of foxp3 was higher in the Reactivated group compared to any of the asymptomatic groups (Figure 4F). This implies that the increase in foxp3 expression is higher than the increase in the total T cell number and thus, the increased proportion of foxp3 cells could be used for distinguishing Reactivated individuals from the rest of the population at a late stage of infection. The role of foxp3 positive cells in reactivation of tuberculosis and its applicability in diagnostics warrants further investigation. It is known that T cell responses are essential in restricting mycobacterial growth in human tuberculosis as well as in various mammalian models for studying tuberculosis. The zebrafish is a newcomer in the field of immunology, and the components of its adaptive immune system have only recently been elucidated in more detail. It has been shown [33] that there are professional antigen presenting cells capable of inducing antigen-specific T cell responses in zebrafish. In our recent publication [34], we have demonstrated the presence of specific and protective immune responses against M. marinum infection in zebrafish. We have shown that vaccination of zebrafish with the Bacillus Calmette-Guérin (BCG) increases survival of adult zebrafish from infection with M. marinum. Furthermore, vaccination of zebrafish with plasmid DNA encoding mycobacterial antigens increases survival, reduces the spreading of bacteria as well as the number of granulomas in M. marinum infection, compared to vaccination with a control pDNA lacking the antigen-encoding sequence. Rag1 (−/−) zebrafish lacking functional adaptive lymphocytes are not protected by the antigen-pDNA vaccine. In the antigen-pDNA -vaccinated zebrafish, interferon gamma expression levels are significantly higher during infection than in the control pDNA vaccinated fish, demonstrating the specificity of the anti-mycobacterial immune response induced in zebrafish. For studies of zebrafish lymphocyte populations, antibody markers for FACS analysis are not available. To characterize the T cell responses of individuals with different infection outcomes, we measured expression levels of marker genes reflecting the total number of T cells and the Th profile from the internal organs of infected zebrafish. We also used the transgenic lck: GFP zebrafish as a tool to isolate T cells and to validate the T cell markers that were used in the study (Figure S2). We showed that the Th marker genes are enriched in sorted lck+ T cells and demonstrated a correlation between the cd3 transcript level and the total number of lck+ cells from lck: GFP reporter fish. These assessments could only be done from uninfected zebrafish, because it is likely that M. marinum infection influences the expression of the GFP reporter gene driven by the lck promoter. It has been previously shown for other pathogenic mycobacteria that one of the mycobacterial virulence strategies is to inhibit host T cell receptor signaling by interfering with the expression and phosphorylation of lck [35]–[37]. Thus, the use of the lck: GFP reporter zebrafish line as a tool to study T cell responses in M. marinum infections would first require careful assessment of the effects of the bacterium on zebrafish lck expression. Instead, we have here relied on the use of several parallel markers reflecting the different Th profiles. In the future, development of new research tools is needed for a more detailed characterization of Th responses and their role in the pathogenesis of mycobacterial infection in the zebrafish model. The results of the current study support the view that both Th1 and Th2 responses are induced in an optimal anti-TB-response. In the well-controlling zebrafish individuals, both Th1 and Th2-type responses were efficiently induced and showed temporal and quantitative differences compared to the Th responses of progressive individuals. The individuals that are not capable of restricting bacterial growth and, subsequently, are likely to develop a progressive disease, maintain a more Th1-biased response at all stages of infection, not reflecting an excessive Th1-type response but, instead, a lack of Th2-type induction. Our study shows that a low-dose M. marinum infection elicits different types of responses in different individuals. It is known that environmental factors, such as nutritional status or infections, influence the differentiation of T helper cells. In addition to environmental factors, it has recently been demonstrated in mice and humans that each mouse strain as well as each human individual has a genetically defined Th1/Th2 bias, and that the characteristic Th phenotype is sustained over the time [23]. Generally, genetic variation in the associated transcription factors, cytokines or cytokine receptors may define the inherent individual Th bias. It is known that there are various single nucleotide polymorphisms in the enhancer regions of human Th differentiation genes, and that these polymorphisms are related to the susceptibility to various disease states [24]. In the studied heterogeneous zebrafish population, the individual outcome of mycobacterial infection can be assumed to be affected by host genetic factors, including the inherent bias in T helper phenotype, as the variation in environmental and bacterial factors is minimal. Based on the results of this study, a Th1-type response is induced equally efficiently in both progressive and well-controlling individuals, but the lack of a Th2-type response causes the disease to progress in the (genetically) susceptible population. However, in a control experiment, in which zebrafish were infected with a high initial dose of M. marinum, we saw that the rapid growth of bacteria may also alter the Th2/Th1 balance tilting it towards Th1. Also, we saw that the high initial dose caused the total cd3 expression levels to remain low suggesting bacterium-induced T cell inhibition. In humans, it is known that M. tuberculosis can cause apoptosis of specific T cells [38] and delayed activation of CD4-positive T cells [39]. It is likely that both the bacteria and the genetic determinants of the host are capable of affecting the T cell responses in mycobacterial infection, and it is challenging to distinguish the contribution of either alone. Also, additional host factors alongside with those related T cell responses are likely to affect the disease outcome. As latent tuberculosis exists in a major part of the human population, its spontaneous reactivation is a serious global threat. Latent tuberculosis, when not initially caused by a resistant strain, can be treated with a 9-month isoniazid monotherapy that reduces the risk of reactivation by 60–90%. However, poor treatment compliance is a common problem in treating this asymptomatic disease, as only half of the patients complete therapy [40]. The poor compliance, in turn, affects the increased antibiotic resistance to isoniazid complicating the treatment of both latent and active tuberculosis. Therefore, it would be of paramount importance to be able to recognize the small population of latently infected individuals with a higher bacterial load and to allocate the treatment to only those who are most likely to benefit from it. At a late time point, at 5 months post infection, there was a zebrafish subpopulation present with a clearly more active disease, as determined by the total bacterial load and mycobacterial dormancy gene (GltA1) expression. The fish with a higher bacterial load had a lower gata3/tbx21 ratio. Based on our results, analysis of the Th1/Th2 ratio from peripheral blood mononuclear cells could provide a correlate of activity of disease among the carriers of latent M. tuberculosis infection. The risk of reactivation is also thought to increase with increasing bacterial loads [25], and thus the Th1/Th2 ratio could have predictive value in evaluating the risk of reactivation of a latent infection. The potential of the Th1/Th2 ratio as a biomarker in the human population warrants further investigation. The high induction of foxp3 expression in spontaneously reactivated individuals is in line with a previous human study showing that quantification of Foxp3 from antigen-induced peripheral blood mononuclear cells can be used to discriminate between latent and active TB [41]. During infection, regulatory T cells (Treg) have an important role in controlling excessive inflammation to prevent tissue damage, but at the same time, their immunosuppressive function can prevent bacterial clearance [42]. The role of Treg cells has been investigated during the early response to TB infection, and there is evidence that M. tuberculosis induces the expansion of antigen-specific Treg cells thus delaying the priming of effector T cells in the lymph nodes and the subsequent arrival of T cells to the infection site [43]. As M. tuberculosis is capable of such exploitation of the Treg response as part of its virulence strategy during the early TB infection, it is plausible that similar pathogen-driven expansion of antigen-specific Treg cells could also play a role in the reactivation of latent TB and the subsequent transmission of the disease. On the whole, the role of Treg cells in reactivation of latent TB is highly interesting and calls for further characterization. The existence of individuals that are able to clear mycobacterial infection illustrates that the optimal immune response to fight TB has already developed during the evolution. Adaptive mechanisms underlying mycobacterial clearance have so far remained enigmatic, and their better understanding will undoubtedly provide valuable knowledge for drug and vaccine development against tuberculosis. The zebrafish model is uniquely suitable for dissecting the natural spectrum of mycobacterial infection in large scale population studies. Analysis of the protective immunity leading to the eradication of bacteria in zebrafish can provide valuable knowledge for the development of new innovative approaches to prevention and treatment of tuberculosis. The importance of Th1-type response in controlling mycobacterial infection is generally recognized because mycobacteria are (facultative) intracellular pathogens. The general – and simplified – paradigm of the reciprocal regulation between Th1 and Th2 responses has led to the idea that Th2 response in tuberculosis might inhibit the bacterial clearance by Th1 immunity. Therefore most tuberculosis vaccines currently under development aim at promoting an efficient Th1 response and inhibiting the induction of a Th2 response [44]. In the studied zebrafish population, 10% of the individuals were able to clear the infection after the activation of adaptive responses (>2 wpi). These clearers had a similar, Th2-biased response as the other individuals in the well-controlling Low and Medium subgroups. On the other hand, inability to induce Th2 responses seems to be a trait that is associated with progressive mycobacterial infection in the zebrafish. Our finding argues against the paradigm of Th2 response not being useful for controlling tuberculosis. If this holds true in human TB, the current therapeutic and preventive approaches promoting Th1 and inhibiting Th2-type response need to be thoroughly reconsidered. For most experiments, adult (5–8 month-old) wild-type AB zebrafish were used. In addition, adult, rag1 (−/−) hu1999 mutant fish and lck: GFP transgenic fish (both from ZIRC) were used. Fish were kept in a flow-through system with a light/dark cycle of 14 h/10 h and were fed with SDS 400 food twice daily. All experiments have been accepted by the Animal Experiment Board in Finland (under the Regional State Administrative Agency for Southern Finland) and were carried out in accordance with the EU-directive 2010/63/EU on the protection of animals used for scientific purposes and with the Finnish Act on Animal Experimentation (62/2006). Permit for the zebrafish facility: LSLH-2007-7254/Ym-23, Permit for experiments: ESAVI/6407/04. 10. 03/2012, PH1267A and ESAVI/733/04. 10. 07/2013. M. marinum (ATCC 927) was cultured as described in [17]. In brief, bacteria were grown at 29°C in standard mycobacterium medium 7H9 (BD) with standard additives to an OD600 of 0. 495–0. 680 Anesthetized fish were intraperitoneally (i. p.) injected with 5 µl of bacteria suspended in sterile PBS using an Omnican 100 30 G insulin needle (Braun, Melsungen, Germany). The bacterial dose was verified by plating on 7H10 (BD) with the standard additives. The low infection dose was 21±7 cfu and the high dose 1783±364 cfu. M. marinum (ATCC 927) was transferred from 7H10 plate into 10 ml of liquid 7H9 medium with standard additives and cultured for 3–4 days at 29°C to an OD600 of 0. 490. Pelleted bacteria were resuspended in PBS corresponding to half of the original culture volume. The bacteria were heat-killed at 100°C for 20 min and thereafter homogenized for 4 min with 4000 rpm using homogenization tubes from Mobio (California, USA) and Mobio PowerLyzer24 bead beater. Samples were plated on 7H10 and LB to verify proper killing. Heat-killed bacteria were injected in a volume of 5 µl i. p using Omnican 100 30 G insulin needles (Braun). Kidney and spleen were collected from a euthanized AB fish in 20 µl of sterile PBS. The organs were gently homogenized by pipetting up and down ∼20 times. 10 µl of this suspension was injected i. p. into an anesthetized recipient rag1 (−/−) fish. For lymphocyte sorting experiments, lck: GFP fish were euthanized and their internal organs collected in ice-cold HBSS supplemented with 2% FBS (both from Life technologies, CA, USA). The tissue was mechanically disrupted by pipetting and passed through a 50 µm cell strainer to prepare single cell suspensions. Cells were washed twice with cold HBSS (+FBS), pelleted at 4°C, 300 g for 5 min and resuspended in 1 ml of the same buffer. 1 ml of Histopaque-1077 (Sigma-Aldrich, MO, USA) was then added under the cell suspension and lymphocytes and other mononuclear cells were enriched by centrifugation at room temperature for 20 min, 400 g. After centrifugation, the middle phase containing the target cells was transferred into a new tube, washed once and resuspended in HBSS (+FBS). Lck+ lymphocytes were sorted with FACSAria I (BD) (purity ≥95% based on GFP expression), collected by centrifugation and RNA was extracted using TRI reagent as described in [17]. The samples for gene expression analysis and mycobacterial quantitation were prepared using TRI reagent for DNA-RNA co-extraction (MRC, OH, USA) as previously described in [17]. RNA samples were treated with DNAse (Fermentas) according to the manufacturer' s protocol. Bacterial loads were measured by q-PCR from DNA samples using SENSIFAST NO-ROX SYBR kit with M. marinum-specific primers as described in [17]. A dilution series of DNA extracted from mycobacterial culture was included in each run to allow absolute quantification. Gene expression was measured by q-RT-PCR using Bio-Rad iScript One-Step RT-PCR Kit with SYBR Green with various primers. Host genes were normalized to glyceraldehyde 3-phosphate dehydrogenase (GAPDH) or to elongation factor 1 alpha (Ef1a), and the mycobacterial dormancy gene GltA1 was normalized to the total bacterial load. q-RT-PCR results were analyzed using the ΔCt method. The induction of host genes was compared to a baseline RNA sample extracted from a pool of healthy, non-infected zebrafish and shown as fold induction compared to average normal levels. GltA1 levels are shown in arbitrary units. Gene accession numbers and primer sequences can be found in Text S1. For Western blotting, fish were collected 4 weeks after infection with 21±7 cfu. The peritoneal cavity of the euthanized fish was emptied and the collected organs were homogenized in 1. 5 ml of TRI reagent (MRC, OH, USA) using the PowerLyzer24 bead beater. RNA-DNA co-extraction was carried out as described in [17]. After DNA extraction, the remaining interphase and organic phase were used for protein extraction according to the TRI reagent manufacturer' s protocol. In brief, proteins were precipitated by adding 3 volumes of acetone and pelleted at 12,000 g for 10 min at 4°C. The protein pellet was washed three times with 0. 8 ml of 0. 3 M guanidine hydrochloride in 95% ethanol supplemented with 2. 5% glycerol (v: v) and once with 1 ml of ethanol containing 2. 5% glycerol (v: v). For solubilization of the protein pellet, 0. 1 ml of 1% SDS per 10 mg of tissue sample was used. For Western blotting, 40 µg of total protein was resolved on a 10% SDS-PAGE gel and blotted onto Amersham Hybond ECL nitrocellulose membrane (GE Healthcare, Little Chalfont, UK). The following primary antibodies were used: anti-Gata-3 (IN) Z-Fish (AnaSpec, California, USA), anti-CXCR-3. 2 (IN) Z-Fish (AnaSpec), anti-GFP antibody NB600-303 (Novus biological, Colorado, USA). In addition, actin was detected from all the membranes with anti-actin (MAB1501) antibody (Millipore, Temecula, USA) for sample normalization. IRDye infrared secondary antibodies (LI-COR Biosciences, Nebraska, USA) and Odyssey CLx (LI-COR) were used for target protein detection and Image Studio software (LI-COR) was used for protein quantitation. A representative image of the blots showing 10 individuals can be found in the Supplementary material (Figure S4). Statistical analysis was carried out using the GraphPad Prism software (5. 02). For determination of statistical significance of differences between the different groups, a non-parametric one-tailed Mann-Whitney test was used, if not stated otherwise. P-values<0. 05 were considered significant. For estimating the predictive value of gata3/tbx21 and foxp3 expression for activity of the disease a ROC analysis was carried out with a confidence interval of 95%. AUC (area under curve) value of 0. 5 indicates no connection and 1. 0 indicates a perfect marker.

Tuberculosis is a common and potentially lethal lung disease spread worldwide. One third of the world' s population is estimated to be infected with Mycobacterium tuberculosis, yet most individuals develop a latent disease which has the potential to reactivate. Some are thought to be able to clear the infection. The current vaccine does not give adequate protection against the disease, and due to incorrect use of antibiotics, resistance to treatment has substantially increased. There is an urgent need for novel treatment approaches, such as modulation of the host' s immune response. However, the ideal immune response against tuberculosis is unknown. In addition, more accurate diagnostic tools are needed for distinguishing the high risk individuals among latent patients so that treatment could be given to those that are most likely to benefit from it. In this study, we used the Mycobacterium marinum-zebrafish model to study the T cell responses in mycobacterial infection. Utilizing the natural heterogeneity of the zebrafish population, we found associations between the disease severity (bacterial load) and the type and magnitude of T cell responses. Our results on typical T cell signatures are useful as diagnostic biomarkers as well as provide new understanding needed for therapeutic approaches based on immunomodulation.

Abstract Introduction Results Discussion Materials and Methods

bacterial diseases infectious diseases mycobacteria medicine and health sciences medical microbiology microbial pathogens biology and life sciences microbiology bacterial pathogens

2014

Adequate Th2-Type Response Associates with Restricted Bacterial Growth in Latent Mycobacterial Infection of Zebrafish

Herpesviruses include many important human pathogens such as herpes simplex virus, cytomegalovirus, varicella-zoster virus, and the oncogenic Epstein–Barr virus and Kaposi sarcoma–associated herpesvirus. Herpes virions contain a large icosahedral capsid that has a portal at a unique 5-fold vertex, similar to that seen in the tailed bacteriophages. The portal is a molecular motor through which the viral genome enters the capsid during virion morphogenesis. The genome also exits the capsid through the portal-vertex when it is injected through the nuclear pore into the nucleus of a new host cell to initiate infection. Structural investigations of the herpesvirus portal-vertex have proven challenging, owing to the small size of the tail-like portal-vertex–associated tegument (PVAT) and the presence of the tegument layer that lays between the nucleocapsid and the viral envelope, obscuring the view of the portal-vertex. Here, we show the structure of the herpes simplex virus portal-vertex at subnanometer resolution, solved by electron cryomicroscopy (cryoEM) and single-particle 3D reconstruction. This led to a number of new discoveries, including the presence of two previously unknown portal-associated structures that occupy the sites normally taken by the penton and the Ta triplex. Our data revealed that the PVAT is composed of 10 copies of the C-terminal domain of pUL25, which are uniquely arranged as two tiers of star-shaped density. Our 3D reconstruction of the portal-vertex also shows that one end of the viral genome extends outside the portal in the manner described for some bacteriophages but not previously seen in any eukaryote viruses. Finally, we show that the viral genome is consistently packed in a highly ordered left-handed spool to form concentric shells of DNA. Our data provide new insights into the structure of a molecular machine critical to the biology of an important class of human pathogens. Herpes Simplex Virus 1 and 2 (HSV-1 and HSV-2) are important human pathogens. It is estimated that approximately 90% of the world’s population are infected with one or both viruses [1]. HSV-1 is the primary cause of cold sores and HSV-2 of genital herpes. These conditions are both highly contagious, and HSV-2 is amongst the most common sexually transmitted infections. Infection with HSV is lifelong, owing to the ability of herpesviruses to enter a latent state with periodic reactivations [2]. HSV can also cause more serious conditions including keratitis, which may lead to loss of sight [3], and a potentially fatal encephalitis [4]. The herpesvirus family includes many other important human pathogens, such as varicella-zoster virus, the cause of chicken pox and shingles; cytomegalovirus, a notable cause of congenital abnormalities; Kaposi sarcoma–associated herpesvirus, which causes cancer in immune-compromised individuals; and Epstein–Barr virus, the cause of infectious mononucleosis that has also been linked to several cancers. Herpesviruses are large double-stranded DNA viruses, having genomes up to 240 kbp. The viral DNA is packaged in a complex T = 16 icosahedral capsid that is 1,250 Å in diameter [5,6]. The DNA-containing capsid, or nucleocapsid, is embedded in a proteinaceous layer known as the tegument that is in turn surrounded by a host-derived lipid envelope. The viral envelope is studded with glycoproteins that mediate viral attachment and entry. HSV virions enter host cells by fusing their envelopes with the host cell plasma membrane, allowing the nucleocapsid and tegument to enter the cytoplasm [7]. The nucleocapsid traffics along microtubules to the microtubule-organising centre, and from there, to the nucleus [8]. The nucleocapsid then docks to a nuclear pore complex, through which it injects its genome into the nucleus [9,10]. DNA egress from the capsid is through a unique portal-vertex, located at an icosahedral 5-fold symmetry axis. The portal-vertex is also the means by which the viral DNA is packaged into capsids within the nucleus [11]. Virion morphogenesis commences in the nucleus with the formation of the procapsid; an icosahedrally symmetrical spherical shell assembly of capsomeres that are hexamers (hexons) and pentamers (pentons) of the major capsid protein pUL19 (VP5) [12]. Heterotrimers of pUL38 (VP19C) and pUL18 (VP23) —termed triplexes—along with the scaffold protein pUL26, direct procapsid assembly, which nucleates around the dodecameric portal formed by pUL6 [13]. Procapsid maturation—angularisation and expulsion of the scaffold protein—occurs as the viral genome is pumped into the shell [14]. Replication of the viral genome results in formation of a concatemer, from which unit-length genomes are packaged into procapsids by the portal (pUL6) and terminase complex consisting of pUL33, pUL28, and pUL15 [15,16]. It has been noted that herpesviruses are structurally and biologically similar to DNA-containing tailed bacteriophages, and it has been suggested that these two viral groups share a common ancestry. This is based on the observation of fold conservation in the major capsid protein [17] and their similar capsid assembly and DNA-packaging strategies. Sequence analysis also indicates that pUL15 is a homolog of the phage terminase large subunit. By analogy, pUL15 is predicted to have ATPase activity, powering the translocation of the genome into the capsid through the portal, and endonuclease activity cleaving the DNA when a cleavage signal is detected [18]. pUL28 is known to bind viral DNA and is equivalent to the small terminase subunit of bacteriophages [19]. The capsid-associated tegument complex (CATC—previously termed CCSC and CVSC), is composed of pUL17, pUL25, and pUL36 and binds to the triplexes and hexons about the icosahedral 5-fold vertices of mature capsids within the nucleus [5,20–25]. Notably, it has been observed that pUL25 is essential for retention of DNA within the nucleocapsid [26,27]. Moreover, pUL25 has also been shown to be important for genome release [28]. The mature nucleocapsid leaves the nucleus by budding through the nuclear membrane via an envelopment/de-envelopment step [29]. Cytoplasmic capsids acquire further tegument proteins in the cytoplasm and are enveloped by budding into plasma membrane–derived lipid vesicles, from which they are released by exocytosis at the cell surface [30]. The structure of the HSV particle has been the subject of investigation for over 30 years [31], using cryoEM and icosahedral 3D reconstruction to determine the high-resolution features of the nucleocapsid [5,25,32,33] and lower-resolution tomography to investigate the nature of asymmetric features such as the portal-vertex and viral envelope [34,35]. Attempts to resolve the structure of the portal-vertex have been largely unsuccessful, however. This is because the herpesvirus portal-vertex is similar in size and mass to the penton-vertex, unlike bacteriophages, in which the portal-vertex is marked by the presence of a substantial tail assembly. Moreover, in herpes virions, the subtle differences between the portal-vertex and penton-vertices are obscured by the tegument layer. Finally, the high symmetry of the viral capsid dominates attempts to align particle images for asymmetric reconstruction. Here, we show the structure of the portal-vertex of HSV-1 at 8 angstroms resolution, revealed by focussed-classification [36,37] and 3D reconstruction of cryoEM images of purified virions. These data reveal that the usual pUL19 penton is replaced by a unique 5-fold symmetrical assembly. This feature displays five well-defined coiled-coil motifs, each made up of two α-helices, arranged perpendicular to the capsid surface about the 5-fold symmetry axis. It appears to be anchored to the virion by interactions with triplex-like structures that occupy the position normally taken up by peripentonal Ta triplexes, immediately about the 5-fold vertex. The CATC assembly is still present and, similarly to penton associated CATC, is bound to the Tc triplexes, forming a bridge across the periportal triplex-like structures towards the 5-fold axis. We interpret our data as showing that the pUL25 C-terminal domains are positioned differently to those seen at penton-vertices, giving rise to a small tail-like assembly that crowns the unique 5-fold vertex. Strong density was seen to extend through the portal-vertex structures that we interpret as DNA. This suggests that the trailing end of the packaged genome remains engaged in the portal-vertex, ready for release through the nuclear pore. The portal itself is not well resolved owing to a mismatch between the C5 symmetry imposed in calculating our reconstruction and the C12 symmetry of that feature. Finally, our reconstruction also reveals the arrangement of packaged DNA within the virion, which is clearly resolved as a left-handed spool arranged in concentric layers. We provide the highest-resolution view to date of a critical component of the herpesvirus virion. The portal-vertex is a molecular machine responsible for both packaging and release of the viral genome, in one of the most important groups of viral pathogens to infect humans. Furthermore, our focussed classification approach demonstrates the power of modern image-processing algorithms to break the shackles of symmetry that have limited our understanding of virus structural biology for so long. A relaxed symmetry (C5) reconstruction was calculated yielding a map at a resolution of 7. 7 Å (Fig 1b, S4 Fig, S3 Data). This revealed a uniquely structured capsid-associated tegument assembly at the portal-vertex (Fig 1f–1h, S1 Movie), giving rise to the tail-like features previously described at lower resolution, termed the PVAT [35]. At intermediate resolution, we see that the portal-vertex CATC density closely resembles that seen at the penton-vertices. The five-helix bundle, comprising pUL17, pUL25, and pUL36 is well resolved but is rotated approximately 6° counter-clockwise relative to that of the penton-vertex. In penton-vertices, this part of the CATC is anchored to the capsid by binding to two triplexes, the peripentonal Ta triplex, and the Tc triplex. At the portal-vertex, the CATC still binds to the Tc triplex; however, the Ta triplex is replaced with a globular density that appears almost twice the size of a normal triplex (pink arrow, Fig 1e and 1h). At this resolution, it is unclear whether this feature is made up of a heterotrimer of pUL38 and pUL18, plus an additional component, or whether the triplex has been entirely substituted by another protein. A candidate for this additional protein density is pUL36. This gene product was recently shown to be a component of the CATC, contributing two copies of its C-terminal domain to the five-helix bundle; however, only a small proportion of that large protein has thus far been accounted for (47 of 3,164 amino acids) [5,21]. As well as the novel triplex-like density, we see major differences in the arrangement of the C-terminal domains of pUL25 at the portal-vertex. Peripentonal CATC complexes have two globular densities that bind to penton pUL19: one on top and one to the side of each copy of the major capsid protein (black arrows, Fig 1c). Portal-vertex CATC also shows 10 globular densities that we attribute to the pUL25 C-terminal domains. These are however arranged to form two star-shaped tiers to make up the tail-like PVAT. The distal (outermost) tier being rotated approximately 36° relative to the proximal one (Fig 2, S2 Movie). As was seen in penton vertices of our icosahedral reconstruction and published structures [5,21], the C-terminal domains of pUL25 are less well resolved than the CATC five-helix bundle; in particular, the distal tier is only visible at lower threshold levels and is more clearly seen in unsharpened maps, suggesting that this feature is not rigidly constrained. Sharpening is a process of weighting data across different resolution ranges to compensate for the loss of high-resolution information during imaging and data-processing—effectively, it down-weights low-resolution features to reveal finer details. Sharpened maps are typically shown at a higher isosurface threshold level, however density arising from flexible regions may only present low-resolution features and can disappear in sharpened maps. Atomic coordinates for the CATC five-helix bundle (extracted from PDB 6CGR [5]) and the C-terminal domain of pUL25 (PDB 2F5U [38]) were docked to the PVAT density. The CATC bundle fitted the density well, suggesting that the assembly, although slightly displaced, is not radically different from that at penton vertices (Fig 1h). Given the limited resolution in the density that we attribute to the pUL25 C-terminal domains, however, docking of these coordinates is only shown to illustrate our interpretation; there is insufficient detail to produce a reliable fit (Fig 2, S2 Movie). A central slice through the 5-fold symmetrical reconstruction of the herpes simplex virion reveals the internal features of the portal-vertex and the packaged DNA (Fig 3a, S3 Movie). Lying just inside the capsid shell, we can see poorly defined density that we attribute to the portal protein pUL6 (Fig 3c and 3d). Single-particle 3D reconstruction of this assembly has previously shown it to have cyclic symmetry, forming oligomers ranging from undecamers to tetradecamers. Authentic pUL6 portals assemble as dodecamers [39]. Thus, a symmetry mismatch between this structure and the C5 capsid leads to incoherent averaging and explains the lack of high-resolution features seen in our reconstruction. Nonetheless, we can segment this feature from our density map to highlight its position and gross morphology (Fig 3f and 3g). Running through the centre of the portal-vertex along the 5-fold axis, we see strong density that extends through the portal and up against the inner surface of the PVAT. This is highlighted with a white arrow in Fig 3a and in S3 Movie. We interpret this as being the trailing end of the viral DNA, retained in the portal-vertex and ready to be ejected upon infection of a new host cell. A similar feature has been shown in the bacteriophages Ø29 and Spp1, in which specific interactions between stopper proteins and the DNA hold the end of the DNA within the tail assembly [40,41]. Lying between the portal and the pUL25 PVAT density, a novel 5-fold symmetrical assembly replaces the usual pUL19 penton. This structure is composed of five protein subunits that are largely α-helical. In each subunit, a well-resolved two-helix coiled coil extends radially and is approximately 10 nm in length (Fig 3e). This structure is anchored to the capsid through an interaction with the Ta triplex–like structures that are arranged about the portal-vertex. At this resolution, we see no evidence of an interaction with the major capsid protein. The tapered end of the portal inserts into this assembly; however, owing to the symmetry mismatch, we are unable to identify specific contacts between the two structures. The identity of the pentameric portal-vertex protein is unclear. However, as the locations of all the components of the icosahedral capsid shell and the pUL6 portal protein are known, together with the CATC proteins, pUL17, and most of pUL25, a candidate for this density is also the inner tegument protein, pUL36. As noted above, pUL36 is known to be a CATC component; however, this represents only a small proportion of this large protein, leaving the rest unaccounted for. Another possible candidate for this novel assembly is the terminase protein pUL33. In tailed phages, terminase complexes usually comprise two proteins: a small subunit that mediates binding of the terminase to the viral DNA and a large subunit (having both endonuclease and ATPase activity) that powers the translocation of the genome into the capsid through the portal and then cleaves the DNA when the head is full or a cleavage signal is detected. The HSV-1 terminase is made up of three proteins: pUL15, pUL28, and pUL33 [16]. pUL28 is functionally equivalent to the small terminase subunit [19], and pUL15 shows sequence similarity to the large terminase subunit of phage. The function of pUL33 is rather less well defined. It is predicted to be largely α-helical in structure [42], and thus we do not rule out the possibility that the novel pentameric portal-vertex protein could be pUL33. By analogy to bacteriophage terminase complexes, however, the terminase assembly is expected to detach from the portal following cleavage of the concatamer, remaining associated with the unpackaged DNA, ready to engage the portal in another empty procapsid. Furthermore, proteomics analysis has failed to detect terminase components in purified HSV-1 virions [43]. Further biochemical and structural analysis of isolated nucleocapsids and virions is therefore required to unambiguously identify this protein. We have applied a novel image-processing approach to determining the structure of an important asymmetric feature of the icosahedral herpesvirus nucleocapsid. Our data provide new insights into the organisation and composition of this important molecular machine. The herpesvirus portal-vertex is the unique site through which the viral genome both enters and exits the nucleocapsid. It is therefore a critical assembly in both virion morphogenesis and initiation of infection. We show that the PVAT, a tail-like structure previously described at low resolution, is likely composed of the C-terminal domain of pUL25, a protein known to be critical to DNA retention. We have also described two novel structures. The first, a hitherto unknown pentameric assembly, lays between the portal and PVAT in place of the pUL19 penton and has a clear two-helix coiled-coil structure. The second is a large assembly that occupies the space usually taken up by the Ta triplex in peripentonal vertices. We have suggested possible identities for both structures; however, further biochemical analysis, coupled with higher-resolution structure determination, are required to provide a definitive description. In agreement with previous lower-resolution studies that employed electron tomography, we show that the portal motor, composed of 12 copies of pUL6, lays against the inner edge of the capsid shell. Finally, our data show that HSV-1 genomic DNA is packaged as a left-handed spool arranged in concentric layers. This suggests that the process of genome translocation combined with interactions between the inner surface of the capsid and successive layers of DNA lead to a highly reproducible packing process. This is perhaps not as remarkable as one might initially surmise. Herpesviruses package their genomes to extremely high density. It is critical, then, that the genome should be able to reliably achieve this high density and, with equal reliability, eject the complete genome through the nuclear pore to initiate infection. Herpes simplex virus was propagated in BHK cells grown in roller bottles containing growth medium (GMEM, 10% FCS, 10% tryptose phosphate broth). Confluent cells were infected with HSV-1 strain 17syn+ at a m. o. i. of 0. 002 pfu/cell. Cells were incubated at 37 °C for 3 days, then harvested by shaking into the medium. Cells and media were centrifuged at 1600× g for 10 minutes to remove cell debris. The supernatant was then transferred to a new tube and spun at 17,000× g for 2 hours. The pellet was then gently resuspended on ice overnight by overlaying with 2 ml GMEM. The resuspended material was moved to a new tube and clarified by spinning at 200× g for 10 minutes. The supernatant was layered onto the top of a 5%–15% Ficoll gradient prepared in GMEM and centrifuged at 26,000× g for 2 hours. The opaque band containing the virions was collected by side puncture of the tube using an 18-gauge needle. Collected virions were diluted in GMEM and pelleted at 40,000× g for 1 hour. The pellet was washed gently in PBS and then allowed to resuspend in 50–100 μl PBS by incubating on ice for at least 1 hour. HSV virions were prepared for cryoEM by plunge freezing into liquid ethane. Four μl of purified virions was loaded onto a freshly glow-discharged holey-carbon support film (R2/2 Quantifoil) in an FEI Vitrobot mk. IV vitrification robot. The grid was immediately blotted for 3 seconds and then plunged into liquid nitrogen–cooled liquid ethane. Grids were imaged at the United Kingdom national cryoEM facility (electron bioimaging centre—eBIC) at Diamond Light Source, Harwell, in an FEI Titan Krios cryotransmission electron microscope at a nominal magnification of 81,000×. Images were recorded as ‘movies’ on a Falcon III camera operated in integrating mode with sampling of 1. 78 Å/pixel. Movies were recorded as 12-second exposures, giving a total of 480 detector frames that were integrated into 40 fractions, with a dose rate of 1. 95 e/Å2/fraction. All image processing was performed using Relion 2. 1 [46] on a GPU workstation running Linux CentOS 7. 3,702 micrograph movies were processed to correct for particle movement using Motioncor2 [47], and the defocus for each motion-corrected micrograph was estimated using GCTF [48]. A small subset of particles was picked and subjected to 2D classification to produce a template for automated particle picking, which yielded a total of 12,431 virion images for further processing. Particles were initially extracted with 2× binning. After 2D classification of this dataset 7,476 particles were selected for ab initio calculation of a starting model, followed by 3D classification (imposing full icosahedral symmetry in both cases). This led to the definition of a final dataset of 6,069 virion images that were taken forward for 3D refinement with full icosahedral symmetry. Once completed, a new dataset of particles with 1. 5× binning was extracted and used to calculate the final icosahedral reconstruction. Following icosahedral reconstruction, focused classification was performed to identify the unique portal-vertices [36,37]. A cylindrical mask was prepared in SPIDER [49] to cover a single 5-fold symmetry axis. A metadata file (STAR file) was generated to expand the symmetry for our dataset, i. e. , for each virion image, 60 orientations were defined, corresponding to the 60 symmetry-related views of the icosahedral object. To speed up the calculation, a new dataset was extracted from the raw micrographs with 5× binning. These data were then subjected to masked 3D classification, with a T value of 20, to reconstruct a single 5-fold vertex and classify the data into self-similar classes. During this process, orientations and origins were not refined. A total of 10 classes were calculated, one of which was identified as containing the unique portal-vertex. To calculate our final C5 symmetric reconstruction, we used 1. 5×-binned particle images. Resolution assessment was performed in Relion, using the postprocessing task to mask the density maps and calculate the ‘gold-standard’ Fourier shell correlation. A B-factor was estimated and applied to each reconstruction [50], which was then interpreted by visualization in UCSF Chimera [51]. Docking of atomic coordinates was performed using the ‘fit model in map’ function of UCSF Chimera. Segmentation was performed using the ‘Segger’ plugin in UCSF Chimera. The C5 reconstruction has been deposited in the EM databank with accession code EMD-4347. Motion-corrected micrographs (raw data) are deposited in EMPIAR with accession code EMPIAR-10189.

The herpesvirus family includes many important human pathogens such as herpes simplex viruses that cause cold-sores and human cytomegalovirus, a major cause of congenital abnormalities. Several herpes viruses are known to cause cancer. Herpes viruses assemble enveloped virus particles (virions) that incorporate a large DNA-containing icosahedral capsid. Virion assembly commences in the nucleus of an infected cell, where the viral genome is pumped into preassembled capsids by a portal motor that is located at a unique 5-fold symmetry axis: the portal-vertex. We have used cryogenic electron microscopy and 3D image reconstruction to solve the structure of the portal-vertex of herpes simplex virus type 1 at 8 Å resolution. Our structure reveals the presence of several previously unknown features including a novel pentameric assembly that exhibits a coiled-coil motif comprising two α-helices. This substitutes the penton, a pentamer of the major capsid protein that is seen at other 5-fold symmetry axes (penton-vertices). Our data allowed us to postulate an identity for the small tail-like structure previously identified at the portal-vertex and termed the portal-vertex–associated tegument (PVAT). Finally, our 3D reconstruction reveals that the viral DNA is packaged within the capsid as a left-handed spool that is arranged in concentric shells. Our data present a structural view of a molecular machine that plays a critical role in the replication cycle of an important family of human pathogens.

Abstract Introduction Results Materials and methods

herpes simplex virus medicine and health sciences pathology and laboratory medicine pathogens microbiology nucleocapsids viral structure viruses dna viruses herpesviruses herpes simplex virus-1 medical microbiology microbial pathogens viral packaging viral replication short reports virions virology viral pathogens biology and life sciences organisms

2018

Structure of the herpes simplex virus portal-vertex

Many plant bacterial pathogens including Pseudomonas species, utilize the type III secretion system (T3SS) to deliver effector proteins into plant cells. Genes encoding the T3SS and its effectors are repressed in nutrient-rich media but are rapidly induced after the bacteria enter a plant or are transferred into nutrient-deficient media. To understand how the T3SS genes are regulated, we screened for P. savastanoi pv. phaseolicola (Psph) mutants displaying diminished induction of avrPto-luc, a reporter for the T3SS genes, in Arabidopsis. A mutant carrying transposon insertion into a gene coding for a small functional unknown protein, designated as rhpC, was identified that poorly induced avrPto-luc in plants and in minimal medium (MM). Interestingly, rhpC is located immediately downstream of a putative metalloprotease gene named rhpP, and the two genes are organized in an operon rhpPC; but rhpP and rhpC displayed different RNA expression patterns in nutrient-rich King’s B medium (KB) and MM. Deletion of the whole rhpPC locus did not significantly affect the avrPto-luc induction, implying coordinated actions of rhpP and rhpC in regulating the T3SS genes. Further analysis showed that RhpC was a cytoplasmic protein that interacted with RhpP and targeted RhpP to the periplasm. In the absence of RhpC, RhpP was localized in the cytoplasm and caused a reduction of HrpL, a key regulator of the T3SS genes, and also reduced the fitness of Psph. Expression of RhpP alone in E. coli inhibited the bacterial growth. The detrimental effect of RhpP on the fitness of Psph and E. coli required metalloprotease active sites, and was repressed when RhpC was co-expressed with RhpP. The coordination between rhpP and rhpC in tuning the T3SS gene expression and cell fitness reveals a novel regulatory mechanism for bacterial pathogenesis. The function of RhpP in the periplasm remains to be studied. Many Gram-negative bacterial pathogens rely on the T3SS for successful infection of their hosts [1]. The T3SS is encoded by a cluster of hrp/hrc genes that are essential for the induction of a hypersensitive response (HR) in resistant and nonhost plants and pathogenicity in susceptible plants [2]. The T3SS functions as a conduit to deliver an array of effector proteins into plant cells [3]. The effectors interfere with the host defense systems and contribute to bacterial pathogenicity [4]. Some effectors elicit HR and disease resistance in plants containing cognate disease resistance genes, and are therefore named avirulence proteins [5]. The expression of hrp/hrc genes and the effector genes (together called T3SS genes hereafter) is coordinately regulated by various environmental and host factors [6]. T3SS genes are expressed at a very low level when grown in nutrient-rich medium, but quickly induced to high levels after the bacteria are infiltrated into plants or cultured in nutrient-deficient medium that is believed to resemble the environment of the plant intercellular spaces where the bacteria proliferate during infection [7,8]. As depicted in S1 Fig, the T3SS genes in P. syringae and P. savastanoi are activated by the extracytoplasmic function (ECF) -family alternate σ factor HrpL that recognizes a hrp box motif conserved in the promoters of many T3SS genes [9–11]. In turn, transcription of hrpL is controlled by a σ54-dependent promoter in an alternate σ factor RpoN-dependent manner [12]. Activation of hrpL also requires HrpR and HrpS, two homologous enhancer-binding proteins [13]. The hrpR and hrpS genes are in the same operon proceeded by the hrpR promoter [14]. HrpR and HrpS form a heterodimer that binds the hrpL promoter and induces hrpL transcription via interaction with the RpoN RNA polymerase holoenzyme [15,16]. The hrpRS operon is moderately expressed in KB and further induced in MM and in the plant [14–16]. A number of genes, including hrpA, aefR, and at least two two-component systems, gacAS and rhpRS, have been reported to regulate the transcription of the hrpRS operon [17–21]. hrpA encodes the T3SS pilus structural protein. Mutation of hrpA severely reduces the transcription of hrpRS, hrpL and T3SS genes [17]. AefR is a regulator of quorum sensing and epiphytic traits, and an aefR mutation reduces the hrpR promoter activity [18]. The GacAS system regulates multiple biological processes in various bacterial species, including motility, virulence, quorum sensing, and production of toxins, antibiotics, exopolysaccharides, and biofilms [19,20]. A mutation in the response regulator gene gacA severely reduces the expression of hrpRS and the downstream cascade genes [20]. Thus far, the mechanisms by which HrpA, AefR, and GacAS regulate the hrpRS expression remain unknown. RhpRS was identified as a negative regulator of the T3SS genes [21]. The response regulator RhpR, upon phosphorylation, binds a putative inverted repeat motif in the hrpR promoter and represses hrpRS transcription [21,22]. Dephosphorylation of RhpR reduces the binding affinity of RhpR to the hrpR promoter, and consequently the hrpRS promoter is de-repressed [21–23]. RhpS is a sensor kinase with dual activities. Under T3SS inducing conditions, RhpS presumably acts as a phosphatase that renders RhpR in the dephosphorylation state, which derepresses the hrpRS operon [21–23]. Another level of control of T3SS gene expression involves the stability of the HrpR protein, which is regulated by ATP-dependent protease Lon [24,25]. In a lon mutant, HrpR is stabilized, which elevates the expression of downstream genes [24,25]. The activity of HrpS is repressed by HrpV, a T3SS negative regulator that physically interacts with the HrpS protein [26]. HrpV-mediated repression can be suppressed by HrpG, a chaperone-like protein that interacts with HrpV [27]. The interaction of HrpG with HrpV liberates HrpS from the HrpV-repression and consequently up-regulates the T3SS cascade [27]. To investigate how Pseudomonas bacteria activate the T3SS genes in plants, a P. savastanoi pv. phaseolicola (Psph) mutant was isolated with diminished induction of avrPto-luc in Arabidopsis. We named the mutant gene rhpC (regulator of hrp, chaperone). rhpC is located immediately downstream of a putative metalloprotease gene that we called rhpP (regulator of hrp, protease). Here, we report the molecular and functional interactions between RhpP and RhpC in regulating the T3SS genes and bacterial fitness. avrPto is a Pseudomonas T3SS effector gene carrying a typical hrp-box promoter [10]. In a previous study, we fused the avrPto promoter with the firefly luciferase gene luc in a broad host plasmid and constructed a reporter system for the T3SS gene expression in plants and MM [28]. To identify bacterial genes regulating the T3SS gene expression in plants, we constructed a transposon insertion mutant library in a Psph NPS3121 strain carrying the avrPto-luc reporter gene and screened for mutants with diminished avrPto-luc induction in Arabidopsis att1 plants [18,21]. att1 was previously isolated by our group, and it supported several-fold higher induction of avrPto and hrpL promoters than did the wild type Arabidopsis [28]. att1 plants provide a quick and sensitive reporter system for the assay of Pseudomonas T3SS gene regulation in plants [18,21,28]. The Psph NPS3121 strain is a non-host bacterium to Arabidopsis Col-0, and it does not multiply in the wild-type (WT) Col-0 and att1 plants [28,29]. Using the avrPto-luc reporter and att1 plants, we previously isolated a number of T3SS regulatory genes in the Psph NPS3121 strain, including aefR, rhpRS, hrpL, hrpR, and hrpS [18,21]. One Psph mutant we identified carried a transposon insertion in PSPPH2198 [30], which we later called rhpC because of its putative role as a molecular chaperone (Fig 1A). The mutant showed reduced induction of avrPto-luc not only in the att1 plant but also in MM (Fig 1B). A deletion mutant of rhpC (ΔrhpC) was subsequently constructed, and like the transposon insertion mutant, ΔrhpC also showed a reduction of the avrPto-luc activities in att1 and MM (Fig 1C and 1D). When inoculated into the wild-type Arabidopsis Col-0 plants, the ΔrhpC mutant also showed lower induction of avrPto-luc than did the wild-type strain (Fig 1E). The bacterial numbers of the wild-type Psph and the ΔrhpC mutant were similar and did not change clearly after inoculation into the Col-0 and att1 plants (S2 Fig). Therefore, the reduced avrPto-luc activity in the ΔrhpC mutant reflected the poor induction of the reporter gene in the plants. When inoculated into host bean plants, the ΔrhpC mutant displayed reduced bacterial growth (Fig 1F) and disease symptoms (Fig 1G). The avrPto-luc induction in Arabidopsis plants and MM as well as the bacterial growth and disease symptoms in bean plants were complemented by a broad host plasmid with a constitutive promoter expressing the wild-type rhpC gene in the ΔrhpC mutant (Fig 1C–1G). rhpC encodes a protein of 99 amino acid residues [30]. BlastP analysis identified homologous proteins of unknown function in a wide range of bacterial species (S3 Fig). A cytoplasmic localization of the RhpC protein was predicted by both CELLO [31,32] and PredictProtein (https: //predictprotein. org/). This small protein has an acidic isoelectric point (pI = 4. 73) and predominantly β-sheet and α-helical secondary structures (S4 Fig). rhpC is 20 bp downstream of PSPPH2199 (Fig 1A). PSPPH2199 encodes a protein of 356 amino acid residues [30]. BlastP analysis indicated that PSPPH2199 belongs to the metalloprotease of the M4 family [33]. We therefore named PSPPH2199 as RhpP. RhpP homologues exist not only in Pseudomonas (S5 Fig) but also in a wide range of other bacteria (S6 Fig). Proteins of this family have three Zn2+-binding amino acid residues, including two histidine residues in the HEXXH motif and a unique glutamate residue toward the C-terminal from the motif [33]. Protein alignment indicated that these residues (His176, His180 and Glu200) were all conserved in RhpP (S7 and S8 Figs). Additional residues at the RhpP active site were 146,147,173,177,191,240,241, and 279 (S7 and S8 Figs). Interestingly, all bacterial strains, except one (Desulfomonile tiedjei DSM 6799), that carry the RhpC homologue also carry the RhpP homologue (S5, S6 and S8 Figs), and the two genes are arranged in the same manner as in the Psph strain, suggesting co-evolution of the rhpPC locus. However, many bacterial species, including the plant bacterial pathogens Xanthomonas and Pectobacterium, carry only the RhpP homologue without RhpC (S5 and S6 Figs). The close proximity of rhpC and rhpP genes in the bacterial genome led us to test whether the two genes are co-transcribed into a polycistronic RNA. Total RNA was extracted from the NPS3121 strains cultured in KB and MM and reverse-transcribed with a primer derived from the 3' end of rhpC (S9 Fig). RT-PCR analysis using one primer derived from the 5' end of rhpP and another primer derived from the 3' end of rhpC (S9 Fig) produced a DNA fragment of the expected size (Fig 2A), indicating that the two genes are transcribed into a polycistronic RNA. The same RNA samples were also reverse-transcribed using primers specific for rhpC and rhpP, respectively. RT-PCR with rhpC- and rhpP-specific primers also detected the respective transcripts (Fig 2A). The different intensities of the RT-PCR products from the KB and MM cultures suggested possible differential regulation of rhpC and rhpP in response to different nutrient conditions (Fig 2A). To test this possibility, real-time PCR was performed to analyze the levels of rhpP and rhpC transcripts after reverse-transcription of the RNA samples with random primers, using 16S rRNA and rpoD as internal references. The rhpP RNA was more abundant in KB than in MM (Fig 2B and 2E), and rhpC showed the opposite expression patterns compared to rhpP (Fig 2C and 2F). The ratio of the rhpC/rhpP transcripts in KB was much lower than the ratio in MM (Fig 2D and 2G), suggesting that rhpP and rhpC were expressed not only as polycistronic RNA, but also in separate forms. The transcriptions of rhpP and rhpC were likely controlled by different promoters that were differentially regulated by nutrient conditions. Polycistronic genes are often functionally related. To test whether rhpP and rhpC are related in function, deletion mutants of the whole rhpPC locus and rhpP were created by homologous recombination. The resulting mutants were tested for avrPto-luc induction and bacterial virulence. Compared to the ΔrhpC mutant, the ΔrhpPC strain displayed better induction of avrPto-luc in both the att1 plants (Fig 3A) and MM (Fig 3B). In the host bean plants, the ΔrhpPC mutant displayed better bacterial growth and stronger disease symptoms than did the ΔrhpC mutant (Fig 3C and 3D). However, bacterial growth and disease symptoms in the host plants were repressed when a plasmid-borne rhpP under a constitutive promoter was expressed in the ΔrhpPC mutant (Fig 3C and 3D). Deletion of rhpP alone did not significantly affect the bacterial virulence, as shown by the similar disease symptoms induced by the wild-type strain and the ΔrhpP mutant in the host bean plants (Fig 3D). These results suggested that RhpP, when present alone in Psph, compromised the induction of the T3SS genes and bacterial virulence, and the presence of RhpC repressed the negative effect of RhpP. The sequence of rhpC predicts a cytoplasmic protein. To test this, we tagged RhpC at the C-terminus with HA and expressed the gene in the ΔrhpC mutant strain using a broad host plasmid. Bacterial protein was separated into periplasmic, membrane, and cytoplasmic fractions. Western blot analysis indicated that RhpC, like the cytoplasmic control protein RopA [34], was exclusively localized in the cytoplasm (Fig 4A). Metalloproteases of the M4 family in bacterial pathogens are largely secreted to the extracellular milieu by their host bacteria [35]. We therefore examined whether RhpP was also a secreted protein. RhpP was FLAG-tagged at the C-terminus in the broad host vector pHM1 under a constitutive promoter and expressed in the wide-type Psph strain. The resulting bacterium was cultured in KB and MM. The bacterial cells and culture media were then separated by centrifugation, the protein in the culturing media was precipitated by TCA, and the presence of RhpP-FLAG was examined using the anti-FLAG antibody. PSPPH_1783, which is homologous to the P. aeruginosa out-secreted protein PlcN/PlcH through the type II secretion pathway, was used as the secretion control in KB [36]. The T3SS effector AvrPto was used as a secretion control in MM [37]. Unlike the control proteins that were detected in the culture media, no RhpP-FLAG protein was detected in KB or MM, even when 500 mL of culture medium was used for protein precipitation (Fig 4B), indicating that RhpP was not secreted to the outside of the bacterial cells at the tested conditions. Sequence analysis of RhpP did not predict a membrane localization signal. The other possible compartment for RhpP localization would be the cytoplasm or periplasm. We therefore separated the bacterial cells into periplasmic and cytoplasmic fractions using osmotic shock. AmiC, a protein that is dispersed throughout the new-born cells and is secreted to the periplasm in divisional cells [38], was used to indicate the protein fractionation in periplasm. The cytoplasmic protein RpoA [34] was used as control of protein fractionation in cytoplasm. Western blot analysis indicated that the RhpP protein was present in both the cytoplasm and the periplasm in the WT strain (Fig 4C). As RhpC inhibited the function of RhpP, we asked whether RhpC affects the localization of RhpP. We therefore introduced the pHM1: : rhpP-FLAG plasmid into ΔrhpP, ΔrhpC, and ΔrhpPC mutants and examined the RhpP localization. RhpP was present in both the periplasm and the cytoplasm of the ΔrhpP mutant, but was detected only in the cytoplasm of the ΔrhpC and ΔrhpPC mutants (Fig 4C), suggesting that RhpC is required for the translocation of RhpP to the periplasm. We further tested whether RhpP translocation to the periplasm was through the general secretion pathway. As shown in Fig 4D, mutations of gspD and gspE genes of the general secretion pathway did not affect the RhpP localization in the Psph strain. The rhpP-FLAG gene expressed by the pHM1 plasmid under a constitutive promoter produced stronger protein signal in the WT Psph strain than in the ΔrhpC mutant (Fig 5A), suggesting that RhpC stabilized RhpP in the bacterium. This result and the functional interactions between RhpC and RhpP in regulating the T3SS genes and bacterial virulence raised the question whether RhpC physically interacts with RhpP. We tested this using the protein pull-down assay. RhpP-FLAG was expressed in the WT Psph strain using a plasmid and purified using anti-FLAG affinity gel. The E. coli BL21 strain expressing RhpC-GST or GST was sonicated, and the protein was purified with glutathione sepharose. RhpC-GST or GST on glutathione sepharose was further incubated with the RhpP-FLAG protein eluted from the anti-FLAG affinity gel. Western blot analysis showed that RhpP-FLAG was specifically pulled down by RhpC-GST (Fig 5B), indicating the physical interaction between RhpP and RhpC. Because RhpP is a putative metalloprotease, we hypothesized that, in the absence of RhpC, RhpP represses the T3SS gene expression by destabilizing the TTSS regulatory proteins. To test this, we compared the levels of three major T3SS regulatory proteins, HrpL, HrpR, and HrpS, in the WT, ΔrhpC, ΔrhpP, and ΔrhpPC strains. All these proteins were expressed under a constitutive promoter in the pHM1 plasmid. Western blot analysis revealed a lower level of HrpL protein in the ΔrhpC mutant than in other strains (Fig 6A), which was consistent with the lower T3SS gene expression in this strain. The levels of HrpR protein were similar in all the strains (Fig 6B), and so were the levels of HrpS protein (Fig 6C). The reduced level of HrpL protein in the ΔrhpC mutant might have resulted from degradation of HrpL by RhpP. To test this possibility, we purified HrpL-FLAG and RhpP-FLAG using anti-FLAG affinity gel. The purified proteins were incubated together in PBS buffer, and the levels of HrpL-FLAG protein were examined after 3 h of incubation. In the presence of PBS buffer only, HrpL-FLAG was stable (Fig 6D). However, the level of HrpL-FLAG was significantly reduced after incubation with RhpP-FLAG (Fig 6D). The reduction of HrpL-FLAG was much suppressed when EDTA, a metalloprotease inhibitor, was added with RhpP-FLAG, or when HrpL-FLAG was incubated with RhpPH279T, a RhpP mutant of the protease active site (Fig 6D). These results indicated that RhpP was an active metalloprotease for HrpL. To test whether RhpC had any effect on the RhpP protease activity, RhpC was added to the reaction. Co-incubation of RhpC with RhpP stabilized the HrpL-FLAG protein (Fig 6E), indicating that RhpC blocked the RhpP protease activity. We consistently observed smaller colonies of the ΔrhpC mutant than the WT, ΔrhpP, and ΔrhpPC strains on the KB plate. In liquid KB or MM culture, the ΔrhpC mutant also displayed slower growth, compared to the other three strains (Fig 7A and 7B). When RhpP was tagged with MBP in the pMAL-p2X plasmid and transformed into the E. coli BL21 strain, the resulting colonies were very small on the LB plate. Liquid LB medium inoculated with these small colonies stayed clear, even after 24 h of culture (Fig 7C). These results indicated that RhpP, when presented alone, was detrimental not only to the native bacterium but also to other bacterial species. To determine whether the detrimental effect of RhpP was related to the protease activity, amino acid residue His279 at the putative active site, was mutagenized to Asp and Thr. E. coli BL21 strains expressing the mutant recombinant proteins (RhpPH279D-MBP and RhpPH279T-MBP) showed normal growth in liquid LB culture (Fig 7C). The RhpPH279D and RhpPH279T proteins were also compared to RhpP when expressed in the ΔrhpPC strain. Unlike RhpP, neither RhpPH279D nor RhpPH279T affected the bacterial growth (Fig 7D; S10 Fig). Western blot analysis indicated that the RhpPH279D and RhpPH279T proteins were as stable as RhpP in the ΔrhpPC strain (Fig 7E). These results suggested that the negative effect of RhpP on bacterial growth was mediated by the protease activity. As RhpC inhibited the RhpP protease activity, and the Psph strain harboring the rhpPC locus was normal, we speculated that the presence of RhpC could repress the detrimental effect of RhpP on bacterial growth. This possibility was further tested by the co-expression of RhpC and RhpP in the E. coli BL21 strain. pMAL-p2X plasmid expressing RhpP-MBP, when introduced alone into BL21, severely inhibited the bacterial growth. However, when this plasmid was introduced into the BL21 strain expressing RhpC, the resulted bacterium grew normally both on the LB plate and in the LB liquid medium (Fig 7F). These results suggested that RhpC suppressed the detrimental effect of RhpP on bacterial fitness, probably by inhibiting the protease activity of RhpP in the cytoplasm. Given that RhpC interacts with RhpP and inhibits its protease activity, we wondered whether mutation of the active sites of RhpP interferes with the RhpC interaction. This was tested by protein pull-down assay between GST-RhpC and RhpP mutant proteins. In addition to RhpPH279D and RhpPH279T, we created another mutant with the three conserved zinc-binding residues all changed to Ala. As shown in Fig 8A, all the mutant proteins were pulled down specifically by GST-RhpC, indicating that the mutations of the active sites did not interrupt the interaction with RhpC. We also tested whether the mutations affected the RhpP protein translocation to the periplasm. As shown in Fig 8B, the mutant proteins, like the wild type RhpP, were also translocated to the periplasm. In the search for genes regulating the T3SS genes in Psph, we identified the mutant of the rhpC gene with diminished T3SS gene induction and reduced pathogenicity in the host plant. rhpC is downstream of rhpP which encodes a putative metalloprotease. RhpC is a cytoplasmic protein that interacts with RhpP and facilitates the translocation of RhpP to the periplasm. The expression of RhpP alone in Psph and E. coli inhibited bacterial growth, and the inhibition was abolished by mutation of the conserved protease activity sites, suggesting that RhpP alone is an active protease in the bacterial cytoplasm. Purified recombinant RhpP protein can degrade HrpL in vitro, further showing RhpP as an active protease. The ΔrhpC mutant exhibited a reduced level of HrpL protein, the direct regulator of the T3SS genes. The reduction in HrpL was consistent with the reduced T3SS gene induction in the ΔrhpC mutant. Interestingly, the ΔrhpC mutant did not show a reduction in the HrpS and HrpR protein levels, suggesting that RhpP has substrate specificity. Nonetheless, HrpL is unlikely to be the only substrate for RhpP in the Psph cytoplasm, because hrpL- mutant did not show any growth defect in KB and MM [13,39]. In addition, RhpP probably degraded the E. coli proteins as well because expression of the RhpP-MBP recombinant protein in E. coli caused a severe growth defect, and the detrimental effect in E. coli required the protease active sites of RhpP. The reduced growth of the ΔrhpC mutant in the host plants was likely the result of reduced T3SS gene expression plus the reduced fitness of the bacterial cells. Protein localization analysis revealed that, in the WT Psph strain, an RhpP protein of the same size was present in the cytoplasm and the periplasm but not in the culture medium. This is different from the known RhpP homologues such as thermolysin from Bacillus thermoproteolyticus [40], elastase from Pseudomonas aeruginosa [41], Prt1 in Pectobacterium cratovorum [42], and many other M4 metalloproteases from other bacteria, which are all secreted proteins to extracellular milieu [35,43–45]. These extracellular proteases are encoded as inactive preproproteases with an N-terminal signal peptide that is followed by a propeptide and a catalytic domain. The signal peptides mediate the translocation of these preproproteases through the Sec pathway to periplasm, where the signal peptides are removed by the periplasmic-localized signal peptidase [35,43]. Immediately following the signal peptides are propeptides which are also called intramolecular chaperones [35,43]. Propeptides assist in the folding of the catalytic domains into active conformation [44,45]. Upon the completion of protein folding, propeptides are cleaved either by autoprocessing or with the help of other proteases [44]. The cleaved propeptide stays in contact with the catalytic domain, which stabilizes the substrate protein, prevents premature activation of the protease, and facilitates translocation of the protease through the outer membrane [46,47]. Alignment of RhpP with the out-secreted M4 metalloproteases showed a conserved C-terminal catalytic domain preceded by an N-terminal peptide of 67 amino acid residues. SignalP analysis did not find a signal peptide for the Sec pathway in the N-terminal sequence (S11 Fig). The propeptide sequences presented in the known M4 metalloproteases were not found in RhpP (S7 Fig). Consistently, mutations of the general secretion pathway genes gspD and gspE did not block the translocation of RhpP to the periplasm. Characterization of rhpC led us to propose that RhpC is a specific chaperone for RhpP. First, in almost all sequenced bacterial genomes carrying an RhpC homologue, the rhpC homologous gene is always coupled by the rhpP homologous gene, suggesting co-evolution of the two genes. Second, RhpC is a small cytoplasmic protein (molecular mass = 11 kDa) with a relatively low isoelectric point and a C-terminal helical structure. These characteristics are also shared by many other chaperones [48,49]. In addition, RhpC is required for RhpP protein stability and translocation from the cytoplasm to the periplasm. The presence of RhpC inhibited the RhpP protease activity and the detrimental effect of RhpP on bacterial fitness, suggesting that RhpC is required for the proper function of RhpP. The interaction of RhpC with RhpP in the protein pull-down assay provided further evidence that RhpC is the chaperone of RhpP. Based on these characteristics, RhpC appears to be a functional mimic of prepropeptides of the out-secreted metalloproteases. Although RhpC can inhibit the protease activity of RhpP, mutations of the protease active sites (His176, Glu177, His180, and His279) in RhpP did not abolish the interaction with RhpC, suggesting that these active sites are probably not at the RhpC-interacting surface. These RhpP mutant proteins were also translocated to the periplasm in the presence of RhpC. It is possible that the RhpC interaction changes RhpP to an inactive conformation that is compatible with the translocation through the bacterial inner membrane. Gram-negative bacteria have five classes of protein secretion systems spanning both the inner membrane and the outer membrane that deliver cytoplasmic proteins to the outside of the bacterial cell, including type I secretion system (T1SS), T2SS, T3SS, T4SS, and T6SS [50]. However, only two classes of protein transporters have been reported that span the inner membrane and deliver proteins to the periplasm, including the Sec apparatus and the Tat system (twin-arginine translocation system) [50,51]. The involvement of the Sec pathway in RhpP translocation was ruled out because deletion of gspE (secE) did not block the RhpP translocation to the periplasm. Proteins targeting to the Tat machinery usually have signal peptides that contain a conserved twin-arginine motif [51]. TatP analysis predicted a low Tat-targeting signal and a peptide cleavage site between A49 and R50 in the N-terminus of RhpP (S12 Fig). However, the same molecular weight of RhpP detected by Western blot in cytoplasm and periplasm did not support the cleavage of the signal peptide. Because RhpC interacts with RhpP and is required for RhpP translocation, we wondered if RhpC has a signal peptide for the Tat system and carries RhpP to the Tat system in a" piggy-back" manner. However, we did not find the twin-arginine signal peptide in RhpC. Instead, we found significant similarity between RhpC and TatB of the Tat system (S13 Fig). The functional significance of this sequence similarity is unknown. Whether RhpP passes through the cytoplasmic membrane through the Tat system remains to be investigated. Real-time PCR analysis indicated that the rhpPC locus is subjected to complex regulation in response to nutrient conditions. This locus is transcribed not only into rhpPC polycistronic RNA but also into discrete rhpC RNA. The rhpP (or rhpPC) RNA is more abundant in KB than in MM, opposite to rhpC that is expressed at a higher level in MM than in KB. The ratio of rhpC/rhpP transcripts is also much higher in MM than in KB. Given the facts that RhpP in cytoplasm can degrade the HrpL protein and that the RhpP protease activity can be repressed by RhpC, we propose that the reduced rhpP transcripts and induced rhpC transcripts in MM serve as double protection that restricts the detrimental RhpP activity to a minimal level in the cytoplasm, which in turn secures the high induction of the T3SS genes. It is worthwhile to mention that the negative regulation of RhpP on the T3SS gene induction is obvious only when RhpC is absent, because deletion of the rhpPC locus did not significantly change the induction of avrPto-luc in the plant and MM. Thus, it is the coordination between rhpP and rhpC that plays a critical role in regulating the T3SS gene induction. It appears that RhpC, as a chaperone, keeps the function of RhpP under control in the cytoplasm by inhibiting the RhpP protease and facilitating its translocation to the periplasm. RhpP may play an important role in periplasmic protein quality control by degrading the functionally abnormal proteins of bacterium or imported peptides from environment. We conducted a preliminary analysis to compare the proteomes of the wild-type Psph and ΔrhpC strains, which revealed ~30 periplasmic proteins that were >1. 5 fold more abundant in the ΔrhpC mutant (S1 Table). Most of these periplasmic proteins are ABC transporters involved in transportation of various small molecules. Whether any of these differentially expressed proteins are direct substrates of RhpP remains to be studied. Nonetheless, the presence of RhpP protein in the periplasm and cytoplasm in the wild-type Psph strain implies that the regulation of T3SS genes is coordinated with the periplasmic activities. RhpC plays a key role in protecting the T3SS regulation pathway and bacterial fitness from the detrimental effect of RhpP by directly inhibiting the RhpP protease activity in bacterial cytoplasm and targeting RhpP to the periplasm. The coordination of the two-component rhpPC module represents a novel mechanism underlying the fine regulation of T3SS genes in response to growth conditions. Arabidopsis att1 mutant plants [28] were used for the screening of Psph mutants with diminished induction of the avrPto-luc reporter. Bean (Phaseolus valgaris cv. Red Kidney) [52] was used for pathogenicity assays. Growth of plant materials was as described previously [21]. All Psph mutants were derived from the NPS3121 strain. E. coli DH5α was used for constructing all plasmids. E. coli BL21 was used for expression of recombinant proteins. NPS3121 and its derivatives were grown at 28°C in KB medium [53] containing appropriate antibiotics. E. coli strains were cultured in Luria-Bertani (LB) medium at 37°C. Antibiotics for selection of Psph strains were rifampicin, 25 mg/liter; kanamycin, 10 mg/liter; gentamicin, 10 mg/liter; and spectinomycin, 50 mg/liter. Antibiotics for selection of E. coli were ampicillin, 100 mg/liter; kanamycin, 50 mg/liter; spectinomycin, 100 mg/liter; and gentamicin, 20 mg/liter. Psph NPS3121 strain carrying the pHM2: : avrPto-luc reporter plasmid was used for construction of the EZ Tn5<KAN-2> (Epicentre, Madison, WI, U. S. A.) transposon insertion library, as described by Xiao et al. [21]. Briefly, electro-competent cells were mixed with transposon and transposase as instructed by the manufacturer. Following electroperation, the NPS3121 mutant library was plated on KB medium containing 10 mg of kanamycin per liter (selection for EZ: : TN<KAN-2> transposon) and 10 mg of spectinomycin per liter (selection for pHM2: : avrPto-luc plasmid). To screen the mutants regulating avrPto-luc reporter in plant, the NPS3121 mutant colonies were grown overnight at 25°C in liquid KB medium containing spectinomycin and kanamycin. The bacteria were washed twice with sterile water, and then suspended in sterile water to optimal density at 600 nm (OD600) = 0. 5 for inoculation. Six hour after injection into Arabidopsis att1 plants, the inoculated leaves were excised and sprayed with 1 mM luciferin dissolved in 0. 01% Tween-20, and the luciferase activity was determined using a cooled charge-couple device (CCD, Roper Scientific, Trenton, NJ, U. S. A.). A total of 6,000 colonies were screened. The transposon insertion sites were determined by a two stage semi-degenerated PCR, as described by Xiao et al [21]. To confirm the transposon insertion site in rhpC- mutant, gene-specific primer rhpC-R and transposon-specific primer Kan2-SP3 were used to PCR amplify the transposon-flanking DNA. The PCR product was sequenced to determine the transposon insertion site. Bacteria containing the avrPto-luc reporter gene were grown in KB with appropriate antibiotics to OD600 = 2. The bacteria were washed twice with sterile water, and then resuspended in sterile water to OD600 = 0. 5. To measure the reporter gene activity in plant, bacteria at OD600 = 0. 5 were injected into Arabidopsis att1 plants. The inoculated leaves were excised 6 h after injection and sprayed with 1 mM luciferin dissolved in 0. 01% Tween-20, and the luciferase activity was determined using a cooled CCD (Roper Scientific, Trenton, NJ, U. S. A.). To measure the reporter gene activity in MM, bacteria were grown in KB with appropriate antibiotics overnight to OD600 = 2. 0, washed three times with sterile water, resuspended in MM with appropriate antibiotics to OD600 = 0. 1, and then cultured at 28°C with constant shaking for 6 h before the measurement. Bacterial culture (100 μl) was mixed with 1 μl of 1 mM luciferin in a 96-well plate, and the luciferase activities were determined using a cooled CCD (Roper Scientific, Trenton, NJ, U. S. A.). The luciferase activities of all measurements were normalized to the bacterial numbers. An F test (P < 0. 05) was conducted to all quantitative experiments. Preparation of bacterial inoculum and bacterial growth assay were as previously described [21]. Psph NPS3121 wild type and mutant strains at 2×105 CFU/ml were hand-injected into the primary leaves of 10-day-old bean plants. Bacterial growth were measured at 0,2, and 4 days after inoculation. Disease symptoms on bean leaves were photographed 7 days after inoculation. Oligo primers and plasmids used in this study are listed in S2 and S3 Tables, respectively. pBluescript-HA was modified from pBluescript-SK (+) in our previous study [21] and used for HA-tagging of genes at the 3′ end. Broad host plasmids pML122 [54] and pHM1 [55] with strong constitutive promoters were used to express genes in NPS3121 strain. To construct the pML122: : rhpC-HA plasmid, the rhpC gene from the NPS3121 strain was PCR-amplified using the RhpC-HA primer pair. The PCR product was digested with HindIII and NheI and cloned into the pBluescript-HA plasmid predigested with the same enzymes. After sequence confirmation, the pBluescript: : rhpC-HA plasmid was digested by HindIII and BamHI, and the insert was subsequently cloned into pML122 predigested with the same enzymes, resulting in pML122: : rhpC-HA. To construct the pHM1: : rhpP-FLAG plasmid, the rhpP gene of the NPS3121 strain was PCR-amplified using the RhpP-FLAG (containing FLAG-tag DNA sequence) primer pair. The PCR product was digested with EcoRI and HindIII and cloned into the pGEM-T plasmid predigested with the same enzymes. After sequence confirmation, the pGEM-T: : rhpP-FLAG plasmid was digested by EcoRI and HindIII, and the insert was subsequently cloned into pHM1 predigested with the same enzymes, resulting in pHM1: : rhpP-FLAG. To construct the pHM1: : hrpL-FLAG plasmid, the hrpL gene from the NPS3121 strain was PCR-amplified using the HrpL-FLAG (containing FLAG tag DNA sequence) primers. The PCR product was digested with HindIII and EcoRI and cloned into the pGEM-T plasmid predigested with the same enzymes. After sequence confirmation, the pGEM-T: : hrpL-FLAG plasmid was digested by HindIII and EcoRI, and the insert was subsequently cloned into pHM1 predigested with the same enzymes, resulting in pHM1: : hrpL-FLAG. To construct the pHM1: : hrpS-FLAG plasmid, the hrpS gene from the NPS3121 strain was PCR-amplified using the HrpS-FLAG (containing FLAG tag DNA sequence) primers. The PCR product was digested with EcoRI and HindIII and cloned into the pGEM-T plasmid predigested with the same enzymes. After sequence confirmation, the pGEM-T: : HrpS-FLAG plasmid was digested by EcoRI and HindIII, and the insert was subsequently cloned into pHM1 predigested with the same enzymes, resulting in pHM1: : hrpS-FLAG. To construct the pHM1: : hrpR-HA plasmid, the hrpR gene from the NPS3121 strain was PCR-amplified using the HrpR-HA (containing HA-tag DNA sequence) primers. The PCR product was digested with PstI and HindIII and cloned into the pGEM-T plasmid predigested with the same enzymes. After sequence confirmation, the pGEM-T: : hrpR-HA plasmid was digested by PstI and HindIII, and the insert was subsequently cloned into pHM1 predigested with the same enzymes, resulting in pHM1: : hrpR-HA. To construct the pHM1: : PSPPH_1783-FLAG, pHM1: : AmiC-FLAG, pHM1: : RpoA-FLAG plasmid, the PSPPH_1783, AmiC (PSPPH_5159), RpoA (PSPPH_4567) were amplified from the NPS3121 strain by PCR and cloned into pHM1 using a ClonExpress II One Step Cloning Kit (Vazyme Biotech, Nanjing, China). To construct the pGEX3X: : rhpC-GST plasmid, the rhpC gene from the NPS3121 strain was PCR-amplified using the RhpC-GST primers. The PCR product was digested with BamHI and EcoRI and cloned into the pGEM-T plasmid predigested with the same enzymes. After sequence confirmation, the pGEM-T: : rhpC plasmid was digested by BamHI and EcoRI, and the insert was subsequently cloned into pGEX3X predigested with the same enzymes, resulting in pGEX3X: : rhpC-GST. To construct the pMAL-p2X: : rhpP-MBP plasmid, the rhpP gene from the NPS3121 strain was PCR-amplified using the RhpP-MBP primers. The PCR product was digested with EcoRI and BamHI and cloned into the pGEM-T plasmid predigested with the same enzymes. After sequence confirmation, the pGEM-T: : rhpP-MBP plasmid was digested by EcoRI and BamHI, and the insert was subsequently cloned into pMAL-p2X predigested with the same enzymes, resulting in pMAL-p2X: : rhpP-MBP. To construct the ΔrhpP mutant, a 1. 5-kb DNA fragment upstream of the rhpP open reading frame (ORF) was PCR-amplified using primers RhpP-FlankA (EcoRI and BamHI). A 1-kb DNA fragment downstream of rhpP ORF was PCR-amplified using primers RhpP-FlankB (BamHI and HindIII). The PCR products were digested with EcoRI and BamHI and BamHI and HindIII, respectively, and were ligated into pGEM-T, resulting in pGEM-T: : rhpPFlankA and pGEM-T: : rhpPFlankB. After sequence confirmation, the pGEM-T: : rhpPFlankA and pGEM-T: : rhpPFlankB plasmids were digested by EcoRI and BamHI and BamHI and HindIII, and the inserts were subsequently cloned into pK18mobsacB [56], resulting in pK18mobsacB: : rhpPFlankAB. This plasmid was then transformed into E. coli S17-1 which was used as a donor strain, and NPS3121 was used as a recipient strain. Both recipient strain and donor strain were cultured overnight, and then 50 μl of each were mixed and incubated for 2 days at 28°C. The bacteria were then collected and plated onto the KB plate containing rifampicin and kanamycin. After 2 days incubation, the rifampicin and kanamycin resistant colonies were single crossover merodiploid transconjugants. Pick several colonies and suspend the cells in a new tube and plate the suspension onto the KB/ rifampicin 10% sucrose plate. The sucrose rifampicin resistant colonies were regarded as the deletion mutant and further verified by PCR and sequencing. To construct the ΔrhpC mutant, a 1. 5-kb DNA fragment upstream of rhpC ORF was PCR-amplified using primers RhpC-FlankA (EcoRI and BamHI). A 1-kb DNA fragment downstream of rhpC ORF was PCR-amplified using primers RhpC-FlankB (BamHI and HindIII). The PCR products were digested with EcoRI and BamHI and BamHI and HindIII, respectively, and were ligated into pGEM-T, resulting in pGEM-T: : rhpCFlankA and pGEM-T: : rhpCFlankB. After sequence confirmation, the pGEM-T: : rhpCFlankA and pGEM-T: : rhpCFlankB plasmids were digested by EcoRI and BamHI and BamHI and HindIII, and the inserts were subsequently cloned into pK18mobsacB, resulting in pK18mobsacB: : rhpCFlankAB. The remaining procedures were the same as described for construction of the ΔrhpP mutant. To construct the ΔrhpPC mutant, a 1. 5-kb DNA fragment upstream of rhpP ORF was PCR-amplified using primers RhpP-FlankA (EcoRI and BamHI). A 1-kb DNA fragment downstream of rhpC ORF was PCR-amplified using primers RhpC-FlankB (BamHI and HindIII). The PCR products were cloned into pK18mobsacB, resulting in pK18mobsacB: : rhpCFlankAB. The remaining procedures were the same as described for construction of the ΔrhpP mutant. The same procedures described above were also used for construction of the ΔgspD and ΔgspE deletion mutants, except different primers were used to amplify DNA flanking the corresponding genes. Primers GspD-FlankA and GspD-FlankB were used for construction of ΔgspD, and primers GspE-FlankA and GspE-FlankB were used for construction of ΔgspE. To generate RhpPH279D, RhpPH279T, and RhpPH176A, E177A, H180A mutants, primer sets (RhpPH279D, RhpPH279T, and RhpPH176A, E177A, H180A) containing corresponding mutations and the pGEM-T: : rhpP-FLAG construct as template were used in PCR with the high fidelity DNA polymerase Phusion (Thermo Scientific, Waltham, MA, U. S. A.). The PCR product was digested with DMT enzyme (Transgen Biotech, Beijing, China) and then transformed into E. coli. Mutations in the resulting clones were verified by sequencing. Bacterial RNA was extracted using a modified hot phenol method [57]. RQ1 DNase (Promega, Madison, WI, U. S. A.) treatment was used to remove the contaminating DNA in RNA samples. Real-time PCR assay of gene expression was performed as described [58]. Briefly total RNA (10 μg) was reverse-transcribed using random primers. The reactions were first treated with RNase and then used as template for real-time PCR analyses with primer pairs rhpP-5' /-3 and rhpC-5' /-3 for rhpP and rhpC respectively, and primer pairs 16S rRNA, recA, and rpoD as internal controls. The relative expression levels of rhpP and rhpC were normalized to 16S rRNA, rpoD and recA. E. coli BL21 strains containing pGEX-3X-rhpC and pGEX-3X plasmids were cultured in LB at 37°C to OD600 = 0. 5 before adding 1 mM IPTG. The bacteria were then cultured overnight at 16°C to induce the production of the recombinant proteins. RhpC-GST and GST were purified using the Glutathione Sepharose ™ 4B (GE Healthcare Bio-Science AB, Uppsala, Sweden), following the manufacture’s instruction. The wild type Psph strains containing the pHM1: : rhpP-FLAG, pHM1: : rhpPH279D-FLAG, pHM1: : rhpPH279T-FLAG, and pHM1: : phpPH176A, E177A, H180A- FLAG were used to produce the corresponding RhpP and mutant proteins. The bacteria were cultured in 200 ml KB/rifampicin and spectinomycin at 28°C with constant shaking at 220 rpm. The bacteria were washed once with ice-cold PBS (10 mM Tris-HCl pH8. 0,1 mM EDTA pH 8. 0,150 mM NaCl), sonicated in ice-cold PBS, and then centrifuged at 4°C and 10,000 x g for 30 min. The supernatant was mixed with 1/1000 PBS-pre-washed anti-FLAG M2 Affinity Gel (Sigma Aldrich, St. Louis, MO, U. S. A.) and incubated at 4°C for 2 h with constant shaking. RhpP-FLAG protein bound to the beads was centrifuged, washed three times with PBS, and the protein was then washed off the beads using 3×FLAG peptide (200 μg/ml). The purified protein was examined using Western blot. RhpC-GST or GST bound to the Glutathione Sepharose beads was mixed with the purified RhpP-FLAG protein and incubated at 4°C for 2 h with constant shaking. Aliquot a small volume as input. The remaining of the mixture was centrifuged at 50 x g for 5 min, the precipitate was then washed three times (each time for 5 min) with ice-cold PBS, and then three times (each time for 5 min) with ice-cold wash solution (10 mM Tris-HCl pH8. 0,1 mM EDTA pH 8. 0, and 500 mM NaCl). Protein bound to the beads were eluted with elution buffer (20 mM reduced glutathione, 50 mM Tris-HCL pH 8. 0) and analyzed with Western blot as previously described [21]. The procedures described previously [59,60] were applied for periplasmic and cytoplasmic protein assay. Bacteria cultured in KB or MM were centrifuged and then suspended in ice-cold osmotic solution II (OS II: 20 mM Tris-HCl, pH8. 0; 2. 5 mM EDTA; 2 mM CaCl2) to OD600 = 3. 0. Aliquot 10 μl of the bacterial suspension and slowly add in 1 ml ice-cold osmotic solution I (OS I: 20 mM Tris-HCl, pH8. 0; 2. 5 mM EDTA; 2 mM CaCl2; 20% (w/v) sucrose). After 10 min of incubation in ice, the bacteria were centrifuged and then suspended in 1 ml ice-cold OS II, and continued incubation in ice for 20 min. Periplasmic contents defused into the solution during the incubation was collected by centrifugation. The bacterial pellet was suspended in 1 ml OS II. Cytoplasmic protein were released into solution by sonication and collected by ultracentrifugation. To test protein secretion into the culture medium, bacteria were first cultured in 10 ml KB medium to OD600 = 2, and then transferred into 500 ml KB for continuing culture to OD600 = 1. For secretion assay in KB medium, the bacteria were separated from the KB medium by centrifugation. For secretion assay in MM, the 500 ml KB culture were centrifuged, and the bacteria were washed three times with MM and then suspended in 500 ml MM. After 6 h culture in MM, the bacteria were separated from the MM by centrifugation. The medium was filtrated with the 0. 45 μM membrane to remove the bacterial residuals, and then mixed with equal volume of ice-cold 20% TCA. After 30 min incubation in ice, the mixture was centrifuged at 10,000 x g for 15 mins, and the pellet was washed three times with ice-cold acetone, and then boiled in SDS sample buffer for Western blot analysis. RhpP-FLAG and RhpPH279T-FLAG proteins were produced by pHM1: : rhpP-FLAG and pHM1: : rhpPH279T-FLAG plasmids, respectively, in Psph strain. HrpL-FLAG protein was produced by the pHM1: : hrpL-FLAG plasmid in E. coli BL21 strain. The proteins were purified by the anti-FLAG M2 Affinity Gel and eluted using 3×FLAG peptide (200 μg/ml). RhpC-GST and GST were purified using glutathione Sepharose beads. The purified HrpL-FLAG protein was adjusted to a concentration of 200μg/ml in PBS buffer, and all the other purified were adjusted to a concentration of 400μg/ml in PBS buffer. For each reaction, 50 μl of purified HrpL-FLAG protein was mixed with equal volume of PBS buffer or other purified proteins. EDTA was added to a final concentration of 5mM. Aliquot of 50 μl was taken from each reaction and mixed with SDS sample buffer as 0 h control, and the remaining volumes were incubated at 28°C for 3 h before examined by Western blot analysis. SDS-PAGE and Western blot analysis was conducted according to the procedures described previously [21]. Anti-FLAG antibody and anti-HA antibody were purchased from Sigma Aldrich (St Louis, MO, USA) and Abmart (Shanghai, China), respectively. Homologous sequences of RhpC and RhpP were downloaded from EnsemblBacteria [61] (http: //bacteria. ensembl. org/index. html) and the Pseudomonas Genome [62] (http: //pseudomonas. com/) databases with RhpC and RhpP from P. s. pv. phaseolicola 1448A as queries. All the sequences were aligned with ClustalW, and the UPGMA phylogenetic tree were then constructed with MEGA5 using default parameters [63]. The isoeletric point of RhpC protein was calculated with 15 methods included in Isoelectric Point Calculator (IPC, http: //isoelectric. org/calculate. php). The secondary structure of RhpC was predicted with PredictProtein (https: //predictprotein. org/), the most widely used server for structure prediction. SignalP [64] (http: //www. cbs. dtu. dk/services/SignalP/) and TatP [65] (http: //www. cbs. dtu. dk/services/TatP/) were used to predict the signal peptide and possible Tat-targeting peptide in RhpP proteins based on the classifications of bacteria. Two widely used online servers, CELLO [31,32] (http: //cello. life. nctu. edu. tw/) and PredictProtein (https: //predictprotein. org/), were used to predict the subcellular location of RhpC.

The induction of the type III secretion system (T3SS) is of great importance to the pathogenesis of bacterial pathogens in host plants. Pseudomonas savastanoi pv. phaseolicola (Psph) causes halo blight disease on beans. We discovered that the bicistronic genes in the rhpPC locus of Psph act coordinately to regulate the T3SS gene expression, bacterial fitness, and pathogenicity. rhpP encodes a metalloprotease that can degrade the key T3SS regulator protein HrpL and reduce bacterial fitness. rhpC encodes a chaperone that inhibits the RhpP activity and mediates translocation of RhpP to the periplasm. The level of rhpP RNA is high in KB but decreases in MM, but the rhpC RNA is low in KB but increases in MM. The elevated rhpC/rhpP transcript ratio in MM plus the inhibition of RhpC on RhpP activity in cytoplasm provide double insurance that warrants high induction of the T3SS genes in MM and bacterial fitness. The coordination between rhpP and rhpC reveals a new mechanism regulating bacterial pathogenicity, and may provide an important target for controlling bacterial pathogens.

Abstract Introduction Results Discussion Materials and methods

2019

Two components of the rhpPC operon coordinately regulate the type III secretion system and bacterial fitness in Pseudomonas savastanoi pv. phaseolicola

The surveillance of influenza activity is critical to early detection of epidemics and pandemics and the design of disease control strategies. Case reporting through a voluntary network of sentinel physicians is a commonly used method of passive surveillance for monitoring rates of influenza-like illness (ILI) worldwide. Despite its ubiquity, little attention has been given to the processes underlying the observation, collection, and spatial aggregation of sentinel surveillance data, and its subsequent effects on epidemiological understanding. We harnessed the high specificity of diagnosis codes in medical claims from a database that represented 2. 5 billion visits from upwards of 120,000 United States healthcare providers each year. Among influenza seasons from 2002-2009 and the 2009 pandemic, we simulated limitations of sentinel surveillance systems such as low coverage and coarse spatial resolution, and performed Bayesian inference to probe the robustness of ecological inference and spatial prediction of disease burden. Our models suggest that a number of socio-environmental factors, in addition to local population interactions, state-specific health policies, as well as sampling effort may be responsible for the spatial patterns in U. S. sentinel ILI surveillance. In addition, we find that biases related to spatial aggregation were accentuated among areas with more heterogeneous disease risk, and sentinel systems designed with fixed reporting locations across seasons provided robust inference and prediction. With the growing availability of health-associated big data worldwide, our results suggest mechanisms for optimizing digital data streams to complement traditional surveillance in developed settings and enhance surveillance opportunities in developing countries. Seasonal influenza represents an important public health burden worldwide, and even within a single year, there is substantial variation in disease burden across populations [1–3]. On the other hand, pandemic influenza, which has the potential to cause millions of fatalities, is characterized by even more uncertainty in spatio-temporal risk. Traditional influenza surveillance is guided by the World Health Organization’s global standards for the collection of virological and epidemiological influenza surveillance data [4]. Epidemiological surveillance systems play an important role in our understanding of influenza dynamics and are used to identify seasonal influenza disease burden, severity, epidemic onset and seasonality, but they often suffer from reporting delays and limited, opportunistic sampling of the population. Sentinel surveillance for influenza-like illness (ILI) is one such system that passively estimates influenza morbidity. Select general practitioners or health care facilities (“sentinels”) report aggregate counts of ILI to a centralized public health agency as an efficient means of collecting high quality data by focusing resources on a few population-representative sites [5]. The European Influenza Surveillance Network (EISN) collates sentinel ILI data from over 30 European countries, while the U. S. Centers for Disease Control and Prevention’s (CDC) ILINet surveillance system recruits roughly 2,000 sentinel physicians to submit reports on the percentage of patient visits with ILI weekly throughout the year [6–8]. While such sentinel surveillance systems are sufficient to provide situational awareness of national-level influenza activity, the coarseness of such data limit its use in local decision-making. Additionally, WHO recommends that choice in sentinel sites should consider population representativeness, geographical representation, patient volume, feasibility, and the data needs and goals of the surveillance system [4]. However, few ILI surveillance systems meet these criteria as they are limited by few incentives (e. g. , data feedback from higher-level agencies, additional support for laboratory testing) and hampered by the administrative burden of data collection. Indeed, past studies have identified discrepancies across surveillance systems [9], and have investigated strategies to limit practitioner-based biases and improve capture of true population patterns in sentinel surveillance [10–12]. Medical claims represent an alternative potential source of passive ILI surveillance data with larger volume, fewer reporting delays, and finer spatio-temporal resolution than many traditional surveillance systems [13]. Additionally, medical claims data do not require additional administrative burden or voluntary reporting to a surveillance agency. We acknowledge that it may not be possible to combine these medical data streams directly into public health systems without further consideration of the ethical and privacy concerns of integrating health data at fine spatial resolutions [14]. In the meantime, however, we can leverage these features of medical claims and combine them with statistical models to explore the most informative design of passive surveillance systems and to test the robustness of ecological inference from opportunistic samples of health-associated big data. We cannot, however, rely solely on the volume and resolution of big data to address surveillance data gaps; statistical models for ILI surveillance should also utilize information from known factors of spatial heterogeneity in influenza transmission and disease burden. Many studies have examined the relationship of environmental factors [15–21], transmission dynamics [22,23], demography and contact patterns [24–32], immune landscapes [33,34], and influenza type and subtype circulation [6,29,35–39] on influenza disease burden, although few have compared the relative importance of these mechanisms (except [40]). In addition, it is important to consider the possibility that individual patient behavior may bias the reporting of ILI disease burden, thus driving observed spatial heterogeneity. The association between poverty and social determinants [41–46], access to care, care-seeking behavior, and health insurance coverage [47–49], and reported ILI disease burden has been treated extensively elsewhere. Surveillance system design may also contribute to the biased observation of ILI disease burden [11]. Current national sentinel systems in Australia, China, the United States, and Europe capture patients seen by 5% to less than 1% of active physicians in a given population, and while these systems strive to represent population demography, spatial distributions, and patient volume as accurately as possible, this is not always possible [4,50,51]. Theoretical work suggests influenza disease burden detection could be optimized if population coverage, care-seeking rates, and geographic access to care are considered in sentinel site choice [10,11,52,53]. Non-traditional data with surveillance potential such as medical claims may enhance the estimation of attack rates through improved population coverage, better discriminate the duration of heightened epidemic activity and public health need through its real-time reporting, and improve our prediction of surge capacity needs with finer spatial resolution data. We note, however, that consideration of measurement biases is even more important as these digital data streams are opportunistic; they have greater volume and coverage in the population, but their measurement biases are less well studied [14]. Fortunately, non-traditional systems are often accompanied with metadata that provides context about the data coverage and user demographics, thus enabling explicit treatment of these potential flaws. In this study, we developed a Bayesian hierarchical influenza surveillance model that accounts for transmission, environmental, influenza-specific, and socioeconomic factors, as well as measurement processes underlying spatial heterogeneity in reported influenza-like illness across counties in the United States. This model leveraged a large-scale and highly-resolved dataset of passive ILI surveillance from medical claims, and we validated the model results using ILI sentinel surveillance from CDC. Next, we probed the robustness of this ecological inference under limited data availability in order to mimic the potential conditions of real-world sentinel surveillance systems and to improve one primary goal of surveillance —the end-of-season estimation of disease burden. Our results highlight the relative contributions of surveillance data collection and socio-environmental processes to disease reporting, and emphasize the importance of considering surveillance system design and measurement biases when using surveillance data for epidemiological inference and prediction. We validated our surveillance models of medical claims data in two ways. First we compared the model fits to CDC ILI and laboratory-confirmed surveillance data (details in Methods: ‘Model assessment and validation’). We then verified that significant socio-environmental factors identified by our models are consistent with past influenza studies. The outputs of our statistical models provide improved surveillance of U. S. county-level disease burden due to influenza-like illness from 2002 to 2010. In this section, we explore broad temporal and spatial trends of seasonal ILI, the burden of seasonal ILI among children and adults, and the burden of ILI during the 2009 H1N1 pandemic. Leveraging the large volume and spatial resolution of our data, we sought to examine the robustness of our model predictions and inference in order to assess their suitability for disease surveillance and prediction. First, we compared our estimation of epidemic intensity when using analogous models at the county and state spatial units of analysis. These comparisons recall hypothetical scenarios where inference from state-level surveillance data might inform county-level decision making in the absence of resolved county-level data. Next, two model sequences were designed to simulate different flu sentinel surveillance systems —fixed-location sentinels, where the same sentinel locations reported data every year, and moving-location sentinels, where new sentinel locations are recruited each year. A third model sequence considered the specificity of inference and model predictions to certain inclusion of historical data, thus providing insight into the generalization of our model to epidemic forecasting. We examine these applications for the total population epidemic intensity model, and ten replicates were performed for each model with missingness to generalize findings beyond that of random chance. Reliable surveillance systems are at the heart of public health preparedness, mitigation and response. In this study, we opportunistically use an administrative data source to inform influenza spatio-temporal patterns and surveillance design. Our medical claims data represented an average of 24% of all U. S. health care visits to approximately 37% of all health care providers across 95% of U. S. counties during flu season months in our study period (increasing to 38%, 70% and 96%, respectively, by 2009). We pair these data with a Bayesian hierarchical modeling approach which enables “borrowing information”, the efficient incorporation of spatial dependence and group indicators for spatial and temporal random effects. The high resolution and coverage of our data combined with this spatial statistical approach allowed us to contribute to influenza surveillance in three ways: (a) enhance fine-grain mapping of disease burden from influenza-like illness to guide local influenza preparedness and control; (b) inform the future treatment of digital data streams as a measurement process for infectious disease surveillance; and (c) systematically explore surveillance design choices. Moreover, our surveillance model enables the generation of synthetic datasets that capture realistic spatial distributions of ILI, which can be used in models to inform the design of control strategies and surveillance systems. In the process of our model validation, we also consider the relative importance of 16 environmental, demographic, or socio-economic factors in predicting influenza spatial heterogeneity. This makes ours the first large-scale influenza study to simultaneously consider multiple hypotheses across spatial scales (with the exception of work in review by Chattopadhyay et al [40]), and generates a new set of hypotheses on drivers of influenza spread. Our results strengthen the epidemiological link between humidity and influenza transmission and survival in temperate regions by finding strong negative associations between absolute humidity and both epidemic intensity and duration [54,55]. These associations were not simply influenced by the strong spatial dependence of humidity —the relative effect of this predictor remained consistent when we removed the model’s spatial dependence term (Section 2. 3 in S1 Appendix) and considered humidity as the sole model predictor (Fig AO in S1 Appendix). Charu et al. suggests that humidity may not provide additional information beyond a well-calibrated model of human mobility [56], and our work suggests that humidity among other factors is necessary to capture the end-of-season spatial heterogeneity in influenza disease burden. We also observed that higher estimated prior immunity was associated with greater epidemic intensity and longer epidemic durations. As larger epidemics induce more antigenic drift in subsequent seasons, we suggest that this drift renews population susceptibility every season, even on small spatial scales [57]. Finally, while higher vaccination coverage among toddlers was associated with lower epidemic intensity, we were surprised to note that higher vaccination coverage among elderly was associated with longer epidemics. While statistical results may be neither interpreted as causative evidence nor are free from the possibility of spurious associations, future validations of our findings on influenza epidemiology will become more possible as high volume data sources achieve wider availability and tests of multiple hypotheses become more prevalent [40]. From the perspective of surveillance operations, we acknowledge the limitations of including many predictors with disparate data sources in our model; nevertheless, we gained additional epidemiological knowledge from the multiple predictor comparisons and note that all of the data we used were publicly available annually and at the county scale. In the future, comparisons of inference between models may enable us to posit new hypotheses for epidemiological study (e. g. , vaccination of the elderly provides a protective effect among more susceptible and highly connected populations like children) (Fig AI in S1 Appendix). Our model provides fine-scale, high coverage surveillance of ILI in the United States, allowing for a better understanding of influenza spatio-temporal patterns. Through an examination of significant group effects, we observed that South Atlantic states may experience longer and more acute seasons than other parts of the U. S during both seasonal and pandemic influenza scenarios and across ILI surveillance for children and adults. Our results also suggest that county-level spatial dependence and state effects explain a substantial part of the variation in epidemic intensity, while county-level spatial dependence and season effects best capture variation in epidemic duration. The explanatory power of county spatial dependence for surveillance models in both measures adds evidence to the importance of local mobility in the spatial spread and distribution of influenza disease burden [26,56]. Moreover, we posit that state groupings explained variation in epidemic intensity because state-level policy recommendations and laws drive the probability for influenza infection and seeking of insured healthcare. For instance, influenza vaccination guidelines and access to free vaccinations are driven by local policy recommendations, and insurance policies are tied to state-level rules and regulations. Additional evidence for this hypothesis comes from our 2009 pandemic model where state effects also played a large role in explaining the variance in the data. On the other hand, variation in epidemic duration was better captured by season-level effects, and fixed effects that varied more between seasons than within them (e. g. , influenza A/H3 and B circulation) were significant, similar to other studies [1]. We hypothesize that the duration of heightened ILI activity is more closely tied to population-level susceptibility and the identities of the predominantly circulating strains —factors that are likely to vary more across seasons than across space. Our work uniquely captures factors of the measurement process, highlighting biases and disparities in healthcare-based influenza surveillance. We found that locations with greater poverty had lower influenza disease burden, in contrast to previous evidence for heightened rates of influenza-related hospitalizations, influenza-like illness, respiratory illness, neglected chronic diseases, and other measures of poor health among populations with greater material deprivation [43,44,47,58–63]. Differences in socio-economic background may change recognition and therefore reporting of disease symptoms [46,58]. Material deprivation and lack of social cohesion have also been implicated in lower rates of health care utilization for ILI, which would reduce the observation of influenza disease burden in our medical claims data among the poorest populations [44,60]. When we artificially removed counties from our model (fixed-location sentinels) or subset our data into age groups, measurement factors associated with health care-seeking behavior more strongly explained the variation in epidemic intensity among the remaining observations (Fig 4, Fig AI in S1 Appendix). These two results together suggest that statistical inference from opportunistic data samples may avoid some types of reporting biases when the coverage or volume of data achieves a minimum threshold, in response to concerns posed in [14]. Increases to claims database coverage or care-seeking behavior may reduce reporting biases by increasing the representativeness of a given location’s sample, thus highlighting the importance of collecting and using metadata from opportunistic sources of epidemiological data. Equipped with our model, we investigated the impact of surveillance system structure. We present the concept of a network of sentinel locations, in contrast to sentinel physicians or hospitals, which may be composed of administrative units (e. g. , counties) that are chosen for either their representativeness of the larger population or their status as an outlier (e. g. , match or failure to match locations in Fig 4, respectively). The ability for our model to estimate relatively accurate estimates of influenza burden across increasingly missing data suggests that routine sentinel surveillance in fixed locations may be more accurate for interpolating ILI disease burden among uncovered areas than surveillance across changing locations, even when fewer locations may be surveyed. Our framework enables sentinel counties to have flexible physician recruitment strategies, provided that county health departments can achieve target population coverage levels. Moreover, the improved performance of fixed-location surveillance systems is operationally ideal; as counties and physicians are retained as sentinels over long periods of time, we may expect the quality and consistency of reporting to improve. The accuracy of our surveillance model broke down at roughly 70% missingness among sentinels in fixed locations, which translates to fewer than 950 sentinel counties reporting data. While there are fewer sentinel counties than sentinel physicians in ILINet (approximately 2,000), we note that our county data represents aggregate reports from many healthcare providers. Indeed, the volume of visits captured by ILINet corresponded roughly to 5% of reporting counties in our medical claims data, and this level of missingness provided poor disease burden estimates for approximately 10-30% of counties in the best-case sentinel design (i. e. , fixed-locations). Our work contributes to our understanding of optimal population capture through surveillance by suggesting a framework that best maintains surveillance system design over multiple flu seasons [10–12]. Previous work acknowledges that spatial scales of aggregation alter statistical inference and statistically-identified drivers of disease distributions [64,65]. Our aggregated state surveillance models adequately captured the high epidemic intensity risk among counties in the South Atlantic, similar to other studies of spatial scale [66], but they over-estimated epidemic intensity among low-risk states, thus suggesting that these types of surveillance models may be useful for public health preparedness but less optimal for the allocation of limited resources. Nevertheless, we observed that larger discrepancies between state- and county-level surveillance models were associated with greater within-state heterogeneity in disease burden, suggesting perhaps that the spatial aggregation of data may have minimal effects on epidemiological inference and policy-making if populations and socio-environmental determinants are relatively homogeneous within a given spatial unit (Fig Q in S1 Appendix). Overall, state surveillance models seemed more prone to over-estimate than under-estimate county-level disease burden, suggesting that inference from state surveillance data is best limited to populous counties in a given state (Fig P in S1 Appendix). Future work is needed to better understand surveillance-associated aggregation biases in order to expand the utility of aggregate scale surveillance data in local contexts. Given the growing availability of health-associated big data in infectious disease surveillance [13,67], we emphasize the importance of collecting relevant metadata on system coverage and reporting, while considering the ethical and privacy implications of using these data at fine spatial resolutions [14]. In the future, statistical surveillance modeling may become standard methodology to inform the choice of sentinel locations with non-traditional high-volume digital health data, improve the long-term design of disease surveillance systems, and enhance the development of syndromic surveillance in developing countries [68]. Weekly visits for influenza-like illness (ILI) and any diagnosis from October 2002 to April 2010 were obtained from a records-level database of US medical claims managed by IMS Health and aggregated to three-digit patient US zipcode prefixes (zip3s), where ILI was defined with International Classification of Diseases, Ninth Revision (ICD-9) codes for: direct mention of influenza, fever combined with respiratory symptoms or febrile viral illness, or prescription of oseltamivir. Medical claims have been demonstrated to capture respiratory infections accurately and in near real-time [69,70], and our specific dataset was validated to independent ILI surveillance data at multiple spatial scales and age groups and captures spatial dynamics of influenza spread in seasonal and pandemic scenarios [56,71,72]. Please see Section 1 in S1 Appendix for a statement on ethics and data access. We also obtained database metadata from IMS Health on the percentage of reporting physicians and the estimated effective physician coverage by visit volume; these data were used to generate “measurement” predictors (Table 1). ILI reports and measurement factors at the zip3-level were redistributed to the county-level according to population weights derived from the 2010 US Census ZIP Code Tabulation Area (ZCTA) to county relationship file, assuming that ZCTAs that shared the first three digits belonged to the same zip3. These metadata indicated that our medical claims database represented roughly 24% of visits for any diagnosis from approximately 37% of all health care providers across 95% of U. S. counties during influenza season months, averaged over the years in our study period. We performed the following data processing steps for each county-level time series of ILI per population (Section 7 in S1 Appendix): i) Fit a LOESS curve to non-flu period weeks (flu period defined as November through March each year) to capture moderate-scale time trends (span = 0. 4, degree = 2); ii) Subtract LOESS predictions from original data to detrend the entire time series; iii) Fit a linear regression model with annual harmonic terms and a time trend to non-flu period weeks [16]; iv) Counties were defined to have an “epidemic” in a given flu season if at least two consecutive weeks of detrended ILI observations exceeded the ILI epidemic threshold during the flu period (i. e. , epidemic period) [73]. The epidemic period was the maximum length consecutive period where detrended ILI exceeded the epidemic threshold during the flu period. The epidemic threshold was the upper bound of the 95% confidence interval for the linear model prediction. Counties with a greater number of consecutive weeks above the epidemic threshold during the non-flu period than during the flu period were removed from the analysis; v) Disease burden metrics were calculated for counties with epidemics. Two measures of influenza disease burden were defined for each county. For a given season and county: We define attack rate as the sum of population-normalized and detrended ILI during the epidemic period found above (and shifted by one to accommodate the likelihood distribution). Our epidemic intensity measure is defined as the standardized ratio of this attack rate and the expected attack rate. The expected attack rate is calculated as the population-weighted mean of the observed attack rates, and is a model offset described under ‘Model structure’. Epidemic duration was defined as the number of weeks in the epidemic period and counties without epidemics were assigned the value zero. Models and data were processed separately for the 2009 H1N1 pandemic season and for state-level epidemic intensity (details in Sections 2. 5 and S2. 6 in S1 Appendix respectively). Quantifiable proxies were identified for each hypothesis found in the literature, and these mechanistic predictors were collected from probability-sampled or gridded, publicly available sources and collected or aggregated to the smallest available spatial unit among US counties, states, and Department of Health and Human Services (HHS) regions for each year or flu season in the study period, as appropriate (Table 1, Section 6 in S1 Appendix). We selected one predictor to represent each hypothesis according to the following criteria, in order: i) Select for the finest spatial resolution; ii) Select for the greatest temporal coverage for years in the study period; iii) Select for limited multicollinearity with predictors representing the other hypotheses, as indicated by the magnitude of Spearman rank cross-correlation coefficients between predictor pairs. We also compared the results of single predictor models and our final multi-predictor models as another check of multicollinearity (Section 6 in S1 Appendix). For the modeling analysis, if a predictor had missing data at all locations for an entire year, data from the subsequent or closest other survey year were replicated to fill in that year. If a predictor data source was available only at the state or region-level, all inclusive counties were assigned the corresponding state or region-level predictor value (e. g. , assign estimated percentage of flu vaccination coverage for state of California to all counties in California). Predictors were centered and standardized prior to all exploratory analyses and modeling, as appropriate. Interaction terms comprised the product of their component centered and standardized predictors. Data cleaning and exploratory data analysis were conducted primarily in R [74]. Final model predictors are described below, and our hypotheses for each predictor are described in Table 1. All cleaned predictor data are available upon request. We present the most common version of our model structure here. The generic model for county-year observations (for i counties and t years) of influenza disease burden yt is: y t | μ t, τ ∼ f (y t | μ t, τ) (1) where yt = (y1t, …, ynt) ′ denotes the vector of i = 1, …, n county observations across t = 1, …, T years included in the model (Eq 1). We modeled the mean of the observed disease burden magnitude (μt), where f (yt|μt, τ) is the distribution of the likelihood of the disease burden data, parameterized with mean μt = (μ1t, …, μnt) ′ and precision τ (where precision is the inverse of variance), as appropriate to the likelihood distribution. The proposed determinants of disease burden were modeled as: g (μ i t) = E i t + α + ∑ p = 1 m X i t p β p + γ i + ζ j [ i ] + η k [ i ] + ν t + ϕ i + ϵ i t (2) where g (.) is the link function, α is the intercept, there are m socio-environmental and measurement predictors (i. e. , Xt1, …, Xtm), where Xt1 = (X1t1, …, Xnt1) ′, and Eit is an offset of the expected disease burden, such that Eq 2 models the relative risk of disease (μit/Eit) in county i, common in disease mapping [78–80]. Group terms at the county, state j, region k, and season t levels (γi, ζj[i], ηk[i], νt, respectively) and the error term (ϵit) are independent and identically distributed (iid). Geographical proximity appears to increase the synchrony of flu epidemic timing [81,82], while connectivity between cities has been linked with spatial spread in the context of commuting and longer distance travel [83–86]. We modeled county spatial dependence ϕi with an intrinsic conditional autoregressive (ICAR) model, which smooths model predictions by borrowing information from neighbors [87]: ϕ i | ϕ j, - i, τ ϕ ∼ Normal (1 ξ i ∑ i ∼ j ϕ j, 1 ξ i τ ϕ), (3) where ξi represents the number of neighbors for node i, ϕj, −i represents the neighborhood of node i, which is composed of neighboring nodes j (neighbors denoted i ∼ j). The precision parameter is τϕ (Eq 3). The goals of our modeling approach were to i) estimate the contribution of each predictor to influenza disease burden, ii) predict disease burden in locations with missing data, and iii) improve mapping of influenza disease burden. We performed approximate Bayesian inference using Integrated Nested Laplace Approximations (INLA) with the R-INLA package (www. r-inla. org) [88,89]. INLA has demonstrated computational efficiency for latent Gaussian models, produced similar estimates for fixed parameters as established implementations of Markov Chain Monte Carlo (MCMC) methods for Bayesian inference, and been applied to disease mapping and spatial ecology questions [90–94]. Log epidemic intensity was modeled with a normal distribution, and log epidemic duration was modeled with a normal distribution without the offset term in Eq 2. Consequently, we note that all epidemic intensity models examine the relative risk of disease burden, while epidemic duration models examine the duration in weeks. Multi-season models included all terms in Eq 2. Model coefficients were interpreted as statistically significant if the 95% credible interval for a parameter’s posterior distribution failed to include zero. To assess model fit, we examined scatterplots and Pearson’s cross-correlation coefficients between observed and fitted values for the epidemic intensity and epidemic duration total population surveillance models. The epidemic intensity model fit the data well and the Pearson’s cross-correlation coefficient between the observed and fitted mean relative epidemic intensity was R = 0. 86 (Section 2 in S1 Appendix). The epidemic duration model fit relatively well, and the Pearson’s cross-correlation coefficient between the observed and predicted mean number of epidemic weeks was R = 0. 94 (Section 4 in S1 Appendix). We also examined scatterplots of standardized residuals and fitted values; standardized residuals were defined as (y - μ y ^) / σ y ^, where μ y ^ is the fitted value posterior mean and σ y ^ is the fitted value standard deviation. Residual plots for the epidemic intensity and duration models may be found in Sections 2 and 4 in S1 Appendix, respectively. For each disease burden measure, we compared models with no spatial dependence, county-level dependence only, state-level dependence only, and both county and state-level dependence. The goal of the county-level dependence was to capture local population flows, while state-level dependence attempted to capture state-level flight passenger flows (details in Section 2 in S1 Appendix). We determined that models with only county-level spatial neighborhood structure best fit the data after examining the Deviance Information Criteria (DIC) values and spatial dependence coefficients of the four model structures, further supporting evidence in [56]. County-level spatial structure was subsequently used in all final model combinations. We report results from models with county-level dependence only. We assessed the contribution of each set of group effects (i. e. , season, region, state, county, county spatial dependence, observation error) to model fit by comparing the mean precision estimates for the terms, where precision is the inverse of variance. Effects with a smaller precision captured a greater magnitude of variability in the data. We examined the added value of county-level information relative to state-level information by comparing the aggregation bias between county and state surveillance models. Here, we defined aggregation bias as the difference between fitted log epidemic intensity from state and county surveillance models. Positive values mean that the state model overestimates risk relative to the county model, and vice versa. For model validation, we compared model fitted values for epidemic intensity with CDC ILI and laboratory surveillance data, which are derived from approximately 2,000 ILI-reporting sentinel physicians and 100,000-200,000 respiratory specimens annually (details in Section 2 in S1 Appendix). We assessed model robustness through additional cross-validation and out-of-sample validation analyses; the total population epidemic intensity model was refit where 20%, 40%, 60%, 80%, 90%, 95%, and 97. 5% of all county observations were randomly replaced with NAs (sentinels in fixed locations), and where 20%, 40%, 60% and 80% of model observations were stratified by season and randomly replaced with NAs (sentinels in moving locations). We also refit three models where one, three, and five of seven flu seasons were randomly chosen and completely replaced with NAs (inclusion of historical data). To account for variability due to random chance, models were replicated ten times each with different random seeds. For each sequence of missingness, we performed out-of-sample validation by comparing the mean fitted values to the true observed values for all data that were randomly removed across seasons and replicates (Section 3. 4 in S1 Appendix). We then compared the magnitude and significance of socio-environmental and measurement drivers, and the posterior distributions of county-season fitted values. Fitted value distributions were noted as significantly different (i. e. , values did not match) if the interquartile ranges for two fitted values failed to overlap with each other (Section 3. 2 in S1 Appendix). The results described in “Sentinel surveillance design” use methods identical to this analysis and may be interpreted additionally as model sensitivity and robustness. Model estimates of disease burden, summary statistics for predictors, and their associated model codes are openly available on GitHub at https: //github. com/bansallab/optimize-flu-surveillance. All processed predictor data are available upon request.

Influenza contributes substantially to global morbidity and mortality each year, and epidemiological surveillance for influenza is typically conducted by sentinel physicians and health care providers recruited to report cases of influenza-like illness. While population coverage and representativeness, and geographic distribution are considered during sentinel provider recruitment, systems cannot always achieve these standards due to the administrative burdens of data collection. We present spatial estimates of influenza disease burden across United States counties by leveraging the volume and fine spatial resolution of medical claims data, and existing socio-environmental hypotheses about the determinants of influenza disease disease burden. Using medical claims as a testbed, this study adds to literature on the optimization of surveillance system design by considering conditions of limited reporting and spatial aggregation. We highlight the importance of considering sampling biases and reporting locations when interpreting surveillance data, and suggest that local mobility and regional policies may be critical to understanding the spatial distribution of reported influenza-like illness.

Abstract Introduction Results Discussion Methods

medicine and health sciences physicians medical doctors medical personnel influenza atmospheric science spatial epidemiology health care health care providers infectious disease control humidity public and occupational health infectious diseases epidemiology people and places infectious disease surveillance socioeconomic aspects of health professions meteorology earth sciences disease surveillance population groupings viral diseases

2018

Deploying digital health data to optimize influenza surveillance at national and local scales

Among broadly neutralizing antibodies to HIV, 10E8 exhibits greater neutralizing breadth than most. Consequently, this antibody is the focus of prophylactic/therapeutic development. The 10E8 epitope has been identified as the conserved membrane proximal external region (MPER) of gp41 subunit of the envelope (Env) viral glycoprotein and is a major vaccine target. However, the MPER is proximal to the viral membrane and may be laterally inserted into the membrane in the Env prefusion form. Nevertheless, 10E8 has not been reported to have significant lipid-binding reactivity. Here we report x-ray structures of lipid complexes with 10E8 and a scaffolded MPER construct and mutagenesis studies that provide evidence that the 10E8 epitope is composed of both MPER and lipid. 10E8 engages lipids through a specific lipid head group interaction site and a basic and polar surface on the light chain. In the model that we constructed, the MPER would then be essentially perpendicular to the virion membrane during 10E8 neutralization of HIV-1. As the viral membrane likely also plays a role in selecting for the germline antibody as well as size and residue composition of MPER antibody complementarity determining regions, the identification of lipid interaction sites and the MPER orientation with regard to the viral membrane surface during 10E8 engagement can be of great utility for immunogen and therapeutic design. The HIV-1 envelope protein (Env), a hetero-trimer of non-covalently linked gp120 and gp41 subunits, is the target of broadly neutralizing antibodies (bnAbs) [1]. BnAbs recognize several sites of vulnerability on Env [2]. The membrane proximal external region (MPER) of Env is one of its most conserved regions [3] and, hence, the focus of vaccine and therapeutic design efforts. Previously, several models have been proposed for the MPER orientation with respect to the viral membrane and for the mechanism by which MPER antibodies approach their respective epitopes in vivo [4–7]. Several reports suggested that the MPER is relatively inaccessible to antibodies in the pre-fusion conformation and is exposed only transiently after CD4 binding [4,8–10]. Other studies suggested that the MPER is partially laterally inserted into the viral membrane [6] in its pre-fusion form and that neutralization would require antibodies to" extract" the MPER from the membrane [5,7]. Early cryo-electron tomography (cryoET) of native HIV-1 [11] and SIV [12] Env on virions interpreted the MPER and transmembrane (TM) region as three independent helices organized in a tripod-like fashion followed by a turn at Lys683 (the last residue of the gp41 ectodomain; UNIPROT ID: Q70626, HIV1-LW123 numbering). In other cryoET studies [13,14], the gp41 stem was proposed to adopt a compact stalk organization within the trimer suggesting instead an extended, intertwined helical architecture for the MPER and TM region. An extended helical conformation was observed in a recent NMR structure of a construct spanning the MPER and part of the gp41 TM domain that showed no break in helicity at Lys683 [15] and also in the NMR structure of the gp41 TM, which revealed a triple-helix, quaternary TM organization in bicelles [16]. X-ray crystallography and EM [17–19] studies have described in atomic detail the structure of a soluble, stabilized Env construct (BG505 SOSIP. 664 gp140 trimer), but these SOSIP structures lack the MPER, TM and cytoplasmic domains. In a recent 4. 2 Å resolution cryo-EM structure of a native HIV-1 Env trimer (ΔCT) containing the MPER and TM domains in complex with antibody PGT151, the micelle-embedded TM domain could not be resolved, but the structure also suggested that the MPER may be inaccessible in the pre-fusion form of Env [20]. Furthermore, in the presence of MPER antibody 10E8, an 8. 8 Å resolution cryo-EM structure illustrated that the three MPER epitope regions within the trimer form a triple helix [20] and the 10E8-bound Env appears to be elevated off the micelles. This elevation in comparison to the pre-fusion form suggested how the MPER may be engaged by 10E8, but the presence of detergent and the lack of a membrane in that study limited the ability to draw definitive conclusions about 10E8 engagement of native Env on virions. Due to the proximity of MPER to the membrane, MPER binding antibodies are thought to interact with the membrane and, indeed, some have been shown to interact with lipids [21–24]. Several bnAbs target the MPER, with 10E8 and 4E10 neutralizing about 98% of all HIV-1 subtypes tested [25]. 10E8 is the most potent of the MPER antibodies and lacks the polyreactivity of 4E10. 10E8 and 4E10 target the same helical epitope (C-terminal MPER residues 671–683) but differ in their modes of binding [25–27]. Residues 672–674 adopt a 310 helical turn when engaged by 4E10, but are part of a continuous α helix (region 671–683) when bound by 10E8. The 2F5 epitope (residues 656–671) [28,29], which is N-terminal to the 10E8 and 4E10 epitopes, adopts an extended conformation with a type 1 β-turn and the Z13e1 epitope (residue 668–677) [30] consists of linked helical turns. Although these antibodies target linear MPER epitopes on gp41 [25,26,29,30], binding to individual lipids [21] or liposomes [31] also suggests that their complete epitopes include viral membrane components. Indeed, studies of MPER antibodies suggest a role for their complementarity determining region (CDR) H3 in binding to viral membrane [22,24,31–34], but no clear picture has emerged as to how each antibody is oriented with respect to the MPER and viral membrane during engagement. To determine which regions of 10E8 might interact with the HIV-1 membrane, and motivated by our recent findings regarding 4E10 interaction with lipids that compose the viral membrane [23], we undertook a crystallographic study of 10E8 in complex with lipids and an epitope scaffold [35]. We complemented our observations of 10E8-lipid binding with binding and neutralization experiments as well as structure determination of several 10E8 mutants. Our combined results have identified which 10E8 regions interact with the viral membrane and indicated that the MPER likely adopts an upright orientation with respect to the viral membrane during antibody engagement, as we also proposed for 4E10 [23]. Our high-resolution structural study increases our understanding of the relative MPER location, orientation and conformation during MPER antibody binding, and provides insights for the design of immunogens and therapeutic antibodies. Crystal structures of 10E8 in complex with T117v2 epitope scaffold and lipids 06: 0 phosphatidylglycerol (PG) (2. 37 Å resolution) or 06: 0 phosphatidic acid (PA) (2. 62 Å resolution) led to identification of a lipid-binding site at the proximity of the CDRL1 and CDRH3 loops (Fig 1A). Electron density for the lipid head groups and part of the PG acyl tail were observed (S2 Fig). The orientation of the lipid fragments with respect to 10E8 suggests that the hydrophobic lipid tails do not interact with the Fab or T117v2 and are disordered. The 06: 0 PG and 06: 0 PA head groups are bound into a crevice delineated by CDRL1 (Leu28 (L), Arg29 (L), Ser30 (L), His31 (L) and Tyr32 (L) ), FRL3 (Ala66 (L), Ser67 (L) and Gly68 (L) ), and by Trp100b (H), Ser100c (H) and Gly100d (H) at the CDRH3 tip (Fig 1B and 1C; where (L) stand for light and (H) for heavy chains). Glycerol, the cryo-protection component of the crystals, occupies the lipid-binding site in the 10E8-T117v2 structure when exogenous lipids are not added in the crystallization experiments (Fig 1D). Interestingly, the 10E8 light chain displays basic surface patches arising from somatically mutated (Arg17 (L), Arg24 (L), Arg70 (L) ), and germline-encoded (Arg29 (L), Lys51 (L), Arg61 (L) ) residues in a plane with the bound lipid head groups observed in our structures, as well as with K/R683, the last residue of the gp41 ectodomain before the TM region (Fig 2A). Thus, this basic as well as Ser/Thr/Asn rich polar surface (Fig 2B) is likely to interact with polar and negatively charged lipid head groups of the viral membrane [36]. To further investigate and validate that the light chain of 10E8 is involved in binding to the lipid head groups of the membrane upper leaflet, we designed several 10E8 light-chain mutants (Fig 2C): mutant 1 reverts the three somatically mutated arginines to 10E8 germline residues (R17Q (L), R24Q (L), R70T (L) ); mutant 2 replaces most of the lipid-proximal arginine residues with aspartate or glutamate (R24E (L), R29E (L), R70E (L), as well as R17D (L) ); mutant 3 has all the basic residues as well as some of the polar residues on this surface mutated to D or E (R17D (L), R24E (L), R29E (L), K51D (L), N52D (L), S65D (L), S67D (L), R70E (L), S76D (L) ); mutant 4 has R29 (L) and Y32 (L), which flank the lipid head groups, mutated to glutamate (R29E (L) and Y32E (L) ); and mutant 5 has R29 (L) and Y32 (L) mutated to alanine (R29A (L) and Y32A (L) ). Overall, mutants 1 to 3 were constructed to coat the presumed membrane-interacting surface of 10E8 with different amounts of negative charge to investigate binding affinity and specificity of the 10E8 epitope (protein and lipid) and to examine the effects of the mutations on neutralization potency. In addition, mutants 4 and 5 were designed to disrupt the lipid-binding site that we observe in the crystal structures. To investigate the effect of the light-chain surface mutations on 10E8 IgG binding to the T117v2 scaffold, we performed surface plasmon resonance (SPR) experiments. SPR analysis (Table 1 and S3 Fig) shows that, with the exception of mutant 4 for which the residues involved in lipid head group binding were mutated to glutamate (R29E (L) and Y32E (L) ), all other mutants retained picomolar affinity (KD) to the T117v2 scaffold (Table 1). Mutations R29E (L) and Y32E (L) in mutant 4 (1. 2 nM KD for T117v2) resulted in a ~ 40 fold reduction in binding compared to the affinity-matured 10E8 IgG (0. 029 nM for T117v2), while mutation of the same residues to alanine in mutant 5 (~0. 12 nM KD) resulted in only a 4-fold decrease in binding. As 10E8 interaction with the MPER peptide epitope is only with the heavy chain (Fig 1A), these results indicate that, except for mutant 4, all other light-chain mutants do not interfere with 10E8 binding to the MPER-scaffold. Structure determination and neutralization potency experiments with the best binding mutants were then performed to determine if changing the light-chain surface charge by mutation leads to conformational changes or compromises neutralization. Abolition or decrease in neutralization would suggest that the 10E8 light chain is oriented toward and interact with the viral membrane. We focused on mutants 1–3 and 5, as binding to the peptide-scaffolded T117v2 was decreased for mutant 4. X-ray structures of mutants 1–3 and 5 (Table 1) were determined in complex with the T117v2 scaffold at 1. 6 Å (mutant 1) and 2. 0–2. 2 Å (mutants 2,3 and 5) resolution. Superposition of the Cα atoms of the variable domains of mutants and wild-type 10E8 shows nearly identical conformations (Cα r. m. s. d. of 0. 40,0. 23,0. 31 and 0. 39 Å for mutants 1,2, 3, and 5, respectively). Thus, the mutations mainly result in different charge distributions on the light-chain surface (Fig 3A–3E), with the most negative surface observed for mutant 3. Although mutant 1 has a slightly larger r. m. s. d. (0. 4 Å), the differences in the main chain occur mainly at the Fab elbow region. Only in mutant 5 do the R29A (L) and Y32A (L) mutations produce a slight shift in the main-chain for FRL2 (residues 48,49), CDRL2 (50–56), FRL3 (57–72) and CDRL1 (24–32) with maximum Cα differences observed in CDRL2 (1. 3 Å for Lys51 (L) ), FRL3 (0. 9 Å for Gly68 (L) ) and CDRL1 (0. 9 Å between Arg29 (L) and Ala29 (L); S4 Fig). The shift for CDRL2 in mutant 5 may be the direct result of substituting Tyr32 (L) with alanine, which allows the nearby Phe48 (L) side chain to rotate and influence the conformation of Lys51 (L), Asn52 (L) and Asn53 (L). None of the mutations affect binding to T117v2 (in agreement with the SPR study, Table 1) as the mutant CDRH3 conformations are almost identical to wild-type 10E8 (S4 Fig). The four mutant structures show different bound ligands (Fig 3B–3E) depending on the residue mutated and on the components of the crystallization mother liquor and cryoprotectants (n. b. all mutants were crystallized without lipids and in different conditions). Mutant 1 has glycerol from the cryoprotectant in the lipid-binding site (Fig 3B; S2 Table; S5A–S5C Fig), as also observed for the wild-type 10E8-T117v2 complex cryoprotected with glycerol (Fig 1D; S2E and S2F Fig). In mutant 2 for which Arg29 (L) was mutated to glutamate, only water molecules seem to occupy this site when ethylene glycol was used for cryoprotection rather than glycerol. However, in mutant 3, with the same mutation (R29E (L) ), a phosphate from the crystallization condition occupies the site (S5E and S5F Fig) suggesting that replacement of Arg29 (L) with a negative charge might still retain binding to at least a phosphate, which is a component of phospholipid head groups (n. b. glycerol was absent from the buffers used with crystallization of mutant 3, and thus does not compete for phosphate binding to this site). In mutant 5, where R29 (L) and Y32 (L) are mutated to alanine (Fig 3E), glycerol (used for cryoprotection) is not observed at this location in either of the two Fabs in the asymmetric unit, suggesting that the lipid binding site is severely disrupted by these mutations. In parallel, we probed the neutralization potency of 10E8 mutants 1–3 and 5 against a panel of 109 viruses [37] to determine if these mutations in the light chain interfere with neutralization. A decrease in neutralization potency compared to wild type was observed with most of the strains for mutant 1 in which three arginines (two are proximal to the lipid-binding site) were reverted to the germline residues (Fig 3B and Table 2). A more significant decrease in neutralization was observed for mutant 2 (Fig 3C and Table 2). Increasing the amount of negatively charged residues on the light-chain surface in mutant 3 results in loss of neutralization (Fig 3D and Table 2) of most of the viruses that we tested with the breadth dropping to 34%. Interestingly, mutation of only R29 (L) and Y32 (L) to alanine in mutant 5, which was designed to disrupt the observed lipid-binding site (Fig 3E), also abrogated the neutralization of most of the viruses tested (breadth of 44%, Table 2), despite retaining picomolar binding affinity to the epitope scaffolds. Thus, mutation of basic and polar residues on the light-chain surface to acidic residues does not interfere with binding to the MPER peptide epitope, but neutralization potency and breadth decreased as the number of negatively charged residues on the surface increased. Substantial loss of neutralization occurred when the lipid-binding site was disrupted (mutant 5) or when the light-chain surface patch contained mainly acidic residues (mutant 3). We noticed that our 10E8 wild-type IgG shows about ten-fold higher neutralization potency compared with results reported previously [25]. To probe the quality of our 10E8 and mutants IgG preparations, we performed size exclusion chromatography-multi-angle light scattering (SECMALS) analysis of the samples. 10E8 and mutant 1 showed the presence of three major peaks in each sample. Each of these peaks had a molecular weight of approximately 150 kDa, consistent with the expected size of monomeric antibody (S6 Fig). We attribute this behavior to a slow conformational isomeration of 10E8, which results in differential interaction with the size exclusion matrix as previously reported for a solubility-optimized 10E8 mutant (10E8v4) [38]. Thus, our results suggest conformational isomerization for both 10E8 wild type and mutant 1. In contrast, mutants 2,3 and 5 each eluted as a single peak with a molecular weight of approximately 150 kDa. The SECMAL analysis shows that the IgG samples used for neutralization are all monomeric with only insignificant amounts (<0. 5%) of aggregates observed for wild-type 10E8 and mutant 1 (S6 Fig). The structural, neutralization and binding analyses of wild-type and light-chain 10E8 mutants suggest that the 10E8 variable light-chain residues facilitate approach to and interaction with the viral membrane. To obtain further insights into 10E8 binding to gp41 of HIV Env on the viral surface, and encompassing our recent findings on 4E10 binding to the lipids [23], we generated a model of 10E8 bound to the viral membrane-gp41 epitope assembly based on the orientation of the lipids and gp41 observed in our crystal structures (Fig 4A) using the CHARMM force field membrane builder [39]. The viral membrane head groups of the outer leaflet were roughly placed in a plane containing the basic side chains of the 10E8 light-chain, Arg17 (L), Arg24 (L), Arg29 (L) and Arg70 (L), Lys51 (L), the lipid head group, and gp41 Lys/Arg683 (Fig 4A and 4B). Our model suggests that 10E8 epitope includes the gp41 helical peptide (residues 671–683) tilted about 75–80° from the viral membrane surface and lipid head groups of the membrane. The 10E8 light chain would then face the membrane, with which it interacts via CDRH3, CDRL1, and FRL3, and possibly additional residues (Ser/Thr/Arg/Lys) of the light-chain surface (Fig 4B). Indeed, the relatively large percentage of short polar residues, serine and threonine, on this light-chain β-sheet surface form a flat polar region that can perhaps also interact with the head groups of the various lipids composing the viral membrane. Fitting of the model into the experimental EM map of the CD4-bound Env (EMDB-5455 [40]) shows that 10E8 CDRs H1 and H2 might interact with additional regions of Env (Fig 4D), as also suggested by the cryo-EM structure of 10E8 in complex with the native HIV-1 Env trimer (ΔCT) and PGT151 [20]. The MPER region of gp41, although highly conserved and targeted by several very broadly neutralizing antibodies (10E8,4E10 and 2F5), has so far not led to a vaccine immunogen that elicits such antibodies. 4E10 and 2F5 antibodies have been studied in great detail [21,26,27,29]. Their binding to phospholipids prompted the suggestion that such antibodies are rarely produced due to tolerance mechanisms resulting from interaction with ‘self-components’ [41–44]. However, the binding of 4E10 to cardiolipin [44] and its overall reactivity profile appears to differ from those of autoimmune antibodies [22,45]. 10E8, which binds to the same MPER epitope region as 4E10, does not show cross-reactivity with cardiolipin or other autoantigens [25], although binding to cholesterol-rich liposomes was recently demonstrated [31]. Our crystallographic study of 10E8 binding to two phospholipids PG and PA reveals a lipid-binding site in a cavity delineated by CDRL1, CDRH3 and FRL3 (Figs 1A and 4A). In healthy cells, PG and PA are less abundant lipids of the plasma membrane [46]. However, the amount of these lipids increases on the HIV-1 membrane that is acquired from the host cell during budding [36,47–49]. A diversity of ligands (PG, PA, phosphate, glycerol) is observed in our structures in the lipid-binding site and, therefore, other phospholipid head groups could potentially occupy this cleft. The lipid binding site residues are relatively conserved in the 10E8 germline IGLV3-19*01 sequence with only two differences observed for light-chain residues 31 (tyrosine in the germline-encoded sequence and histidine in mature 10E8) and 66 (serine in the germline and alanine in mature 10E8). The presence of the tyrosine and serine at the respective positions in germline does not influence the conformation of CDRL1 and FRL3 in this cavity as comparison with the structure of the unbound germline 10E8 (PDB 5JO5 [50]) shows that the side chains of residues 31 and 66 point away from this cleft and do not change the topology of the lipid-binding site. However, CDRH3, which also delineates this cavity, is not visible in the germline structure as its residues appear to be disordered. Thus, it is not clear if the lipid-binding site is fully recapitulated in the 10E8 germline. In addition to the lipid-binding site, we observed that the 10E8 light-chain surface is coated by several basic (arginine and lysine) and multiple polar (serine and threonine) residues, which together with Lys683, the last residue of the gp41 ectodomain, lie in a plane that roughly coincides with the lipid head groups observed in our structure. Several light-chain mutants show similar binding affinity to the epitope scaffold but with decreased neutralization potency and breadth suggesting that, although binding to the MPER peptide epitope is not compromised, binding to its composite epitope formed by MPER peptide and viral membrane lipids on the virus surface would be affected. Furthermore, the light-chain mutations do not alter the Fab conformation, consistent with the binding experiments, but result in more acidic surface, strongly suggesting that the 10E8 antibody interacts with the viral membrane via this basic-polar light-chain patch. Our data also suggest an upright orientation of the MPER helical epitope of 10E8 with respect to the viral membrane (Fig 4B), which is tilted about 78°±3° from the bilayer during antibody engagement (similar to that observed for 4E10 [23]). This angle is defined by the epitope’s helical axis intersecting the axis of the plane of the lipid head group in the direction from which 10E8 approaches the MPER in the CHARM model. This MPER orientation with respect to the membrane and the angle of antibody approach is also suggested by the recent cryo-EM structure of native JR-FL EnvΔCT at 8. 8 Å resolution that contains the MPER and TM domains in complex with 10E8 and PGT151 antibodies [20]. This cryo-EM structure provided fascinating new insights into the 10E8 interaction with Env and on new features outside the MPER that are part of the 10E8 epitope (e. g. N88 and N625 glycans), although no detailed information was possible for the 10E8 interaction with membrane lipids (or the micelle in the EM study) or for the TM region. Our study suggests that 10E8 interacts with the helical MPER with an angle of approach of ~43°±3° as measured from the viral membrane to the pseudo dyad axis between the variable light and heavy chains (defined by atoms Cβ of Ser179 and Cα of Phe100a, with the light-chain variable region interacting with the phospholipid head groups on the membrane surface. In this model, CDRH3 inserts between the MPER epitope and the lipid head groups with the aromatic residues at its tip located within the hydrophilic region of the lipid bilayer. It is likely that CDRH1, CDRH2 and FRH3 of 10E8 form additional contacts with Env as shown in our MD model fitted into the EM map of the full-length Env bound to CD4 [40] (Fig 4D) and as shown in the JR-FL EnvΔCT-10E8-PGT151 EM structure [20]. In the JR-FL EnvΔCT-10E8-PGT151 EM structure, the presence of PGT151, which binds with a stoichiometry of two Fabs per trimer, leads to a disruption in the symmetry of the spike causing the three 10E8 Fabs to have slightly different orientations compared to each other (i. e. slightly different angle of tilt toward the micelle position and rotations around the dyad axis between the heavy and light chains). In our model, the three 10E8 Fabs bind with the same angle of approach to their respective MPER-viral membrane composite epitope. The 10E8 Fab in the EM structure with the closest orientation to the one that we propose here is tilted only ~8° more towards the predicted position of the membrane surface. Gp120 conformational changes on receptor and co-receptor engagement could also promote 10E8 binding. Comparison of this 10E8 model with a previous model of 4E10 bound to the gp41 epitope-lipid bilayer (Fig 4B and 4C; [23]) shows a similar orientation of the MPER with regard to the viral membrane. 10E8 and 4E10 both interact with the residues on the same face of the helical MPER epitope, but the Fab variable regions that contact the gp41 epitope differ. The two antibodies are rotated by about 90° around their pseudo-dyad axes with respect to each other, with 10E8 interacting with the viral membrane via the light chain and CDRH3, while 4E10 interacts via CDRH1 and CDRH3. 4E10 may also make more extensive interactions with upstream regions of Env (Fig 4E) than 10E8 (Fig 4D) or most likely engage its epitope as fusion intermediate forms after receptor engagement when gp120 regions are no longer in the way. Reduction in the 10E8 interaction with other regions of Env compared with 4E10 may possibly explain its increased neutralization potency. Our combined design, structural and functional study has provided an explanation for how two extremely broad MPER antibodies engage their common epitopes at the stem of the gp41 ectodomain. The location and conformation of MPER with regard to gp41 and the membrane at different stages of viral fusion remains unclear, but the information presented here helps to fill in missing pieces of the dynamic viral fusion process. These structural and functional insights are important for design of therapeutic antibodies and immunogens as HIV vaccines. The information here can enable the MPER to be linked to lipids in an appropriate orientation as in liposomes or chimeric viruses for vaccination and provide information for improving the pharmacodynamic and pharmacokinetic properties of 10E8 for therapeutic applications. Genes for 10E8 IgG and Fabs mutants were synthesized by Genscript, Inc. All antibodies and Fabs were expressed in FreeStyle 293S cells (Invitrogen) and purified as described previously [25]. Briefly about 400 μg heavy-chain and 200 μg light-chain vectors were diluted in 25 ml Gibco Opti-MEM I (Invitrogen) reduced-serum medium, sterile filtered, and mixed with 25 ml final volume Opti-MEM I pre-incubated with 1 ml of 293fectin (Invitrogen). After 30 minutes incubation, the mixture was added to 1 L of cells (about 1. 2x106 cells/ml density) in FreeStyle 293 expression medium (Invitrogen). The Fabs were purified on a lambda light chain Capture Select affinity column pre-equilibrated with 1x PBS buffer. The unbound material was washed out with the same buffer and the bound Fabs were eluted with 0. 1 M glycine, pH 3. 0, and immediately neutralized using Tris pH 8. 0. Fractions were concentrated and buffer exchanged into 20 mM sodium acetate, pH 5. 5, then loaded on a Mono S column (GE Healthcare Life Sciences) equilibrated with the same buffer. The proteins were eluted with a linear gradient of 0 to 50%, 1M KCl in 20 mM sodium acetate, pH 5. 6. The concentrated samples were stored in 1xHBS (150 mM NaCl, 10 mM HEPES, pH 7. 4). T117v2 was expressed and purified as previously described [35]. The purified 10E8 variants were incubated in a 1: 1 molar ratio with T117v2 scaffold and purified by size exclusion chromatography using a HiLoad 16/600 SuperDex 200pg column (GE Healthcare Life Sciences) in 1xHBS buffer. Kinetics and affinities of antibody-antigen interactions were measured on a ProteOn XPR36 (Bio-Rad) using GLC Sensor Chip (Bio-Rad) and 1x HBS-EP+ pH 7. 4 running buffer (20x stock from Teknova, Cat. No H8022) supplemented with BSA at 1mg/ml. The Human Antibody Capture Kit instructions (Cat. No BR-1008-39 from GE Healthcare Life Sciences) were used to prepare chip surfaces for ligand capture. In a typical experiment, about 6000 RU of capture antibody was amine-coupled in all six flow cells of the GLC Chip. Regeneration was accomplished using 3 M magnesium chloride with 180 seconds contact time and injected four times per each cycle. Raw sensograms were analyzed using ProteOn Manager software (Bio-Rad), including interspot and column double referencing, and either Equilibrium or Kinetic fits with Langmuir model, or both, were employed when applicable. Analyte concentrations were measured on a NanoDrop 2000c Spectrophotometer using Absorption at 280 nm. Pseudoviruses were generated by transfection of 293T cells (ATCC) with an HIV-1 Env expressing plasmid and an Env-deficient genomic backbone plasmid (pSG3ΔEnv, (NIH AIDS Reagent Program 11051) ), as described previously [51]. Pseudoviruses were harvested 72 hours post-transfection for use in neutralization assays. Neutralizing activity was assessed using a single round of replication pseudovirus assay and TZM-bl target cell (NIH AIDS Reagent Program 8129). TZM-bl cells were seeded at a density of 5,000 cells/well in half-volume white luminescent 96 well plates (Costar 3688), one day prior to assay. Assay and growing medium was Complete DMEM [Dulbecco' s Modified Eagle Medium (Corning Cellgro MT15013CV) with 200 μM L-glutamine (Gibco 25030081), 100 U/ml Penicillin-Streptomycin (Invitrogen 15140–122), and ten percent fetal bovine serum (Thermo Scientific HyClone SH3091003) ]. To this plate was added pseudovirus, which was preincubated with serial dilutions of antibody for 1 hour at 37°C in duplicate with 25 μl per well final volume. Virus-infected (no serum) and uninfected cell wells were controls on each cell plate. After 24 hours, 75 μl of Complete DMEM was added to each well, bringing the total volume to 100 μl; the plates were replaced in the incubator another 48 hours. Prior to virus signal determination, the liquid medium was removed from the plates, cells were lysed with 45 μl per well Promega Cell Lysis buffer (Product number E1531), and the plates were then shaken for 10 min at 1000 RPM on a Jitterbug Microplate Incubator/Shaker. Thirty μl of Promega Flash substrate (Promega Luciferase 1000 Assay System E4550) was added per well, and luminescence was measured via Synergy 2 Multi-Mode Reader (BioTek). SECMALS analysis was performed by separating approximately 50 μg IgG on a Superdex 200 Increase 10/300 GL column (GE Healthcare Life Sciences) in phosphate buffered saline (140 mM NaCl, 2. 7 mM KCl, 5. 6 mM Na2HPO4,1. 8 mM KH2PO4, pH 7. 4) and measuring UV absorption at 280 nM and multi-angle light scattering on a DAWN HELEOS II system with Optilab T-rEX refractometer (Wyatt Technology). Raw values were background subtracted and normalized to the maximum signal intensity of each injection. Molecular weights were calculated using the ASTRA6 software package (Wyatt Technology). Data were plotted using GraphPad Prism version 7. 0a for Mac (GraphPad Software). X-ray diffraction data were collected at APS 23ID-B and 23ID-D beam lines and at SSRL on 12–2 and 11–1 beam lines (S1 and S2 Tables) and were auto-indexed and processed with HKL-2000 [52] or XDS [53]. Molecular replacement was performed with Phaser [54] using one of the 10E8 Fabs (PDB 4G6F [25]) and the T117 scaffold (PDB 3LF6 [35]) as search models. Model rebuilding in Coot [55] and refinement with Phenix [56] were performed following an initial rigid body refinement step. The refinement cycles, for structures solved between 2. 0 and 2. 6 Å, included refinement of individual atomic coordinates, cartesian simulated annealing, refinement of individual isotropic atomic displacement parameters and optimization of X-ray/stereochemistry and X-ray/ADP weights. For the 1. 6 Å structure of 10E8 mutant 1 in complex with T117v2, refinement of individual atomic coordinates, cartesian simulated annealing, occupancy and individual atomic displacement parameters refinement with anisotropic ADP for protein atoms and isotropic ADP for solvent were performed as well as optimization of X-ray/stereochemistry and X-ray/ADP weights. X-ray diffraction and refinement statistics are reported in S1 Table for the wild-type 10E8 mature-T117v2 complexes bound to lipids and in S2 Table for 10E8 mutant-T117v2 complexes. Structure figures were generated with Pymol [57]. The atomic coordinates and structure factors of 10E8-T117v2 structures have been deposited in the Protein Data Bank, with the accession codes: 5T6L (for 10E8-T117v2) and 5T85,5T80 for co-crystals with 06: 0 PG and 06: 0 PA, respectively and those for 10E8 mutants-T117v2 structures with the accession codes: 5SY8 (for 10E8 mutant 1-T117v2), 5TFW (for 10E8 mutant 2-T117v2), 5T29 (for 10E8 mutant 3-T117v2) and 5T5B (for 10E8 mutant 5-T117v2). A model of the trimeric MPER epitope-transmembrane region of the gp41 was constructed using PDB 2MOM as a template as described previously [23], with the orientation of the MPER epitope modeled base on crystal structures determined in this study. The structural information on the 10E8 Fab and the PG fragment was transferred to the trimeric model by superposing the MPER epitope region of the T117v2 scaffold to the corresponding region in the model. The crystallographic 06: 0 PG lipid fragment was extended to the size of a 1,2-dihexadecanoyl-sn-glycero-3-phospho- (1' -rac-glycerol) (DPPG) molecule by adding the lipid tails in the direction perpendicular to the plane that includes the light-chain surface residues, the head group of the PG fragment, and Lys683. The putative transmembrane region model constructed solely to anchor the lipid bilayer was then used in membrane building with CHARMM [39]. A rectangular box (x = y = 157. 7 Å) was used to generate a heterogeneous bilayer containing 400 lipids on the upper leaflet and 434 lipids on the lower leaflet of the membrane, by replacement method [58]. The HIV-1 membrane lipid composition [47] was taken into account when choosing the composition of the bilayer. The Monte Carlo method was used to place counter potassium ions and NVT (constant volume) ensemble was used during six equilibration steps at a constant temperature of 303 K. The final model has Lys683 of the MPER and the head group of the crystallographically observed lipid embedded into the head group region of the membrane outer leaflet, while the side chains of the residues of the aforementioned light-chain surface are embedded in, or touching, this hydrophilic layer. The model remained stable during equilibration steps.

The trimeric Env glycoprotein located on HIV surface is the target of broadly neutralizing antibodies and is the focus of vaccine and therapeutic approaches to prevent HIV infection. Structural studies with HIV Env trimers have shed light on the complete epitopes of several broadly neutralizing antibodies. However, structural determination of the complete epitopes of the highly cross-reactive MPER antibodies has been technically difficult due to the viral membrane component and that these epitopes are probably only exposed transiently after Env engages CD4. In this study, we structurally characterize the interaction of the broadest and most potent MPER-targeting antibody, 10E8, with viral membrane lipids and scaffolded MPER and propose how 10E8 approaches the MPER-viral membrane epitope during neutralization. Our results indicate that 10E8 interacts with the viral membrane via its light chain and engages MPER in an upright orientation with respect to the HIV-1 membrane. These findings are of interest for design of HIV-1 vaccines and therapeutics.

Abstract Introduction Results Discussion Materials and methods

medicine and health sciences immune physiology crystal structure pathology and laboratory medicine pathogens immunology condensed matter physics microbiology viral structure retroviruses viruses immunodeficiency viruses monomers (chemistry) rna viruses lipid structure crystallography antibodies immune system proteins polymer chemistry lipids solid state physics proteins medical microbiology hiv microbial pathogens chemistry hiv-1 physics biochemistry electron density virology physiology glycerol viral pathogens biology and life sciences physical sciences lentivirus organisms

2017

Lipid interactions and angle of approach to the HIV-1 viral membrane of broadly neutralizing antibody 10E8: Insights for vaccine and therapeutic design

Although it has been known for nearly a century that strains of Trypanosoma cruzi, the etiological agent for Chagas' disease, are enzootic in the southern U. S. , much remains unknown about the dynamics of its transmission in the sylvatic cycles that maintain it, including the relative importance of different transmission routes. Mathematical models can fill in gaps where field and lab data are difficult to collect, but they need as inputs the values of certain key demographic and epidemiological quantities which parametrize the models. In particular, they determine whether saturation occurs in the contact processes that communicate the infection between the two populations. Concentrating on raccoons, opossums, and woodrats as hosts in Texas and the southeastern U. S. , and the vectors Triatoma sanguisuga and Triatoma gerstaeckeri, we use an exhaustive literature review to derive estimates for fundamental parameters, and use simple mathematical models to illustrate a method for estimating infection rates indirectly based on prevalence data. Results are used to draw conclusions about saturation and which population density drives each of the two contact-based infection processes (stercorarian/bloodborne and oral). Analysis suggests that the vector feeding process associated with stercorarian transmission to hosts and bloodborne transmission to vectors is limited by the population density of vectors when dealing with woodrats, but by that of hosts when dealing with raccoons and opossums, while the predation of hosts on vectors which drives oral transmission to hosts is limited by the population density of hosts. Confidence in these conclusions is limited by a severe paucity of data underlying associated parameter estimates, but the approaches developed here can also be applied to the study of other vector-borne infections. Since the Brazilian physician Carlos Chagas discovered the parasite Trypanosoma cruzi in 1909, much research has been devoted throughout the Americas to the study of its transmission and control, primarily in the domestic and peridomestic settings in which it is passed to humans, via triatomine insect vectors of the subfamily Triatominae (Hemiptera: Reduviidae). Although control measures have succeeded in preventing new infections among humans in some areas of Brazil, Uruguay, Chile, and Argentina, the parasite, which is native to the Americas, remains endemic in sylvatic settings as far north as the United States, being limited only by the habitats of the several vector species. In each region, the epidemiology of sylvatic T. cruzi transmission differs in important particulars, as each host and vector species has certain peculiarities—behaviors or immunities—which have led to adaptations in the ways by which the infection is maintained. In the United States, sylvatic hosts (which rapid urbanization often brings into peridomestic settings) include primarily raccoons (Procyon lotor) and opossums (Didelphis virginiana) in the southeast and woodrats (Neotoma micropus) in Texas, although dogs and armadillos have also been cited as significant, and the parasite is also found in skunks, foxes, squirrels, mice, and other Neotoma spp. (Vectors do feed upon birds, reptiles and amphibians as well, but these are refractory to T. cruzi infection [1], and hence incompetent hosts.) There are over 130 species of triatomine vectors, of which 11 are known to inhabit the southern United States, 8 of them in Texas [2]. Two of the most important in the southeastern U. S. [2], [3] are Triatoma sanguisuga, found from central Texas all the way east to islands off the Atlantic coast, and Triatoma gerstaeckeri, associated primarily with woodrat nests and domestic settings from central Texas south into Mexico as far as the state of Queretaro [4]. In addition, there are different strains of T. cruzi circulating in these populations. Strains are classified within six major groups known as Type I and Type IIa through IIe. Of these, only Types I and IIa are known to circulate in the United States [5], and it is widely believed (primarily from experiments in mice, e. g. , [6]–[8]) that the strains circulating in the U. S. are less virulent than those in Latin America, where the incidence of Chagas' disease in humans is much higher: an estimated 16–18 million people (only a handful of autochthonous cases have been diagnosed in the United States [9], though it has also been estimated that as many as half a million people in the U. S. may harbor the parasite, due to migration from Latin America). Among sylvatic hosts in the United States, raccoons and other placental mammals are associated with Type IIa infections, while opossums are associated with Type I infections [5]. T. cruzi may be transmitted in a number of ways. Historically, the primary infection route, especially in South America, has involved the vector' s feeding process, in which a bloodmeal from an infected host can transmit the parasite to the vector, where it lives in the insect' s gut, and defecation by an infected vector on the host following the bloodmeal can result in stercorarian transmission to the host. In sylvatic hosts this may occur when the animal scratches the bite and inadvertently rubs the parasite-contaminated matter into the lesion. However, among humans there have recently been other transmission avenues of greater concern: the parasite can be passed from one human to another through blood transfusion and organ transplants, congenitally from mother to child through the placenta, and oral transmission by consumption of food contaminated by vectors has been blamed for outbreaks in South America. In fact, these avenues of transmission may also be important for sylvatic hosts as well: vertical (congenital) transmission has been verified experimentally among rats [10] and supported by circumstantial evidence among lemurs [11] and other animals, and oral transmission to hosts through their predation upon vectors (raccoons, opossums, and even woodrats are opportunistic feeders that commonly include insects in their diets) has even been suggested by some [12], [13] to be the primary means of T. cruzi transmission to hosts in some cycles in the U. S. Indeed, T. sanguisuga and T. gerstaeckeri are known to be so cautious in their feeding behavior as to avoid climbing up entirely onto hosts during feeding [3], and often defecate 30 minutes or more after feeding ends, making them likely to be rather inefficient at stercorarian transmission to hosts. Both oral and stercorarian transmission to hosts, however, as well as bloodborne transmission to vectors, may be amplified by changes in vector behavior caused by infection with T. cruzi. Many disease vectors are known to increase their feeding rate when infected, due to parasites building up inside their digestive tracts and impeding feeding. This behavior has been verified for one species of triatomine vector and trypanosome [14], but not documented for Chagas vectors and T. cruzi. Many of the still-unanswered questions regarding sylvatic T. cruzi transmission cycles may be exceptionally difficult to address through direct observation in the laboratory and field: for instance, which of the several transmission pathways is really dominant in each cycle? (We may think of a cycle as a specified host, vector, parasite strain, and geographic region, although in practice such cycles communicate with each other, primarily via vector dispersal.) Mathematical models have proven a useful tool in many fields, including ecology and epidemiology, as they can describe, predict, and provide evaluation measures for phenomena which may be difficult to observe directly. Population biology models consisting of dynamical systems (usually systems of differential equations, see, e. g. , [15]), which describe the spread and growth of populations over time, have made notable contributions to disease control beginning notably with Ronald Ross' s study of malaria transmission in the early 1900s [16], for which he later won the Nobel Prize. Such mathematical modeling of T. cruzi transmission has to date involved primarily household-based modeling of vector infestations and human infection (but see below for a notable exception), although in the past decade geospatial models have been developed to describe vector distribution, disease risk, and relevant ecological niches [2], [17]. The ability of mathematical models to explain and predict depends not only on the underlying assumptions about the biological processes (demographic, infection-related and other) used to construct them, but also on knowing the values of certain fundamental parameters, most of which can be observed directly: information such as average lifespan, population density, or the probability of a host becoming infected from consuming an infected vector. For instance, the ability of a given population to invade or persist in a habitat often depends on threshold quantities such as a reproductive number (which can be calculated in terms of these fundamental parameters) being above or below a critical value. The best-known of these is the basic reproduction number for an infection or population [18], [19], denoted, which typically signals persistence of the population precisely when. In practice, however, the parameters' values for a given transmission cycle change seasonally, from one region to another, and even from study to study (especially if sample sizes are small). As a result, the critical link between theoretical models and empirical data provided by parameter estimation requires a broad perspective and familiarity with a range of empirical literature. As noted above, numerous mathematical modeling studies have been published of T. cruzi transmission to humans (e. g. , [20]–[22]), but almost none have been published on the sylvatic transmission cycles that maintain the parasite. Decades of studies have established details of the life cycles of T. cruzi hosts and vectors in the United States, but studies focused on measuring infection parameters are only just beginning to appear (e. g. , [13]). Mathematical models can bridge this gap by facilitating calculation of these parameters using enzootic prevalence observations together with known information on the life histories of host and vector species. The aims of the present study are to estimate values for those measures of host and vector life histories and T. cruzi infection which have been observed directly in the literature via an extensive review, and then to illustrate a method by which other key infection-related parameters can be calculated using mathematical models. One of the important aspects of the sylvatic T. cruzi transmission cycle which models can help investigate is density dependence in the infection rates. (In this paper the term “rate” refers to a frequency per unit of time at which an event occurs. The term “proportion” will be used to refer to ratios which do not involve time, such as disease prevalence.) Infectious disease transmission is driven by contact processes between susceptible and infective individuals, and sylvatic transmission of T. cruzi in particular depends on both the vector-initiated process of taking bloodmeals and the host-initiated process of predation on vectors. The rates at which these two contacts occur depend in part on the host and vector population densities, and in part on the ratio of those densities, due to the saturation that occurs when this ratio is too high or too low. That is, the per capita contact rate is a function of the vector-host density ratio, so that the total contact rate is the product of this function and the respective (host or vector) density. Ratio-dependent contact rates, which were used in epidemiological models as early as Ross' s classic malaria model [16], are also a well-established notion in the study of predator-prey systems [23], [24], and the present study will illustrate how these correspond to the density-dependent effects observed in the transmission of T. cruzi (e. g. , [25]). Saturation in contact processes—the notion that given rates can increase only up to a certain point—has also been studied extensively in the contexts of both predator-prey systems (e. g. , [26]) and mathematical epidemiology (leading to the distinction between mass-action incidence for low densities and standard incidence for high densities). Predation and infection are superimposed in the transmission of vector-borne infections, and empirical studies [25], [27] have observed a corresponding density dependence in which per-vector biting rates decrease at high vector-host ratios. Per capita contact rates thus increase with the density ratio only up to a certain limit, so that the total contact rates (per capita rates multiplied by host or vector density) then become functions of one density or the other alone. When the ratio of vectors to hosts is low, hosts are plentiful relative to vectors, so on the one hand each vector can feed as often as it wants (that is, at its preferred feeding frequency), but on the other hand an average host has a hard time finding vectors to consume, making both contact processes limited by the number of vectors. When the ratio of vectors to hosts is high, however, there are not enough hosts upon which for the vectors to feed at their desired frequency (requiring them to find other blood sources), but the hosts are able to eat until reaching satiation, so that both contact processes are limited by hosts. One recent theoretical study [28] developed a mathematical model for sylvatic transmission of T. cruzi and determined that the way in which the two contact processes saturate can affect not only vector population densities but also whether the infection cycle persists. Another study [29] found that such a model coupled to one involving human infection explained observed domestic prevalence data better than a model of exclusively domestic transmission. In order for a mathematical model to predict the rate at which new infections occur, it is necessary to derive quantities such as threshold density ratios from empirical data, so as to understand in what phase of saturation the causative contact processes are operating. This paper presents a way to do so. This paper derives estimates for the key biological parameters needed to model sylvatic Trypanosoma cruzi transmission cycles in Texas and the southeastern United States involving raccoons, Virginia opossums, woodrats, and the two vector species Triatoma sanguisuga and Triatoma gerstaeckeri. Many of these parameters can be estimated directly via an extensive literature review, but infection and contact rates will be estimated indirectly using estimated prevalence levels and a few properties of some relatively simple dynamical population models. The results will also be used to address the issue of saturation in the two infectious contact processes. The intention is to provide well-informed direct estimates of as many quantities as possible and a method for computing other estimates which can be applied to models designed to address a broad spectrum of questions. An exhaustive literature review was used to derive estimates for basic demographic information on host and vector species, as well as those epidemiological parameters for which direct estimation is possible. The review initiated with a Medline search on “Triatoma sanguisuga”, “Triatoma gerstaeckeri”, or “Trypanosoma cruzi”, together with “United States”—or, for general demographic information on hosts, keywords used were “raccoon”, “opossum” and “woodrat”. From the over 1000 resulting articles, only those (approximately 80) which reported data on one of the quantities estimated in the Results section of this paper were kept. The vast majority of the papers discarded focused exclusively on genetics or microbiology, rather than population biology, and were discarded from the title and abstract; the full text of all other articles was examined for relevant data. Results were found (and kept) in English, Spanish, and Portuguese. References in the sources were then checked manually as well. Gray literature was not specifically sought except for non-Chagas-related demographic information on host species not identified in scientific literature, but was checked when it appeared as a reference in another source. Additional references were added at reviewers' suggestions. Well-established properties of nonlinear dynamical systems models were then used to estimate infection rates based on prevalence and known parameters, and to frame the estimation of the threshold population-density ratios that determine whether host or vector population densities drive each type of infectious contact. (Specific simple models are used as illustrations in the Results section, but the approach outlined can be applied to a wide variety of dynamical systems, and results are not meant to be limited to the models given.) Models were used (and will be discussed) only where necessary to help estimate relevant quantities. In every case, epidemiological quantities were estimated as time-averaged values over an entire year, in order not to allow seasonal fluctuations (which impact both host and vector populations significantly) to prevent study of endemic steady states and prevalence. Basic demographic information on host and vector species is necessary for all modeling of T. cruzi transmission cycles. Numerous studies have published data supporting the estimation of average lifespans for raccoons [30]–[34], opossums [12], [34], [35], and woodrats [36, and references therein]; reproductive rates for raccoons [30]–[32], opossums [34], [37], and woodrats [37]; population densities for raccoons [32], [38]–[47], opossums [40], [41], [48], and woodrats [36], [49], [50]; average lifespans for T. sanguisuga [3], [51] and T. gerstaeckeri [3], [52], [53]; reproductive rates for T. sanguisuga [3], [12], [51] and T. gerstaeckeri [3], [53]; and, in a single case, vector population density [54]. Discussion and development of estimates for these quantities are provided in Text S1. Table 1 summarizes these estimates (including SI equivalents) for the demographic parameters of each species. Vertical transmission of T. cruzi has been widely documented in humans, and estimated to occur with frequency between 1 and 10 percent in Latin America [55]–[58]. Because the parasite is transmitted through the placenta and blood supply to the fetus, vertical transmission is possible among placental mammals, but it is generally not believed to occur among marsupials. A study in Venezuela found a vertical transmission rate among Wistar rats (Rattus norvegicus) of 9. 1% for a strain of T. cruzi isolated from dogs, but none at all for a strain isolated from humans [10]. Another study in Georgia (USA) found that a Type IIa strain of T. cruzi isolate from Georgia was twice as likely to be vertically transferred in mice as a Type I isolate from South America [11]. In the absence of any data on vertical transmission among raccoons, we might reasonably estimate that Type IIa strains are transmitted congenitally roughly 10% of the time (as a proportion,), with Type I strains transmitted as much as an order of magnitude less frequently (say). There is almost no published data on rates of oral infection with T. cruzi (which could be estimated directly by multiplying the predation rate of hosts upon vectors by the probability of infection following consumption of an infected vector), although the possibility of oral transmission has long been documented. Olsen et al. , writing in the early 1960s, referenced a “postulate” that oral transmission was the primary route of infections for opossums in Alabama, with insects consisting of 43% of opossums' diet by mass, and 60% by volume [12]; Roellig et al. recently extended this notion to include raccoons as well [13]. One recent source wrote, “Animals can easily become infected with T. cruzi when an infected triatomine bug is ingested. ” [59] However, despite a significant body of research on what raccoons, opossums and woodrats eat, a literature review revealed no data on how much (or how often) they eat (in order to estimate predation frequency). Rabinovich et al. [60] observed 33 instances of predation when each of 13 female white-eared opossums (Didelphis albiventris) was placed with 10 infected Triatoma infestans for a day, but the rather high predation rate estimate that would result from this data is skewed by the experimental conditions, e. g. , the fact that both opossums and bugs were starved for a period of time prior to the experiment, and the opossums had no other available food. Since predation is opportunistic and there are other insects available to the hosts as well, we will therefore estimate predation to occur for all hosts no more often than one triatomine every 3 or 4 days, which equates to an upper bound of about vectors/yr/host. However, it may also be orders of magnitude lower. (Woodrats are of course much smaller than raccoons and opossums, and hence eat less, but vectors are found much more easily in woodrat nests, at least by humans, so we will assume opportunity balances out total volume.) The probability (or proportion) of infection of a host following consumption of an infected vector can be estimated from three experiments in which uninfected hosts were fed vectors infected with T. cruzi. Yaeger conducted 11 trials of an experiment in which an uninfected Virginia opossum (D. virginiana) was fed two Rhodnius prolixus vectors [61] infected with a Type IIe strain; 3 of these trials resulted in infection, yielding an estimate for of. Roellig et al. [13] conducted 2 trials of an experiment in which an uninfected raccoon was fed 3 R. prolixus vectors infected with strain IIa; both trials resulted in infection (yielding an estimate for of 1). Finally, the aforementioned study by Rabinovich et al. [60] produced its own estimate of 0. 075 for the infection probability of white-eared opossums by eating T. infestans infected with an unspecified strain of T. cruzi (presumably not IIa); since their experiment combined oral and stercorarian transmission (all 6 of the 13 opossums who ate a bug were also verified to have been bitten by at least one other bug, except for the opossum who ate all 10 of the bugs placed with her), it is impossible to disentangle the raw oral transmission data in a way that can be pooled with the other two experiments. Yaeger' s estimate for opossums is precisely twice that of Rabinovich et al. , although the difference is not inordinate. Roellig et al.' s data is based on so few trials that no great significance can be ascribed to the resulting high estimate for raccoons, but it is nevertheless suggestive that the probability of oral transmission may vary significantly by host species and by parasite strain (opossums appear not to become infected when exposed to Type IIa T. cruzi [62], and hence may be more difficult to infect with any Type II strain) —not to mention vector species—which is entirely consistent with the speculation of some biologists that North American strains may have adapted in response to local conditions. Obtaining a single estimate for opossums requires an assumption that differences due to species (D. virginiana vs. D. albiventris), vector species, and possibly parasite strain are negligible, in which case we can take a weighted average of. To estimate oral infection probability for raccoons we are left with either the above 100% estimate or else an average across all host species (including opossums) of. There is likewise no published research on the extent to which infection with T. cruzi increases vector behaviors in T. sanguisuga or T. gerstaeckeri that promote infection. Añez and East [14] found that triatomine bugs of the genus Rhodnius, a common T. cruzi vector in South America, probed or bit an average of 6. 5 times as often when infected with the parasite Trypanosoma rangeli as when uninfected, prior to engorging. This differential behavior may amplify by a factor (say) not only the biting rate of infected vectors but also their availability for predation due to increased mobility driven by hunger, so that the effective vector density for infection behaviors is rather than. However, D' Alessandro and Mandel [63] found no difference in the feeding behaviors of R. prolixus infected by T. cruzi. Although such frequencies can be expected to vary widely by species (of parasite as well as vector), it would be consistent with research on South American species to expect no differential behavior in infected T. sanguisuga or T. gerstaeckeri. In the case where we wish to investigate the possible effects of such an amplification factor, however, it is worth noting Añez and East' s value. Research suggests that in general sylvatic hosts do not suffer mortality from T. cruzi infections, even though high mortality rates have been reported for dogs, and the long-term risks have been verified for humans. Also, the mice which die from T. cruzi infections in laboratory experiments are often injected with considerably higher concentrations than a single horizontal transmission is likely to produce initially. We may therefore assume (following, e. g. , [64]) that in general the sylvatic hosts under study have no significant additional mortality caused by infection with T. cruzi. Table 2 summarizes these parameter estimates. (Table 3 defines additional variables and parameters used in later sections.) Estimation of the per capita infection rates for vector transmission must be made indirectly, as at present there are few published data on both the vector biting rate and the proportion of feedings which result in an infection in each direction (host to vector and vice versa). (Two notable exceptions are [65], which estimated the probability of vector infection per feeding for a specific South American cycle, and [60], which estimated the probability of stercorarian infection of opossums D. albiventris at 0. 06 [95% CI: 0. 023,0. 162] per infected T. infestans bite). Instead, given the long history of established T. cruzi infections in the regions of interest, we shall assume that the parasite has reached endemic equilibrium in the host and vector populations, and use published data to estimate [endemic] prevalence in both host and vector. This will allow us to use the formulas derived from our population dynamics model which express endemic equilibrium prevalence as a function of model parameters, to calculate the infection rates necessary to produce those endemic levels. With prevalence levels and all other parameter values known, it will be possible to solve for the infection rates. But first we must estimate prevalence. Reported prevalences are given in Tables 4–8 for raccoons, opossums, woodrats, T. sanguisuga and T. gerstaeckeri in the southeastern United States and northern Mexico. Asterisks (*) denote studies which published paired estimates of host and vector prevalence. For host prevalence, the method of diagnosis is given as [hemo]culture, serology (IFAT = Indirect Fluorescent Antibody Test, IHA = indirect hemagglutination assay), either (both culture and serological tests were performed, and a single positive is reported as positive), blood smear (BS), or xeno [diagnosis]. The dagger after the citations to Lathrop and Ominsky [66] marks joint prevalence reported for a mixed population of 6 T. sanguisuga and 9 T. gerstaeckeri. As evidenced by Table 4, dozens of studies have reported prevalence figures for the infection of raccoons with T. cruzi in the past fifty years, in states throughout the southeastern quarter of the United States. As observed by several researchers, notably Yabsley et al. [67], the method used to determine infection can have a significant effect on the results: in particular, the parasite is often only found in the blood (by hemoculture or blood smears) during the initial (acute) period of infection, while the immune system takes some time to develop antibodies to T. cruzi, so that serological tests like IFAT and ELISA are more likely to detect chronic infections. It is therefore best to use both methods in order to capture both acute and chronic infections. Most studies reported prevalence based only on blood cultures until about ten years ago, and as can be seen in Table 4 there is a marked difference in the prevalences reported based on hemoculture studies as compared to serological or both. Ten of the sixteen blood-based studies reported prevalences of 15% or less (seven of these reported prevalences of 1. 5% or less, and the mean of all 16 values is under 20%), whereas apart from a single, small-sample (n = 12) zero value, the studies which included serological results reported a mean of over 50% prevalence. There is also some notable geographic variation. Infection rates near the central part of the country appear to be relatively high, with studies from Kentucky, Missouri, Oklahoma and central Tennessee all reporting prevalences of well over 50%, with a total prevalence of 106/163 or 65%. On the other hand, the region directly east of that, from the mountains to the Atlantic, has little or no infection: studies from Maryland, Virginia, West Virginia and even eastern Tennessee adjacent to Virginia all report effectively zero prevalence, the exception being a study of raccoons in the suburban area of Fairfax County, Virginia, near Washington, D. C. , where increased opportunity for foraging results in a higher raccoon population density. Prevalence among raccoons in Georgia and neighboring South Carolina ranges from 33% to 60% except for one hemoculture-based study which reported 22%. Pooling these 7 studies yields an overall prevalence of 351/908 or 38. 7%, heavily weighted by the large study of Brown et al. [68]. Moving west along the Gulf Coast, there is no data apart from Olsen et al.' s study from eastern-central Alabama in the early 1960s until we reach Texas, where there are only two small studies from 1977–1978. We shall take the figure of 24% from central Texas, rather than that of 0% from south Texas, as being representative of prevalence among raccoons in the central and eastern part of the state. Examining the reported prevalences for opossums, there is a clear tendency for the studies which used both blood culture and serology to report higher prevalences (see Table 5), with the exception of the early datum from Texas, which was of such a small sample size (n = 8) that it cannot be claimed to be representative. There is nearly an order of magnitude difference in sample size between the three largest studies [68]–[70] and the next largest, and these three show, on the one hand, nearly identical hemoculture-based prevalences between Texas (16%) and Florida and Georgia (17%, consistent with the more recent Georgia figure of 15. 4% [71]), and, on the other hand, a prevalence that nearly doubles when both hemoculture and serology are taken into account (28% in Georgia [68]). Although some of the smaller studies suggest that in places the prevalence of T. cruzi in opossums may be much higher than this, we shall use Brown et al.' s 28% figure as representative of prevalence in both the southeast and Texas. The four earliest reported prevalences of T. cruzi infection in Texas woodrats are relatively close to each other (ranging from 21. 4% to 34. 9%, see Table 6) but used hemocultures or blood smears rather than serology, which may imply an underestimate; the two reports from west Texas, both serological, are higher but come from much smaller samples. We shall nevertheless pool the data to obtain an overall prevalence of 225/678 or 33. 2%. Very few studies have reported infection prevalence for the vector T. sanguisuga east of Texas (see Table 7). The studies published by Hays, Olsen and their collaborators in the 1960s give prevalences of around 6% in eastern central Alabama, but the two more recent studies in Georgia and Louisiana agree on values an order of magnitude higher. It is likely that infection prevalence does vary by location, but for an overall average we shall pool the two more recent reports, for a total prevalence of 56. 5% in the southeast. In Texas, reported prevalences appear to fluctuate within a range of 17% to 44%. Pooling all but the first two studies (since the second gave no absolute numbers) yields an overall prevalence of 135/543 or 24. 9%. Early studies had T. cruzi prevalence in the vector T. gerstaeckeri varying widely from 5% to 92% (see Table 8), and despite some slight convergence, results continue to fluctuate from 26. 5% to 77. 4%, even among relatively large () samples (we exclude from further discussion the small sample from Queretaro in central Mexico). Since these studies typically collected vectors from woodrat nests, it is likely that there may be considerable variation in infection proportions from one nest to another. The three reports from the state of Nuevo León, Mexico, just south of Texas, also fit within this range. We will therefore pool all studies for which raw data is given (noting that the rate given in Galavíz et al. is close to that in the study by Martínez-Ibarra et al. , on which Galavíz was second author, and that the data in deShazo is likely incorporated into the study by Sullivan et al. given the dates, and the fact that deShazo and Sullivan were the same person), to derive an overall prevalence of 572/1259 or 45. 4%. Note that all collections of vectors in Texas were made from either woodrat nests or peridomestic environments, while collections in the southeast mention association with both raccoons and opossums. This complicates the matter of disentangling the various transmission cycles (for instance, are vectors in raccoon dens in Texas infected at the same level as vectors in nearby woodrat nests?), which may be especially important where different strains of T. cruzi are involved, as with opossums (typically infected with type I) and raccoons (typically infected with Type IIa) in the southeast. In the absence of more complete data, however, we can do no better at present than use these figures as applying across hosts in a given habitat. As a brief aside, we also note reports of prevalence in Texas among the vector Triatoma neotomae, uniquely identified with woodrat nests, of 87. 5% by deShazo [72], 11/17 (64. 7%) by Sullivan et al. [73], 27/31 (87%) by Eads et al. [74], and 2/3 (66. 7%) by Burkholder et al. [54], the latter three of which combine to give an overall prevalence of 40/51 or 78. 4%, significantly higher than that of most other vector species. As the vector' s habitat is confined to one or two regions of Texas, however, we will not consider it further. Table 9 summarizes these prevalence estimates for Texas and the southeast. Most quantities dealing with the T. cruzi infection process itself must be estimated indirectly by inference, since (as illustrated in the previous subsection) little or no published data exists on quantities such as probabilities of infection and even species-specific contact rates. Instead, one can use population models of transmission dynamics to back-calculate the effective infection rates given observed endemic prevalences and the known demographic parameter estimates. The specific calculations and expressions involved are model-dependent—for example, one model may distinguish between oral and stercorarian infection rates for hosts, while another uses a single term with a net host infection rate—but the general idea remains the same: to use equations for the observed endemic equilibrium to solve for the desired parameters. (Note this method assumes that observed infection prevalence represents a steady endemic state.) Table 3 summarizes all model variables and parameters used in modeling discussions in this and the following sections, except for those already defined in Table 2. To illustrate this technique with a minimum of model parameters, we here consider a scenario with a single host and single vector species, each at a constant population density, and only a single (net) route to infection. The simplest vector infection model has the formHere and are the densities of infected hosts and vectors, respectively, as functions of time, and are the host and vector densities as before (here assumed constant over time), and are the respective infection rates, and and are the mortality rates. In each differential equation the first term describes the rate of new infections, and the second describes removal by natural mortality (we assume no recovery from infection). Here for simplicity we use so-called standard incidence to describe the total infection rates, and defer discussion of saturation in the relevant contact processes until the next section. This model is mathematically equivalent to the classical Ross model for malaria transmission [16], although removal of infected hosts here is due to natural death (not recovery as in Ross' s model) and for simplicity the [here constant] vector-host ratio that is explicit in Ross' s model has been absorbed into (the following subsection on saturation in contact processes will address how the infection rates depend on this ratio). If we define proportional infection levels, , then the equilibrium conditions for this model (setting the time derivatives and to zero for the steady state) can be written asWe can solve these equations for the infection rates and, so that in case we know the prevalence levels, (assumed positive) and also the mortality rates, , we can calculate the corresponding infection rates: We can apply this result to the transmission cycle between raccoons and T. sanguisuga in the southeastern U. S. using the prevalence estimates, derived in the previous section and the mortality rates /yr, /yr from Table 1 (assuming opportunistic host predation on vectors does not significantly impact vector mortality), to obtainIf we instead consider opossums (, /yr) and T. sanguisuga in the southeastern U. S. , we get insteadThe fact that in both cases reflects the higher prevalence found in vectors compared to hosts, , consistent with the observation (e. g. , [3]) that T. sanguisuga and T. gerstaeckeri are so cautious as to rarely walk entirely onto a host, therefore making (stercorarian) transmission to hosts much less likely than transmission to vectors through bloodmeals. Note that this model assumes no vertical transmission, and treats all transmission routes (here, stercorarian and oral for the host) as one to produce an estimated overall infection rate. Any such distinctions must be made in the model used to derive the infection rates. For instance, if we wish to take into account vertical transmission of T. cruzi among placental hosts such as raccoons, then we add a corresponding term to the equation for (if hosts are assumed to reproduce according to a logistic law, at a total rate): If we assume the host population to have reached its equilibrium value, then the new term simplifies to, and the differential equation simplifies to its previous form, with replaced by: This means that the only change made in the two expressions for infection rates is to multiply (and hence) by: The vector infection rate is unaffected, but in the case of raccoons infected with Type IIa T. cruzi in the southeastern U. S. , the vertical transmission estimate of for Type IIa yields an estimated horizontal transmission rate of. Similar adaptations can be made for models which distinguish between stercorarian and oral transmission to hosts, or address differential behavior of infected vectors, etc. , although sufficiently complicated models may require solving equilibrium conditions numerically once other parameter values are substituted, if closed-form expressions for endemic equilibria are not available. Finally, in order to complete a model description of T. cruzi transmission dynamics, it is necessary to address the specific forms of the host-vector contact processes that drive infection: host predation upon vectors, which can produce oral transmission, and vector feeding upon hosts, which can produce bloodborne and stercorarian transmission. Here, too, mathematical models can help identify and articulate the key parameters that determine those forms. Since both types of contact processes are predation-driven, we begin with a brief review of considerations from the well-developed area of predator-prey modeling. Mathematical models have enormous predictive and explicative power in the study of biological systems, especially those where the feasibility of large-scale field studies is limited. Dynamical systems have managed to capture the nonlinear contact processes at the heart of many population biology questions in ecology and epidemiology, but their descriptive ability as models hinges on having accurate estimates for the biological parameters that measure key rates and quantities. Any study of the dynamics of sylvatic Trypanosoma cruzi infection must include both demographic and epidemiological information on the hosts and vectors involved. As seen in the preceding sections and Text S1, a thorough literature review is sufficient to determine many of the most basic demographic parameters for the host and vector species that drive T. cruzi transmission in the southeastern quarter of the U. S. , but many aspects of the contact processes which actually cause infection remain poorly understood. Simple dynamical systems models can be used to back-calculate infection rates from data on zoonotic prevalence, as well as to pinpoint what specific biological data needs to be gathered to complete parametrization of the models. In the present study, these data include: vector population densities, the probability of vertical transmission in raccoons and other hosts, the probability of oral infection per host type (and per vector consumed), the (maximum) rate at which hosts consume vectors, the extent to which T. cruzi infection changes the relevant behaviors of the vectors T. sanguisuga and T. gerstaeckeri, infection prevalence among Texas vectors outside woodrat nests and peridomestic sites, and the threshold vector-host density ratios which determine saturation for both contact processes. The rough estimates derived in this paper regarding the latter ratios and suggest that host predation on vectors is saturated in vectors (largely because this predation is opportunistic), and therefore dependent on host density for each host, whereas the vector feeding process is saturated in vectors only for the larger hosts (raccoons and opossums), which have a relatively low population density, and saturated in hosts for the woodrats that are the predominant host from central Texas south to Mexico, since woodrats occur at a higher density and return to the same nests on a long-term basis, making these nests efficient feeding sites for the vectors. Since T. sanguisuga and T. gerstaeckeri are widely believed to be inefficient vectors, the vector feeding process is primarily responsible for prevalence in vectors, and it is therefore interesting to note that T. sanguisuga appears to have a higher prevalence in many parts of the southeast (especially those closest to the center of the U. S.) than T. gerstaeckeri does in Texas, where it has ready access to abundant hosts. It is important to keep in mind, however, that the uncertainty in several parameter estimates (notably the effective vector population density) limits the confidence one can place in the conclusions regarding contact process saturation. Of course, all models, however complex, remain caricatures or sketches of reality, and have their limitations. Dynamical systems models are limited in their predictive power not only by the accuracy of the estimates used for the biological rates that parametrize them, but also by the correctness and completeness of the assumptions that underlie every term in each equation. This paper is meant to connect these theoretical models to the many empirical studies that add detail to our understanding of the T. cruzi infection process in the U. S. Further studies are already in progress developing models that begin to incorporate the multiple infection mechanisms described in this work and the literature reviewed within, as well as the effects of dispersal and migration connecting the various evolving habitats (such as central Texas and the southeastern U. S.) where T. cruzi is in zoonosis. Readers interested in the question of vector feeding preferences for different types of host are referred to the studies [25] and [27].

The parasite Trypanosoma cruzi, transmitted by insect vectors, causes Chagas' disease, which affects millions of people throughout the Americas and over 100 other mammalian species. In the United States, infection in humans is believed rare, but prevalence is high in hosts like raccoons and opossums in the southeast and woodrats in Texas and northern Mexico. The principal U. S. vector species appear inefficient, however, so hosts may be primarily infected by congenital transmission and oral transmission caused by eating infected vectors. Mathematical models can evaluate the importance of each transmission route but require as inputs estimates for basic contact rates and demographic information. We estimate basic quantities via an exhaustive review of T. cruzi transmission in the southern and southeastern U. S. , and use properties of mathematical models to estimate infection rates and the threshold (saturation) population-density ratios that govern whether each infection process depends on host or vector density. Results (based on extremely limited data) suggest that oral transmission is always driven by host density, while transmission to vectors depends upon host density in cycles involving raccoons and opossums, but upon vector density in cycles involving woodrats, which live in higher concentrations.

Abstract Introduction Methods Results Discussion

infectious diseases/neglected tropical diseases ecology/population ecology mathematics/nonlinear dynamics

2010

Estimating Contact Process Saturation in Sylvatic Transmission of Trypanosoma cruzi in the United States

Locomotor strategies in terrestrial tetrapods have evolved from the utilisation of sinusoidal contractions of axial musculature, evident in ancestral fish species, to the reliance on powerful and complex limb muscles to provide propulsive force. Within tetrapods, a hindlimb-dominant locomotor strategy predominates, and its evolution is considered critical for the evident success of the tetrapod transition onto land. Here, we determine the developmental mechanisms of pelvic fin muscle formation in living fish species at critical points within the vertebrate phylogeny and reveal a stepwise modification from a primitive to a more derived mode of pelvic fin muscle formation. A distinct process generates pelvic fin muscle in bony fishes that incorporates both primitive and derived characteristics of vertebrate appendicular muscle formation. We propose that the adoption of the fully derived mode of hindlimb muscle formation from this bimodal character state is an evolutionary innovation that was critical to the success of the tetrapod transition. Studies of a number of fossil forms have provided information on the evolution of the appendicular skeleton of the hindlimbs within early tetrapods [1]–9. These analyses have revealed that the tetrapod transition is characterised by the gradual replacement of a relatively gracile, ventrally located pelvic girdle and laterally positioned fin endoskeleton, with a robust, dorsally positioned, pelvis and hindlimb skeleton. In tetrapods, the pelvis articulates directly with the axial skeleton via the ilium, which extends dorsally to attach to the sacral vertebrae (Figure 1A, B) [1]–[13]. Evolution of the pelvic girdle and hindlimb endoskeleton is associated with critical functional innovations within transitional tetrapods. These include additional structural support through the articulation of the pelvic girdle with the axial skeleton, increased surface area for muscle attachments, and a more lateral and dorsal positioning of the limb articulations with the axial skeleton. These adaptations are all considered essential for the evolution of load-bearing and locomotor-predominating hindlimbs. The fossil record has, in part, charted the evolution of the skeletal framework of the load-bearing limbs of tetrapods [1]–[13]. However, individual fossils can shed little light on how the dramatic alterations of the limb musculature, required to drive locomotion in terrestrial tetrapods, arose as soft tissues are rarely preserved within the fossil record. In order to examine this question it then becomes necessary to uncover the mechanisms that generate limb and fin muscles within extant species present at crucial nodes within the vertebrate phylogeny. Extensive analyses have been undertaken on the formation of limb muscles within two extant amniote tetrapod species, chick and mouse [14]. Both the fore and hindlimb musculature of chick and mouse embryos are generated via an identical process, in which limb myoblasts are derived from the migration of mesenchymal precursor cells. These precursors de-epithelialise from the ventro-lateral, or hypaxial, region of limb-level somites and undergo a long range migration to their final position within the limb mesenchyme (Figure 1E) [15]. During this process, these cells require the expression of a number of specific genes including the homeobox-containing gene Lbx. Lbx expression is important in the context of this study as in amniotes it is a highly specific marker of migratory muscle precursor cells within the limb-adjacent somitic mesoderm and is also maintained within migrating and post-migratory limb myoblasts. Lbx is not only expressed within amniote limb myoblasts but is also functionally required for their formation and correct differentiation. Homozygous Lbx mutant mice fail to form limb muscle normally with extensor muscles of the forelimbs being absent and flexor muscles reduced in size. Hindlimb muscles are also strongly affected, with distal limb muscles more affected than proximal ones [14], [16]–[18]. A similar, lbx1-positive, set of fin muscle precursors—also derived from the migration of mesenchymal precursor cells that originate from pectoral fin level somites [19]—have been shown to generate the appendicular muscle present within the pectoral fin of the teleost, zebrafish (Figure 1D). Thus, the pectoral fin muscle precursors of zebrafish possess molecular and morphogenetic identity to the limb muscle precursors of tetrapod species [19]–[21]. However, how widely within the bony fish phylogeny this mechanism is deployed has yet to be determined. In contrast, we have previously shown that the embryos of the shark Scyliorhinus canicula (a chondrichthyan species basally positioned in the vertebrate phylogeny) utilise a separate process of direct epithelial extension from the embryonic myotome to generate both the hypaxial muscles of the body wall, and secondarily at its most ventral extent, the muscles of the pectoral fins. This process is characterised by the progressive extension of the myotome, via a ventrally displacing epithelial bud, that directly enters the fin to generate the muscle of the pectoral fin without the migration of lbx1 expressing mesenchymal precursors (Figure 1C). Given the basal position of chondricthyans within the vertebrate phylogeny we have defined the mechanism of direct epithelial myotomal extension as the primitive mode of appendicular muscle formation. In this paradigm, the generation of limb myoblasts in amniotes and pectoral fin muscle in zebrafish via lbx-positive migratory mesenchymal precursor myoblasts represents the derived mode of appendicular muscle formation [19]. By contrast, the developmental origin and molecular processes that generate pelvic fin muscle have not been defined in any fish species to date. To understand the changes underlying the evolution of the pelvic fin musculature, we have studied the mechanisms of pelvic fin muscle formation in living fish species positioned at strategic points within the vertebrate phylogeny. Here we reveal that all bony fish species we have examined make pelvic fin muscle using the same developmental process, utilising a myotomal extension to deliver fin muscle precursors adjacent to the forming pelvic fin. Once in position adjacent to the pelvic fin bud, muscle precursors undergo an epithelial mesenchymal transition and are induced to express lbx1 and migrate into the fin mesenchyme to form individual pelvic fin muscles. Collectively, these studies demonstrate that the pelvic fin musculature of bony fish is generated by a novel morphogenetic process that possesses characteristics of both the primitive (epithelial myotomal extension) and derived (lbx1-positive migratory mesenchymal myoblast precursors) modes of vertebrate appendicular muscle formation. We further propose that the adoption of the fully derived mode of hindlimb muscle formation from this bimodal character state was an evolutionary innovation critical to the success of the tetrapod transition. We have compared the mechanism of fin muscle development of two chondrichthyan cartilaginous fish species (bamboo shark, Chiloscyllium punctatum and the chimera, Callorhinchus milii) and three bony fishes, the North American paddlefish, Polyodon spathula, a teleost (zebrafish Danio rerio), and the Australian Lungfish Neoceratodus forsteri. The bamboo shark, Chiloscyllium punctatum and the chimera Callorhinchus milii are basal in the vertebrate phylogeny. C. milii as a Chimaeriform is considered basal within the chondricthyan clade, with the callorhynchids representing the most primitive living members of the Holocephali [22]. Thus, C. milii is a representative of the most primitive extant fish species with paired appendages. Thus, the common developmental features shared by the shark and chimera are expected to represent the primitive state of fin/limbed vertebrates. Of the bony fishes examined, the North American paddlefish, Polyodon spathula, is a living representative of a group of primitive Chondrostean ray-finned (actinopterygian) fish and occupies an important basal position within the bony fish. The Australian lungfish, Neoceratodus forsteri, is the only example of a lobe fin (sarcopterygian) fish for which embryonic material can be obtained, and as such is a critically positioned species for understanding the evolution of the tetrapod transition. Finally, the zebrafish family (Cyprinidae of the order Cypriniformes) represents a teleost and is a genetically tractable established model vertebrate that is amenable to a myriad of powerful molecular techniques allowing greater resolution/depth to the precise nature of the developmental mechanisms occurring. We focused our analysis on the periods of development of these distinct species when the fins are initially formed. We utilised the presence of epithelial buds at the head of myotomal extensions within the fin mesenchyme as an indicator of the primitive mode of fin muscle formation [19]. Conversely, the lack of an epithelial myotomal extension and the expression of lbx1 within mesenchymal migratory fin myoblasts were used as markers for the derived mode of fin muscle precursor migration [19]. Our initial studies focused on confirming the deployment of the derived mode of appendicular muscle formation in the pectoral fin of bony fish other than zebrafish. We undertook this analysis in order to strengthen the phylogenetic assignment of this character state as having arisen prior to the sarcopterygian radiation, as a mosaic distribution of this mode of appendicular muscle formation has been previously reported to occur within the bony fish phylogeny (extensively reviewed in [23], see below). Within both paddlefish and lungfish embryos there was no evidence of a myotomal extension, and lbx-positive pectoral fin muscles were generated discretely within the fin bud (Figure 2, unpublished data). In embryos of these species muscle differentiation occurred within the pectoral fin and generated defined dorsal and ventral fin muscle masses, without any connection to an epithelial somitic extension (Figure 2B). This analysis confirmed the presence of the derived mode of appendicular muscle formation in the pectoral fin of paddlefish and lungfish analogous to that seen in our previous studies on zebrafish pectoral fin muscle development [19]. We next turned our attention to the mechanisms that generate pelvic fin muscle formation. As mentioned above, the morphogenetic and molecular basis for pelvic fin muscle development has not been determined for any fish species to date. We first examined pelvic fin muscle formation in the two chondricthyan species under study. Within both species, the pelvic fin muscles were generated by direct epithelial extension of the myotome, headed by a characteristic migrating epithelial bud (Figure 3A–M). Epithelial buds progressively generate the muscle of the body wall as they extend ventrally from the myotome and, at their most ventral extent, enter the forming pelvic fin mesenchyme to generate the pelvic fin muscles (Figure 3A–M). Furthermore, although lbx1 expression could be detected by antibody labelling within the neural tube of C. milli, a known site of lbx1 expression in zebrafish embryos [19], no expression could be detected within its fin epithelial myotomal extensions (Figure S4). Collectively, the above data strongly suggest that the pelvic fin muscle of chondrichthyan species is generated by the primitive mode of direct epithelial extension of fin-adjacent myotomes, in a process identical to that described to generate the muscles of the chondrichthyan pectoral fin. Similarly, in all three bony fishes examined, myosin heavy chain (MHC) positive cells were detected extending ventrally from the somite towards the future position of the developing pelvic fins (Figure 4D, E, N, O, X, Y). In each case the extension was headed by an epithelial bud that progressively generated the hypaxial muscle of the body wall as it extended towards the level of the pelvic fin bud. However, in contrast to chondricthyan embryos, the myotomal extension of all bony fish examined failed to enter the fin bud mesenchyme. This was despite having arrived at its most ventral extent temporally and spatially coincident with the formation of the adjacent pelvic fin bud (Figure 4E–I, N–Q, X–BB). Furthermore, the first differentiated muscle cells that appeared within the pelvic fins of each species were clearly separate from the myotomal extension, which by this stage lacked any evidence of an epithelial character (Figure 4H, I, P–R, Z–BB). Muscle differentiated within the fin of each species with no evidence of the myotomal extension entering the forming pelvic fin bud (Figure 4G–I, P–R, Z–BB). In contrast to the epithelial extension evident during chondrichthyan paired fin muscle formation, in situ hybridization revealed that lbx1 mRNA is expressed in the tip of the fin adjacent to myotomal extension, and in fin muscle precursors during their short range migration from the extension into the pelvic fin (Figure 4J, S, T, DD, Figure S2). This migration was most clearly seen in the zebrafish by section in situ hybridization where lbx1-positive mesenchymal cells migrated towards the developing fin bud mesenchyme such that at 9 mm TL, two distinct regions of lbx1 expression were evident within the pelvic fin, corresponding to the future dorsal and ventral muscle masses of the pelvic fin (Figure 4T). Taken as a whole, these data suggest that a similar mechanism to that operating in zebrafish generates paddlefish and lungfish pelvic fin musculature. This mechanism is a bimodal character state comprising features of both the primitive and derived modes of vertebrate appendicular muscle formation. Although these morphological and gene expression studies suggest the origin of the pelvic fin muscle lies within the adjacent epithelial myotomal extensions, they do not provide direct evidence for it. The heterochronous development displayed by the pelvic fins of zebrafish (a primitive character of vertebrates with paired appendages shared by most fish [24], [25]), which develop 4 wk after formation of the pectoral fin bud initiates at the end of somitogenesis has precluded an examination of the developmental origin of the cells of the pelvic fin musculature, as no fate mapping strategies have been developed in fish that allow tracking of somite derived cells for this period of time. Somite transplantation has been deployed in other model systems, most successfully by chick embryologists, where it has been used to determine the fates of somitic cells. Historically, the technique was utilized to determine somitic origin of limb myoblasts, a finding that overturned the prevailing hypothesis that limb myoblasts originated from lateral plate mesoderm [15]. In the context of this current analysis somite transplantation has the advantage that genetically distinct donor tissue is indelibly marked and can be used to determine donor tissue contribution to a host structure at any developmental time point. Thus, we developed a transgenic fate mapping strategy that enabled the transplantation of embryonic somitic tissue between different genetically marked strains of zebrafish. In this strategy donor embryos are generated by crossing adults transgenic for mCherry driven by the muscle specific alpha actin promoter Tg (acta1∶mCherry) pc4 with those carrying a transgene that drives GFP off the same promoter Tg (acta1∶GFP) zf13 to generate donor embryos marked both with red and green fluorescence in the nascent myotome. This strategy was necessary because mCherry gives a weaker signal and consequently was not highly visible at the embryonic stages at which transplantation can be carried out. Thus, in order to enable specific dissection of somitic tissue, the brighter GFP construct was crossed into the background of the donor. Isolated donor somites were then transplanted homotypically into a “green only” Tg (acta1∶GFP) zf13 host in which the host level somite had been extirpated. Transplanted embryos were grown to adulthood and donor tissue contribution to the host assessed via mCherry fluorescence. Using this strategy we were able to observe in vivo, and in real time, the fate of the transplanted somitic cells through the entire life span of the fish (Figure 5A, B, Materials and Methods). Under this transplant regime, mCherry positive donor somites formed the ventral somitic extension, which generated the body wall musculature at all transplant levels. In order to examine if any cell type other than muscle was generated by the ventral somitic extension, a triple transgenic fluorescent transplant strategy was developed. In this strategy, a ubiquitously expressed promoter (beta actin) drives mCherry in the donor somite, which is also marked with GFP driven by the alpha actin skeletal muscle-specific promoter. The donor somite is transplanted into a host that is transgenic for BFP also expressed via alpha actin skeletal muscle-specific promoter. In each of 12 transplants performed in this way, only co-expression of both green and red fluorescent protein was ever observed, indicating that the donor somite contained only somitic tissue and that this only ever generated donor-derived muscle in the host (Figure S1). At the level of the pelvic fin, the extension contributed to the pelvic fin muscle on the operated side, with individual somites transplanted at specific somitic levels giving rise to specific muscles within the pelvic fin (n = 6; Figure 5A–T). By contrast, the non-operated (contralateral) side never showed mCherry-positive cells within the pelvic fin muscle masses. Furthermore, somite transplantation anterior (n = 6) and posterior (n = 4) to somites 10 and 11 did not reveal any contribution to the pelvic fin muscles (Figure S3). This procedure revealed that the pelvic fin muscles of zebrafish originate from pelvic fin level somites. Furthermore, in order for a donor somite to contribute to the pelvic fin muscles, the donor tissue has to be present in the most ventral portion of the extension, as transplants where host tissue remained at the most ventral tip ahead of the donor tissue resulted in only host tissue contributing to the pelvic fin muscle (Figure 5J–N). These data therefore illustrate that fin muscle precursors of zebrafish are contained within, and are carried ventrally by, myotomal extension. Once in position adjacent to the pelvic fin bud, muscle precursors undergo an epithelial mesenchymal transition and are induced to express lbx1 and migrate into the fin mesenchyme to form individual pelvic fin muscles. Collectively, these studies demonstrate that the pelvic fin musculature of bony fish is generated by a novel morphogenetic process that possesses characteristics of both the primitive (epithelial myotomal extension) and derived (lbx1-positive migratory mesenchymal myoblast precursors) modes of vertebrate appendicular muscle formation. Through these studies we have morphologically and molecularly defined a developmental mode of fin muscle formation that is intermediate between the primitive mechanism of paired fin muscle formation, evident in chondrichthyan species, and the derived mode, evident in the pectoral fin of bony fishes and the fore and hindlimbs of tetrapods [19]. The existence of such a bimodal character state has been postulated previously, but evidence as to its existence as well as an understanding of its evolutionary significance have both been lacking [23], [26], [27]. Our observations also resolve the previously reported mosaic distribution of the primitive and derived modes of appendicular muscle formation [15], [19], [23], [26]–[34] into a single phylogenetically harmonious framework (Figure 6). Indeed, we think it likely that the existence of this bimodal character state was difficult to resolve with only the tools of simple histology available to the researchers at the time that many of these studies were performed. The existence of an epithelial extension associated with pelvic fin muscle formation may well have caused confusion as to the nature of the mechanisms deployed in both sets of paired fins. In order to locomote on land, the robust, dorsally articulated, pelvic girdle and expanded hindlimb skeletons present in tetrapod species need to be populated by powerful hindlimb muscles that not only support the weight of the whole animal but also coordinate movement. Critically, we believe, removing the requirement for pelvic fin muscle formation to be coupled to the ventral arrival of a myotomal extension provided flexibility in pelvic fin/limb positioning without compromising body wall muscle formation. The possible structural limitations arising from the deployment of myotomal extension are evident if we consider that it is used to generate all hypaxial muscles of the body wall by continuous ventral extension of the somitic epithelial bud. Upon the arrival of the somitic bud at is most ventral extent, the bud undergoes an epithelial to mesenchymal transition to generate the muscles of the pelvic fins. Thus, use of this mode of muscle formation precludes hypaxial muscle formation ventral to the position of the pelvic fin, as continued ventral formation of hypaxial muscle cannot occur in the absence of the somitic bud. In support of this hypothesis a lack of hypaxial muscle formation ventral to the pelvic fin is evident in all the species we examined (Figure 4). Thus, upon the adoption of the fully derived mode of appendicular muscle formation the pelvic fin was released from the constraint of having to be positioned ventral to the hypaxial muscle of the body wall. Consequently, the pelvic fin could be located anywhere in the dorsoventral body axis (a plasticity already evident in the positioning of the pectoral fin of bony fishes), facilitating the dorsal shift in pelvic girdle location necessary for direct articulation with the axial skeleton and the development of load-bearing hind limbs. It would also allow the pectoral and pelvic fins to develop relatively synchronously, a process characteristic of the fore and hind limbs of model tetrapod species, which is in contrast to the primitive condition, described for both Chondrichthyes and Osteichthyes, where the pectoral fin invariably develops prior to the formation of the pelvic fin [24], [25]. Furthermore, the ability to deploy muscle progenitors into the pelvic fin/limb environment at a relatively earlier phase of body plan development, prior to the completion of hypaxial body wall formation, may have facilitated the development of the more complex and physically larger sets of muscle required for terrestrial locomotion. We therefore consider that the novel method of pelvic fin formation we describe in bony fish may represent an important intermediate step in the evolution of tetrapod limb muscle developmental mechanisms. We hypothesise that the adoption of the fully derived mode of hindlimb muscle formation was an evolutionary innovation critical to the success of the tetrapod transition. Data in amphibian species support this notion as several studies, as well as our own unpublished observations, have failed to detect epithelial extensions associated with the formation of pelvic fin muscle in both Amblystoma puncatum and Xenopus laevis [35], [36], despite early controversy as to the presence or absence of epithelial extension in these species [37], [38]. Histological and gene expression studies have revealed that Xenopus hindlimb muscle precursors express markers associated with migratory limb muscle precursors of amniotes, and differentiate discretely within the limb bud, devoid of an association with an epithelial bud [39], [40]. Furthermore, recent studies have shown that lbx expression is associated with migratory limb muscle precursors in both fore and hind limbs of the direct developing frog Eleutherodaclylus coqui [41]. Collectively, these studies suggest that amphibians adopted the fully derived mode of limb muscle formation during the tetrapod transition. Whole-mount immunohistochemistry and in situ hybridization on shark, paddlefish, lungfish, and zebrafish embryos and larvae were carried out as described [19]. Cryostat, wax sections, and counterstains were carried out as described [19]. Some C. milli sections were obtained from museum specimens archived in ethanol and required extensive antigen retrieval to detect MHC expression. Sections were incubated in sodium citrate buffer (10 mM Sodium Citrate, 0. 05% Tween 20, pH 6. 0) at 95°C for 40 min, cooled for 20 min, and sections rinsed in PBS 0. 05% Tween 20 for 2×2 min before incubation with antibody. Primary antibodies used were: anti-myosin 1∶200 (A4-1025, Developmental Studies Hybridoma Bank) and anti-LBX1 1∶1000 (ab90839, Abcam). Antibody binding was visualized by standard techniques [19]. Donor (Tg (acta1∶mCherry) pc4) and host (Tg (acta1∶GFP) zf13) [42] embryos were stage matched from syncronous spawnings and transplantations undertaken at the 15 somite stage. Donor embryos were either singly transgenic for α-actin-mCherry or doubly transgenic for both α-actin-mCherry and α-actin-GFP. Use of the double transgenic donors greatly aided the initial dissection of donor somites, as the slow maturation rate of the mCherry produced donor somites that were only weakly fluorescent at the initial transplantation stage. Somites from the donor animal were collected on ice following dissection and pancreatin treatment in DMEM medium. Host embryos were embedded in 1% agarose with 0. 016% tricaine (pH7) and submerged in DMEM medium. One or two somites were removed from the required position by dissection with flame sharpened tungsten needles. Surgery involving transplantation of two consecutive somites gave a greater probability of transplanting the entire somite, including the ventral aspect required for pelvic fin muscle formation. It also led to greater transplant survival. The donor somite (s) was then aligned and inserted into the extirpated somite region and the embryo was allowed to recover for 2 h before dissection from the agarose and rearing in E3 medium containing antibiotic (1,000 U/mL Penicillin-G 1,000 µg/mL Streptomycin) for 2 d. Fish were then reared under standard laboratory conditions for 6 wk and the transplant observed regularly under a dissecting fluorescent microscope. The Tg (bact2∶mCherry) pc3, Tg (acta1∶mCherry) pc4, and Tg (acta1∶EBFP2) pc5 transgenic lines were created using the Tol2kit [43]. The vectors used for transgenesis were assembled from appropriate combinations of the entry clones p5E-acta1 [44], p5E-bact2, pME-mCherry, pME-EBFP2, p3E-polyA, and the destination vector pDEST-tol2-pA2. We generated pME-EBFP2 by PCR subcloning from pBAD-EBFP2 [45]. The primers used for PCR amplification were: EGFP/EBFP2_F1_pME 5′- GGGGACAAGTTTGTACAAAAAAGCAGGCTggaccatggtgagcaagggcgaggagctgtt -3′ and Flouro-STOP-pME 5′- GGGGACCACTTTGTACAAGAAAGCTGGGTgttacttgtacagctcgtccatgc -3′ (Gateway sites shown in upper case). We identified fragments of the lungfish, paddlefish, and bamboo shark lbx1genes, encoding the homeodomain, from complementary DNA pools prepared from embryos of each species by using degenerate PCR primers previously described [19]. Nucleotide and amino acid alignment of lungfish, paddlefish, bamboo shark, and zebrafish LBX to each other and human LBX proteins is included in the Supporting Information section. The lungfish, paddlefish, and bamboo shark lbx1 fragments isolated exhibit 79%, 82%, and 82%, respectively, of sequence identity over a 182-bp region of the homeodomain. Lungfish, paddlefish, and bamboo shark lbx1 sequences have been submitted to Genbank accession nos. EU937814, EU937815, and EU937816. Impregnated females of Callorhinchus milii were line caught during breeding season in Western Port Bay, Victoria, Australia. They were transferred to holding tanks for a month while they laid eggs in captivity. Eggs were labelled with the deposition date and were opened at regular intervals, staged, and fixed in the laboratory with 4% PFA using standard procedures.

The transition of vertebrates from water to land is a fundamental step in the evolution of terrestrial life. Innovations that were critical to this transition were the evolution of a weight bearing pelvis, hindlimbs and their associated musculature, and the development of the “rear wheel drive” strategy that predominates in terrestrial locomotion. The fossil record can reveal how the skeletal framework of the load-bearing limbs of tetrapods (animals descended from fish) has evolved, but as soft tissues are rarely preserved within the fossil record, it can shed little light on how the accompanying dramatic alterations of the limb musculature arose developmentally. To examine this question we determined the mechanisms that generate fin muscles within larvae of living species representing several clades of fish across the vertebrate phylogeny. Using this comparative approach and a novel somite transplantation technique in zebrafish, we determine that the pelvic fin muscles of bony fish are generated by a bimodal mechanism that has features of limb/fin muscle formation in tetrapods and primitive cartilaginous fish. Using these data, we propose a unifying evolutionary hypothesis on the origins of the muscle of the paired fins and limbs, and speculate that the adoption of tetrapod mode of hindlimb muscle formation was also an evolutionary innovation critical to the success of the tetrapod transition.

Abstract Introduction Results Discussion Material and Methods

animal models developmental biology zebrafish model organisms biology evolutionary biology evolutionary developmental biology

2011

Development and Evolution of the Muscles of the Pelvic Fin

Zika virus (ZIKV) and Dengue virus (DENV) are often co-endemic. The high protein-sequence homology of flaviviruses renders IgG induced by and directed against them highly cross-reactive against their antigen (s), as observed on a large set of sera, leading to poorly reliable sero-diagnosis. We selected Domain III of the ZIKV Envelope (ZEDIII) sequence, which is virus specific. This recombinant domain was expressed and purified for the specific detection of ZEDIII-induced IgG by ELISA from ZIKV-RT-PCR-positive, ZIKV-IgM-positive, flavivirus-positive but ZIKV-negative, or flavivirus-negative sera. We also assessed the reactivity of ZEDIII-specific human antibodies against EDIII of DENV serotype 4 (D4EDIII) as a specific control. Sera from ZEDIII-immunized mice were also tested. Cross-reactivity of IgG from 5,600 sera against total inactivated DENV or ZIKV was high (71. 0% [69. 1; 72. 2]), whereas the specificity and sensitivity calculated using a representative cohort (242 sera) reached 90% [84. 0; 95. 8] and 92% [84. 5; 99. 5], respectively, using a ZEDIII-based ELISA. Moreover, purified human IgG against D2EDIII or D4EDIII did not bind to ZEDIII and we observed no D4EDIII reactivity with ZIKV-induced mouse polyclonal IgGs. We developed a ZEDIII-based ELISA that can discriminate between past or current DENV and ZIKV infections, allowing the detection of a serological scar from other flaviviruses. This could be used to confirm exposure of pregnant women or to follow the spread of an endemic disease. Zika virus (ZIKV) was isolated in 1947 in the Zika forest in Uganda [1] and has been responsible for sporadic cases for several decades. Epidemics in Yap State, Micronesia (2007) [2], French Polynesia (2013) [3], and more recently the Americas 2015 [4], have dramatically changed its status. The association of ZIKV infection with Guillain-Barré syndrome [5] and severe outcomes during pregnancy, including microcephaly in fetuses and neonates [6,7], have been brought to light during the last outbreaks. ZIKV can be sexually transmitted. This is uncommon for flaviviruses, which are arthropod-borne viruses (arbovirus) [8]. Infective ZIKV particles have also been found in breast milk but whether neonatal infection or perinatal transmission can occur is unclear [9]. Due to the ZIKV threat, the World Health Organization (WHO) declared a Public Health Emergency of International Concern on the 1st February 2016 [10]. According to WHO recommendations, the diagnosis is based on detection of the ZIKV genome by real-time reverse transcription PCR, serology, and neutralization assays, such as plaque-reduction neutralization tests (PRNT) [11]. However, induced antibodies can show high cross-reactivity between antigens of the same family, as Flaviviruses are phylogenetically very close. Moreover, Dengue virus (DENV) and ZIKV can co-circulate [12]. In addition, cross-seroneutralization of DENV and ZIKV, described recently [13], further complicates the development of relevant target antigens for reliable serological diagnosis. Several immunodiagnostics based on IgM detection have been developed [14,15]. However, most cases are asymptomatic [16] and ZIKV-IgG detection is relevant for dating or confirming infections, especially those of pregnant women, retrospective diagnosis for evaluating transmission intensity to decide on the use of a new prophylaxis, or confirming future ZIKV protection by vaccination. Domain III of the ZIKV-envelope protein (EDIII) shares 29% amino-acid identity with DENV-EDIII and 90% of EDIII-antibodies (EDIII-Abs) elicited by ZIKV infection are virus specific [17]. ZIKV EDIII-Abs tested in ELISA do not bind to DENV-2 or West-Nile virus (WNV) -EDIII [18]. Studies in mice have shown that treatment with EDIII-Abs protect against ZIKV infection [19] by neutralizing the virus [18]. In this context, we produced a recombinant ZIKV-EDIII protein (ZEDIII) and assessed its recognition by IgG. We constructed four representative panels of reference sera, (1) ZIKV-RT-PCR-positive, (2) ZIKV-IgM-positive, (3) flavivirus-positive but ZIKV-negative, and (4) flavivirus-negative, from more than 5,000 sera serodiagnosed at the French National Reference Center for Arboviruses (NRC). The objective was to evaluate a ZEDIII-based ELISA for its specificity (ability to detect IgG raised after infection by ZIKV, but not other flaviviruses) and sensitivity (ability to detect ZIKV-specific IgG in the serum of a patient confirmed to be ZIKV-positive). We thus designed three experiments to evaluate: (1) the specificity and sensitivity of recognition by the IgG of cohort sera, (2) potential purified DENV-IgG cross-reactivity, and (3) the specificity of ZEDIII-induced IgG. The specificity of the assay represents its ability to not detect false-positive sera, that is positivity due to cross-reactive antibodies that are able to recognize a common antigen with the same affinity shared by various viruses. As French National Reference center, all samples used in this study were send to the laboratory for arboviruses diagnosis including the specificity of the antibodies response against different flaviviruses, as Zika virus. For this study, samples were analyzed anonymously. All samples sent in the laboratory are associated with a file with clinical data, travel, date of symptoms and also the consent to used the end of tube for technic development or comparison of diagnosis technics. All sera were submitted to the French National Reference Center for arboviruses (NRC, Marseille) for routine diagnosis and stored at −20°C in an anonymized biobank before testing. No specific sampling dedicated to the study was performed. There were no legal or ethical restrictions for sample use. Associated documents contained the following information: date of birth, gender, date, nature of sampling, place of stay and return date, symptom onset date, clinical symptoms, and clinical diagnosis results. Ethical approval of the ZIFAG cohort was given by the “Comité de Protection des Personnes Sud-Mediterranée” corresponding to the “Etude descriptive prospective de la maladie a virus Zika au sein de la communaute de defense des Forces Armees en Guyane—ZIFAG” and was registered on 29 February 2016 as RCB: 2016-A00394-47. Written consent was obtained from all participants [20]. A total of 5,600 sera from the NRC Arbovirus serum bank were serodiagnosed by ELISA using inactivated virus as antigen. This group was selected to follow antibody cross-reactivity between ZIKV and DENV. Eighty-one sera were not tested for the ZIKV-IgM response. A set of sera was selected (n = 242) from the 5,600 of the NRC Arbovirus serum bank, as well as a set from the ZIFAG cohort, and divided into four groups based on their genome detection, reaction against inactivated arboviruses, and epidemiological data (Table 1, SD1 Fig). The “ZIKV-RT-PCR-positive” group, from the ZIFAG one-year clinical follow-up cohort, is composed of 43 sera collected in French Guyana during the 2016 ZIKV outbreak [20,21]. The ZIKV genome was detected by RT-PCR during the acute phase and tested sera were obtained during convalescence: 6 to 195 days post symptoms onset (DPSO). Eighty-four percent of this group (n = 36) was sampled during the first month post symptoms onset: two sera were obtained less than 10 days DPSO and 34 between 13 and 25 DPSO. The other sera (n = 7) were obtained between 35 and 195 DPSO [21]. The “ZIKV-IgM-positive” group was composed of 50 randomly selected sera, collected mainly from the Caribbean Islands or another DENV-endemic area during the 2015–2016 ZIKV epidemic (total n = 241), based on the following criteria: ZIKV-IgM-positive, DENV-IgM negative, and ZIKV-IgG positive, all tested sera neutralizing ZIKV in PRNT assays. “Flavivirus” sera (n = 99) group was selected to be flavivirus-positive but not ZIKV-positive (total n = 797). All were flavivirus-IgG positive and sampled between 2013 and 2014 in Guadeloupe, Martinique, or Saint Martin, DENV-endemic areas, where and when ZIKV was not circulating, assuming that these patients were not ZIKV infected. The “Negative” sera group was composed of 50 negative sera that were both IgM- and IgG-negative for DENV, WNV, Chikungunya virus, Encephalitis St Louis virus, ZIKV, Toscana virus, and Rift Valley fever virus. The ZIKV-IgM-positive, flavivirus and negative groups were representative of the entire parent groups in terms of their mean age and immune responses. We determined the sensitivity of the ZEDIII-based ELISA with the ZIKV-RT-PCR-positive group and compared it to that of the ZIKV-IgM-positive group, and the specificity to that of the flavivirus-positive group. The serology of the samples was tested upon their arrival to the arbovirus NRC and then stored at -20°C. The antibodies were stable over time and no difference in response was observed several months or years after freezing. Sequences of the Asian ZIKV strain from the French Polynesia outbreak of 2013, DENV2, the most frequently detected serotype by the NRC for Arbovirus, and DENV4, the DENV strain phylogenetically closest to ZIKV (accession numbers AHZ13508, P09866, and M29095, respectively), were selected for EDIII production using the structure of the ZIKV envelope protein (PDB 5JHM). The ZEDIII, D2EDIII, and D4EDIII coding sequences, optimized for production in E. coli, were synthesized in fusion with a His-tag coding sequence at their 5’ end by GenScript. The sequences were then cloned into pET-24a or pET-19b plasmids (Novagen). The recombinant proteins were produced in E. coli T7 Iq Express (New England Biolabs) and purified under denaturing conditions prior to in vitro refolding according to a protocol described previously [22]. The purity, conformation, and homogeneity of folding of the protein were controlled by size-exclusion chromatography coupled to multi-angle light-scattering analysis (SEC-MALS) and Coomassie blue-stained SDS PAGE gels. IgG were purified from pools of 20 sera from the ZIKV-positive group against ZEDIII, and 20 positive sera from the flavivirus-positive but ZIKV-negative groups against D2EDIII or D4EDIII (25 μL per sera for a total of 500 μL per pool). Antibody purification was carried out in two steps, first on a protein-G column and then on a column with immobilized recombinant EDIII. For the immobilization of EDIII, 1 mg of recombinant ZEDIII, D2EDIII, or D4EDIII in 2. 5 mL phosphate-buffered saline (PBS) was covalently bound to an NHS-activated column following the manufacturer’s protocol (GE Healthcare). Four milligrams of protein-G-purified antibodies from EDIII positive sera in 4 mL were then loaded onto the EDIII-coupled columns. After washes with PBS, EDIII-specific antibodies were eluted from the column with 1 mL 0. 1 M glycine (pH 2. 7) and neutralized with 100 μL 1 M Tris (pH 9). Vero cells were inoculated with 0. 01 MOI ZIKV (African ZIKV strain, accession number ArB41644) or DENV2 (Martinique DENV2 98–703 strain of 1998, accession number AF208496) and grown in Dulbecco’s modified Eagle medium (DMEM) complemented with 2% heat-inactivated fetal calf serum (FCS) at 37°C in 5% CO2 for 3. 5 (ZIKV) or 7 (DENV2) days. The time of growth depended on the virus. Culture supernatants were centrifuged, and viral particles precipitated with polyethylene glycol 6000 (PEG 6000) and NaCl. The precipitates were washed and resuspended in PBS/Hepes solution. This viral solution was inactivated with beta-propiolactone. Six-to-eight-week old female AG129 mice, weighing approximately 20 g at the start of the study, were subcutaneously injected with 10 μg ZEDIII in 20 μL PBS on day 0 and received a boost on days 15 and 30. Blood was collected on day 46 via the retro-orbital route. All procedures were in accordance with the guidelines set by the Noble Life Sciences (NLS) Animal Care and Use Committee. Noble is fully accredited by the Association for the Assessment and Accreditation of Laboratory Animal Care International (AAALAC). Veterinary care for all lab animals was in accordance with the Public Health Service Policy, U. S. Dept. of Agriculture (USDA) and AAALAC International requirements. All mice were bred at B&K Universal Laboratories under specific pathogen-free conditions. All experiments were carried out at Integrated BioTherapeutics, Inc. 96-well plates (Maxisorp, NUNC-Immuno Plate) were coated with 200 ng/well of recombinant protein diluted in PBS and incubated over night at 4°C. After removing antigen, wells were blocked with blocking buffer (3% milk in PBS) for 1 h at 25°C. After two washing steps with washing buffer (1% Tween 20 in PBS), they were incubated with 100 μL diluted human serum samples (1: 500), diluted purified human IgG (0. 15 or 0. 06 μg/mL), or sera from vaccinated mice diluted (1: 25) in dilution buffer (3% milk and 0. 1% Tween 20 in PBS), then incubated 1 h at 25°C. After four washing steps, the plates were incubated with horseradish peroxidase-conjugated anti–human IgG (1: 10,000) (Jackson Laboratories, Immuno Research) or horseradish peroxidase-conjugated anti–mouse IgG (1: 40,000) (Sigma-Aldrich) diluted in dilution buffer for 1 h at 25°C. After four washing steps, TMB (KPL, TMB microwell Peroxidase Substrate System) was added and the reaction stopped after 5 min with 50 μL 0. 25M H2SO4 per well. The optical density (OD) was read at 450 nm and the final values, OD ratio (ODr), obtained by dividing the average OD of duplicate wells from that of the corresponding blank non-coated wells. The threshold of positivity was calculated following the equation: m + 3σ (m = mean, σ = standard deviation, α (error) = 1%) on the ODr of the negative-sera group. Self-recognition of human purified IgG ODr was normalized to 100% for each EDIII. 96-well plates (Maxisorp, NUNC-Immuno Plate) were coated with inactivated viruses diluted (1: 200) in PBS and incubated over night at 4°C. After removing antigen, wells were blocked with blocking buffer (3% milk and 1% sodium azide in PBS) for 1 h at room temperature (RT) in the dark. After two washing steps with washing buffer (0. 05% Tween 20 in PBS), they were incubated with 100 μL human serum samples diluted (1: 500) in dilution buffer (3% milk, 0. 05% Tween 20, and 1% sodium azide in PBS), then incubated 1 h at 41°C. After four washing steps, the wells were incubated with human secondary antibody, horseradish peroxidase-conjugated anti–human IgG (1: 10 000) (Jackson Laboratories, Immuno Research), diluted in dilution buffer and incubated 1 h at 41°C. After four washing steps, TMB (KPL, TMB microwell Peroxidase Substrate System) was added and the reaction stopped after 5 min with 50 μL 0. 25M H2SO4 per well. The optical density (OD) was read at 450 nm and final values were obtained by dividing the average OD of duplicate wells from that of the corresponding blank wells coated with negative antigen. The threshold of positivity was fixed to an ODr of 3. IgG responses of human sera samples were tested against NS1 (Euroimmun kit) following the manufacturer’s protocol (Euroimmun). Ambiguous results were considered to be negative. Statistical analyses were performed with Prism 6 (GraphPad). IgM and IgG responses to DENV and ZIKV were correlated and R-values determined by the Spearman rank correlation test with n = 5,600. The IgG response data of the selected groups (ZIKV-RT-PCR-positive, n = 43; ZIKV-IgM-positive, n = 50; Flavivirus, n = 99; Negative, n = 50) were plotted showing the median, 25th, and 75th percentiles. The non-parametric Kruskal-Wallis test with Dunn’s correction was used to identify the differences between the medians of the IgG responses with two-tailed and alpha = 0. 01 for each test. The two-tailed Z-test with alpha = 0. 05 was used to compare the difference in sensitivity of the ZEDIII- and NS1-based ELISAs according to the group or cohort used. The overlap of the IC 95% was assessed to compare the difference of misdiagnoses between the ZEDIII- and NS1-based ELISAs. The non-parametric Wilcoxon rank-sum test was used to compare the difference of the mouse IgG responses to ZEDIII or D4EDIII. The study was performed in compliance with the STAndards for Reporting of Diagnostic accuracy (STARD) statement [23]. IgM cross-reactivity [95% Confidence Interval] between related flaviviruses (Fig 1A, Table 2) represented 23. 8% [20. 4; 27. 2] of the sera (n = 121) that reacted against at least one virus (n = 606). In contrast, IgG cross-reactivity [95% Confidence Interval] (Fig 1B) represented 71. 0% [69. 1; 72. 2] of the sera (n = 1,485) that reacted against at least one virus (n = 2,091). This cross-reactivity was evaluated by calculating the Spearman r correlation (p < 0. 0001; r (IgM) = 0. 4191, r (IgG) = 0. 7986). The r value for IgG indicates a strong correlation between the recognition of ZIKV and DENV (Table 1). We obtained similar results concerning the cross-reactivity of IgM (11. 6% [8. 3; 14. 9]; n = 42) and IgG (79. 7% [77. 4; 82]; n = 971) against WNV and ZIKV (Table 3). IgG from the sera of the ZIKV-positive group and the flavivirus-positive but ZIKV-negative group (colored dots, Fig 1B) cross reacted with DENV and ZIKV similarly to all 5,600 sera: 68. 7% versus 71% of IgG, showing that these two panels are representative of the 5,600 sera (Table 1). The ZEDIII amino-acid sequence shares 46. 3% and 47. 2% identity and 64. 8% homology with the amino-acid sequences of D2EDIII and D4EDIII, respectively (Fig 2). The DENV2 and DENV4 EDIII amino-acid sequences share 61% identity and 78. 1% homology. The purified EDIIIs analyzed by SEC-MALS and Coomassie blue-stained SDS Page gels showed the elution of a pure protein in a single and symmetric peak with an average molecular weight of 14. 5 kDa and a polydispersity of 1. 01 (Fig 3A and 3B). The presence and importance of the conformational epitopes for ZEDIII recognition was highlighted by the loss of recognition of ZIKV-positive sera against chemically and thermally denatured ZEDIII and a gain of recognition of flavivirus-positive but non ZIKV-positive or negative sera (Fig 3C). We investigated the reactivity of IgG of the four sera groups against inactivated DENV, ZIKV, ZEDIII, and NS1 proteins (Fig 4 and [21]) by ELISA. The ZIKV-RT-PCR-positive, ZIKV-IgM-positive, and flavivirus groups recognized DENV with medians of 14. 4,16. 6, and 12. 3 ODr, respectively, with no statistical difference (K = 5. 02, p = 0. 0811), whereas no negative serum samples recognized DENV (median ODr of 1. 1). The ZIKV-RT-PCR-positive, ZIKV-IgM-positive, and flavivirus groups recognized ZIKV (median ODr of 8. 6,7. 3, and 2. 8, respectively). The levels of recognition between the ZIKV-RT-PCR-positive or ZIKV-IgM-positive groups and flavivirus group were statistically different (Mean rank diff. = 89. 61, p < 0. 0001; Mean rank diff. = 64. 98, p < 0. 0001) but similar for the ZIKV-RT-PCR-positive and ZIKV-IgM-positive groups (Mean rank diff. = 24. 63, p = 0. 0993). The negative group did not recognize ZIKV (median ODr of 1. 1). The difference between the ability of the ZIKV-RT-PCR-positive and flavivirus groups (median ODr of 2. 2 and 1. 1, respectively) or ZIK-IgM-positive and flavivirus groups (median ODr of 3. 6 and 1. 1, respectively) to recognize ZEDIII was highly significant (Mean rank diff. = 103. 3, p < 0. 0001, Mean rank diff. = 123. 2, p < 0. 0001, respectively), whereas there was no statistical difference between the flavivirus and negative sera (median ODr of 1. 1 and 1. 2; respectively, Mean rank diff. = -13. 75, p > 0. 9999) or the ZIKV-RT-PCR-positive and ZIKV-IgM-positive sera (median ODr of 2. 2 [1. 7; 4. 0] and 3. 6 [2. 0; 8. 5]; respectively, Mean rank diff. = -19. 82, p > 0. 9999). The median of the OD obtained for the ZIKV-RT-PCR-positive and ZIKV-IgM-positive sera against ZEDIII was 0. 12 (IQR [0. 09; 0. 24] min = 0. 06; max = 1. 96) and 0. 29 (IQR [0. 17; 0. 64] min = 0. 07; max = 2. 10) respectively, whereas the median OD obtained against the blank was 0. 05 (IQR [0. 05; 0. 06] min = 0. 05; max = 0. 08) and 0. 08 ([0. 07; 0. 10], min = 0. 05, max = 0. 23), respectively. With a calculated positive threshold based on the negative-group values (m + 3σ) of 1. 54 (m = 1. 17 and σ = 0. 13), and tolerating a 1% error rate, one serum (2%) of the negative sera was positive against ZEDIII and 49 (98%) were negative. The smallest OD value considered to be positive with this positive threshold was 0. 085, with a blank of 0. 055, and the highest OD value was 2. 12. The sensitivity was calculated as the number of positive sera from the ZIKV-RT-PCR-positive or ZIKV-IgM-positive groups, of which 39 of 43 (90. 7%) or 46 of 50 (92. 0%), respectively, were positive against the ZEDIII-based assay (Table 4) and four (9. 3% or 8. 0%, respectively) did not react. The sensitivity [95% confidence interval] was thus 90. 7% [82. 0; 99. 4] or 92. 0% [84. 5; 99. 5], respectively. There was no difference in sensitivity between these two groups (Z-test, p = 0. 2219). The specificity was calculated as the negative sera in the flavivirus group (n = 99) or in both the flavivirus group plus the negative group (n = 149). Ten sera (10. 1%) of the flavivirus group or 11 sera (7. 4%) of the flavivirus group plus the negative group were positive against ZEDIII and 89 or 138 were negative (89. 9% and 92. 6%, respectively). Thus, the specificity [95% confidence interval] was 89. 9% [84. 0; 95. 8] or 92. 6% [84. 4; 100]. These groups were also tested against NS1 protein using the Euroimmun kit. We determined a specificity of 83. 8% [76. 8; 91. 2] with the flavivirus group and up to 89. 3% [84. 3; 94. 3] with the flavivirus group plus the negative group and a sensitivity of 88. 4% [78. 8; 98. 0] with the ZIKV-RT-PCR-positive group and 96. 0% [90. 6; 100. 0] with the ZIKV-IgM-positive group (Table 4). There was no statistical difference in sensitivity between these two groups (Z-test, p = 1. 3535). The specificity obtained with the NRC ELISA against ZIKV was 46. 5% [36. 7; 56. 3]; the sensitivity was 100% with the ZIKV-RT-PCR-positive group and 96. 0% [90. 6; 100] with ZIKV-IgM-positive group. We compared the reliability of the ZEDIII- and NS1-based ELISA by calculating the positive predictive value (PPV), the false discovery rate (FDR), the negative predictive value (NPV), and the false omission rate (FOR) within a ‘hypothetical population’ with a given prevalence of 33. 5% (50/149), composed of ZIKV-IgM-positive (sensitivity identical to that of the ZIKV-RT-PCR-positive group) and flavivirus groups (n = 149; Table 5). The likelihood of being positive when detected as positive (PPV) was 82. 1% [72. 1; 92. 1] with the ZEDIII- and 75. 0% [64. 4; 85. 6] with the NS1-based ELISA, whereas the likelihood of being negative when detected as negative (NPV) was 95. 7% [91. 6; 99. 8] and 97. 6% [94. 3; 100] with the ZEDIII- and NS1-based assays, respectively. The likelihood of being positive when not detected (FOR) was 4. 3% [0. 2; 8. 4] and 2. 4% [0; 5. 7] with the ZEDIII- and NS1-based ELISAs, respectively, and the likelihood of being negative when detected as positive (FDR) was 17. 9% [7. 9; 27. 9] and 25. 0% [14. 4; 35. 6] with the ZEDIII- and NS1-based ELISAs, respectively. We further characterized the lack of cross reactivity of anti-EDIII IgG of the three sub-groups selected from the ZIKV-positive and flavivirus-positive panels (Table 6). IgG purified against D2EDIII and D4EDIII cross-reacted with D4EDIII and D2EDIII (39. 7 ± 4. 2% and 47. 0 ± 4. 2% respectively, Fig 5A). Conversely, D2EDIII and D4EDIII-purified IgG did not recognize ZEDIII. ZEDIII-purified IgG recognized ZEDIII well (ODr = 15. 9). We next addressed the immune response induced by the recombinant ZEDIII. The median ODr of the sera obtained after the immunization of eight mice by ZEDIII was 2. 25 and 1. 02 against ZEDIII and D4EDIII, respectively. The difference between the ability of the sera to recognize the two recombinant proteins was significant (W = 36, p = 0. 0078). None of the five ZEDIII-positive sera from mice (n = 8) recognized D4EDIII (Fig 5B), with the sequence closest to that of ZEDIII (Fig 2). To date, no broad antibody cross-reactivity study has been performed for flavivirus with a particular emphasis on ZIKV, despite the often high co-endemicity of flaviviruses. Our study included more than 5,000 samples, providing sufficient power to observe statistically significant cross-reactivity results and select sera for the design of reference groups of patients. First, we observed low IgM cross-reactivity and high IgG cross reactivity against inactivated ZIKV, DENV, and WNV. This is generally observed with flavivirus-positive human sera, resulting in low diagnostic reliability of IgG-based assays [14,24]. The use of peptides or recombinant proteins can partially improve the specificity of IgG-based diagnostic assays. Indeed, ZEDIII is specifically recognized with high sensitivity by anti-ZIKV IgG and can distinguish between DENV and ZIKV in late phase or post-infection. IgM detection shows relatively good specificity (only 23. 8% cross-reactivity) permitting the determination of the infecting flavivirus, as already shown [25]. However, the IgM response does not persist. Thus, we studied ZEDIII recognition by IgGs, the most persistent antibodies. We divided the set of sera into four groups based on various characteristics that allowed us to confirm their membership in each group. First, we determined the sensitivity of our assay using the reference group, the ZIKV-RT-PCR-positive group, which recognized the ZEDIII domain with the same sensitivity (90. 7% [82. 0; 99. 4]) as the selected ZIKV-IgM-positive group (92. 0% [84. 5; 99. 5]), allowing us to validate the ZIKV-IgM-positive group as a ZIKV-positive group. Thus, IgM that recognized only the total inactivated ZIKV, but not DENV (56. 3% of ZIKV-infected patients), were induced by ZIKV, allowing the diagnosis of a recent ZIKV infection. Second, IgG from the sera of “flavivirus” patients, selected from 2013 to 2014 in Guadeloupe and Martinique, recognized all flaviviruses where and when ZIKV was likely not circulating. The large size of our group and the rational to select the sera allowed us to assemble relevant panels of samples suitable to support a robust and accurate study of ZIKV IgG detection using recombinant ZEDIII antigen. The specificity for ZEDIII reached 89. 9% and sensitivity 92. 0%, whereas IgG presented cross-reactivity against total inactivated viruses (71% of IgG recognized both DENV and ZIKV). This makes recombinant EDIII one of the most robust tools for diagnosis [14,24]. Of note, most of the 10 false-positive sera from the flavivirus groups gave a value close to the positive threshold. The ODr of only one serum sample was 6. 26. Based on denaturation experiments (Fig 3), correct conformational folding of the EDIII protein is required for specific recognition. Indeed, chemical and heat denaturation of the protein led to the loss of recognition in ELISA assays, likely due to the loss of conformational epitopes, thereby decreasing specificity. Given the high DENV seroprevalence observed in the area from which the ZIKV-IgM-positive group (mean age = 40 years) originated (93. 5% of the Caribbean island population (mean age = 38 years), [26]) it is likely that they were infected with DENV prior to ZIKV infection, although we have no way of knowing whether the antibodies that recognized ZEDIII were actually induced by DENV. We thus purified IgG against immobilized EDIII domains to discriminate between IgG induced by and directed against ZIKV from that induced by DENV but cross-reacting to both ZIKV and DENV. No IgG from the selected pools purified against D2EDIII or D4EDIII recognized ZEDIII. Such a lack of cross-reactivity underscores the high specificity of the immune response against this antigenic domain. It is likely that the anti-ZEDIII antibodies were induced by ZIKV infection and not DENV infection, as we were unable to observe recognition of ZEDIII by the sera of patients who had never been infected by ZIKV. Thus, the high sensitivity obtained with our assay is due to ZIKV-induced IgG recognition and not potential cross-reactive IgG. The binding experiments developed to validate the low cross-reactivity in the ZEDIII-based ELISA showed that IgG purified against D4EDIII partially recognized D2EDIII, and vice-versa. This result is not surprising, as the experimental conditions were designed to force cross-reactivity further than in a classical sero-diagnostic test with high IgG concentrations. Instead, the experiment validates ZEDIII as a relevant antigen for sero-diagnosis. We further validated the high specificity of the humoral immune response against EDIII by assessing the ability of ZEDIII-immunized mouse sera to recognize recombinant EDIII. The sera of ZEDIII-positive mice did not recognize D4EDIII. As the D4EDIII amino-acid sequence is that closest to ZEDIII, it is possible that the result would be the same with any other flavivirus EDIII. Finally, we designed experiments using DEDIII-purified IgG to show that the observed sensitivity and specificity were due to only ZIKV-IgG binding to ZEDIII. The epitopes present on the surface of DEDIII were only recognized by IgG that did not recognize ZEDIII. IgG induced by ZEDIII immunization recognized only epitopes present on the surface of ZEDIII. Thus, the epitopes recognized by ZEDIII-IgG are highly different from those carried by D4EDIIIand the EDIII epitopes are virus specific. We hypothesized that WNV induced IgG will not recognize ZEDIII. To date, the most robust and specific ZIKV sero-diagnostic test is the PRNT, which is costly and time-consuming. ELISA is also commonly used but usually serves as a first-line screen prior to PRNT, due to its low specificity. Indeed, we showed that 71% of IgG diagnoses were not reliable with an ELISA performed with inactivated ZIKV and that 48% of flavivirus sera were ZIKV-cross-reactive. We assessed our ZEDIII-based-ELISA and the NS1-based-ELISA (Euroimmun Assay) on the selected sera of the four groups. The ZEDIII- and NS1-based-ELISAs for IgG were 89. 9% to 92. 6% and 83. 8% to 89. 3% specific (calculated with the flavivirus group (n = 99) or the flavivirus group plus the negative group (n = 149), respectively), and had a sensitivity of 92. 0% and 96. 0%, respectively (Table 4). The sensitivity determined for the NS1-based ELISA varies significantly, depending on the cohort used: from 70. 7% with the cohort of Matheus et al. [21] to 88. 4% (ZIKV-RT-PCR-positive group) or 92. 0% (ZIKV-IgM-positive group) (Z-test between the ZIKV-RT-PCR-positive or ZIKV-IgM-positive groups and the cohort of Matheus et al. , p = 2. 5257 and p = 4. 4080, respectively). With our assay, the sensitivity varied non-significantly from 90. 7% (ZIKV-RT-PCR-positive group) to 92. 0% (ZIKV-IgM-positive group). The sensitivity of the ZEDIII-based ELISA can thus be considered stable. The ZIKV-IgM-positive group sera were collected in a DENV-endemic area, where the ZIKV infection is a second flavivirus infection, whereas this infection was the first for 65. 6% of the ZIKV-RT-PCR-positive group. Thus, the difference in sensitivity could be due to this immunological scar. Specificity is an important criterion for flavivirus diagnosis, as flaviviruses can induce cross-reactive IgG and high DENV prevalence is observed in the ZIKV endemic area. The specificity of the ZEDIII-based-ELISA was in the same range as that of the commercial test: within a 33. 5% prevalence population, the probability of misdiagnosis would be 9. 4% for the ZEDIII-based ELISA versus 12. 1% for the NS1-based assay, which was not significantly different when considering the IC 95% overlap. However, unlike the NS1-based ELISA, the sensitivity of our assay did not depend on the cohort tested. Relative to the NS1-based ELISA, our assay would lead to the misdiagnosis of 1. 9% additional positive patients but 7. 1% fewer negative patients. The recombinant refolded ZEDIII domain could thus be a good target for diagnosis. According to Matheus et al. [21] and our study (data not shown), IgG raised against ZIKV and ZEDIII were still present, with at least 50% of the maximum ODr, after 300 DPSO and up to several years after infection [27]. This minor limitation could allow us to carry out retrospective seroprevalence studies of the ZIKV outbreak, whereas detecting ZIKV-induced IgG with ZEDIII could be used to detect recent past infections in populations at risk. A specific diagnosis must also be reliable for pregnant women or those planning to become pregnant to indicate their ZIKV immunization status, and not necessarily at the first stages of putative ZIKV infection, to orient medical care, as ZIKV can cause miscarriage [28,29]. Recombinant EDIII proteins have already been used as IgG-targets in ELISA and Microsphere ImmunoAssay to detect the immune response of horses and humans in seroprevalence studies [30–33] and, in particular, ZEDIII to link pathology to infection in humans [34,35]. However, no study has yet clearly shown the limits, specificity, and sensitivity of a ZEDIII-based ELISA. Our study reinforces precedent EDIII findings and shows the use of recombinant flavivirus protein-based ELISAs, such as that based on ZEDIIIE, to be potentially reliable, offering opportunities to develop rapid, inexpensive, and specific first-line assays. Antibodies specific to the infecting flavivirus are able to protect against a new infection by neutralization [13,36], whereas cross-reacting antibodies have been extensively linked to antibody-dependent enhancement, increasing viremia [37]. This phenomenon has been observed for both DENV- and ZIKV-induced antibodies [38]. Our results suggest that ZEDIII, by inducing a strong and specific immune response, could also be a safe model for the development of vaccines.

The serological detection of Zika virus (ZIKV) is a challenge, as ZIKV infection generally leads to an immune response with a high level of cross reactivity against related viruses, such as Dengue virus. Although seroneutralization assays are the gold-standard to address specificity, a rapid and cost-effective detection assay with good specificity and sensitivity could be used for first-line screening. We used a large cohort to define a set of human reference sera to validate an ELISA based on a recombinant ZIKV antigen. The assay showed 90% specificity and 92% sensitivity, providing a good basis for the development of diagnostic assays. Characterization of both DENV-EDIII-purified human and murine IgG induced by ZIKV infection or ZEDIII, respectively, confirmed the good specificity of the antigen.

Abstract Introduction Materials and methods Results Discussion

dengue virus medicine and health sciences enzyme-linked immunoassays immune physiology pathology and laboratory medicine pathogens immunology microbiology viruses rna viruses antibodies immunologic techniques cross reactivity research and analysis methods immune system proteins proteins medical microbiology microbial pathogens immunoassays recombinant proteins biochemistry arboviruses flaviviruses viral pathogens physiology biology and life sciences organisms zika virus

2019

High specificity and sensitivity of Zika EDIII-based ELISA diagnosis highlighted by a large human reference panel

NF-κB and inflammasomes both play central roles in orchestrating anti-pathogen responses by rapidly inducing a variety of early-response cytokines and chemokines following infection. Myxoma virus (MYXV), a pathogenic poxvirus of rabbits, encodes a member of the cellular pyrin domain (PYD) superfamily, called M013. The viral M013 protein was previously shown to bind host ASC-1 protein and inhibit the cellular inflammasome complex that regulates the activation and secretion of caspase 1-regulated cytokines such as IL-1β and IL-18. Here, we report that human THP-1 monocytic cells infected with a MYXV construct deleted for the M013L gene (vMyxM013-KO), in stark contrast to the parental MYXV, rapidly induce high levels of secreted pro-inflammatory cytokines like TNF, IL-6, and MCP-1, all of which are regulated by NF-κB. The induction of these NF-κB regulated cytokines following infection with vMyxM013-KO was also confirmed in vivo using THP-1 derived xenografts in NOD-SCID mice. vMyxM013-KO virus infection specifically induced the rapid phosphorylation of IKK and degradation of IκBα, which was followed by nuclear translocation of NF-κB/p65. Even in the absence of virus infection, transiently expressed M013 protein alone inhibited cellular NF-κB-mediated reporter gene expression and nuclear translocation of NF-κB/p65. Using protein/protein interaction analysis, we show that M013 protein also binds directly with cellular NF-κB1, suggesting a direct physical and functional linkage between NF-κB1 and ASC-1. We further demonstrate that inhibition of the inflammasome with a caspase-1 inhibitor did not prevent the induction of NF-κB regulated cytokines following infection with vMyxM013-KO virus, but did block the activation of IL-1β. Thus, the poxviral M013 inhibitor exerts a dual immuno-subversive role in the simultaneous co-regulation of both the cellular inflammasome complex and NF-κB-mediated pro-inflammatory responses. The nuclear factor κB (NF-κB) family comprise a set of related transcription factors that regulate multiple cellular pathways associated with immune responses, inflammation, apoptosis, cell growth and survival [1], [2]. The mammalian NF-κB family members include NF-κB1 (p105 and p50), NF-κB2 (p100 and p52), RelA (p65), RelB and c-Rel, which become activated by upstream signals from diverse immune receptors, such as ligand-triggered Toll-like receptors (TLRs), interleukin-1 receptor (IL-1R), tumor necrosis factor receptor (TNFR) and antigen receptors [3], [4]. However, their activation is tightly regulated by another family of proteins known as inhibitors of NF-κB (IκBs), which are also associated with the regulation of these diverse cellular processes [5]. In unstimulated cells, NF-κB proteins remain inactive in the cytoplasm, usually as homodimers or as heterodimers with RelA, RelB or c-Rel and complexed with the inhibitory IκBs. The IκBs are in turn regulated by another group of regulatory proteins called IκB kinases (IKKs). The IKK complex is composed of two catalysis subunits (IKKα and IKKβ) and a regulatory subunit IKKγ, or NEMO (NF-κB essential modulator) [6], [7], [8]. Upon ligand engagement, the receptors initiate signal transduction events that lead to the activation of IKK complex by activated cellular protein kinases like NF-κB-inducing kinase (NIK), mitogen-activated protein/extracellular signal-regulated kinase1 (MEKK1), transforming growth factor-β (TGFβ) - activated kinase 1 (TAK1), MEKK2 or MEKK3. Activated IKK phosphorylates IκBα using specific serine residue within the IκBα proteins, triggering their ubiquitination via ubiquitin ligase [9]. The IκBα protein is then degraded by the 26S proteasome, allowing the release and translocation of the active NF-κB dimer into the nucleus to upregulate targeted gene transcription events [10], [11]. Subversion of the host cell NF-κB signaling is a key strategy adapted by many pathogens because this pathway regulates the expression of multiple cellular host response proteins such as anti-viral cytokines, chemokines and the presentation of viral antigens to cytotoxic T lymphocytes [12], [13], [14]. Amongst viruses, members of the poxvirus family are known to target and downregulate NF-κB signaling in multiple ways. One strategy is to directly block the primary immune ligands, particularly cytokines, chemokines or IFNs, and thus prevent binding and activation of their cognate receptors. In many cases these viral inhibitors (called viroceptors) are secreted homologs of corresponding cellular receptors [15], [16]. Another strategy is to deploy viral proteins to target the components of the intracellular NF-κB signaling pathways [17]. For example, vaccinia virus (VACV), the prototypic member of the poxviridae family and the most extensively studied poxvirus, encodes multiple proteins to inhibit the intracellular NF-κB pathway at different steps of the signaling cascade. For example, VACV-encoded proteins A52R and A46R inhibit the interleukin 1 receptor (IL-1R) and toll-like receptor (TLR) signaling pathway linked to the NF-κB complex [18], [19]. A52R inhibits NF-κB activation by interaction with interleukin 1 receptor-associated kinase 2 (IRAK2) and tumor necrosis factor receptor associated factor (TRAF6) [20]. B14R was recently shown to bind the IKK complex and prevent phosphorylation of the IKKβ activation loop [21]. Another VACV protein, N1L, also targets the IKK complex and interferes with both NF-κB and interferon regulatory factor 3 (IRF3) signaling [22]. Among the other VACV proteins, K1L inhibits NF-κB activation by blocking degradation of IκBα [23] while M2L downregulates ERK-mediated NF-κB induction in infected cells [24]. Another poxvirus, molluscum contagiosum virus (MOCV), also encodes multiple proteins, for example MC159 and MC160, which regulate the NF-κB signaling pathways [25], [26], [27]. Myxoma virus (MYXV) is a member of the leporipoxvirus genus of the poxviridae family and causes lethal disease myxomatosis in European rabbits (Oryctolagus cuniculus) [28]. MYXV encodes diverse secreted and intracellular immunomodulatory proteins to overcome host immune defenses [29]. However, unlike most of the members of the orthopoxvirus genus, MYXV does not express any predicted secreted proteins that directly bind and neutralize pro-inflammatory cytokines such as IL-1β and IL-18, which are regulated by a series of cellular multi-protein complexes collectively termed inflammasomes [30], [31], [32]. Instead, the MYXV-encoded protein M013, which was first identified as a member of the PYRIN domain (PYD) superfamily, was shown to function as an intracellular inhibitor of inflammasome activation [33]. The cellular PYD-containing proteins mediate protein-protein interactions with components of signaling pathways involved in the regulation of apoptosis, NF-κB activation and pro-inflammatory cytokine production [34], [35]. Our initial study showed that a MYXV construct deleted for M013L (vMyxM013-KO) was significantly attenuated in host rabbits because of decreased virus dissemination and enhanced inflammatory responses at the tissue sites of virus infection [33]. The vMyxM013-KO virus was unable to productively infect rabbit monocytes or lymphocytes due to an abortive phenotype, compared to parental MYXV. M013 protein was also shown to directly interact with apoptosis-associated speck-like protein containing CARD-1 (ASC-1), a component of the cellular inflammasome complex and inhibited caspase-1 activation and the processing of pro-inflammatory cytokines IL-1β and IL-18 [33]. In the present study, we report that infection of human THP-1 monocytic cells with the vMyxM013-KO virus, unlike the parental MYXV, unexpectedly induced rapid and dramatic secretion of diverse pro-inflammatory cytokines such as TNF, IL-6 and MCP-1, all of which are regulated by NF-κB. THP-1 cells infected with vMyxM013-KO virus, but not the parental MYXV, activated the IKK kinases and degradation of IκBα at very early time of infection, resulting in the activation and nuclear translocation of NF-κB. We further demonstrate that the expressed M013 protein alone interacts directly not only with ASC-1, but also with host NF-κB1 and inhibits the translocation of p65 to the nucleus. We conclude that the viral PYD-containing M013 protein can simultaneously bind and co-regulate key components from both the inflammasome complex and NF-κB-mediated signaling pathways. Recombinant TNF was purchased from Biosource. Rabbit polyclonal antibodies (pAb) for IκBα, phospho-IκBα, p65, NF-κB1, IKKα/β, phospho-IKKα/β, were purchased from Cell Signaling Technology. HRP-conjugated goat anti-rabbit and anti-mouse IgG antibodies were purchased from Jackson Lab. Phorbol-12-myristate-13-acetate (PMA), LPS (lipopolysaccharide), and caspase-1 inhibitor zVAD-fmk were purchased from Sigma. ERK1/2 inhibitor U0126 and PI3kinase inhibitor LY294002 were purchased from Cell Signaling technology. Cloning strategies used are based on the Gateway cloning technology (Invitrogen). The M013L ORF of myxoma virus was PCR amplified from the virus genome using Pfu Ultra polymerase (Stratagene) and cloned in the Gateway entry vector pDONR221 (Invitrogen) using BP Clonase enzyme mix (Invitrogen). M013L ORF lacking the PYD (1–81 amino acids) was constructed by PCR amplification using appropriate primers and cloned in the entry vector pDONR221. Both human RelA and NF-κB1 cDNAs (Open Biosystems) were PCR amplified and cloned in pDONR222 vector (Invitrogen). The resultant entry clones were subsequently cloned in expression vectors pANT7_cGST, pANT7_nHA [36], pcDNA3. 1MycHis and pDEST40 (Invitrogen) using the LR Clonase II enzyme mix (Invitrogen). BGMK, 293T, BSRT7/5 (BHK cells expressing T7 polymerase) [37] cells were cultured in DMEM supplemented with 10% heat-inactivated fetal bovine serum (FBS), 2 mM L-glutamine, 100 U/ml penicillin and 100 µg/ml streptomycin (pen-strep). THP1 cell line was cultured in RPMI 1640 medium (Lonza) supplemented with 10% FBS and pen-strep. For differentiation, THP1 cells were stimulated for 18 h (hour) with 100 ng/ml PMA. Construction of a wild-type myxoma virus (vMyx-gfp) that expresses a GFP cassette under the control of a synthetic VACV early/late promoter was described previously [38]. Construction of the vMyxM013-KO virus was previously described [33]. Viruses were purified by centrifugation through a sucrose cushion and two successive sucrose gradient sedimentations as described previously [39]. HeLa cells (1×104) were seeded at 50% confluence onto 96-well plates the day before being transfected. Cells were transfected according to the manufacturer' s protocol using 0. 5 µl Lipofectamine 2000 reagent (Invitrogen) and 0. 2 µg DNA, in serum-free minimum essential medium (MEM), per each well. The DNA was a 1∶1∶1 mix of NF-κB reporter vector to constitutive expression vector to viral protein expression vector. The NF-κB reporter vector used was the Stratagene (La Jolla, CA) pNFκB-Luc plasmid, which gives inducible NF-κB-dependent expression of firefly (Photinus pyralis) luciferase driven by a synthetic promoter comprising a TATA box preceded by five direct repeats of the sequence 59-TGGGGACTTTCCGC-39 containing the NF-κB binding element first identified in the kappa light chain gene enhancer [40]. The constitutive expression vector used was the Promega (Madison, WI) pUCbased pRL-TK vector, which gives low-level constitutive expression of sea pansy (Renilla reniformis) luciferase from the promoter of the herpes virus thymidine kinase gene. M013L ORF was cloned in pcDNA3. 1MycHis vector (Invitrogen) that gives a constitutive level of protein expression from the CMV promoter [33]. At 6 h post-transfection, FBS was added to the medium for a final concentration of 10%. At 48 h post-transfection, the cells were treated with 20 ng/ml TNF for 6 h. To determine luciferase values, the Promega Dual-Luciferase Reporter Assay System was used according to the manufacturer' s instructions, with slight modifications. Briefly, the medium was removed and the cells were washed once with 50 µl PBS and the cells were lysed in 20 µl of 1× passive lysis buffer (Promega) for 15 min at room temperature. The lysates were then analyzed for firefly and sea pansy luciferase activity. For luciferase value determination, 100 µl of luciferase assay reagent II (Promega) were added to the cell lysate and analyzed for firefly luciferase activity using the Appliskan multimode microplate reader (Thermo Scientific). Then 100 µl of Stop & Glo reagent (Promega) were added for determination of sea pansy luciferase activity. The relative fold changes were determined by normalizing the ratios of firefly to sea pansy luciferase activities in each transfected cell group to the value obtained for non-transfected cells. For detection of protein, cells were harvested at different time points, washed with PBS and stored at −80°C or processed immediately with RIPA lysis buffer (50 mM Tris, 150 mM NaCl, 0. 1% SDS, 0. 5% sodium deoxycholate, 1% NP40,1 mM PMSF, protease inhibitor cocktail (Roche) ). Protein samples were separated on SDS-PAGE gels and transferred to PVDF membrane (GE Healthcare) using a semidry transfer apparatus (Fisher). Membranes were blocked in TBST (20 mM Tris, 150 mM NaCl, 0. 1% Tween-20 pH 7. 6) containing 5% non-fat dry milk for 1 hr at room temperature and then incubated overnight with primary antibody at 4°C. The membranes were washed three times, 15 minutes each with TBST and incubated with HRP-conjugated goat-anti-mouse (1∶5000) or goat-anti-rabbit (1∶5000) secondary antibody in TBST containing 5% non-fat dry milk for 1 hour at room temperature with gentle agitation. The membranes were washed three times, 15 minutes each with TBST, and the signal was detected following the application of chemiluminescence substrate (Pierce) and exposure to X-ray film (Eastman Kodak). THP-1 cells were plated in multi-well plates in the presence of PMA (100 ng/ml) for 12–18 h. The following day, media were replaced with fresh media and the cells were either infected with WT myxoma and vMyxM013-KO viruses (at MOI of 3) or treated with LPS and the supernatants were collected at different time points. Whenever mentioned, inhibitors were added one hour before viral infection. The level of secreted cytokines TNF, IL-1β, IL-6, IL-12 and MCP-1 were determined using ELISA assay kits (eBioscience) following manufacturer protocol. The rabbit reticulolysate coupled transcription and translation (TnT) system (Promega) was used according to manufacturer' s protocol for expression of proteins in vitro. Protein expression was confirmed by running 2. 5 µl of the TnT reaction on SDS-PAGE gels followed by western blot analysis using anti-GST antibody (Neomarkers) for the GST-tagged protein and anti-HA antibody (Santa Cruz) for the HA-tagged protein. The Amplified Luminescent Proximity Homogeneous Assay (AlphaScreen™; PerkinElmer) [41], [42] is a bead-based technology for screening biomolecular interactions in a microplate format [43]. The assay was performed in 384-well white opaque optiplates (Perkin Elmer) in a total volume of 25 µl. BSR T7/5 cells were used for expression of tested proteins. Cells were seeded onto 24-well plates one day before transfection. Cells were then transfected according to the manufacturer' s protocol using 2 µl Lipofectamine 2000 reagent (Gibco BRL) and 0. 8 µg total DNA (equal amounts of host and viral DNA), in serum-free MEM, per each well. The expression vectors used were either the pcDNA3. 1MycHis plasmid (Invitrogen), for the host proteins, or the pANT7_cGST plasmid (Harvard Institute of Proteomics), for the viral protein. Expression from the pcDNA3. 1MycHis plasmid is under the control of either CMV or T7 RNA polymerase promoter and the expressed proteins are fused at their C-terminal with both Myc and His tags. On the other hand, expression from the pANT7_cGST plasmid is under the control of T7 RNA polymerase promoter and expressed proteins are fused at their C-terminal with a glutathione-S-transferase (GST) tag. Forty eight hours post transfection, cells were collected in 500 µl PBS, pelleted and then lysed by suspending in 20 µl lysis buffer (100 mM NaCl, 100 mM Tris, pH 8. 0,0. 5% NP-40, containing 25 µl/ml Roche complete protease inhibitor). The cellular extracts were then separated following centrifugation for 5 min at 13,000 rpm. All dilutions were made in phosphate buffered saline (PBS; 3. 2 mM Na2HPO4,0. 5 mM KH2PO4,1. 3 mM KCl, 135 mM NaCl, pH 7. 4) containing 0. 1% bovine serum albumin (BSA). Five µl of anti-GST acceptor beads (PerkinElmer, 6760603C), diluted 50-fold (20 µg/ml final concentration of beads in assay) in 0. 1% BSA/PBS assay buffer, and 5 µl of cell extract were mixed to a final volume of 20 µl with 0. 1% BSA/PBS assay buffer, and incubated at room temperature for 1. 5 hour. Subsequently, 5 µl of Nickel Chelate donor beads (PerkinElmer, AS101D), diluted 50-fold in 0. 1% BSA/PBS assay buffer (20 µg/ml final concentration of beads in assay), were added to a final volume of 25 µl and incubation was continued for another 1. 5 hour at room temperature. All additions and incubations were performed under subdued lighting conditions due to the photosensitivity of the beads. The plate was then read on an EnVision multiwell plate reader (PerkinElmer). A modified Enzyme-linked immunosorbent assay (ELISA) was used for in-vitro protein-protein interaction studies. The 384-well ELISA plates were coated with a rabbit polyclonal anti-GST antibody (Neomarkers), diluted 1 to 400 in coating buffer (0. 138 M NaCl, 0. 0027 M KCl, pH 7. 4), overnight at 4°C in 25 µl volume. Following incubation, the antibody was removed and wells were blocked overnight with 5% non-fat milk in PBS. Blocking buffer was removed and wells were washed 5 times with 100 µl washing buffer (PBS containing 1% BSA and 0. 05% Tween 20). Both viral and host proteins were expressed, either individually or in combination, using the TnT in vitro expression system following manufacturer' s protocol. Either 1. 25 µl or 2. 5 µl (for individually expressed or co-expressed proteins, respectively) of each TnT reaction was then applied, per well, to the anti-GST antibody coated plate and incubated for 2 hr at room temperature. The wells were then washed 5 times with washing buffer. Assembly of the protein complex in the wells was then assessed through the incubation of HRP-conjugated rat anti-HA antibody (1∶500 dilution) (Roche) in 25 µl volume for 2 hr at room temperature. The unbound antibody was removed by washing the wells 5 times with washing buffer. Binding of the second antibody (HRP-conjugated anti-HA) to the protein complex was then detected by applying 50 µl of TMB substrate. The reaction was stopped by adding 25 µl of 2 N H2SO4 to each well. The plate was read at 450 nm using a multi-well plate reader. HeLa cells (3×104) plated on coverslips were either mock transfected or transiently transfected with 2 µg of pcDNA3. 1M013L or pDEST40M013ΔPYD DNA. Cells were mock treated or treated with 20 ng/ml TNF for 30 min at 37°C at 24 hr post transfection. Coverslips were then washed with PBS and fixed with 2% paraformaldehyde (Sigma) for 12 minutes, permeabilized with 1% NP40 (Sigma) and blocked with 3% BSA in PBS. Coverslips were incubated with both rabbit anti- p65 (sc-372; Santa Cruz) and mouse anti-Myc (sc-40; Santa Cruz) or mouse anti-V5 (Invitrogen) diluted 1∶100 in PBS containing 3% BSA for 30 minutes at 37°C. Coverslips were incubated in these secondary antibodies: goat anti-rabbit-Texas-Red and goat anti-mouse- fluorescein isothiocyanate diluted at 1∶750 in PBS containing 3% BSA for 30 minutes at 37°C in the dark. Following staining, coverslips were mounted on microscope slides with 7. 5 µl Vecta-shield mounting media with DAPI (4′, 6 diamidino-2-phenylindole; Vector Laboratories, Burlingame, CA, USA) for visualization of nuclei. Cells were visualized using the 40X objective of an Olympus DSU-IX81 Spinning Disc Confocal/Deconvolution fluorescent microscope. Xenografted THP-1 tumors were generated by implanting THP-1 cells (5×106 in 100 µl PBS) subcutaneously (s. c) on the left hind leg of 6–8 week old female NOD-SCID mice (Charles River). Tumor growth was monitored every other day. On day 6 after tumor cell implantation, a single dose of MYXV or vMyxM013-KO (1×108 ffu virus in 100 µl volume) was injected directly in the tumor. The animals of the control group were injected with PBS into the tumor. After 48 h following virus injection, the animals were sacrificed and both blood and tumor tissues (suspended in PBS) were collected. The levels of secreted human cytokines in serum and tumor tissues were determined by ELISA as described before. All animal studies were performed in compliance with the regulations of the University of Florida Animal Care Services in accordance with the guidelines set by the Association for Assessment and Accreditation of Laboratory Animal Care. The vMyxM013-KO virus was markedly attenuated in infected rabbits due to enhanced inflammatory responses at sites of infection, suggesting that M013 was involved in regulating host inflammatory responses to the virus infection [33]. Since many MYXV-encoded modulators are able to recognize and inhibit host proteins from a wide variety of species, particularly if the host protein targets are well conserved, this phenomenon was further investigated in cultured human monocytic THP-1 cells, where the secretion of the inflammasome-controlled cytokines IL-1β and IL-18 was previously shown to be inhibited by M013 protein. The PYD domain of M013 was critical for the protein/protein interaction with host ASC-1 protein, the common adapter protein of inflammasome complexes, and prevented caspase-1-mediated activation and secretion of these cytokines in response to MYXV infection [33]. It has been reported that some cellular PYD-containing regulatory proteins can co-modulate the function of both inflammasomes and NF-κB [44], [45]. To determine whether M013 might also regulate the secretion of other pro-inflammatory molecules under NF-κB control, human THP-1 cells were infected with either wild-type (WT) MYXV or vMyxM013-KO virus and the cell supernatants were harvested at various time post infection to measure various indicator cytokines and chemokines by ELISA. The THP-1 cells were first treated with PMA to promote monocytic differentiation and then infected with the test viruses. When checked for virus replication in THP-1 cells both the test viruses were substantially compromised in production of progeny virions (data not shown). As a positive control, THP-1 cells were treated with LPS, a known TLR-mediated activator of NF-κB and inducer of many pro-inflammatory cytokines. In order to monitor early inflammatory cytokine responses, the supernatants were collected starting from 15 minutes after addition of virus to the cells. As shown in Fig. 1A, infection of THP-1 cells with vMyxM013-KO virus significantly increased IL-1β secretion within one hour, compared to control MYXV, indicating that the vMyxM013-KO virus was uniquely unable to block the inflammasome/caspase 1-mediated response to the virus infection. When tested for induction of an NF-κB controlled cytokine such as TNF, this cytokine level was also significantly increased within one hour, specifically in case of vMyxM013-KO virus infection but not WT-MYXV (Fig. 1B). The secretion levels of both cytokines quickly rose and co-ordinantly reached a maximum within 2 to 4 hours of infection. LPS treatment of uninfected THP-1 cells also co-induced the secretion of IL-1β and TNF to comparable levels (data not shown), suggesting that infection with vMyxM013-KO virus was as comparably robust at inducing these two cytokines as the triggering of TLR4 activation with LPS. To assess whether infection with the M013L-minus variant of MYXV induced broader effects on other pro-inflammatory cytokines, the infected THP-1 cell supernatants were also tested for IL-6, IL-12 and the CC chemokine MCP-1 by ELISA. Compared to TNF and IL-1β, the secretion of IL-6 and MCP-1 by THP-1 cells infected with vMyxM013-KO virus was first detected much later, and these were only measurable after 12 hours post infection (Fig. 1C and D). The second wave of vMyxM013-KO virus-mediated cytokine induction might be the effect of other cytokines like TNF or IL-1β. However, the control WT-MYXV that expresses M013 at this time point induced only minimal secretion of any of these cytokines. In case of LPS treatment of THP-1 cells, the induction of IL-6 and MCP-1 also began after a delay, but was first detectable after 8 hours of treatment (data not shown). Also, we did not observe any detectable induction of IL-12 and type I interferon (IFNα and IFNβ) secretion following infection of THP-1 cells with either MYXV or vMyxM013-KO viruses (data not shown). Analysis of the cytokine mRNA levels using real-time PCR revealed a similar induction of cytokine mRNA levels following infection of THP-1 cells with vMyxM013-KO (data not shown). These results indicate that, in addition to promoting inflammasome activation, infection of THP-1 cells with the M013-knockout MYXV also specifically triggered the induction of various early pro-inflammatory cytokines and chemokines that are regulated by NF-κB. We next examined whether vMyxM013-KO virus can also induce comparable pro-inflammatory cytokine production in vivo, using THP-1 derived monocytic tumors established by engraftment into immunodeficient NOD/SCID mice. MYXV is currently being developed as an oncolytic therapeutic treatment for human cancer, and the WT virus replicates similarly within xenografted human tumor tissues in situ as it does within normal rabbit tissues [46]. Thus, to model MYXV behavior within susceptible tissues for which the appropriate cytokine reagents are currently available, engrafted THP-1 cells were used to generate monocytic tumors in NOD-SCID mice, which were first detected at 6 days after subcutaneous injection on the left hind leg. At this point, samples of either MYXV or vMyxM013-KO virus were injected intra-tumorally on day 7 and tissues were collected after 48 h post-virus injection for cytokine induction measurements. Significantly, we observed that tumors infected with vMyxM013-KO virus, but not WT-MYXV or mock PBS-injected control tumors, selectively induced significant levels of human TNF (Fig. 1E). However, we were not able to detect newly induced human IL-1β from any of these tissue samples (data not shown). Due to the localized nature of the THP-1 tumors and the virus injection route, we also could not detect any selectively induced human cytokines in the serum samples from any of the mice at this time point. This result suggests that vMyxM013-KO virus, but not WT-MYXV, can specifically induce NF-κB mediated TNF directly within susceptible tissue sites in vivo as well as in vitro. The transcription factor NF-κB has been shown to play an essential early role in virus-induced expression of diverse pro-inflammatory cytokines, interferons and chemokines in infected cells. We next investigated how the innate NF-κB pathway was activated in response to vMyxM013-KO virus infection, which then leads to the secretion of pro-inflammatory cytokines under NF-κB control. Activation of NF-κB is usually regulated by the inhibitor IκBα, which is itself often the immediate target of viral subversion. In response to upstream stimulatory signals such as virus infection, IκBα becomes phosphorylated by cellular kinases and then is degraded by the proteasome following ubiquitination. This loss of IκBα repression then triggers the translocation of NF-κB to the nucleus to initiate target gene transcription. To determine whether increased NF-κB-driven gene transcription occurs specifically in response to vMyxM013-KO virus infection, and whether this resulted from loss of IκBα, we investigated the degradation of IκBα in virus-infected THP-1 cells by Western blot analysis. THP-1 cells were infected with WT-MYXV or vMyxM013-KO virus, harvested at different time points and the level of IκBα was assessed. By one hour post infection, we observed a transient reduction in IκBα uniquely in the cells infected with vMyxM013-KO virus, but the IκBα levels were fairly similar at later time points following either WT-MYXV or vMyxM013-KO virus infection (Fig. 2A). We then investigated the first hour post-infection in greater detail, and observed that within the first 15 minutes of vMyxM013-KO virus infection of the THP-1 cells, the cellular pool of IκBα was almost completely degraded (Fig. 2B, 2nd row). However, in WT-MYXV -infected cells the level of IκBα remained essentially the same as mock infected THP-1 cells (Fig. 2B, 1st and 4th rows). Following this rapid degradation, the level of IκBα in the vMyxM013-KO virus infected cells then recovered and returned to normal levels within 2 hr post infection. As a comparison, we also measured the kinetics of LPS- induced degradation of IκBα in THP-1 cells (Fig. 2B, 3rd row). In LPS-treated cells, the degradation of IκBα was first detected later, beginning at 30 min and the protein level continued to decrease up to 2 h post treatment, which is dramatically different from the rapid degradation induced following infection with vMyxM013-KO virus. As the degradation of IκBα is mediated by the kinetics of its phosphorylation, we next investigated the phosphorylation state of IκBα in virus-infected THP-1 cells by Western blot analysis. The induction of phosphor-IκBα (p-IκBα) was readily detected at 0. 5 and 2 hours after infection of THP-1 cells with vMyxM013-KO virus but not WT-MYXV (Fig. 2C). This suggests that degradation of IκBα in vMyxM013-KO-infected THP-1 cells is mediated by its phosphorylation, as it is by other nonviral inducers of NF-κB activation. It is evident that vMyxM013-KO virus can induce early pro-inflammatory responses in THP-1 cells by rapidly inducing the phosphorylation and degradation of IκBα, which should trigger the nuclear translocation of NF-κB and activation of target gene transcription. We next investigated whether the multi-component IKK complex becomes activated in THP-1 cells in response to vMyxM013-KO infection and subsequently degraded the IκBα. The virus-infected THP-1 cells were harvested at different time points, and examined for IKKα/β or p-IKKα/β by Western blot analysis. In response to vMyxM013-KO virus infection, we observed a rapid elevation in the phosphorylation levels of IKKα and IKKβ, as compared to the WT-MYXV infections (Fig. 3A). The level of total IKKα/β protein was very similar when comparing the two viruses (Fig. 3B). However, the antibody used to detect the phosphorylated IKK also reacted with a non-specific protein in all the test samples. Based on this observation we conclude that early phosphorylation of IKK by vMyxM013-KO virus mediates the activation of NF-κB pathway. We next investigated the mechanism of NF-κB-mediated activation of pro-inflammatory cytokine production in the virus-infected THP-1 cells. Among the NF-κB family members, NF-κB1 (p105 and p50) and RelA (p65) have so far been linked with virus-induced expression of pro-inflammatory target gene expression. In unstimulated cells, NF-κB1-p105 remains in the cytoplasm as a dimer with RelA/p65, c-Rel or p50. In response to activating stimuli, p105 becomes phosphorylated by kinases and is degraded to release the associated Rel or p50 subunit [47]. The p50 and p65 heterodimeric complex then translocates to the nucleus after degradation of IκBα and initiates transcription of target pro-inflammatory genes. To deduce how M013 might perturb this activation pathway, we first examined the cleavage of p105 in virus-infected THP-1 cells. The cells were infected with either WT-MYXV or vMyxM013-KO and harvested at 30 min and 2 hr for detection of protein by Western blot analysis. Infection of THP-1 cells with vMyxM013-KO virus induced the rapid degradation of NF-κB1-p105 and the NF-κB1-p50 form can be detected (Fig. 4A). However, in the WT-MYXV infected THP-1 cells, both p105 and p50 can be readily detected. The level of the p65 subunit, however, in case of both virus infections remained essentially unchanged (Fig. 4B). Thus, the expression of the M013 protein protects the p105 form of NF-κB1 from the degradation associated with the canonical NF-κB activation pathway. In order to determine whether the RelA/p65 complex is correctly translocated to the nucleus, we prepared cytoplasmic and nuclear fractions of mock and virus-infected THP-1 cells (30 min after virus infection) and Western blot analysis was performed. In response to vMyxM013-KO virus infection, the majority of the p65 protein had migrated to the nucleus during early time points of infection (Fig. 4C, right panel). The level of p50 also increased in the nuclear fraction of vMyxM013-KO virus infection compared to the mock and WT-MYXV infection (Fig. 4C, left panel). As expected, no NF-κB1/p105 was detected in the nucleus (not shown). As loading controls, levels of histone H1 (nuclear extracts; Fig. 4C left middle panel) and actin (cytoplasmic extracts; Fig. 4C right bottom panel) were examined. These results indicate that, once THP-1 cells become activated by infection with vMyxM013-KO virus, RelA/p65 rapidly translocates to the nucleus, with kinetics similar to that triggered by many other nonviral inducers. In THP-1 cells, vMyxM013-KO virus infection can co-ordinantly induce inflammasome-mediated secretion of IL-1β, and IL-18, as well as NF-κB -mediated secretion of various other pro-inflammatory cytokines and chemokines. In order to determine whether these two pathways were activated independently or through common mechanism, we used a specific inhibitor of inflammasome activation to differentiate the two pathways. Inflammasome-mediated secretion of IL-1β depends on activation of caspase-1, which cleaves the pro-IL1β to produce active IL-1β for secretion. Treatment of THP-1 cells with caspase-1 inhibitor, zVAD-fmk, alone caused no induction of IL-1β but when combined with the infection with vMyxM013-KO virus resulted in the dramatic inhibition of IL-1β secretion at early time points, starting in the first hour post-infection (Fig. 5A). When the same samples were also tested for TNF secretion, we observed that inhibition of caspase-1 with this inhibitor had no effect on vMyxM013-KO virus-mediated induction of TNF (Fig. 5B). This suggests that vMyxM013-KO virus-mediated early induction of IL-1β secretion depends on inflammasome activation of caspase 1, but in case of TNF, the induction is mediated by NF-κB and is independent of caspase-1. In order to test which cellular signaling pathway (s) might be involved in the activation of NF-κB in response to vMyxM013-KO virus infection, we used the kinase inhibitors U0126, an ERK1/2 inhibitor and LY294002, a PI3 kinase inhibitor. THP-1 cells were pretreated with these inhibitors for one hour and infected with WT-MYXV or vMyxM013-KO viruses. The supernatants were collected at different time points after infection and tested for secreted IL-1β and TNF by ELISA. Both the kinase inhibitors had no effect on viral gene expression in this cell line when compared to the untreated cells (data not shown). Treatment of THP-1 cells with U0126 totally inhibited the vMyxM013-KO virus-induced secretion of TNF, starting from early times of infection (Fig. 6A). However, when the samples were tested for IL-1β secretion, induction of this cytokine was also inhibited and after two hours the cytokine level remains constant. This suggests that U0126 did not totally abrogate the inflammasome-mediated secretion of processed IL-1β from pre-existing stores of precursor pro-IL-1β, but did effectively inhibit the NF-κB-mediated production of precursor pro-IL-1β, which subsequently is processed by inflammasome. When tested the uninfected and infected THP-1 cells, U0126 totally inhibited the phosphorylation of ERK1/2 and IKKα/β (data not shown). The PI3 kinase inhibitor LY294002 did not inhibit the induction of either TNF or IL-1β in response to vMyxM013-KO virus infection, indicating that this pathway is not critical for induction of either cytokine in response to vMyxM013-KO virus infection (Fig. 6B). This suggests that vMyxM013-KO virus mediated activation of NF-κB is dependent on ERK1/2 kinase signaling but not the PI3 kinase pathway. To investigate the direct effects of MYXV M013 protein on NF-κB activation, the expression of the Firefly luciferase reporter gene linked to a NF-κB dependent promoter was used to quantify induction of the NF-κB signaling pathway. As an internal control, the expression of a co-transfected Renilla luciferase gene driven by a constitutively active promoter (thymidine kinase, tk promoter) was also examined. HeLa cells were co-transfected with the plasmids expressing the reporter genes and M013L (under CMV promoter) and then the cells were stimulated with TNF for induction of NF-κB-dependent reporter gene expression. The Firefly/Renilla Luciferase expression ratio was increased significantly (about 12 fold) upon stimulation with TNF but this level was reduced (about 50%) in the presence of transfected M013L (Fig. 7A). Even with the caveats of incomplete transfection efficiencies, these results demonstrated that PYD-containing M013 protein specifically inhibits NF-κB-regulated gene expression even in the absence of any other MYXV gene products. This result is in contrast to another report claiming that a related PYD-containing poxviral protein from Shope fibroma virus, called S013, induced the activation of NF-κB activity in a transient transfection assay [48]. We next tested whether the expressed M013 protein inhibited the nuclear translocation of p65/RelA after treatment with TNF. HeLa cells were mock transfected or transiently transfected with an expression plasmid that expressed Myc-His tagged M013 protein. Immunostaining of cells using anti-p65 antibody identified the major location of this protein in the cytosol (Fig. 7B, panels a–d), however, when the cells were induced with TNF, the p65 protein efficiently translocated to the nucleus (Fig. 7B, panel e–h). In the absence of TNF stimulation, Myc-tagged M013 protein was located mostly in the cytoplasm and formed characteristic punctate bodies in the expressing cells (Fig. 7B, panels i–l). The cells which expressed M013 protein uniquely exhibited the property that TNF-induced nuclear translocation of p65/RelA was blocked (Fig. 7B, panels m–p), suggesting that the direct inhibitory effect of M013 protein on NF-κB relocation to the nucleus. The ability of M013 to inhibit the TNF-induced nuclear translocation of p65/RelA was further confirmed using a construct lacking the PYD (Fig. 7C, panels a–h). These results suggest that M013 protein in the absence of other viral proteins can inhibit NF-κB-mediated gene transcription by blocking the translocation of NF-κB p65 to the nucleus. To better understand how M013 protein prevents translocation of RelA/p65 to the nucleus, we examined the physical interactions between various NF-κB members and M013 protein using in vivo and in vitro protein-protein interaction methods. In vivo interaction was studied in transfected cells using the AlphaScreen method [41], [42]. Plasmids expressing GST-tagged M013 protein and Myc/His tagged host proteins under a T7 promoter were co-transfected in T7 polymerase-expressing BSRT7/5 cells and tested using specific AlphaScreen beads. Interaction of M013 protein with human NF-κB1 p105 protein was detected by emission flourescence whereas a control poxviral protein of a similar size (MYXV M063) or vector control (Fig. 8A) did not. Using the same method, however, we did not detect any interaction between M013 and RelA/p65. The interaction of GST-tagged M013 and Myc/His tagged NF-κB1 p105 was also confirmed by GST pulldown assay after co-expression of the corresponding plasmids in the BSRT7/5 cells (data not shown). The protein/protein interaction between M013 and NF-κB1-p105 following expression in reticulocyte lysates was also confirmed using an ELISA method. GST-tagged M013 protein and HA-tagged host proteins were individually (as control) or co-expressed using the in vitro TNT expression system. The GST-tagged M013 viral protein was bound to the ELISA plate coated with anti-GST antibodies and the interacting HA-tagged host protein was detected using HRP conjugated anti-HA antibody. Using this method, we could detect the interaction of M013 with NF-κB1-p105, while RelA produced essentially no signal above background (Fig. 8B). In addition, the pyrin domain was specifically shown to be required for this interaction because when this domain was deleted from M013, no binding with NF-kB1 was detected (Fig. 8C). This confirmed that M013 protein can associate directly with NF-κB1-p105 in the absence of any other viral proteins. The MYXV-encoded M013 protein was first characterized as a functional member of the cellular PYRIN domain (PYD) -containing protein family that regulates the function of host inflammasome complex [33]. The PYD domain is one of four subfamilies of the death-domain fold superfamily of evolutionary conserved protein-protein interaction domain containing proteins [34], [49]. PYD-containing proteins mediate homotypic and heterotypic interactions among family members and even proteins with different domains, from other families. These diverse interactions allow PYD-containing proteins to be associated with cellular processes like apoptosis, immunity, inflammation, differentiation and proliferation [34]. In humans, at least 23 PYD containing proteins have been identified to date that fall in four groups: 1- receptor proteins or pathogen recognition receptors (PRRs), that recognize pathogen associated molecular patterns (PAMPs) known as PAN, NALP (NACHT, LRR and PYD containing proteins), PYPAF, NOD (nucleotide-binding oligomerization domain), Caterpiller and NLR (NOD-like receptors) proteins [35], [50], [51]; 2- adapter proteins like ASC-1 [52], [53]; 3-regulatory proteins, for example cellular PYD-only proteins (cPOPs) cPOP1 and cPOP2 [44], [45], [54] and, 4- IFN-inducible proteins known as 200-amino-acid repeat (HIN-200) gene family [55]. The PYD-containing host proteins are predominantly expressed in leukocytes and tissue sentinel cells and are integrated with pathogen sensors that trigger the innate immune responses. The common adapter protein ASC-1 contains bifunctional domains PYD and CARD (C-terminal caspase-recruitment domain) which interacts with the PYD-containing NALP proteins and CARD-containing caspase-1, respectively, to activate pro-caspase-1 in the NALP3-inflammasome complex [53]. ASC-1 also regulates apoptosis and NF-κB signaling responses by interaction with multiple other proteins in those pathways [52]. ASC-1, in turn, is regulated by cellular and viral regulatory proteins, collectively called POPs [33], [44], [45], [48], [54], [56]. The cellular regulatory proteins, cPOP1 and cPOP2, control the function of adapter protein ASC-1 by mediating PYD-PYD interactions. In addition to ASC-1, cPOPs also can inhibit NF-κB activation [44], [57]. M013 protein of MYXV is an example of a functional viral POP (vPOP). M013 interacts directly with ASC-1 protein and modulates caspase-1 activation and IL-1β/IL-18 processing in cultured monocyte-derived cells, such as THP-1 cells, and in infected rabbit tissues [33]. A closely related vPOP protein from Shope Fibroma virus was also shown to block the activation of pro-caspase-1 in transient transfection assay and subsequent processing of pro-IL-1β [48]. Furthermore, infection of THP-1 cells with vMyxM013-KO virus induced the activation of caspase-1 and processing and secretion of pro-inflammatory cytokines IL-1β and IL-18 [33]. Here, we demonstrate that vMyxM013-KO infection of THP-1 cells also induces other cytokines and chemokines (TNF, IL-6, and MCP-1) that are regulated by the NF-κB pathway, in addition to the cytokines controlled by the caspase 1/inflammasome complex. Both cPOP1 and cPOP2 have been shown to also exert inhibitory properties for NF-κB signaling [44], [45]. On the other hand, PYD-containing pathogen receptors (PRRs) have diverse effects on the function of NF-κB. Some of these are reported to have NF-κB-mediated pro-inflammatory roles such as PYPAF1/NALP3 [58], whereas others such as PAN2/PYPAF4 and PAN1/PYPAF2 inhibited NF-κB [59]. The vMyxM013-KO virus-mediated activation of NF-κB reported here suggests that the M013 vPOP protein has a co-inhibitory role against both the inflammasome complex and cellular NF-κB signaling, and binds distinct protein targets from each pathway, namely ASC-1 and NF-κB1. One of the novel findings in this study is the rapid induction of NF-κB-mediated inflammatory responses against vMyxM013-KO virus, which is repressed by the WT-MYXV. Infection of THP-1 cells with vMyxM013-KO virus induced the formation of NF-κB p50/p65 complex, which migrates to the nucleus to induce the transcription of pro-inflammatory mediators. Indeed, transient expression of M013 protein alone inhibited the TNF-induced NF-κB activation by blocking the translocation of p65 to the nucleus. The host cPOP2 also blocks the NF-κB activation in the same fashion [45]. Although in case of POP2, the mechanism of this inhibition is not known, POP1 interacts with IKKα, which likely affects the downstream phosphorylation of IκBα [44]. M013 protein, on the other hand, interacts directly with NF-κB1/p105 and we postulate that this complex prevents the subsequent nuclear translocation of p50/p65 complex at a point downstream of IKK activation. For example, the interaction between M013 and NF-κB1 might interfere with the degradation of NF-κB1/p105, and therefore prevent the release of active p50 and the formation of the active p65/p50 heterodimer that subsequently translocates to the nucleus. Thus, the vMyx-M013KO-induced phosphorylation of IKK and degradation of IκBα raises the possibility that deletion of MYXV-M013 from MYXV triggers change (s) in the cellular sensing and/or responses to virus infection such that it hinders the ability of the other viral gene products to prevent the degradation of IκBα. Very recently, it has been shown that a different family of ankyrin-repeat containing viral proteins, encoded by orthpoxviruses also interacts with NF-κB1/p105 and inhibits the activation of the NF-κB pathway [60]. This suggests that the NF-κB1/p105 is a more common target for viral modulation than previously appreciated. It is not yet determined which protein domain of NF-κB1 is involved in interaction with the M013 vPOP. Since NF-κB1 itself lacks any PYD domain, the interaction is predicted to be heterotypic, which has been previously observed in case of other PYD-containing proteins [52]. It is also possible that the PYD of M013 may interact with multiple signaling molecules in the NF-κB pathway, based on the reported ability of PYD to enter into heteromeric complexes and as demonstrated for ASC-1 [52]. It is also probable that MYXV encodes multiple proteins to inhibit the NF-κB pathway at different stages, as reported in the case of VACV and MOCV [22], [23], [24], [25], [27]. The MYXV-encoded ankyrin repeats (ANKs) containing protein M150 has previously been proposed to be regulator of NF-κB function [61]. Also called myxoma nuclear factor (MNF), M150 protein of MYXV migrated to the nucleus in response to TNF treatment and co-localized with NF-κB p65 [61] but to date no interacting host protein has yet been reported for MNF/M150. From our observations we propose that the single PYD-containing vPOP protein M013 is able to simultaneously associate with components from both the inflammasome complex (i. e. ASC-1) and NF-κB signaling pathway (i. e. NF-κB1) to efficiently downregulate both these arms of the innate cellular pro-inflammatory responses to virus infection. In the cellular context, PYD-containing proteins are very important in diverse processes, as mutations in the part of the genome that encodes certain cellular PYD family members are connected with hereditary diseases such as familial Mediterranean fever (FMF), familial cold autoinflammatory syndrome (FCAS), Muckle-Wells syndrome (MWS), and Neonatal-onset multisystem inflammatory disease (NOMID) [62], [63]. For example, the vast majority of FMF-associated mutations are located in the C-terminal B30. 2 (SPRY) domain of the MEFV (Mediterranean Fever) gene-encoded protein product pyrin. Any of these mutations in fact disrupt the regulatory role of pyrin in caspase-1 activation and IL-1β production [64], [65]. In case of MYXV, the PYD domain of the viral protein is important for the pathogenicity of the virus in vivo and the lack of M013 expression induced a rapid early inflammatory cytokine response (featuring caspase-1-dependent induction of IL-1β and IL-18) in the infected host lesions [33]. The successful co-inhibition of both inflammasome activation and NF-κB -mediated inflammatory signaling responses by a single viral immunomodulator vPOP, M013, represents a unique bi-functional strategy deployed by poxviruses to dampen multiple early innate immune responses to the infecting virus. Next, it is critical to better understand the mediator used by the host cells to sense the infecting vMyxM013-KO virus to rapidly induce both the inflammasome and NF-κB pro-inflammatory responses and how the M013 vPOP protein intercepts these signals.

Myxoma virus (MYXV), a rabbit-specific poxvirus pathogen, encodes diverse immunomodulatory proteins that can collectively overcome essentially all of the host immune defenses. MYXV-encoded protein M013, a member of the cellular PYRIN domain-containing superfamily of proteins, was previously shown to be important for virus virulence by blocking inflammasome regulated pro-inflammatory cytokine secretion. Here, we report that, in addition to targeting the ASC-1 protein of the host cell inflammasome complex, M013 also blocks activation of NF-κB signaling pathway by interacting with NF-κB1 and preventing nuclear translocation of the transcription factor RelA/p65. MYXV virus lacking a functional M013L gene (vMyxM013-KO) induced the early activation of NF-κB signaling pathway in human monocytic cells, causing the secretion of antiviral pro-inflammatory cytokines in vitro and in vivo. These results demonstrate that MYXV protein M013 is a multifunctional immunosubversive protein that co-regulates both the inflammasome complex and the NF-κB signaling pathway simultaneously.

Abstract Introduction Materials and Methods Results Discussion

immunology/immunomodulation virology virology/immune evasion virology/effects of virus infection on host gene expression

2009

Co-Regulation of NF-κB and Inflammasome-Mediated Inflammatory Responses by Myxoma Virus Pyrin Domain-Containing Protein M013

The Hippo pathway regulates organ size, stem cell proliferation and tumorigenesis in adult organs. Whether the Hippo pathway influences establishment of stem cell niche size to accommodate changes in organ size, however, has received little attention. Here, we ask whether Hippo signaling influences the number of stem cell niches that are established during development of the Drosophila larval ovary, and whether it interacts with the same or different effector signaling pathways in different cell types. We demonstrate that canonical Hippo signaling regulates autonomous proliferation of the soma, while a novel hippo-independent activity of Yorkie regulates autonomous proliferation of the germ line. Moreover, we demonstrate that Hippo signaling mediates non-autonomous proliferation signals between germ cells and somatic cells, and contributes to maintaining the correct proportion of these niche precursors. Finally, we show that the Hippo pathway interacts with different growth pathways in distinct somatic cell types, and interacts with EGFR and JAK/STAT pathways to regulate non-autonomous proliferation of germ cells. We thus provide evidence for novel roles of the Hippo pathway in establishing the precise balance of soma and germ line, the appropriate number of stem cell niches, and ultimately regulating adult female reproductive capacity. The Hippo pathway is a tissue-intrinsic regulator of organ size, and is also implicated in stem cell maintenance and cancer [1,2, 3]. An outstanding question in the field is whether the Hippo pathway regulates proliferation of cells comprising stem cell niches during development in order to ensure that adult organs have an appropriate number of stem cells and stem cell niches [4]. The adult Drosophila ovary is an extensively studied stem cell niche system. In this organ, specialized somatic cells regulate the proliferation and differentiation of germ line stem cells (GSCs) throughout adult reproductive life [reviewed in 5]. The fact that GSCsare first established in larval stages raises the question of how the correct numbers of GSCs, and their associated somatic niche cells, are achieved during larval development. To date, only the Ecdysone, Insulin and EGFR pathways have been implicated in this process [6,7, 8]. Here, we investigate the role of the Hippo pathway in regulating proliferation of somatic cells and GSC niche precursors to establish correct number of GSC niches. Our current understanding of the Hippo pathway is focused on the core kinase cascade and upstream regulatory members. The Hippo pathway’s upstream regulation is mediated by a growth signal transducer complex comprising Kibra, Expanded and Merlin [9,10,11,12] and the planar cell polarity regulators Fat [13,14,15] and Crumbs [16,17]. Regulation of Hippo signaling further upstream of these factors appears to be cell type-specific [18]. When the core kinase cascade is active, the kinase Hippo (Hpo) phosphorylates the kinase Warts (Wts) [19,20]. Phosphorylated Wts then phosphorylates the transcriptional coactivator Yorkie (Yki), which sequesters Yki within the cytoplasm [21]. In the absence of Hpo kinase activity, unphosphorylated Yki can enter the nucleus and upregulate proliferation-inducing genes [21,22,23,24]. The Hippo pathway affects proliferation cell-autonomously in the eye and wing imaginal discs, glia, and adult ovarian follicle cells in Drosophila [18,19,20,25,26], as well as in liver, intestine, heart, brain, breast and ovarian cells in mammals [27,28,29,30,31,32]. Hippo pathway is often improperly regulated in cancers of these tissues, which display high levels and ectopic activation of the human ortholog of Yki, YAP [27,28,33,34]. Upregulation of YAP is also commonly observed in a variety of mammalian stem cell niches, where YAP can be regulated in a Hippo-independent way to regulate stem cell function [reviewed in 4]. Interestingly, germ line clones lacking Hippo pathway member function do not cause germ cell tumors in the adult Drosophila ovary, which has led to the hypothesis that Hippo signaling functions only in somatic cells but not in the germ line [35,36]. More recently, it has become clear that the Hippo pathway can regulate proliferation non-autonomously: Hippo signaling regulates secretion of JAK/STAT and EGFR ligands in Drosophila intestinal stem cells [37,38,39], and of EGFR ligands in breast cancer cell lines [31], and the resulting changes in ligand levels affect the proliferation of surrounding cells non-autonomously. How autonomous and non-autonomous effects of the Hippo pathway coordinate differentiation and proliferation of multiple cell types has nonetheless been poorly investigated. Moreover, most studies address the Hippo pathway’s role in adult stem cell function, but whether Hippo signaling also plays a role in the early establishment of stem cell niches during development remains unknown. Here we use the Drosophila larval ovary as a model to address both of these issues. Adult ovaries comprise egg-producing structures called ovarioles, each of which houses a single GSC niche. The GSC niche is located at the anterior tip of each ovariole, and produces new oocytes throughout adult life. The niche cells include both GSC and differentiated somatic cells called cap cells [40]. Each GSC niche lies at the posterior end of a stack of seven or eight somatic cells termed terminal filaments (TFs). Somatic stem cells located close to the GSCs serve as a source of follicle cells that enclose each developing egg chamber during oogenesis [5]. All of these cell types originate during larval development, when the appropriate number of stem cells and their niches must be established. The larval ovary thus serves as a compelling model to address issues of homeostasis and stem cell niche development. TFs serve as beginning points for ovariole formation and thus establish the number of GSC niches [41]. TFs form during third instar larval (L3) development by the intercalation of terminal filament cells (TFCs) into stacks (TFs) (Fig. 1A; [41]). TFCs proliferate prior to entering a TF, and cease proliferation once incorporated into a TF [42]. The morphogenesis and proliferation of TFCs during the third larval instar (L3) is regulated by Ecdysone and Insulin signaling, and by the BTB/POZ factor bric-à-brac (bab) [6,8, 41,43,44,45]. Intermingled cells (ICs) arise from somatic cells that are in close contact with the germ cells (GCs) during L2, and proliferate throughout larval development [46] (Fig. 1A). ICs regulate GC proliferation and differentiation and are thought to give rise to escort cells in the adult niche [6,8, 47]. Both Insulin and EGFR signaling promote the proliferation of ICs [6,48]. Finally, larval GCs give rise to GSCs and early differentiating oocytes. GCs proliferate during development and do not differentiate until mid-L3, when the GSCs are specified in niches that form posterior to the TFs [6,8, 49], and the remaining GCs begin to differentiate as oocytes. GCs secrete Spitz, an EGFR ligand, and promote proliferation of ICs [49]. In addition, activation of Insulin and Ecdysone signaling in ICs regulates timing of early GC differentiation and cyst formation [6,8], though the identity of the IC-to-GC signal is unknown. ICs can non-autonomously regulate the proliferation of GCs both positively and negatively through Insulin and EGFR signaling respectively [6,7, 49]. We previously showed that hpo and wts regulate TFC number in a cell-autonomous manner [50]. Here we demonstrate a role for canonical Hippo pathway activity in regulating both TFCs and ICs. We also provide evidence for three novel roles of Hippo pathway members in ovarian development: First, in contrast to a previous report suggesting that yki did not play a role in determining GC number [35], we show that non-canonical, hpo-independent yki activity regulates proliferation of the germ line. Second, we show that Hippo signaling regulates homeostatic growth of germ cells and somatic cells through the JAK/STAT and EGFR pathways. Third, we show that the Hippo pathway interacts with the JAK/STAT pathway to regulate TFC number, and with both the EGFR and JAK/STAT pathways to regulate IC number autonomously and GC number non-autonomously. These data elucidate how Hippo pathway-mediated control of ovarian development establishes an organ-appropriate number of stem cell niches, and thus ultimately influences adult reproductive capacity. To determine whether Hippo signaling regulates proliferation of the GSC niche precursor cells, we first examined the expression pattern of Hippo pathway members in the larval ovary. Throughout larval development, Hpo was expressed ubiquitously in the ovary (S1A–C Fig.), and Yki was expressed in all somatic cells of the ovary (S1F–H Fig.). However, the subcellular localization of Yki was dynamic during ovariole morphogenesis, and different in distinct somatic cell types. We observed nuclear Yki expression in newly differentiating TFCs (identified by Engrailed expression and elongated cellular morphology) (Fig. 1B–B’, arrowhead; S2I Fig.), while late stage TFs had very little detectable nuclear Yki (Fig. 1C–C’, arrowhead; S2I Fig.). Since Yki localization in the nucleus indicates low or absent Hippo pathway activity [21], these data suggest that Hpo signaling may promote TFC and TFC-progenitor proliferation before TF formation, and then suppress proliferation in TFCs that have entered TFs. This is consistent with previous reports of the somatic proliferative dynamics of the larval ovary [42,48]. We also assessed Yki activity by analyzing expression of the downstream target genes expanded (ex) [21], diap1 (also called thread) [21] and bantam [22]. ex-lacZ (Fig. 1F–G, S2J Fig.) and diap1-lacZ (S2A–B, K Fig.) were expressed in early TFCs, but ceased expression once TFCs were incorporated into a TF. The bantam-GFP sensor is a GFP construct containing bantam miRNA target sites, such that low or absent GFP expression indicates bantam expression and activity. The sensor was not expressed in early differentiating TFCs, but was expressed in TFCs within a TF (S2E–F, L Fig.). These data are consistent with the subcellular localization of Yki in TFCs. Yki activity reporters were also expressed in cap cells of the GSC niche, which are immediately posterior to TFs (Fig. 1G–G’, yellow arrowhead). In ICs, strong cytoplasmic and nuclear expression of Yki was observed throughout development (Fig. 1D–D’; S2I Fig.). Likewise, all Yki activity reporters examined were expressed in ICs (Fig. 1H–H’, S2C–C’, G–G’, J–L Figs.), consistent with continuous proliferation of these cells throughout larval development. The expression patterns described above, and our previous observation that knockdown of hpo or wts increased TFC number [50], suggested that the Hippo pathway regulates TFC proliferation. To further test this hypothesis, we manipulated activity of the core Hippo pathway members hpo, wts and yki in somatic cells using the bric-à-brac (bab) and traffic jam (tj) GAL4 drivers [51,52]. bab: GAL4 is strongly expressed in TFCs during L3 but only weakly in other somatic cell types [50,51]. tj: GAL4 is expressed primarily in somatic cells posterior to the TFs, including ICs, to a lesser extent in newly forming TF stacks during early and mid L3, and in posterior TFCs in late L3 (S3A–D Fig.) [53]. We note that the expression of Tj in intercalating TFCs is not detected with the Traffic-Jam antibody (S3E Fig.). Antibody staining against Hpo and Yki was used to confirm effectiveness of the RNAi-mediated knockdown under both GAL4 drivers (S1D–D’, I–I’ Fig. ; see Methods for further details of RNAi validation in these and subsequent experiments). Lowering Hippo pathway activity in somatic cells by expressing RNAi against hpo or wts under either GAL4 driver significantly increased TFC number (student’s t-test was used for this and all other comparisons: p<0. 05; Fig. 2A, S1 Table). We previously showed that TFC number correlates with TF number [50]. Accordingly, driving RNAi against either hpo or wts in somatic tissues significantly increased TF number (p<0. 05; Fig. 2B, S1 Table). Conversely, decreasing Yki activity in somatic cells by expressing yki RNAi under either driver significantly reduced both TFC number and TF number (p<0. 05; Fig. 2A–B, S1 Table). Somatic overexpression of hpo or yki under the bab: GAL4 driver resulted in larval lethality, likely due to the known pleiotropic expression of bab in multiple non-ovarian tissues [51]. However, tj: GAL4-driven overexpression of yki or hpo was viable. Using the tj: GAL4 driver, we found that somatic yki overexpression resulted in a significant increase in both TFC number (p<0. 05; Fig. 2A, S1 Table) and TF number (p<0. 01; Fig. 2B, S1 Table), while somatic hpo overexpression led to a significant reduction in both TFCs and TFs (p<0. 05; Fig. 2A–B, S1 Table). A null allele of the Hippo pathway effector expanded (ex1 [54]) and a gain of function allele of yorkie (ykiDB02 [33]) both led to a significant increase in TFC number (S4A Fig. , S2 Table), consistent with results obtained from RNAi treatments. As larval TF number corresponds to the number of GSC niches in the adult (ovariole number) [50], we asked if Hpo signaling might play a role in determining ovariole number. We quantified ovariole number in adults with RNAi-mediated knockdown of Hippo signaling pathway members hpo, wts, salvador (sav), Merlin (Mer), or ex in somatic cells. In all cases adult ovariole number was significantly increased (p<0. 01; Fig. 2C). Conversely, yki knockdown under tj: GAL4 significantly decreased ovariole number (p<0. 01; Fig. 2C). Adult females of all reported somatic knockdown and overexpression experiments were viable and did not have defects in adult ovarian structure. Adult females expressing hpo RNAi under the tj: GAL4 driver, which had significantly more ovarioles than controls (Fig. 2B, S1 Table), also laid significantly more eggs than controls (Fig. 2D). Conversely, adult females expressing yki RNAi under tj: GAL4 driver laid significantly fewer eggs than controls, and some appeared to be entirely sterile (Fig. 2D). This shows that by regulating somatic gonad cell number in the larval ovary, the Hippo pathway can influence adult female reproductive capacity. We next asked whether the Hippo pathway also influenced the proliferation of ICs, which do not contribute to TF formation but are in direct contact with germ cells and are thought to give rise to somatic stem cells or escort cells [6,8, 47]. Larval-pupal transition (LP) stage ICs were identified by antibody staining against Traffic Jam, which is specific to ICs at this stage of development (Lin et al. , 2003). Altering Hippo pathway activity in somatic cells had the same overall effects on IC number as on TFC number: knocking down hpo or wts or overexpressing Yki resulted in a significant increase in IC number (hpo or wts RNAi: p<0. 05 for bab: GAL4 and p<0. 01 for tj: GAL4; yki overexpression: p<0. 01 for both drivers; Fig. 3A, D–I, M–N S3 Table). Conversely, RNAi against yki or overexpression of hpo in the soma significantly reduced IC number (p<0. 01; Fig. 3A, J–L, S3 Table). As observed for TFC number, IC numbers in ex1 or ykiDB02 backgrounds were significantly increased (S4 Fig. , S2 Table), consistent with the RNAi data. Ovarian morphogenesis, including TF, ovariole and GSC niche formation, was normal in most cases (Fig. 3D–N). However, the 150% increase in IC number caused by yki overexpression correlated with failure of swarm cell migration in some ovaries (Fig. 3M–N, arrowhead; n = 2/10), suggesting that excessive proliferation of ICs above a certain threshold cannot be accommodated by the ovary, leading to disrupted ovariole morphogenesis. Because the tj: GAL4 and bab: GAL4 drivers are expressed in both ICs and TFCs (albeit at varying levels), we could not use these tools to determine whether ICs and TFCs influence each other’s proliferation non-autonomously. Thus, we tested the utility of ptc: GAL4, which is expressed in ICs (albeit at low levels) but not in TFCs (S5A–C Fig.), and hh: GAL4, which is expressed in a subset of TFCs during and after TF stacking but not in ICs (S5D–F Fig.), for this purpose. However, when we drove RNAi against hpo or wts using either driver, we did not observe any significant changes in IC, TFC or TF number compared to controls (S5G–H Fig. ; S5 Table). This is likely due to the facts that (1) hh: GAL4 expression in TFCs arises after TFC proliferation has essentially completed (S5D–F Fig.); and (2) ptc: GAL4 expression is extremely weak in ICs (S5A–C Fig.). We therefore cannot rule out the hypothesis that TFC or IC proliferation has a non-autonomous influence on the other of these two somatic cell types. Having observed apparently canonical Hippo pathway activity in the somatic gonad cells, we next asked whether this pathway operated similarly in germ cells, and found a number of significant differences. First, unlike the dynamic expression of Yki in somatic ovarian cells, we detected only extremely low levels of Yki in GCs throughout development (Figs. 1E, E’, S1F–H, S2I). The bantam-GFP sensor also suggested low or absent Yki activity in GCs (S2H–H’, L Fig.). However, we did observe expression of the expanded-lacZ (Figs. 1I–I’, S2J) and diap1-lacZ (S2D–D’, S2K Figs.) reporters in the GCs. We thus performed functional experiments to evaluate the roles of Yki and other Hpo pathway members in GCs. We disrupted Hippo pathway activity in GCs using the germ line-specific driver nos: GAL4 (S3E–H Fig.). In contrast to the overproliferation of somatic cell types observed in the experiments described above, driving RNAi against hpo or wts in the germ line did not significantly change GC number (Fig. 4A; S3 Table). However, driving yki RNAi in the germ line significantly reduced GC number (p<0. 01; Fig. 4A), and a second independent RNAi line [55,56] yielded similar results (p<0. 05; S3 Table). Conversely, overexpression of yki in GCs led to a significant increase in GCs (p<0. 01, Fig. 4A). Although hpo RNAi had no effect on GC number (Fig. 4A, S3 Table), hpo overexpression significantly decreased GC number (p<0. 01; Fig. 4A). Interestingly, we observed a non-autonomous increase in ICs in when yki was overexpressed in GCs, but not in the other experimental conditions (p<0. 05, S3 Table). To validate our findings from the hpo and yki RNAi experiments, we induced hpo [57] and yki [21] null mutant GC clones in L1 larvae and compared the clone sizes (number of cells per clone) of homozygous mutant clones and their homozygous wild type twin spot clones in late L3 ovaries. Consistent with our RNAi analysis, hpoBF33 clones were not significantly different in size from controls (Fig. 4F, H), but ykiB5 clones were significantly smaller than controls (p<0. 01; Fig. 4F, I). Taken together, both RNAi and clonal analysis data suggest that yki but not hpo is involved in regulating GC number. We therefore sought further evidence that yki activity in the germ line was independent of hpo. The FERM domain protein Expanded can bind to Yki independently of Hpo or Wts to sequester Yki to the cytoplasm of Drosophila eye imaginal disc and S2 cells [58], or alternatively can bind to and sequester Yki by forming a complex with Hpo and Wts in Drosophila wing imaginal discs [59]. To determine if one of these mechanisms might be operating in GCs, we knocked down ex alone, or hpo, wts and ex together in GCs. We did not observe significant changes in GC number under either condition (Fig. 4A; S3 Table), suggesting that these phosphorylation-independent mechanisms do not regulate Yki in GCs. Consistent with this hypothesis, we found that overexpression of ykiS168A, an allele of Yki that is impervious to Wts-mediated phosphorylation [60], also significantly increased GC number (Fig. 4A). To our knowledge, the only other identified hpo-independent mechanism of yki regulation in Drosophila is via the kinase Hipk, which phosphorylates Yki and induces nuclear translocation in Drosophila wing imaginal discs [61]. However, knocking down hipk in GCs also did not affect GC number (Fig. 4A). Finally, we asked if Yki might still operate together with the transcription factor Scalloped (Sd) /TEAD in germ cells, as has been shown in somatic cells of Drosophila and mammals [24,62,63]. Knocking down sd in GCs significantly reduced GC number (p<0. 01, Fig. 4A), suggesting that a Yki/Sd complex could play a role in GC proliferation. A reduction in GC number could be caused by altered GC proliferation, or by premature differentiation of GCs into oocytes, as has been observed for loss of function mutations in members of the Ecdysone and Insulin signaling pathways [6,8]. To ask if altered yki activity was causing changes in GC number by affecting the timing of oocyte differentiation, we assayed for fusome morphology, an indicator for early cyst cells, in ovaries expressing RNAi against yki or overexpressing yki in GCs (S6 Fig.). We observed no overt signs of early differentiation of PGCs and fusome morphology was similar to controls, suggesting that the reduction of GC number induced by yki knockdown in GCs is likely due to reduced GC proliferation. Given our finding that Hippo signaling pathway members regulate autonomous proliferation of both somatic and germ line cells, we asked if this pathway might also coordinate non-autonomous proliferation of both cell types. Such a mechanism might be expected to operate in order to adjust the numbers of one cell type in response to Hippo signaling-mediated changes in the other, which would ensure an appropriate number of operative stem cell niches [8]. To test this hypothesis, we analyzed GC number in conditions where Hippo pathway activity was altered in the somatic cells. Non-autonomous positive regulation of GC number by ICs has been documented, but only in ways that also affect GC differentiation [49]. Whether ICs can positively regulate GC proliferation without affecting their differentiation thus remains unknown [6,8]. We found that increasing somatic cell number by driving hpo or wts RNAi in the soma also significantly increased GC number (p<0. 01, p = 0. 06 respectively; Fig. 3B–I; S3 Table). Strikingly, GC number increased in precise proportion to the IC number increase, whether this increase was as little as 15% (bab: GAL4>>wtsRNAi; Figs. 3C, S7; S3, S6 Tables) or as much as 70% (tj: GAL4>>hpoRNAi; Figs. 3C, S7; S3, S6 Tables), resulting in a consistent ratio of ICs to GCs (Figs. 3C, S7; S6 Table). However, increasing IC number by 150% via somatic overexpression of yki prompted only a 10% increase in GC number (p<0. 05, Fig. 3B). In this condition, the GC: IC ratio was significantly lower than controls (Figs. 3C, S7; S6 Table), and GC: IC proportions were not maintained (Figs. 3C, S7). These results suggest that the Hippo pathway can maintain homeostatic growth of the larval ovary by regulating the number of GCs to accommodate changes of up to 70% in the number of ICs. However, further overproliferation of ICs cannot be matched by proportional GC proliferation. We then asked if somatic Hippo signaling could also non-autonomously compensate for decreases in IC number via a proportional reduction in GC number. We found that somatic yki RNAi significantly decreased IC number (p<0. 05), but did not significantly decrease GC number (p = 0. 29, Fig. 3B), thus disrupting the GC: IC ratio (Figs. 3C, S7). However, reducing IC number via hpo overexpression in the soma yielded a marginally significant decrease in GC number (Fig. 3A, B). These results suggest that the Hippo pathway’s role in non-autonomous proliferation of GCs is primarily operative in cases of somatic cell overproliferation, but that to accommodate significant decreases in IC number by reducing GC numbers, Hippo signaling is not always sufficient and additional mechanisms may be required. The latter may include insulin signaling [6]. Finally, we asked which signaling pathways Hippo signaling might interact with in the ovary to regulate proliferation. We also asked whether these pathways were the same or different in distinct somatic cell types (ICs and TFCs). First, we considered the EGFR pathway. The Hippo pathway interacts with the EGFR pathway to regulate non-autonomous control of proliferation in other organs [31,38,64,65,66]. Moreover, the EGFR pathway is known to regulate IC number and to non-autonomously regulate GC number [7,49]. We therefore asked whether the Hippo pathway interacted with EGFR signaling in the larval ovary. In wild type larval ovaries we observed, as previously reported [49], that pMAPK (a readout of EGFR activity) is expressed predominantly in ICs (Fig. 5A, white arrowhead) and in some TFCs (Fig. 5A, red arrowhead), but not in GCs (Fig. 5A, yellow arrowhead). When we knocked down hpo in the soma, we detected significantly increased pMAPK expression in the ovary (p<0. 01; Fig. 5B–C), most notably in ICs at mid L3 and late L3 stages (Fig. 5B, white arrowhead), and additionally in some TFCs (Fig. 5B, red arrowhead). These results suggest that in wild type ovaries Hippo pathway activity may limit EGFR activity in somatic cells. In order to assess the consequences of hpo/EGFR pathway interactions, we conducted double-RNAi knockdowns of hpo and either the EGFR receptor (egfr) or the EGFR ligand spitz (spi) in the soma using the tj: GAL4 driver. To validate the RNAi constructs, we expressed egfrRNAi or spiRNAi under tj: GAL4, and observed significant reduction in pMAPK levels in L3 ovaries (p<0. 05, S8A Fig.). In both hpo and egfr or spi double-RNAi knockdowns, TFC number was not significantly different from hpo single knockdowns (Fig. 5D; S4 Table). In addition, TFC number was not altered when we knocked down egfr or spi alone in the soma (S4 Table). This suggests that the Hippo pathway does not regulate TFC number via EGFR signaling, consistent with the limited pMAPK expression observed in TFCs (Fig. 5A, red arrowhead). In contrast, and consistent with the strong pMAPK expression in ICs (Fig. 5A, white arrowhead), tj: GAL4-mediated double knockdown of hpo and egfr partially rescued the hpo RNAi-induced overgrowth of ICs (p<0. 05; Fig. 5D; S4 Table). However, these ovaries still had 35% more ICs than wild type controls (p<0. 01; Fig. 5E). Double knockdown of hpo and spi yielded no significant difference in IC number compared to hpo single knockdowns (Fig. 5D). IC number was unaltered by knockdown of egfr or spi alone (S4 Table). In contrast to the TFCs, the Hippo pathway thus appears to interact with EGFR signaling to regulate IC number. Finally, we quantified GCs to test whether the EGFR-Hippo signaling interaction in ICs could non-autonomously regulate GCs. Double knockdown of hpo and egfr, which significantly reduced IC number relative to hpo RNAi alone (p<0. 05; Fig. 5D; S4 Table), also resulted in significantly fewer GCs (p<0. 05; Fig. 5D; S4 Table), completely rescuing the hpo RNAi-induced GC overproliferation (Fig. 5E; S4 Table). Double knockdown of hpo and spi did not alter IC number relative to hpo RNAi alone (p = 0. 24; Fig. 5D; S4 Table), but resulted in near-significant reduction of GCs (p = 0. 054), also yielding a complete rescue of the hpo RNAi-induced overproliferation (Fig. 5E; S4 Table). Because the degree of hpo RNAi rescue was greater in GCs than in ICs in the hpo/spi double knockdown, the homeostatic balance of these cell types was no longer maintained (S5F Fig.). As previously reported [7,49], we observed a significant increase in GC number when we knocked down egfr alone, but not spi alone, in the soma (S4 Table). Taken together, these results indicate that hpo interacts in the soma with egfr signaling, likely through an additional ligand along with spi, to regulate both IC number autonomously and GC number non-autonomously. Another characterized interacting partner of the Hippo pathway in various somatic tissues is the JAK/STAT pathway [37,38,39,67,68,69]. We therefore asked whether these two pathways also interact to regulate autonomous and/or homeostatic proliferation in the larval ovary. First, we used detection of Stat92E as a readout of JAK/STAT activity [70,71,72]. We observed strongest Stat92E expression in posterior somatic cells, including ICs, in wild type ovaries (Fig. 6A–A’). Knocking down hpo in the soma led to significantly higher Stat92E levels (p<0. 01; Fig. 6B–C), suggesting that, similar to its interaction with EGFR, Hippo pathway activity normally limits JAK/STAT pathway activity in the larval ovary. Next, we asked if RNAi against either the JAK/STAT receptor dome or the ligand unpaired (upd1) could rescue the effects of hpo RNAi in the soma. While there are three upd orthologues in Drosophila [73,74,75], we focused on upd1, as it is known to regulate GC proliferation in the testis [76] and thought to be a specific yki target in polar cells [77], which are derivatives of the somatic cells of the ovary. Expressing dome or upd1 RNAi under tj: GAL4 significantly reduced Stat92E levels in the larval ovary (p<0. 05 for dome, p = 0. 06 for upd1; S8B Fig.), confirming functionality of these RNAi lines. In TFCs, double knockdown of dome and hpo, but not of upd1 and hpo, completely suppressed the hpo single knockdown phenotype (p<0. 05; Fig. 6D; S4 Table). Knocking down dome alone in the soma significantly decreased TFC number (p<0. 05; S4 Table), supporting the hypothesis that JAK/STAT signals positively regulate TFC proliferation. These data suggest that Hippo signaling regulates TFC proliferation via interactions with the JAK/STAT pathway, and that a ligand other than upd1 mediates this interaction. We next counted IC and GC number to determine whether JAK/STAT-Hippo pathway interactions regulate ICs proliferation autonomously, and/or GC proliferation non-autonomously. Both dome/hpo or upd1/hpo double knockdowns partially rescued the overproliferation caused by knockdown of hpo alone (p<0. 01; Fig. 6D; S4 Table), but these ovaries still had significantly more ICs than wild type controls (p<0. 05; Fig. 6E). Both double knockdown conditions also completely rescued the non-autonomous increase in GC number caused by hpo RNAi (Fig. 6D; S4 Table). Similar to our experiments on the EGFR pathway, we observed abnormal IC: GC ratios in the hpo and dome RNAi single knockdowns (Figs. 6F, S7; S6 Table). In summary, Hippo signaling interacts with JAK/STAT signaling via upd1 to regulate IC: GC homeostasis. We have shown that canonical cell autonomous Hippo signaling regulates proliferation of two key somatic cell types, TFCs and ICs. Because TFCs form TFs, which are the beginning points of each GSC niche, the number and stacking of TFCs can ultimately influence adult ovariole number and thus reproductive capacity [50,78,79]. We previously showed that the differences in ovariole number between D. melanogaster and closely related Drosophila species results from changes in TFC number [48,50]. This suggests Hippo and JAK/STAT pathway members as novel potential targets of evolutionary change in ovariole number variation. Indeed, loci containing many of these genes have been previously identified in QTL analyses of genomic variation correlated with ovariole number variation [80]. Our study thus provides novel experimental validation of previous quantitative genetics approaches to understanding the genetic regulation of ovariole number. In TFCs, Hippo signaling regulates proliferation by interacting with dome but not upd1, suggesting that one or both of upd2 or upd3 act as ligands for JAK/STAT signaling in this context. Alternatively, a role for upd1 in TFC number regulation may have been obscured by our use of the tj: GAL4 driver, since this driver is restricted to cells posterior to TFCs in L3. A potential source of JAK/STAT ligands that would not have been captured by our experiments could be the anterior somatic cells that are in close contact with TFCs. While TFCs establish the number of niches, ICs appear to communicate with and regulate the number of GCs that can populate those niches. We hypothesize that TFCs and ICs do not regulate each other’s proliferation non-autonomously. However, we cannot test this hypothesis directly, as to our knowledge, no GAL4 drivers currently exist that are exclusively expressed in only TFCs or only ICs. Nevertheless, a number of lines of evidence support this hypothesis. First, reducing Hippo pathway activity in a subset of TFCs had no effect on IC number (S5H Fig. ; S5 Table). Second, a double knockdown of egfr and hpo under the tj: GAL4 driver reduced IC number but had no effect on TFC number (Fig. 5D). Third, loss of germ cell-less (gcl) function leads to reduced GC and IC numbers [49], but has no effect on ovariole number [81]. Given that ovariole number is largely determined by TFC number [50], it is likely that gcl ovaries have reduced ICs but not reduced TFCs. However, we note that both TFCs and ICs respond to hormonal cues provided by Ecdysone and Insulin signaling [6,8]. This suggests that growth of these somatic cell types may be accomplished through their response to systemic hormonal cues, rather than through non-autonomous effects of one somatic cell type on another. While the Hippo pathway regulates proliferation of both ICs and TFCs, each cell type had a unique pattern of Hippo pathway activity during larval development, suggesting that the upstream regulatory cues of Hippo signaling are different for TFCs and ICs. In Drosophila, glial cells and wing disc cells activate the Hippo pathway using different combinations of upstream regulators [18], indicating that the Hippo pathway can interact with a unique set of upstream regulatory genes depending on the cell type. Addressing these cell type-specific differences in Hippo pathway activation in future studies will elucidate how the Hippo pathway is regulated locally during development of complex organs to establish organ size. Another notable difference between Hippo pathway operation in ICs and TFCs is its differential interactions with the EGFR and JAK/STAT pathways in distinct ovarian cell types. In Drosophila intestinal stem cell development and stem cell-mediated regeneration [37,38,39,68], as well as in eye imaginal discs [64,67], the Hippo pathway regulates proliferation of these tissues via interactions with both the EGFR and JAK/STAT pathways. In contrast, the Hippo pathway acts in parallel with but independently of both pathways to regulate the maturation of Drosophila ovarian follicle cells [25,36]. We do not know what mechanisms determine whether the Hippo pathway interacts with EGFR signaling, JAK/STAT signaling, or both in a given cell or tissue type. One mechanism that may be relevant, however, is the differential activation of specific ligands. For example, in the Drosophila eye disc, Hippo signaling interacts genetically with EGFR activity induced by vein, but not by any of the other three Drosophila EGFR ligands [31]. Similarly, constitutively active human YAP can upregulate transcription of vein, but not the other three EGFR ligands, in Drosophila wing imaginal discs [31]. That fact that spiRNAi driven in the soma does not rescue the hpoRNAi overproliferation phenotype in the ovary may indicate that other ligands, such as vein, are required for this EGFR-Hippo signaling interaction, or that the relevant EGFR ligands are expressed by GCs rather than the soma. Our results suggest that the larval ovary could serve as a model to examine whether differential ligand use within a single organ could modulate Hippo pathway activity during development. Previous reports [35,36] suggested that the Hippo pathway components were dispensable for the proliferation of adult GSCs. In contrast, we observed that yki controls proliferation of the larval GCs, albeit independently of hpo and wts. These contrasting results are likely due to the fact that Sun et al. [35] sought to detect conspicuous germ cell tumors in response to reduced Hippo pathway activity, whereas we manually counted GCs and in this way detected significant changes in GC number in response to yki knockdown or overexpression. Although hpo, wts, ex or hipk RNAi (Fig. 4A) and hpo null clones (Fig. 4F) suggested that yki activity in GCs was independent of the canonical Hippo kinase cascade, overexpression of hpo in GCs did decrease GC number (Fig. 4A). Taken together, our data suggest that although sufficiently high levels of hpo are capable of restricting Yki activity in GCs, hpo does not regulate yki in GCs in wild type ovaries. A growing body of evidence shows that hpo-independent mechanisms for regulating Yki are deployed in stem cells of multiple vertebrate and invertebrate tissues. For example, in mammalian epidermal stem cells, YAP is regulated in a Hpo-independent manner by an interaction between alpha-catenin and adaptor protein 14–43 [82]. Similarly, the C-terminal domain of YAP that contains the predicted hpo-dependent phosphorylation sites is dispensable for YAP-dependent tissue growth in postnatal epidermal stem cells in mice [83]. Other known Hpo-independent regulators of Yki include the phosphatase PTPN14 and the WW domain binding protein WBP2, which were identified in mammalian cancer cell lines [84,85]. The flatworm Macrostomum ligano displays a requirement for hpo, sav, wts, mats and yki in regulating stem cell number and proliferation, although it is unknown whether yki operates independently of the core kinase cascade in this system [86]. In contrast, however, in the flatworm Schmidtea mediterranea, while yki plays a role in regulating stem cell numbers, hpo, wts and Mer appear dispensable for stem cell proliferation [87]. We hypothesize that, as in many other stem cell systems, the Drosophila germ line may use Yki regulators that are not commonly used in the soma to regulate proliferation. Further investigation into the Yki interacting partners in GCs will be needed to understand how Yki may be regulated non-canonically in establishing stem cell populations. One of the most striking aspects of growth regulation in the larval ovary is the homeostatic growth of ICs and GCs during development. This homeostatic growth is critical to ensure establishment of an appropriate number of GSC niches that each contain the correct proportions of somatic and germ cells. We have summarized the available data on the molecular mechanisms that regulate the number of ICs and GCs (Fig. 7A) and our current understanding of how these mechanisms operate within and between the cell types that comprise the GSC niche (Fig. 7B). Previous work has shown that these mechanisms include the Insulin signaling and EGFR pathways. Insulin signaling function in the soma regulates differentiation and proliferation both autonomously in ICs and non-autonomously in GCs [6] (Fig. 7A, B). The EGFR pathway regulates homeostatic growth of both IC and GC numbers as follows: GCs produce the ligand Spitz that promotes survival of ICs, and ICs non-autonomously represses GC proliferation via an unknown regulator that is downstream of the EGFR pathway [49] (Fig. 7A, B). Our results add four critical new elements to the emerging model of soma-germ line homeostasis in the larval ovary (Fig. 7B, blue elements). First, yki positively and cell-autonomously regulates GC number independently of the canonical Hippo signaling pathway. Second, canonical Hippo signaling negatively and cell-autonomously regulates TFC number via JAK/STAT signaling, and IC number via both EGFR and JAK/STAT signaling. Third, JAK/STAT signaling also negatively regulates IC and TFC number in a cell-autonomous manner. Finally, Hippo signaling contributes to non-autonomous homeostatic growth of ICs and GCs in at least two ways: (1) Yki activity in GCs non-autonomously regulates IC proliferation; and (2) Hippo signaling activity in ICs non-autonomously regulates GC proliferation through the EGFR and JAK/STAT pathways. The latter relationship is, to our knowledge, the first report of a non-autonomous mechanism that ensures that GC number increases in response to increased IC number, without negatively affecting GSC niche differentiation or function. Finally, we note that although IC number and GC number had been previously observed to affect each other non-autonomously [6,49], our experiments shed new light on the remarkable degree to which specific proportions of each cell type are maintained, and demonstrate the Hippo pathway’s involvement in this precise homeostasis. This proportionality was not maintained, however, in Hippo/ EGFR or Hippo/JAK/STAT pathway double knockdowns (Figs. 7, S7). This suggests that Hippo pathway-mediated proportional growth of ICs and GCs requires activity of not only the EGFR pathway, as previously reported [49], but also of the JAK/STAT pathway in the soma. The proportional growth of these cell types maintained by the Hippo-EGFR-JAK/STAT pathway interactions we describe here suggests that the soma releases proliferation-promoting factors to the GCs, and that the GCs can process these signals to maintain optimal proportionality. Similarly, when GC number increased via yki overexpression in GCs, we noticed that IC number increased non-autonomously. Achieving specific numbers and proportions of distinct cell types within a single organ, and linking these processes to final organ size and function, are largely unexplained phenomena in developmental biology and organogenesis. By using the larval ovary as a system to address these problems, we have shown not only that the Hippo pathway is involved in these processes, but also that it can display remarkable complexity and modularity in regulating stem cell precursor proliferation and adjusting organ-specific stem cell niche number during development. Flies were reared at 25°C at 60% humidity with food containing yeast and in uncrowded conditions as previously described [50]. The following RNAi lines from the Bloomington Stock Center (B) [56] or the Vienna Drosophila RNAi Center (VDRC) [55] were used for knockdown: B33614 (UAS: hpoRNAi), B34064 (UAS: wtsRNAi), B34067 (UASykiRNAi), VDRC104523 (UAS: ykiRNAi), VDRC109281 (UAS: exRNAi), VDRC43267 (UAS: egfrRNAi), VDRC19717 (UAS: domeRNAi), B35363 (UAS: hipkRNAi), B35481 (UAS: sdRNAi). For overexpression of hpo or yki we used w*; UAS: hpo/TM3 Sb [19] and w*; UAS: yki/TM6B [21] (courtesy of D. Pan, Johns Hopkins University). GAL4 lines used were: w; P{GawB}bab1Pgal4–2/TM6, Tb1 (bab: GAL4, B6803), P{UAS-Dcr-2. D}1, w1118; P{GAL4-nos. NGT}40 (nos: GAL4, B25751), y w; P{w+mW. hs = GawB}NP1624 (tj: GAL4, Kyoto Stock Center, K104–055), y w hs: FLP122; Sp/CyO; hh: GAL4/TM6B (hh: GAL4, courtesy of L. Johnston, Columbia University), w; P{w+mW. hs = GawB}ptc559. 1 (ptc: GAL4, B2017). For GAL4 expression domain analysis, GAL4 lines were crossed to w; P{w+mC = UAS-GFP. S65T}T2 (B1521). For clonal analysis of hpo and yki null alleles, the following lines were used: w1118; P{ry+t7. 2 = neoFRT}42D P{w+mC = Ubi GFP (S65T) nls}2R/CyO (B5626), w1118; P{ry+t7. 2 = neoFRT}42D P{w+t* ry+t* = white-un1}47A (B1928), P{ry+t7. 2 = hsFLP}1, w1118; Adv[1]/CyO (B6), hsFLP12 w*; P{ry+t7. 2 = neoFRT}42D ykiB5/CyO [21] (courtesy of D. Pan, Johns Hopkins University), and y* w*; P{ry+t7. 2 = neoFRT}42D hpoBF33/CyO (y+) [57] (Courtesy of J. Jiang, University of Texas Southwestern Medical Center). For analysis of cell type numbers in flies homozygous for loss of function Hippo pathway alleles, we used ex1 (B295; [54]) and y*w*eyFLP; FRT42D ykiDBO2 / CyO (Courtesy of K-L Guan, UCSD; [33]). Validation of RNAi lines was provided by data from a number of independent experiments, as follows: (1) Immunohistochemistry against Hpo or Yki showed that RNAi against these genes reduced protein levels to levels indistinguishable from background in whole mounted larval ovaries (S1D, I Fig.). (2) Germ line clones of null alleles of hpo (hpoBF33 [57]) or yki (ykiB5 [21]) had the same effect on germ cell number as RNAi against these genes driven in the germ line (Fig. 4F). (3) A null allele of expanded [54] had the same effect on TFC number, GC number and IC number as RNAi against Hippo pathway activity (S4 Fig. , S2 Table). (4) Two different yki RNAi lines had the same effect on GC number (S3 Table). (5) Expression of pMAPK and Stat92E in the larval was reduced by RNAi against egfr or spi and dome or upd1, respectively (S8 Fig.). In addition, the wtsRNAi and domeRNAi lines we used here have been independently validated by other studies [88,89]. Larvae were all reared at 25°C at 60% humidity. Larval fat bodies were dissected in 1xPBS with 0. 1% Triton-X, and fixed in 4% PFA in 1xPBS for 20 minutes at room temperature or overnight at 4°C. For tissues stained with the rat-Hippo antibody (courtesy of N. Tapon, London Research Institute), fat body tissue was fixed in freshly made PLP fixative [17] for 20 minutes. Tissues were stained as previously described [50]. Primary antibodies were used in the following concentrations: Mouse anti-Engrailed 4D9 (1: 50, Developmental Studies Hybridoma Bank), guinea pig anti-Traffic Jam (1: 3000–5000, courtesy of D. Godt, University of Toronto), rabbit anti-Vasa (1: 500, courtesy of P. Lasko, McGill University), rabbit anti-Yorkie (1: 400, courtesy of K. Irvine, Rutgers University), rat anti-Hippo (1: 100, courtesy of N. Tapon, London Research Institute), chicken anti-Beta-galactosidase (1: 200, Abcam), mouse anti-Alpha spectrin 3A9 (1: 5, Developmental Studies Hybridoma Bank), rabbit anti-dpErk (1: 300, Cell Signaling), rabbit anti-Stat92E (1: 200, courtesy of E. Bach, New York University). We used goat anti-guinea pig Alexa 488, anti-mouse Alexa 488, Alexa 555, and Alexa 647, anti-rabbit Alexa 555, Alexa 647, anti-rat Alexa 568, and anti-chicken Alexa 568 at 1: 500 as secondary antibodies (Life Technologies). All samples were stained with 10 mg/ml Hoechst 33342 (Sigma) at 1: 500 to visualize nuclei, and some samples were stained with 0. 1 mg/ml FITC-conjugated Phalloidin (Sigma) at 1: 200 to visualize cell outlines. For GAL4 crosses, we crossed virgin females carrying the GAL4 construct with males carrying the UAS construct, and analyzed F1 LP stage larvae. Samples were imaged with Zeiss LSM 700,710 or 780 confocal microscopes at the Harvard Center for Biological Imaging. Each sample was imaged in z-stacks of 1 μm thickness. For expression level analysis, laser settings were normalized to the secondary only control conducted in parallel to the experimental stain. Expression levels were quantified using Image J (NIH) and were normalized to nuclear stain intensity to control for staining level differences between samples. White immobile pupae were collected from uncrowded tubes (<100 larvae) for cell number analysis. All cell counts were obtained manually using Volocity (Perkin Elmer) after samples were randomized and coded to prevent bias; cells stained with Vasa were counted for germ cell number, and cells stained with Traffic Jam were counted for intermingled cell number. TF number and total TFC number were collected as described in [50]. Experimental crosses were compared to parental GAL4 and RNAi strains using a student’s t-test with unequal variance performed in Microsoft Excel. Changes in number were not considered significant unless p values were significant for both parental strains. For crosses where one or both parents were heterozygous for balanced GAL4 and/or UAS elements, sibling data from F1s carrying balancer chromosomes, rather than parental data, was collected as a control. Adult ovariole number was counted in mated females that were 3–5 days post hatching from uncrowded vials kept in 25°C at 60% humidity. Adult ovaries were dissected in 1xPBS containing 0. 1% Triton-X, and ovariole number was counted under a dissecting microscope by teasing apart ovariole strands using a tungsten needle. F1 ovariole number was compared to the ovariole number of siblings carrying balancer chromosomes for bab: GAL4, and to the tj: GAL4 parental line for the tj: GAL4 crosses. Adult fecundity was measured by placing three females and one male in a vial for 24 hours, and counting total egg number per vial. Five replicates (vials) were performed for each treatment. The egg count was divided by the number of females to obtain the average egg number per female per 24 hours. P0 flies were mated (for ykiB5 clones: w1118; P{ry+t7. 2 = neoFRT}42D P{w+mC = Ubi GFP (S65T) nls}2R/CyO x hsFLP12 w*; P{ry+t7. 2 = neoFRT}42D ykiB5/CyO; for hpoBF33 clones: P{ry+t7. 2 = hsFLP}1, w1118; P{ry+t7. 2 = neoFRT}42D P{w+mC = Ubi GFP (S65T) nls}2R/CyO x y* w*; P{ry+t7. 2 = neoFRT}42D hpoBF33/CyO (y+); for control w clones: P{ry+t7. 2 = hsFLP}1, w1118; P{ry+t7. 2 = neoFRT}42D P{w+mC = Ubi GFP (S65T) nls}2R/CyO x w1118; P{ry+t7. 2 = neoFRT}42D P{w+t* ry+t* = white-un1}47A) and F1 eggs were collected for 8–12 hours at 25°C. L1 larvae were heat shocked at 37°C for 1 hour 36–48 hours after egg laying. Late L3 to LP stage ovaries were dissected, stained with 10 mg/ml Hoechst 4333 (Sigma) at 1: 500, FITC-conjugated anti-GFP (1: 500, Life Technologies), and rabbit anti-Vasa (1: 500, courtesy of P. Lasko, McGill University), and imaged. GFP-negative mutant GC clone size (number of cells per clone) and GFP++ wild type twin spot clone size were counted manually.

During development, organ growth must be carefully regulated to make sure that organs achieve the correct final size needed for organ function. In organs that are made of many different types of cells, this growth regulation is likely to be particularly complex, because it is important for organs to have appropriate proportions, or relative numbers, of the different kinds of cells that make up the organ, as well as the correct number of total cells. One method that cells use to regulate organ growth is a signaling pathway called the Hippo pathway. However, Hippo signaling has been studied, to date, primarily in organ systems that are made up of one cell type. In this study, we examine how Hippo signaling can work to regulate the proportions of different types of cells, as well as the total number of cells in an organ. To do this, we used the developing ovary of the fruit fly as a study system. We found that (1) Hippo signaling regulates the proliferation of many different cell types of the ovary; and (2) Hippo signaling activity in one cell type influences proliferation of other cell types, thus ensuring appropriate proportions of different ovarian cell types.

Abstract Introduction Results Discussion Materials and Methods

2015

The Hippo Pathway Regulates Homeostatic Growth of Stem Cell Niche Precursors in the Drosophila Ovary

Validation of elimination of trachoma as a public health problem is based on clinical indicators, using the WHO simplified grading system. Chlamydia trachomatis (Ct) infection and anti-Ct antibody responses (anti-Pgp3) have both been evaluated as alternative indicators in settings with varying levels of trachoma. There is a need to evaluate the feasibility of using tests for Ct infection and anti-Pgp3 antibodies at scale in a trachoma-endemic country and to establish the added value of the data generated for understanding transmission dynamics in the peri-elimination setting. Dried blood spots for serological testing and ocular swabs for Ct infection testing (taken from children aged 1–9 years) were integrated into the pre-validation trachoma surveys conducted in the Northern and Upper West regions of Ghana in 2015 and 2016. Ct infection was detected using the GeneXpert PCR platform and the presence of anti-Pgp3 antibodies was detected using both the ELISA assay and multiplex bead array (MBA). The overall mean cluster-summarised TF prevalence (the clinical indicator) was 0. 8% (95% CI: 0. 6–1. 0) and Ct infection prevalence was 0. 04% (95%CI: 0. 00–0. 12). Anti-Pgp3 seroprevalence using the ELISA was 5. 5% (95% CI: 4. 8–6. 3) compared to 4. 3% (95%CI: 3. 7–4. 9) using the MBA. There was strong evidence from both assays that seropositivity increased with age (p<0. 001), although the seroconversion rate was estimated to be very low (between 1. 2 to 1. 3 yearly events per 100 children). Infection and serological data provide useful information to aid in understanding Ct transmission dynamics. Elimination of trachoma as a public health problem does not equate to the absence of ocular Ct infection nor cessation in acquisition of anti-Ct antibodies. Trachoma is a disease caused by Chlamydia trachomatis (Ct). Repeated ocular infections [1] result in inflammation leading to conjunctival scarring, trichiasis (in-turned eyelashes which touch the eye) and ultimately corneal opacity (CO). The intervention strategy for trachoma is the World Health Organization (WHO) -endorsed SAFE strategy (S: Surgery for trachomatous trichiasis (TT); A: Antibiotics to clear Ct infection; F: Facial cleanliness and E: Environmental improvement to reduce transmission of Ct) [2,3]. Successful implementation of this strategy has resulted in a reduction in trachoma prevalence across many endemic countries [4–7]. WHO has set a goal of global elimination of trachoma as a public health problem by 2020 [8]. For trachoma elimination to be validated, countries must provide evidence that three criteria have been met. First, each previously-endemic evaluation unit (EU; populations of 100,000–250,000 people) must have reached and sustained, for at least two years, a prevalence of trachomatous inflammation—follicular (TF) in 1–9-year-olds of less than 5%. Second, each previously endemic EU must have reached a prevalence of TT previously unknown to the health system in ≥15-year-olds of less than 0. 2%. Third, there must be an appropriately-resourced system to identify and manage incident trichiasis cases [9]. The WHO elimination thresholds for trachoma are based on clinical diagnostic indicators [9], using the simplified grading system [10]. However, TF has been shown to correlate poorly with Ct infection in low prevalence settings [11–14]. A follicular inflammatory response is known to persist for many weeks after infection has been cleared [15,16]. The presence of follicles deep to the upper tarsal conjunctiva is not a sign unique to Ct infection; a number of non-chlamydial pathogens including Haemophilus influenzae may elicit a similar response [14,17]. As such, the positive predictive value of a clinical diagnosis of TF for Ct infection can be reduced in low prevalence settings [18] where other aetiologies may account for a high proportion of TF. In the context of trachoma elimination, a lack of specificity of TF as an indicator will make it increasingly difficult to ensure that EUs are correctly categorised as endemic or not and that valuable resources are not wasted by unnecessarily prolonging interventions [19]. There are also concerns over the inter-grader agreement for diagnosis of TF, which becomes increasingly difficult to demonstrate [20] as trachoma prevalence decreases. As a result, there is a considerable interest in exploring whether and how alternative indicators could provide more objective evidence of elimination of trachoma as a public health problem, or be used as tools for post-validation surveillance [21]. Tests for anti-Ct antibody and Ct infection have been evaluated as alternative markers in settings with varying levels of trachoma [22–26]. In general, there has been very little or no Ct infection identified in areas where TF prevalence is below the elimination threshold [25–28]. Nucleic acid amplification tests (NAATs) including polymerase chain reaction (PCR) are highly specific and sensitive for ocular Ct infection [29,30]. The Cepheid GeneXpert platform is an automated, cartridge-based NAAT platform used widely across Africa for detection of Mycobacterium tuberculosis [31] that can detect Ct infection using different primers [29]. While a good test for Ct infection may have advantages over a proxy indicator, such as a sign of eyelid inflammation, collecting and analysing conjunctival swabs can be time-consuming, require specialist resources and personnel, and be potentially cost-prohibitive for national eye care or neglected tropical disease programmes [30]. The presence of anti-Ct antibodies, measured by multiplex bead array (MBA) [32,33], enzyme-linked immunosorbent assay (ELISA) [32,34,35] or lateral flow assay [32,36,37], may reflect cumulative exposure to Ct and when evaluated against age, represent transmission intensity over time [23,25,38,39]. Studies to date have predominantly focused on the detection of antibodies to Pgp3 [23,25,38], a conserved Ct plasmid protein found in both urogenital and ocular serovars [40]. The prevalence of anti-Pgp3 antibodies correlates fairly well with the prevalence of TF [22,23,25,26,38,39]. In post-elimination settings, the prevalence of Pgp3 seropositivity in children has shown either no increase with age or only minimal increases with increasing age [22,25,26,38]. The feasibility of generating district-level data for Ct infection and anti-Ct antibodies and how to interpret them for programmatic decision-making is still to be determined. A better understanding of the age-prevalence profiles in the post-elimination setting is also needed [9]. In 2015–2016, Ghana conducted a set of population-based surveillance surveys that demonstrated that all EUs previously endemic for trachoma had maintained the elimination threshold of <5% TF in the absence of large scale antibiotic treatment [41]. We integrated ocular swabs and DBSs into the surveys, providing an opportunity to evaluate the feasibility of using tests for Ct infection and anti-Pgp3 antibodies at scale in a trachoma-endemic country. We also compared antibody data collected by ELISA in Ghana to MBA data run at the Centers for Disease Control and Prevention (CDC), USA. These data were also used to evaluate whether measures of infection or Pgp3 antibody response have added value for understanding transmission dynamics in the peri-elimination setting. The study was approved by the Ghana Health Service (GHS) Ethics Review Committee (Reference GHS-ERC: 03/07/15) and the London School of Hygiene & Tropical Medicine Research Ethics Committee (Reference 10285). CDC involvement was determined not to constitute engagement in human subjects research, as CDC staff had no interaction with study participants. Written informed consent was sought from caregivers of all children who participated in this study. Children who were able to provided verbal assent. Individuals with active trachoma were given 1% tetracycline eye ointment. The study was conducted in the Northern and Upper West regions of Ghana, Fig 1. Surveys were conducted between November 2015 and April 2016. Baseline assessments of trachoma prevalence were conducted in all 18 EUs between 1999 and 2003 [41]. A total of four EUs had a TF prevalence (in some cases combined TF/TI prevalence) of above 10% in children aged 1–5 years, five had a prevalence of 5–9. 9% and nine had a prevalence of less than 5%. Based on WHO recommendations [42], GHS implemented the SAFE strategy, delivering EU-wide mass drug administration (MDA) of azithromycin in the EUs with TF prevalence >10% and antibiotic distribution targeted at community level in the EUs with TF prevalence <10%). In 2008, impact surveys were conducted and all 18 EUs were declared to have reached or maintained the TF elimination threshold [43]. In 2011, GHS implemented a surveillance strategy that involved annual community and school screening for detection of TF and TT [41]. Eight communities identified during impact surveys or surveillance to have TF ≥5% were given three years of azithromycin MDA. A series of two-stage cluster-sampled population-based surveys were conducted in all 18 EUs as part of the Ghana pre-validation surveillance process [41]. A sub-set of nine EUs had additional indicators collected and evaluated, the results of which are the focus of this paper. Clinical, antibody and infection data were collected from six EUs (Table 1 and Fig 1). An additional three EUs were sampled for clinical and antibody data only (Table 1 and Fig 1). The EUs selected were chosen to represent a range of baseline TF prevalence and provide geographical spread. Infection data were not collected from the additional three EUs because of financial and time constraints related to the analysis. The primary sampling unit was a community (village), selected with probability proportional to population size, and the secondary sampling unit was the household, selected using compact segment sampling. All children aged between one and nine years residing in the selected households were eligible for inclusion. The sample size calculations and sampling criteria were based on TF parameters and have been detailed elsewhere [41]. Survey data were collected electronically using a secure Open Data Kit-based Android smartphone application (LINKS, Task Force for Global Health, Atlanta, GA, USA; https: //linkssystem. org) [44]. Data were uploaded to a cloud-located server with password-protected access only available to identified study investigators. All graders were certified using Global Trachoma Mapping Project (GTMP) methodologies, described elsewhere [20]. Due to the expected low prevalence of trachoma in Ghana, the graders were trained and certified by examining children in Sokoto, Nigeria, where a number of districts still have TF prevalence estimates above the elimination threshold [7]. Each grader had to achieve a minimum kappa score of 0. 7 for TF in an inter-grader agreement test with a grader trainer who had been certified by the GTMP. Children aged 1–9 years were assessed for all five signs of trachoma (TF, trachomatous inflammation–intense (TI), trachomatous scarring (TS), TT and CO) as per the WHO simplified grading criteria [10]. Ocular swabs for infection testing were collected by passing a dry sterile polyester-tipped swab horizontally along the upper tarsal conjunctiva of the left eye, at least three times, rotating the shaft 120° with each pass. Control procedures were put into place to avoid field contamination, in particular washing hands at each new household, changing gloves between each examinee, and ensuring the end of the swab, once it had touched the conjunctiva, was placed directly into and broken off within a tube, which was sealed without further swab contact. Negative controls were taken after every 50 swabs by passing a clean swab in the air within five centimetres of a child’s eyes. Collected swabs were kept cool in the field and then refrigerated at 4°C for up to one week before being shipped on ice packs to Noguchi Memorial Institute for Medical Research (NMIMR) in Accra, where they were stored at -20°C until the time for analysis. Specimens were limited to one freeze/thaw cycle to reduce potential DNA degradation [29]. Dried blood spots (DBSs) were collected for serological testing. After cleaning with an alcohol-soaked swab, the participant’s finger was pricked using a sterile single-use lancet and the blood collected directly onto filter paper (Trop-Bio, Townsville, Australia). The filter paper had six projections, each calibrated to collect 10 μL of blood. The filter papers were air-dried in the shade then individually packed in sealable plastic bags and stored in a larger (gallon-size) sealable plastic bag with desiccant. DBSs were refrigerated at 4°C for up to one week before being shipped at ambient temperature to Accra and stored at -20°C until analysis. PCR and ELISA were done at NMIMR in Accra, Ghana. MBA analysis was performed at the CDC in Atlanta, USA. Ocular swabs were analysed for the presence of Ct DNA using the GeneXpert IV machine (Cepheid, Sunnyvale, USA) and GeneXpert CT/NG Assay (Cepheid, Sunnyvale, USA). Swabs were eluted using sterile diethylpirocarbonate (DEPC) water and pooled into groups of five samples as per a published pooling strategy [45]. The individual samples that made up a positive pool were tested separately to identify the positive sample (s). Results were reported as Ct-positive, negative or indeterminate (invalid, error or no result). The GeneXpert can produce an invalid result if there is failure of the sample adequacy control, which requires human DNA in the sample, or specimen processing control, indicating that amplification was inhibited [30]. Indeterminate pools were re-tested using a new aliquot of the specimen and a new cartridge. Control swabs collected in the field were analysed individually. Two Ct positive and two Ct negative processing controls were run at the beginning of each week. DBSs were tested for antibodies to the Ct antigen Pgp3 using the semi-quantitative ELISA assay, described elsewhere [32,34]. Briefly, Immulon 2HB 96-well plates (ThermoFisher Scientific, Waltham, MA) were sensitized with 50 μL of Pgp3 antigen (500 ng/mL concentration) overnight at 4°C. DBSs and serum samples were diluted 1: 50 in PBS containing 0. 3% Tween-20 and 5% milk powder (PBST-milk) and stored overnight at 4°C. The next day, wells were washed with PBST (0. 3% Tween-20 in PBS) and then blocked with PBST for one hour. Sample (50 μL) was added to wells and incubated for two hours at room temperature. Wells were then washed with PBST and incubated with 50 μL anti-human IgG conjugated to horseradish peroxidase (HRP) (1: 10,000 dilution) (Southern Biotech, Birmingham, AL) to detect bound antibody. After four washes with PBST, 50 μL of 3,3′, 5,5′-tetramethylbenzidine (TMB) developing reagent (KPL, Gaithersburg, MD) was added to the wells and the reaction was stopped with 50 μL 1N H2SO4 after the predetermined interval. The optical density (OD) at 450 nm was read using an ELx808 Absorbance Microplate Reader (Biotek, Winooski, USA). OD values were corrected for background absorbance by subtracting the average OD of the two wells containing PBST-milk. The blanked OD values for all samples and controls were then normalised against the 200 U standard included on the same plate [34]. For the MBA, samples were tested in single-wells with Pgp3-coupled beads, as previously described [46,47]. Briefly, one DBS extension was diluted 1: 320 in PBS containing 0. 5% casein, 0. 3% Tween 20,0. 5% polyvinyl alcohol, 0. 8% polyvinylpyrrolidone, 0. 02% sodium azide and 3 μg/mL E. coli extract (Buffer B). Coupled beads (2500 per antigen) were incubated with 40 μL of diluted sample per well in a 96-well filter plate (Millipore, Billerica, MA) for 1. 5 hours. Wells were washed three times with PBS containing 0. 05% Tween 20 (PBST2) and incubated with 50 ng biotinylated mouse anti-human IgG (Southern Biotech, Birmingham, AL) and 20 ng biotinylated mouse anti-human IgG4 (Southern BioTech) for 45 minutes to detect bound antibody. Wells were washed three times with PBST2 and incubated with 250 ng phycoerythrin-labelled streptavidin (Invitrogen, South San Francisco, CA) for 30 minutes to detect bound secondary antibody. After three washes with PBST2, wells were incubated for 30 minutes with 0. 5% BSA, 0. 05% Tween 20,0. 02% sodium azide in PBS to remove any non-specific binding. After one wash, wells were suspended in 125 μL of PBS and read on a Bio-Plex 200 instrument (Bio-Rad, Hercules, CA) equipped with Bio-Plex manager 6. 0 software (Bio-Rad). The median fluorescence intensity (MFI) with the background from the blank well (Buffer B alone) subtracted out (MFI-bg) was recorded for each antigen for each sample. All samples were analysed masked to demographic and examination findings. Only individuals with complete serological, infection and clinical data (or serological and clinical data, in EUs where ocular swabs were not collected) were included in the analysis. The dataset was presumed to be self-weighted but the analysis was adjusted (using STATA’s svy command) for the cluster sampling methodology. Individuals were classified as seropositive or seronegative based on normalised OD values on the ELISA platform, and on the MFI-bg, after a log (x + 1) transformation, for the MBA. This transformation took into account that the MFI-bg included values of zero. The seropositive cut-off was defined using a finite mixture model based on maximum likelihood methods [34], with the threshold for seropositivity set as the mean of the Gaussian distribution of the seronegative population, plus four standard deviations [38]. To examine force of infection (FoI), the rate at which susceptible individuals acquire infection, the seroconversion rate (SCR), the rate at which seronegative individuals become seropositive, was estimated using a simple reversible catalytic model (RCM) fitted to seroprevalence in yearly age groups, using maximum likelihood estimates. Evidence for a change in SCR over time was explored by comparing two models using the profile likelihood method; the first model assumed constant transmission over time and the second assumed a potential change in the FoI at a specified time point [48]. Statistical analysis was conducted using R 3. 4. 0 [49] and STATA 12. 0 [50]. The data were adjusted for age and gender based on the Ghana 2010 census [51]. The adjusted cluster-summarised mean prevalence was calculated for all the data and at the level of the EU. Bootstrap estimation was used to determine confidence intervals around prevalence estimates, based on 10,000 iterations and taking the 2. 5th and 97. 5th centiles. For the serology data, chi-square tests were used to determine univariate associations. The non-parametric test for trend was used to determine an increase in seropositivity with age. Positive univariate associations of seropositivity at the individual level (age, EU, baseline TF endemicity) and gender (included a priori) were included in multivariate logistic regression models. The likelihood ratio test was used to determine the model of best fit. Regression was used for analysis of associations between continuous variables at the level of the EU. The geometric mean antibody titre was calculated using a log (x+1) transformation to take into account zero values. A measure of cluster-level heterogeneity was determined by calculating the intra-cluster correlation coefficient (ICC) and design effect (DE). The ICC reveals how strongly observations in the same cluster resemble each other. The DE is the ratio of the variance in the collection of observations amassed using cluster sampling to the variance assuming the same sample size had been generated using simple random sampling [52]. The ICC and DE for this dataset were determined using STATA. Overall 96. 0% of children resident in the selected households were examined, 3. 6% were absent at the time of the survey and 0. 4% refused to participate. A total of 11,730 DBSs were collected across nine districts, analysed by ELISA and matched to demographic and clinical data, Table 1. A total of 10,902 DBSs were also analysed by MBA. The number of samples taken by age group are detailed in Table 2. A total of 7,688 ocular swabs were taken across six districts, analysed and matched to demographic and clinical data. Overall, 50. 2% of individuals surveyed were male and the median age was 5 years old. Across all EUs, 1. 0% of individuals (n = 112; 95%CI: 0. 8–1. 2) had TF in one or both eyes; of those, 67. 9% (n = 76) had bilateral TF. There was no evidence of an association between TF and age (z = -0. 48; p = 0. 63). No TT was identified in children. The median age of individuals with TF was 4. 5 years and 53. 6% were female. The overall cluster-summarised mean TF prevalence was 0. 8% (95%CI: 0. 6–1. 0), with an EU-level range of 0. 5–1. 1%, Table 3. Four infections were identified, giving a cluster-summarised mean prevalence of 0. 04% (95%CI: 0. 00–0. 12), Table 3. The four samples were from two different clusters, one from a community in Bole and the other three from separate households of a single community in Zabzugu-Tatale. Both of these EUs had baseline TF prevalences of 5–9. 9% in children aged 1–5 years. The median age of those infected was 6. 5 years. A total of 83 samples (1. 1%) had indeterminate PCR results. All control swabs were negative for Ct DNA. The seropositive cut-off (four standard deviations from the mean of the seronegative population) for the ELISA was 1. 091 OD450nm and for the MBA was 5. 188 for the log of the MFI-bg. The overall cluster-summarised mean seroprevalence was 5. 5% (95%CI: 4. 8–6. 3) by ELISA and 4. 3% (95%CI: 3. 7–4. 9) by MBA. Pgp3 seropositivity by ELISA differed by EU (F stat = 3. 61; p = 0. 001), with the highest seroprevalence in Bole/Sawla-Tuna-Kalpa (8. 2% 95%CI: 5. 7–11. 1) and lowest in Tolon-Kumbungu (2. 5%; 95%CI: 1. 5–3. 6) (Table 3). This pattern in EU seropositivity was also reflected by the MBA results. Seropositivity increased with age; this association held when analysing results from either platform (p<0. 001) (Fig 2). The model of best fit for the RCM was a singular SCR for the time period studied. Overall, there was a low seroconversion rate of 1. 3 yearly events per 100 children (95%CI: 1. 1–1. 6), using the ELISA data. A similarly low SCR was reported using MBA data, with 1. 2 yearly events per 100 children (95% CI: 0. 9–1. 6). Pgp3 seropositivity was also associated with baseline EU-level TF prevalence (p<0. 001) after controlling for potential confounders. The highest proportions of seropositive children were in those EUs that had a baseline TF prevalence of 5–9. 9% in children aged 1–9 years. In the multi-variate model, there was an interaction observed between age and baseline TF prevalence: EUs with lower baseline TF prevalence estimates had comparatively greater odds of having older children who were seropositive by ELISA as opposed to younger children (p<0. 001) (Table 4). Analysis of the seropositivity data generated by the MBA resulted in the same conclusions. The overall geometric mean of normalised ODs was 0. 49 for the ELISA and 13. 58 MFI-bg for the MBA platform. The geometric mean antibody titre increased with age (p<0. 001). The strongest antibody responses (top 10% antibody titres of all seropositive individuals) were detected in children from the EUs with the highest seroprevalence estimates (for ELISA data: R2 = 0. 68, p = 0. 006). An analysis of heterogeneity of data for the three indicators suggests some variability within clusters (Table 3). The overall DE for clinical data was quite low at 1. 81 (EU-level range 0. 80–2. 88) and an ICC of 0. 01 (cluster-level TF prevalence range 0–7. 1%). The DE was higher for serologic data, 2. 49 for the ELISA (2. 30 for MBA data) with a corresponding ICC of 0. 03 (cluster-level seroprevalence range 0–32. 1%). There was variation in DE across EUs, ranging from 0. 92 to 4. 25 for the ELISA data and 0. 72 to 3. 75 for the MBA data, with the highest values in Zabzugu-Tatale. In the two clusters that were found to have infection, there was also high seropositivity (>15% seroprevalence using the ELISA and MBA). The high heterogeneity of the serology data (in Zabzugu-Tatale) was largely driven by one cluster that had infection (5. 4%) and the highest proportion of seropositive individuals (32. 1%). After removing that cluster from the DE calculations, the DE for (ELISA) serology dropped from 4. 25 to 1. 97 (3. 75 to 0. 90 for the MBA data). In EUs where infection was detected the DE was 3. 01 (Zabzugu-Tatale) and 1. 01 (Bole/Sawla-Tuna-Kalpa). Ghana has met the active trachoma criterion for elimination of trachoma as a public health problem, a measure based on TF parameters [41,53]. The collection of alternative indicators in pre-validation surveillance surveys allowed us to generate a more complete understanding of transmission dynamics in this setting. As evidenced in this study, elimination of trachoma as a public health problem does not equate to the absence of ocular Ct infection nor cessation in acquisition of anti-Ct antibodies. There are a number of potential explanations for this finding. Infection was detected at very low levels (0. 04%) and from limited sites, similar to findings from other studies in analogous settings which found no or very low prevalence of Ct [18,23,26,27]. These infection cases could be false positives, given the specificity of the assay [29,30,45], or a result of cross-contamination in the swab collection or analysis process. Equally, we cannot rule out the possibility that these were urogenital strains, acquired, for example, by transfer to children’s eyes, as a result of poor parental hand hygiene. However, the triangulation of serological, clinical and infection data in the communities suggests these are true Ct cases, whether ocular or urogenital strains. A cross-sectional study cannot tell us whether infection in these communities is transient or it is persistent and a potential risk for recrudescence. It is necessary to follow-up these select communities to determine if there is evidence of continued ocular Ct transmission over time and provide an opportunity to evaluate a model for post-validation surveillance. Pooling of ocular swab samples was used in this study, a process known to be particularly cost-efficient in settings with low infection prevalence [45,54–56]. However, there are some concerns pooling can reduce the sensitivity of the test. Evidence suggests the impact would be minimal and likely to affect those individuals with lower ocular bacterial load [57], who are likely to be less important as drivers of transmission [58,59]. A slight loss in sensitivity would be tolerable where programmatic decision-making relied on EU-level classification; the specificity of the test, however, is critical. The GeneXpert machine is a relatively simple, mobile platform which is closed and self-contained, minimising potential cross-contamination [30], and was successfully used in Ghana. However, even with a simple platform to use and with pooling of samples for analysis, a key limitation of routinely including tests for infection in programmatic surveys will be the time and cost required to analyse the samples using NAATs. Overall seroprevalence in children aged 1–9 years was similar here to estimates reported for other elimination or post-intervention settings [25,38]. The seroprevalence data by age allows estimation of transmission intensity over time. In Ghana there was evidence of a statistically significant increase in seropositivity with age (p<0. 001). Studies in post-MDA settings in Tanzania and The Gambia found a similar increase in seropositivity with age [25,38]. Other studies in settings where elimination thresholds have been sustained over a number of years have found no association of seropositivity with age [22,26], which may be a reflection of interruption of transmission but also potentially a lack of power to be able to detect low rates of seroconversion. The data from Ghana suggested a history of low level ongoing seroconversion (1. 2 to 1. 3 events per 100 children per year) across the Northern and Upper West regions. Ghana stopped district-level antibiotic MDA at least eight years before the time of this study, and the single SCR reported likely reflects this, as any significant change in SCR would probably have occurred before most children enrolled in this study were born. Therefore, the SCR likely reflects a stable FoI in a post-elimination setting. These findings reinforce the idea that very low levels of on-going Ct transmission are not inconsistent with trachoma elimination as a public health problem. As some estimates suggest that more than 150 infections are needed over a lifetime to develop TT [1], the level of transmission we estimated in Ghana is highly unlikely to be of public health concern. It is difficult to directly compare the seropositivity rates across studies because there is currently no agreed standard methodology for defining the threshold used to determine seropositivity [34]. We used an internally calibrated approach, which has the advantage that it does not rely on external positive or negative controls from different populations. This methodology could artificially define a seronegative and seropositive group [60] and inflate the number of “low intensity” positives. We used a conservative cut-off threshold of four standard deviations from the mean of the seronegative population to increase specificity and reduce likelihood of this outcome. Another difficulty in interpreting the serological data is that antibodies to Pgp3 do not distinguish between urogenital and ocular infection, and Ct exposure could have occurred at birth through ocular or respiratory infection from a mother with genital Ct [61]. This is an important consideration, however, although there is a paucity of data on the prevalence of sexually transmitted infections in Ghana, where it is documented it is reported to be relatively low [62–64] and urogenital Ct infection is not believed to be high in the Northern and Upper West regions of the country. If vertical transmission was a major driver of antibody acquisition, then it could be expected that the gradient of the age seroprevalence curve would be flat or even negative with increasing age [38]. It is noted that the use of an RCM is a simplification of real-world transmission dynamics, and using data from a single cross-sectional study to determine two linked RCM parameters is problematic. However, the SCR estimated here is similar to that generated in other studies in comparable environments [38]. Longitudinal data would help to generate better estimates of SCR [65]. In this study, we compared data obtained from the Pgp3 ELISA run in-country to MBA data generated at CDC. While the overall prevalences differed, the qualitative conclusions drawn from the data did not change and in particular the SCR estimates from the two methodologies did not differ significantly. While quantitative data collected on different platforms are difficult to compare, the SCR data presented here suggest a robustness of anti-Pgp3 antibody data. ELISA is relatively inexpensive and the protocol has been shown to be implemented effectively in trachoma-endemic countries. A key advantage of the MBA platform is that it allows for integrated serosurveillance of multiple pathogens [66], however it is unlikely to be widely available in the near future in trachoma-endemic countries, many of which are reluctant to export biological samples for analysis. A rapid test such as a lateral flow assay [36] might be the best format for application to trachoma elimination. Infection and in particular serological data provide useful insights into transmission dynamics. Even if an EU meets trachoma elimination targets, this may not reflect complete interruption of transmission of Ct infection.

Trachoma is a disease caused by Chlamydia trachomatis (Ct). Validation of elimination of trachoma as a public health problem is based on clinical indicators. Antibody and infection data may provide a better understanding of transmission dynamics in elimination settings. Dried blood spots (DBSs) for antibody testing and ocular swabs for Ct infection testing were integrated into the pre-validation trachoma surveys conducted in the Northern and Upper West regions of Ghana in 2015 and 2016. Ct infection was detected using the GeneXpert PCR platform and the presence of anti-Ct antibodies were detected using both the ELISA and multiplex bead array (MBA). Very little infection was identified (0. 04%). The conclusions from the ELISA and MBA testing were similar, with evidence of an association between increasing seroprevalence and age in 1-9-year olds. Infection and serological data provide useful insights into transmission dynamics. Even if an EU meets trachoma elimination targets, this may not reflect complete interruption of transmission of ocular Ct infection.

Abstract Introduction Methods Results Discussion

medicine and health sciences enzyme-linked immunoassays immune physiology pathology and laboratory medicine immunology tropical diseases geographical locations bacterial diseases eye diseases sexually transmitted diseases neglected tropical diseases antibodies immunologic techniques africa research and analysis methods public and occupational health immune system proteins infectious diseases chlamydia infection serology proteins immunoassays people and places biochemistry ghana physiology ophthalmology biology and life sciences trachoma

2018

Serological and PCR-based markers of ocular Chlamydia trachomatis transmission in northern Ghana after elimination of trachoma as a public health problem

Isogenic bacteria can exhibit a range of phenotypes, even in homogeneous environmental conditions. Such nongenetic individuality has been observed in a wide range of biological processes, including differentiation and stress response. A striking example is the heterogeneous response of bacteria to antibiotics, whereby a small fraction of drug-sensitive bacteria can persist under extensive antibiotic treatments. We have previously shown that persistent bacteria enter a phenotypic state, identified by slow growth or dormancy, which protects them from the lethal action of antibiotics. Here, we studied the effect of persistence on the interaction between Escherichia coli and phage lambda. We used long-term time-lapse microscopy to follow the expression of green fluorescent protein (GFP) under the phage lytic promoter, as well as cellular fate, in single infected bacteria. Intriguingly, we found that, whereas persistent bacteria are protected from prophage induction, they are not protected from lytic infection. Quantitative analysis of gene expression reveals that the expression of lytic genes is suppressed in persistent bacteria. However, when persistent bacteria switch to normal growth, the infecting phage resumes the process of gene expression, ultimately causing cell lysis. Using mathematical models for these two host–phage interactions, we found that the bacteria' s nongenetic individuality can significantly affect the population dynamics, and might be relevant for understanding the coevolution of bacterial hosts and phages. Fifty years ago, studies on the heterogeneity of genetically uniform populations demonstrated the importance of single-cell individuality for understanding a number of phenomena, including enzyme induction and radiation resistance [1–3]. The importance of heterogeneity is evident in the response of bacterial populations to antibiotic treatments, in which most bacteria are rapidly killed, but small subpopulations nevertheless persist [4]. Recently, a renewed interest in the persistence phenomenon has revealed that nongenetic heterogeneity might be one of the main reasons for the failure of antibiotic treatment in infections such as tuberculosis, in which a single persistent bacterium can restart an infection [5,6]. Persistence is typically observed through the monitoring of the survival fraction of a bacterial population exposed to antibiotics. A curve showing an initially rapid killing of the bacteria, followed by a significantly reduced killing rate indicates the presence of a persistent subpopulation (Figure 1B, solid curve). When cells grown from this persistent subpopulation are subjected again to antibiotics, the same biphasic killing curve is obtained, suggesting that the persistent subpopulation is not genetically different from the original population. Several mutants with high persistence (hip) to a range of antibiotics were isolated [7,8]. Previously, we directly observed single persister cells in hip strains and determined that persistence is due to an inherent heterogeneity of growth rates in the E. coli population that existed before the antibiotic treatment [9]. Two different processes generate persister cells in the population. In Type I persistence, persister cells are generated when the culture reaches stationary phase [10]. Once transferred to fresh medium, the inoculum contains both normal and persister cells. While the normal cells present in the inoculum start growing within half an hour, persister cells remain dormant for periods that may exceed a day. Because Type I persisters appear at stationary phase and not during the subsequent exponential growth, their number depends on the size of the inoculum from stationary phase [9,11]. Type I persisters exit their dormant state stochastically and switch to normal growth [9]. In contrast, Type II persisters are continuously generated during exponential growth and do not require a starvation signal. The equations describing the dynamics of switching between the normal and persister states have been described for both persistence types [9]. In our present work, we focus on Type I persistence, which has been identified as a major factor of persistence to antibiotics in wild-type (wt) E. coli, as well as in Staphylococcus aureus and in Pseudomonas aeruginosa [11]. In view of the fact that phenotypic individuality plays an important role in bacterial persistence to antibiotics, we were curious to examine its possible involvement in the context of the interaction between bacteria and phages. The abundance of phages in various ecological niches [12] suggests that they represent one of the most common stresses that bacteria have encountered during evolution. Recent studies have indeed revealed the important role played by phages in the evolution [13] and ecology of bacteria [13]. For example, phages are believed to play a pivotal role in the control of bacterial populations in oceans [14]. Our goal was to study the effect of the dormant subpopulation of Type I persisters on the bacterial–phage interaction. We wished to determine whether bacteria persistent to antibiotic treatment are also protected from phage-mediated lysis. For this purpose, we compared the wt E. coli MG1655 to its hipA7 derivative. The hipA7 mutation was shown to confer a high level of Type I persistence [15]. It consists of two point mutations located in the hipA gene of the hipBA operon [10], which acts as a toxin–antitoxin module [16,17]. We adapted our methodology, originally developed for the study of persistence to antibiotics, to investigate the nongenetic individuality in the interaction between E. coli and the λ-phage The λ-phage is a temperate phage that infects E. coli by injecting its DNA into the bacterium. Upon infection, it chooses between lysogeny, in which the phage embeds its genome into the bacterial chromosome and is carried on as prophage in the division process, or lysis, in which the phage activates genes that enable its replication and lysis of the host [18]. Activation of the lytic pathway in lysogenic bacteria, namely prophage induction, can be triggered by stressful conditions such as DNA damage. Using long-term single-cell observations, we studied the ability of persister bacteria to survive phage assaults. We investigated both the induction of prophages and lytic infections. We first compared prophage induction in wt and in the hipA7 mutants with increased persistence to antibiotic treatment. Prophage induction was studied in populations of lysogenic bacteria carrying prophage λcI857KnR. The λcI857KnR phage carries temperature-sensitive mutations in CI, the lytic pathway repressor, that lead to the inactivation of repression at nonpermissive temperatures (42 °C) and to the initiation of the lytic pathway [19]. This system makes possible the controlled induction of prophages (Figure 1C). We subjected MG1655 (wt) λcI857KnR and MG1655A7 (hipA7) λcI857KnR lysogenic strains to prophage induction at 42 °C. The fraction of survivors versus the duration of heat induction is shown in Figure 1A. Similar to the persistence to antibiotic treatments, the survival under prophage induction was much higher for the hipA7 strain than for the wt. A substantial percentage of the hipA7 lysogenic bacteria persisted, even after several hours at 42 °C. When the progeny of these persisters was again exposed to 42 °C, similar killing and survival rates were observed, indicating that persisters to phage induction did not originate from resistant mutants (Text S1). These results were substantiated by single-cell observations: Figure 2A shows the time course of a typical experiment, in which we exposed hipA7 and wt lysogenic bacteria to 42 °C under the microscope, and followed their response, simultaneously. The activation of the lytic pathway was monitored by measuring the fluorescence level of green fluorescent protein (GFP) expression under the control of pR′-tR′, the promoter of the late lytic genes [20]. Whereas the wt population was massively induced and lysed within the first hour following heat induction (Video S1), only 40% of the hipA7 population (Video S2) was induced and lysed during the total observation time (Figure 2B). After 80 min at 42 °C, the temperature was shifted back to 32 °C. Bacteria that survived the heat induction did not express the lytic pathway genes, and started dividing, forming microcolonies (Figure 2A). We observed, under the microscope, that those cells were neither growing nor dividing during the heat-induction treatment. Bacteria that did respond to phage induction in the hipA7 strain followed the same dynamics of induction as wt cells, as seen by monitoring the increase in fluorescence expression from the pR′-tR′-gfp fusion (Figure 2C). Thus, the response to prophage induction of hipA7 lysogenic bacteria reveals two subpopulations: persister bacteria, which are not induced and survive prophage heat induction, and normally growing bacteria, which follow the same induction dynamics as the wt cells and lyse. As expected from Type I persistence, survival following prophage induction decreased with the inoculum' s size, falling below 0. 1% for an inoculum of 103 bacteria (Figure S1). Furthermore, when we exposed the survivors following prophage induction to ampicillin treatment, we observed the expected low killing rate (Figure 1B), indicating that persisters to prophage induction are also persistent to ampicillin. High persistence was also observed for the induction of the wt λ-phage by UV irradiation (Figure S2), when UV-irradiated lysogenic bacteria were subsequently grown under illumination to allow for photoreactivation [21] (Text S1). Collectively, these results show the correspondence between persistence to antibiotics and to prophage induction, and suggest that dormancy protects persisters from both stresses. This view is supported by previous studies on the influence of the host physiology on prophage induction [22]. We now turn to the response of the hipA7 strain to infection by the λ-phage. Are persisters as protected from phage infection as they are from prophage induction? To answer this question without involving the additional complication of the lysis–lysogeny switch, we subjected hipA7 and wt strains to λcI60 infection (Figure 1D). The cI60 mutation allows infection and lysis, but prevents the formation of lysogens by disabling CI repression. Survival to λcI60 infection was measured by counting colony forming units (CFU) on plates, as well as by observation under the microscope. In contrast to the higher survival of hipA7 bacteria to prophage induction, after infection, we measured low bacterial survival in both the wt and hipA7 strains, to the rate of 0. 09% ± 0. 11% and 0. 04% ± 0. 03%, respectively. Under the microscope, we observed lysis in more than 90% of the cells, for both strains, over a total observation time of several hours. We found that the observed killing was due to the initial exposure to the phage during adsorption and not to a secondary infection process after plating (see Text S1). Thus, λcI60 phages are able to infect and kill persister bacteria, despite their dormancy. How do phages kill persister bacteria? Are they able to activate the lytic pathway during the persister' s dormancy, or do they initiate the lytic pathway only upon switching from dormant to growing cells? If the latter is true, the onset of the lytic pathway activation in persister bacteria should follow the same time dependence as the exit from dormancy. We followed the dynamics of lytic gene expression in single infected persister cells by tracking the GFP fluorescence expression from the λ late promoter pR′-tR′ fused to gfp. In addition, we simultaneously measured the dynamics of persisters' exit from dormancy. Initially, we identified persisters as follows: we subjected hipA7 λcI857KnR lysogenic bacteria to 30-min heat induction at 42 °C, as described above, in order to rapidly induce and kill normally growing bacteria. We then measured the exit of persisters from dormancy by identifying their first division with phase-contrast microscopy. Simultaneously, we measured the onset of fluorescence due to lytic pathway activation in persisters infected with lytic phages (Figure 1E). To overcome the immunity of λcI857KnR to superinfection, we used a different immunity phage, λimm434cI3003, that can superinfect λcI857KnR lysogenic bacteria and, similar to λcI60, does not form lysogens [23]. Figure 3A shows the time course of a typical experiment, in which we observed hipA7 λcI857KnR lysogens after 30-min exposure to 42 °C without infectious phages (upper panel), and with λimm434cI3003 superinfection (lower panel). Without superinfection, persister cells exited dormancy and formed microcolonies, as shown above (Figures 2A and 3A, upper panel); with superinfection, no survival was detected, but we observed a late onset of fluorescence in persister bacteria (Figure 3A, arrows in lower panel), indicating that the phage lytic cycle was activated. Using automated image analysis, we quantified the timing of onset of fluorescence activation. The histograms for superinfected and noninfected lysogens are presented in Figure 3B. A rapid activation of fluorescence was observed in both conditions, identifying normally growing bacteria. However, after about 1 h, persister cells that were superinfected began increasing their GFP expression from the pR′-tR′ promoter and lysed. These late-activation events of the lytic promoter lasted until the end of our experiments (Video S3). In Figure 3C, we now compare the late onset of fluorescence, in superinfected cells, to the timing of exit from dormancy, in persisters without superinfection. The correspondence of the timing of these events is striking. Furthermore, the percentage of single cells showing late GFP fluorescence activation was similar to that of noninfected persisters that exited dormancy and divided, within the same observation time (Figure 3D). These results show that only when infected persisters exit dormancy, activation of the phage genes starts and lysis follows. This suggests that, despite the large physiological differences between normally growing and stationary bacteria [24,25], phages are able to infect and eventually to lyse their hosts in both physiological states, showing the perfect adaptation of phages to their host. This behavior is reminiscent of the strategy observed for several phages that can infect starved bacteria, or in phages of sporulating bacteria. For example, under nutrients limitation, T4 phages can enter a state of “hibernation” and activate the phage program only once nutrients become available. Similar to pseudolysogeny, these phenomena might provide an explanation for the maintenance of lytic phages in the wild [26]. We have shown that although persister λcI857KnR lysogens can survive the transient heat induction by being in a dormant state, they cannot survive infection, since the effect of the phage comes into action once persisters switch to normal growth. The difference between the two stresses probably originates from the reversibility of CI857 inactivation under heat induction: bacteria that were dormant during exposure to 42 °C recover an active repressor at the end of the heat treatment [27,28] and therefore survive prophage heat induction. This view is supported by the increase in persistence to prophage induction with UV by photoreactivation, which reverses the UV damage. In contrast, the infection process in phages that do not form lysogens is irreversible: phages might be able to inject their genome into persister bacteria and activate the lytic pathway as soon as bacteria exit from dormancy. Thus, persisters enable differentiating between reversible and irreversible stresses. Our observation that persister cells can survive prophage induction implies that persistence should have a clear effect on the population dynamics of bacteria and phages. This can be shown in a model of the dynamics of two competing populations under cyclic exposure to prophage induction: a bacterial population with higher persistence should quickly overcome a low-persistence strain (Figure 4A). In order to test this prediction, we performed a competition experiment in which a high-persistence strain (hipA7) and a low-persistence strain (wt), both carrying the λcI857KnR prophage, were mixed and exposed to cycles of prophage-inducing conditions. The results of three independent competition experiments are plotted in Figure 4B; as predicted by the model, the high-persistence strain rapidly took over the low-persistence strain, showing the advantage conferred by persistence in prophage-inducing conditions. These observations, together with evidence of increased prophage induction in biofilms [29] and the role of prophages on biofilm development [30], suggest that persistence might have been selected under the pressure of prophage induction to benefit both bacteria and prophages. Whereas the degree of persistence was seen to dramatically affect population dynamics in prophage induction, it would seem that its effect on lytic infection should be insignificant, as we observed no increase in the survival of the high-persistence population. However, the extremely long delay observed in the lysis of the subpopulation of persister cells may affect the long-term dynamics of the interaction between phages and their hosts. In order to evaluate the effect of this delay, we introduced it in a mathematical model of the bacterial–phage interaction. Modeling of the population dynamics of the host–phage system is usually based on predator–prey models such as the Lotka-Volterra equations. However, the oscillations predicted from the model are not always found experimentally, and many systems end up either in extinction or stability. Several additions to the Lotka-Volterra equations have been studied, among them logistic growth and time delays. One class of models that has been extensively studied considers the stabilizing effect of spatial heterogeneity on the population dynamics [31,32]. In our case, we do not have spatial heterogeneity; however, the switching between persister and normal cells can be mapped on an effective “migration” model between two “patches” with different conditions. Patch 1, in which growth of the prey is fast (normal bacteria), and patch 2, in which growth of the prey is slow (persisters). Preys can travel between the patches with rates that correspond to the switching rates between persister and normal states. Predators (phages) diffuse rapidly between the patches. A thorough analysis of the equilibrium points of the general equations describing similar processes for symmetrical diffusion between patches as well as their stability has been done in [33]. The analysis for nonsymmetrical diffusion, more appropriate for the description of persistence, is beyond the scope of this work and left for future work. We show below simplified dynamics, illustrating the potential effects of Type I persisters on the host–phage interaction. Type I persisters are generated at stationary phase. When diluted in fresh medium, they switch to normal growth on a time scale of 10 h in our system. No additional persisters are formed. The equations describing this system take into account the unidirectional migration from persister to normal states: where n represents normal bacteria; p, persister bacteria; pi, infected persister bacteria; μ, growth rate of normal bacteria; b, switching rate from p to n; α, rate of phages attachment; λ, phages; burst, burst size, and d, dilution/death rate of phages Persisters (p) are as susceptible to infection as normal bacteria (n). Infected persisters (pi) release phages only upon switching to the normally growing state, at a rate constant b, as measured in our experiments. Here, we model a phage infection starting from a small number of phages, typically of the order of the burst size, as expected from a typical infection. Free phages are rapidly diluted out. Simulations of infected populations without and with persisters are shown in Figure 5A and 5B, respectively. For similar initial conditions, the bacterial population undergoes very large oscillations in the absence of persisters, whereas these oscillations are partially reduced by the slow release of phages by persisters. Type I persisters generated at stationary phase alter the dynamics by continuously releasing phages on a slow time scale, thus preventing sustained large oscillations. For wt E. coli, a Type I persister fraction of nearly 1% has been measured [11]. This fraction was sufficient to reduce the amplitude of the oscillations by nearly two orders of magnitude in our simulations. This indicates that population dynamics of wt populations can be affected by the host heterogeneity and that this heterogeneity should be taken into account in complete models of the predator–prey interaction. We have shown that the nongenetic individuality in the exit from stationary phase found in populations that persist intensive antibiotic treatments can dramatically affect the interaction between bacteria and λ-phages. Dormancy protects persister cells from prophage induction. Thus, persistence to prophage induction could prevent the eradication of lysogenic bacterial populations by prophage induction, and benefit both bacterial hosts and phages in conditions that trigger prophage induction, such as in biofilms and in stressful environments. However, we have shown that lytic phages are able to infect persistent cells, wait for their switching back to the normally growing state, and then eliminate them by lysis. This implies that bacteria persistent to antibiotic treatments might be targeted by phages [34]. Furthermore, we show that the heterogeneity of the host can significantly alter the predator–prey population dynamics and might be relevant in various other systems. Our results suggest that the ability of phages to infect bacteria under various physiological conditions enables them to benefit from the host heterogeneity under both lysogenic and lytic conditions. Further studies on the influence of nongenetic individuality on the host–phage interaction might shed light on the coevolution of this system. LB was used as growth medium for all experiments. Serial dilutions were performed in this medium for bacterial plating. Ampicillin (Sigma) was used at 100 μg/ml; tetracycline (tet) at 12. 5 μg/ml; and kanamycin at 30 μg/ml. The penicillinase (Sigma) stock was suspended at 5,000 U/ml in sterile water and kept at 4 °C. The stock solution was diluted 1: 40 in water and filter sterilized before use. The plasmid carrying pR′-tR′-gfp fusion was described in the work of Kobiler et al. [20]. Briefly, a plasmid with pBR322 origin of replication and ampicillin resistance was used; the pR′-tR′ region from the λ genome was amplified by PCR using primers 5′-CGCGGATCCGATGCAAAGGCGTCGGCTATTC and 5′-GCGTCTAGACGCCGACTCTTCACGATTATC, and then inserted upstream of the GFP coding sequence (CDS) on the plasmid. Bacteria and phages strains are summarized in Text S1. Single colonies were diluted into fresh LB with the appropriate antibiotics. For infection experiments, 0. 2% maltose was added. Cells were grown overnight (O/N) at 32 °C, with shaking. Cultures were diluted 1: 50 into fresh LB with the appropriate antibiotics and grown for 1 h. To keep growth conditions constant between the different experiments, cultures were transferred to a 42 °C bath for 30 min (unless indicated otherwise), with shaking. MG1655 (λcI857KnR) and MG1655A7 (λcI857KnR) lysogens were incubated at 42 °C for the required time. At the indicated time points, samples were taken and plated at serial dilutions on LB-kanamycin (LB-Kan) plates. Experiments were repeated at least three times. The MG1655A7 high-persistence strain and the wt strain MG1655 both carrying the λcI857KnR lysogen were mixed in a ratio of 1: 20 and exposed to cycles of prophage-inducing conditions. The total number of surviving bacteria was determined by plating and CFU counting on LB-Kan plates. The number of MG1655A7 bacteria in the population was determined by plating and CFU counting on LB-Kan+tet plates, before and after each heat cycle. Each prophage induction cycle was followed by growth O/N in culture and freezing at −80 °C in 15% glycerol. In order to test the neutrality of the tetracycline resistance marker, a MG1655 λcI857KnR lysogenic strain with the same tetracycline resistance marker was constructed (MG21 λcI857KnR). The competition experiment in the same conditions showed the neutrality of the marker. MG1655/pR′-tR′-GFP and MG1655A7/pR′-tR′-GFP bacteria were grown as described above. Cells were concentrated twice by centrifugation at 5,000 g for 5 min and resuspended in fresh LB to an approximate concentration of 1010 cells/ml. A total of 108 cells were transferred into an Eppendorf tube, and 109 phages (or the same volume of LB for the control) was added. After 30-min incubation on ice for adsorption, the infected cells were transferred to 37 °C for 5 min to facilitate DNA injection. Serial dilution plating was carried out and CFU determined. Alternatively, two rounds of centrifugation at 2000 g for 10 min and washing were done before plating. Observation chambers, A polydimethylsiloxane (PDMS) square mold was cut out of cured Sylgard 184 (Dow Corning) layer (thickness: approx. 2 mm). The mold was filled with melted LB-agarose with appropriate antibiotics. Bacteria (1 μl) were put on a coverslip (#1. 5) and covered with the solidified LB-agarose inside the PDMS mold. The whole chamber was sealed with a thin layer of PDMS to avoid dehydration of the LB-agarose, without blocking oxygenation. Time-lapse microscopy. The PDMS chambers were monitored using a Leica DMIRBE inverted microscope system with incubator box (LIS), automated stage, and shutters. Autofocus and image acquisition was done using custom macros in ImagePro/Scope-Pro (Media Cybernetics) to control the microscope, stage, shutters, and camera. Multiple different locations were monitored in parallel for phase contrast and fluorescence on the same chamber. Images were acquired using a 40×/1. 5 or 63× long-range air objective and a cooled charge-coupled device (CCD) camera (Orca; Hamamatsu) and processed with ImagePro and ImageJ (http: //rsb. info. nih. gov/ij/). Fluorescence images were acquired with minimal excitation to minimize bleaching and photodamage. This was checked in situ by exposing one of the multiple locations as control to only 1/10 of the integrated illumination dose. No significant difference was noticed at the control location. Unless otherwise indicated, microscopy was carried out at 32 °C. Transient heat induction under the microscope was done on a custom-made heated stage insert warmed to 42 °C. Single cells microscopy experiments analysis. Analysis of the videos was done with ImageJ, by tracking cells in sequential frames and measuring their mean fluorescence. Simulations of Lotka-Volterra equations with and without persisters were performed using Matlab software (The MathWorks).

Persistence of subpopulations of bacteria to antibiotic treatments is a major problem in recurrent infections. Unlike resistance, which is passed on to the next generations, persistence is a transient trait characterized by slow growth or dormancy. It has been suggested that the existence of both persister and non-persister bacteria within a given population might constitute a general strategy that bacterial populations use to cope with an ever-changing, stressful environment. Here, we studied the influence of persistence on the interaction between bacterial populations and viruses that infect bacteria, called phages. We found that persistence provides a clear advantage for lysogenic bacteria—in which the phage DNA has integrated into the host DNA but remains mostly inactive—as they enter the reversal of this state, typically in response to environmental stress. This suggests that persistence might have evolved in lysogenic bacteria under stressful conditions. In contrast, persister bacteria do not survive infections by lytic phages—which replicate until they cause the host cell to burst—any better than non-persister bacteria, but release the infectious phages on a significantly longer time scale. Mathematical analysis reveals that this host heterogeneity might substantially affect host–phage population dynamics and could be relevant for other predator–prey systems.

Abstract Introduction Results and Discussion Materials and Methods

microbiology

2008

Nongenetic Individuality in the Host–Phage Interaction

Geographic partitioning is postulated to foster divergence of Helicobacter pylori populations as an adaptive response to local differences in predominant host physiology. H. pylori' s ability to establish persistent infection despite host inflammatory responses likely involves active management of host defenses using bacterial proteins that may themselves be targets for adaptive evolution. Sequenced H. pylori genomes encode a family of eight or nine secreted proteins containing repeat motifs that are characteristic of the eukaryotic Sel1 regulatory protein, whereas the related Campylobacter and Wolinella genomes each contain only one or two such “Sel1-like repeat” (SLR) genes (“slr genes”). Signatures of positive selection (ratio of nonsynonymous to synonymous mutations, dN/dS = ω > 1) were evident in the evolutionary history of H. pylori slr gene family expansion. Sequence analysis of six of these slr genes (hp0160, hp0211, hp0235, hp0519, hp0628, and hp1117) from representative East Asian, European, and African H. pylori strains revealed that all but hp0628 had undergone positive selection, with different amino acids often selected in different regions. Most striking was a divergence of Japanese and Korean alleles of hp0519, with Japanese alleles having undergone particularly strong positive selection (ωJ > 25), whereas alleles of other genes from these populations were intermingled. Homology-based structural modeling localized most residues under positive selection to SLR protein surfaces. Rapid evolution of certain slr genes in specific H. pylori lineages suggests a model of adaptive change driven by selection for fine-tuning of host responses, and facilitated by geographic isolation. Characterization of such local adaptations should help elucidate how H. pylori manages persistent infection, and potentially lead to interventions tailored to diverse human populations. Helicobacter pylori chronically infects billions of people worldwide, typically for decades. Most H. pylori reside on gastric epithelial cell surfaces and in the overlying mucin layer, a tissue that turns over rapidly, is infiltrated by inflammatory cells after infection, and is buffeted by gastric acidity on its luminal side. The gastric mucosa is hostile to most bacterial species, and constitutes an unstable niche to which only the Helicobacters among bacterial taxa have become well adapted, in part perhaps through effective management of host responses to infection [1–4]. Great genetic diversity, geographic differences in predominant genotypes, and rapid evolvability are hallmarks of H. pylori populations [5–9]. Independent isolates generally differ by some 2% or more in DNA sequence of any metabolic (housekeeping) gene, with most such differences being synonymous (protein sequences unchanged). Phylogenetic analyses of H. pylori housekeeping gene sequences revealed differences in predominant genotypes between East Asian, European, and African populations that are far greater than those seen with most other pathogens. Such patterns reflect a combination of mutation, recombination, selection to retain gene function, and random genetic drift, which itself likely stems from H. pylori' s highly localized (preferentially intrafamilial; nonepidemic) mode of transmission, and a resulting relative lack of H. pylori gene flow between well-separated human populations [6,7, 10–12]. Correlated with this geographic partitioning of H. pylori populations are striking differences in predominant clinical consequences of infection. To illustrate, duodenal ulcer, which is typically associated with excess gastric acidity (hyperchlorhydria), is far more common in India than in Japan, whereas gastric cancer, which is typically associated with hypochlorhydria, is far more common in Japan than in India [13,14]. The near universality of H. pylori infection until very recently, the extraordinary chronicity of infection, and geographic differences among H. pylori populations all have contributed to an idea that H. pylori may have co-evolved with its human hosts [15]. Geographically isolated populations are also more likely to adapt to differences in local environment [16,17]. Human genetic or physiological traits that diverged during our evolution, that differ geographically, and that are important to H. pylori could have selected for adaptive changes in cognate H. pylori genes. The virulence-associated cagA, and vacA genes provide examples in which evolutionary dynamics are likely to have been shaped by local differences in host physiology. CagA and VacA proteins each enter target cells and affect several normal cellular signal transduction pathways, with strengths and specificities that vary geographically. For example, East Asian and Western type CagA proteins differ most in sequences of domains responsible for phosphorylation and in resulting interactions with host SHP-2 phosphatase, an intracellular regulator of various cell proliferative, morphogenetic, and motility signaling pathways [18,19]. Similarly, highly active “s1, m1”–type alleles of the vacA toxin gene predominate in Japan and Korea, whereas nontoxigenic s2, m2–type alleles are common in the West [20,21], and a recombinant s1, m2 form predominates in coastal China [22]; the “m” region of VacA determines the cell type specificity of toxin action. We suggest that these geographic differences reflect types of selection pressures that predominate (d) in the various human populations, either currently or in centuries past, superimposed on the random genetic drift that figures so importantly in geographic partitioning of housekeeping genes; and that such patterns may be common among genes whose products interact with host components. Furthermore, the extraordinary chronicity of H. pylori infection suggests a possible need for effective management and potentially even exploitation of host responses. For example, although inflammatory responses help protect potential hosts against casual pathogen encounters, H. pylori is thought also to use metabolites leached from inflammation-damaged host tissues for its nutrition [23]. In addition, many strains use host sialylated glycolipids, synthesized during the inflammatory response, as receptors for adherence [24]. In this framework, much of H. pylori-induced gastric pathology might reflect how host signaling pathways are modulated by contact with the bacterium or its secreted products. Sequenced H. pylori genomes contain a gene-family whose encoded proteins are likely secreted, and contain two or more copies of a degenerate 34–36 amino acid repeat motif that is characteristic of eukaryotic “Sel1” proteins, which themselves help regulate diverse signal transduction pathways [25]. The proteins bearing this repeat are typically built of several consecutive α/α motifs, the antiparallel α-helices of the motifs being connected by a short loop [26]. Five of the nine members of this “SLR” (for Sel1-like repeat) protein family are rich in cysteine residues and had been designated as “Helicobacter cysteine rich proteins” (Hcp) [27,28]; The α-helices of each SLR repeat are bridged by a disulfide bond, which is a unique feature of Hcps [26]. Although the in vivo function of H. pylori SLR proteins is not known, some Hcps bind β-lactam compounds [27,29], which suggests possible interactions with immunomodulatory peptidoglycan fragments, that could affect the innate immune response [30]. High antibody titres against four SLR proteins [Hp0211 (HcpA), Hp0235 (HcpE), Hp0336 (HcpB), and Hp1098 (HcpC) ] were found in H. pylori–infected people [28], indicating in vivo expression and immune recognition. Furthermore, recombinant HcpA elicited IL12-dependent IFN-γ secretion in a naïve mouse splenocyte model [30], and HP1117 elicited protective antibodies during mouse infection [31]. Only one or two slr homologs are found in members of the closely related Campylobacter and Wolinella genera, whereas strains of H. acinonychis (from big cats) and of the nongastric mouse pathogen H. hepaticus (implicated in liver cancer) contain seven and six slr homologs, respectively (Figure 1A). It is appealing to imagine that this expanded family of secreted proteins affects bacterial–host interactions during chronic Helicobacter infection of mucosal tissues. Here, we posited that if geographically isolated H. pylori populations had adapted to local differences in host physiology caused by factors such as host nutrition, genotype, or infection by other pathogens that in turn impacted on responses to H. pylori, these adaptations would leave an imprint of natural selection on the affected genes, superimposed on H. pylori' s overall population genetic structure. For protein-coding genes, selection pressures and adaptive evolution can be detected and examined by comparing rates of fixation of synonymous (silent; dS) versus nonsynonymous (amino acid altering; dN) mutations in the population [32,33]. The ratio dN/dS (= ω) indicates whether amino acid change is unaffected, inhibited, or promoted by natural selection. Considering that most synonymous substitutions have little if any effect on fitness, dS is often equated to the rate of neutral nucleotide substitution. Under neutral evolution, one would expect dN and dS to be equal (ω = 1), and functionally critical genes (e. g. , housekeeping genes, responsible for intracellular metabolic functions) to show very low dN (ω < 1) [32]. In certain genes, however, nonsynonymous substitutions are in excess (ω > 1) because changes in their encoded proteins are advantageous and thus have been selected in particular environmental contexts. This is often termed positive selection (sometimes also called diversifying or Darwinian selection). Many such substitutions are likely to have been selected specifically to change the activity or structure of encoded proteins [34]. Since differences in local conditions (e. g. , host features) can lead to geographic differences in patterns of selection [16], evolutionary rates of amino acid substitutions are likely to vary among H. pylori lineages; and an elevated dN (ω >>1) in any specific lineage would indicate adaptive evolution. Such adaptive evolution tends to be episodic, in that it operates sporadically, and affects only a few amino acid positions in the protein [34]. Consequently, methods that estimate average dN and dS (summed over all codons and all lineages) often fail to detect adaptive evolution [33]. Here, we applied codon-based models of sequence evolution in conjunction with maximum-likelihood (ML) computational methods [35,36] that are particularly useful for detecting adaptive changes at specific sites in proteins, to study the evolutionary dynamics of the H. pylori slr gene family. We found that most slr family members had experienced positive selection, and accumulated adaptive mutations in specific H. pylori lineages, preferentially affecting specific surface-exposed sites in encoded proteins. These outcomes suggested selection for management and fine-tuning of host responses during chronic infection. Our results illustrate the utility of population-based phylogenetic strategies for identifying human population-specific adaptive determinants of H. pylori. This study began with subtractive hybridization (as in [37,38]) to find genes or alleles that differed markedly between representative Japanese versus Western strains. One recovered clone contained a fragment of gene hp0519. Subsequent sequencing of hp0519 alleles from representative strains identified two in-frame deletions: a 24-bp segment that was absent from the Japanese strain (Δ24) but present in US reference strain J99 (nt 133–156), and a 15-bp segment in the Japanese strain that was absent from J99 (Δ15) (nt 640–654). PCR tests indicated that 70 of 87 Japanese strains carried Δ24 15+ type alleles, and only 14 carried the reciprocal 24+ Δ15 type allele, whereas 45 of 47 Spanish strains tested carried 24+ Δ15, the allele type that was uncommon in Japanese strains. Remarkably, 24 of 28 Korean strains tested also carried 24+ Δ15 type alleles, not the Δ24 15+ type that predominated in Japan. This difference in hp0519 pattern seemed extraordinary because Japanese and Korean strains were closely related in sequences of all other slr genes tested (see below); they were also closely related in sequences of housekeeping genes (Figure S2B; and Dailide and Berg, unpublished), which should be subject only to purifying selection to maintain function within the bacterial cell. Such relatedness was expected given the proximity of Japan and Korea and the shared history of their peoples. Inspection of the Hp0519 protein sequences using the SignalP (http: //www. cbs. dtu. dk/services/SignalP) and SMART (simple modular research tool) (http: //smart. embl-heidelberg. de/) programs identified an N-terminal signal sequence and three SLRs with 40% sequence similarity to the human Sel1L protein (Tables S1 and S2; Figure S1A). Hp0519 belongs to the cluster of orthologous group (COG0790), which contains eight other family members in the genome of H. pylori strain 26695 ([39], and seven in the genomes of strains J99 [40] and HPAG1[41]) (http: //www. ncbi. nlm. nih. gov/COG) (Figure 1A). Each COG0790 family member is predicted to encode a secreted protein with two or more SLRs (Table S2). Seven of these slr genes are present in each of the three sequenced genomes—hp0160 (jhp0148/hpag0158), hp0211 (jhp0197/hpag0212), hp0235 (jhp0220/hpag0238), hp0519 (jhp0468/hpag0493), hp0628 (jhp0571/hpag0610), hp1098 (jhp1024/hpag1036), hp1117 (jhp1045/hpag1055) — (prefixes “hp” in strain 26695; “jhp” in strain J99; “hpag” in strain HPAG1) (Figure 1A). Each genome also contains strain-specific slr genes: hp0336 in 26695, and jhp0318 (similar to hpag0339) and jhp1437 in J99. Reciprocal BLAST analysis revealed sequence similarities with domains of human Sel1L, ranging from 38% with jhp1437 to 51% with hp1117 (Figure S1), an intriguing pattern, even though such homologies do not by themselves demonstrate functional equivalence. A strain of the related H. acinonychis encodes seven SLR homologs [42], six of which are nearly identical in amino acid sequence to SLRs of H. pylori (Figure 1A); this suggests either equivalent selection of slr gene–related function in these two species or recent interspecies transfer between H. pylori and H. acinonychis, which occurs readily in culture or in vivo [43]. A strain of the related nongastric pathogen H. hepaticus encodes six SLR homologs [44], which are relatively less related to those of H. pylori. In contrast, strains from genera most closely related to Helicobacter, Campylobacter, and Wolinella [45] each contain only one or two slr genes (Figure 1A). The slr genes are organized into discrete repeating units, which should be prone to duplication events [32]. Amino acid identities of 25%–70% are seen among SLR motifs in different members of this protein family in H. pylori (Table S3). Three of the six H. hepaticus slr genes have close homologs in H. pylori (hh1827 and hp0235; hh0718 and hp0628; and hh0816 and hp0519). These features suggest a possible Helicobacter-lineage–specific slr gene family expansion (duplication) after Helicobacters diverged from Campylobacter and Wolinella, and near the time of H. pylori and H. acinonychis versus H. hepaticus divergence. Also tenable is a model of separate gene family expansions in H. pylori and H. hepaticus, and even in H. acinonychis, since the corresponding slr homologs have different chromosomal locations (flanking genes) in these three species (unpublished data). When a gene family' s expansion is adaptive, the sequences of individual members should reflect selective forces that operated during and after this expansion. In general, paralogs that subsequently suffer inactivating mutations tend to be lost from the population over time [32,46]. More important evolutionarily are the paralogs that diverged, and acquired new functions, or optimized or subdivided complex ancestral functions. Purifying selection (ω < 1) predominates in the evolution of genes whose roles remained constant, whereas positive selection (ω > 1) predominates in cases of genes whose functions have diverged [32,46]. Accordingly, we determined ω values to test if slr gene family expansion was accompanied by functional divergence of paralogs, and, more generally, examined selection pressures that operated on this gene family. We applied two codon-based models of sequence evolution to obtain ML estimates of selective pressures during slr gene family expansion, starting with sequences of the eight and nine slr genes in the three sequenced H. pylori genomes (Figure 1B). The simplest one-ratio model assumes the same ω for all branches, whereas the free-ratio (FR) model allows ω to vary among branches [33]. These models are nested, and hence can be compared using a standard likelihood ratio test (LRT). ML estimates were computed under varying conditions using different sets of initial values for ω and κ to confirm optimal algorithm convergence. Regardless of underlying assumptions, the FR model fit the data significantly better than the M0 model (−InL (FR) = 11286. 639; −InL (M0) = 11356. 182; χ2 = 140. 062, degrees of freedom = 47; p < 0. 00001; initial ω = 2, κ = 2; equilibrium codon frequencies estimated as free parameters). This suggested that ω varied significantly among individual branches of the slr gene-family phylogeny (Figure 1B; Table S4). Strong positive selection was evident in several branches, again indicating that slr gene family expansion was driven by selection for functional divergence among paralogs. Given the multiple slr genes in these three Helicobacter species, slr family expansion might have occurred well before H. pylori became widespread in humans: that is, in ancient non-human hosts, possibly reflecting generalized selection pressures during mucosal colonization. Alternatively, because distributions of slr genes in the three sequenced H. pylori genomes vary, these expansions could have been more recent, especially given the ease with which gene duplications can arise [47,48], possibly facilitated in H. pylori by its lack of a MutHSL DNA repair system [49], and/or induced by reactive metabolites generated during infection [50]. In either case, evidence of functional divergence among H. pylori slr homologs driven by positive selection makes it appealing to imagine their products affecting traits important for H. pylori mucosal colonization. With this perspective, and prompted by the nonrandom geographic distribution of hp0519 indels, we determined DNA sequences of six slr genes present in most East Asian, Western European, and African H. pylori strains using isolates from a representative strain collection. Sequences of Japanese and Korean alleles of housekeeping genes are typically intermingled in the same clusters (Figure S2B; and Dailide and Berg, unpublished). Therefore, we asked if hp0519 single nucleotide polymorphisms (SNPs) also differed geographically by sequencing a 322-bp hp0519 segment internal to these “24” and “15” indels in 78 strains. These internal hp0519 sequences contained many SNPs, which fell into separate Japanese (n =20) and Korean (n = 16) allele clusters (Figure 2A), in accord with the PCR-based indel results. Permutation–randomization tests of this 322-bp internal segment suggested great genetic differentiation, perhaps reflecting separation of Japanese island and Korean mainland populations (FST = 0. 5, p < 0. 001; Table S5), and a critical difference in the forces that had operated in these two regions. To better understand the evolutionary forces driving this divergence, we next determined full-length hp0519 sequences (approximately 873 bp) from African, European, and East Asian strains (n = 27), chosen randomly from the larger dataset shown in Figure 2A, and reconstructed an ML phylogeny with these data. This revealed Japanese–Korean allele separation (bootstrap support = 100), as expected (Figure 3A). Further pairwise permutation-randomization tests of all 27 full-length hp0519 sequences also showed genetic differentiation among H. pylori subpopulations in various geographic regions (FST > 0. 5, all comparisons; Table S5). Additional pairwise comparisons of five Korean and five Japanese full-length hp0519 sequences revealed 56 fixed differences (sites at which all sequences in one population differed from all sequences in a second population), versus 33 polymorphisms shared between them. Both sets showed unique polymorphisms (i. e. , sites polymorphic in one set, monomorphic in the other; 68 in Japanese and 42 in Korean, respectively). This inverse relationship between fixed and shared polymorphisms suggests either ancient separation of Korean and Japanese hp0519 alleles or selection for accelerated accumulation of SNPs in at least one population. Application of the McDonald–Kreitman test for adaptive evolution showed that the ratio of nonsynonymous changes to synonymous changes among fixed differences (48/8) was significantly higher than that among polymorphic differences (68/42) (p < 0. 001, Fisher' s exact test and G-test) (Figure 2B). This outcome suggested that the accumulation of nonsynonymous substitutions in Japanese versus Korean hp0519 alleles had been driven by positive selection. In contrast, equivalent tests of hp0519 sequences from other populations suggested divergence between them due mostly to random genetic drift. ML phylogenies of the hp0518 and hp0520 genes that flank hp0519 from six representative Korean and six representative Japanese strains did not show any distinction between these two East Asian populations (Figure S2A). This indicated that divergence of Japanese versus Korean hp0519 alleles was due to selection on hp0519 itself, not due to linkage to an even more highly selected gene. Selection pressures on hp0519′s individual codons and branches of its phylogenetic tree were next studied in detail using three groups of codon-based models of sequence evolution and ML-based LRTs: 1) site-specific models (SSMs), which examine variation in selection pressures across codons and assume a single ω across the phylogeny [35]; 2) lineage-specific models (LSMs), which allow ω to vary among lineages, while assuming a single rate across all codons [51]; and 3) lineage-site–specific models (LSSMs), which allow ω to vary both among codons and across the phylogeny [36]. SSMs confidently identified 18 sites under positive selection (ω2 = 3. 515; Bayesian probability > 0. 99) (Figure 2C; Table S6A), which suggested different selective pressures at different sites in hp0519. [Equivalent site-specific positive selection was also detected in hp0519 codons by the single-likelihood ancestor counting (SLAC) and fixed-effects likelihood (FEL) methods hosted at http: //www. datamonkey. org (unpublished data) ]. Previous work has shown that codon-based models implemented, in M7 and M8 models, in particular, are usually not adversely affected by recombination in bacterial datasets [52–54]. However, because the ML approach used here explicitly assumes a phylogenetic tree when estimating selection pressures, we also assessed if the extensive recombination typically seen in H. pylori populations could have produced a false-positive signal for positive selection. This entailed repeating the analysis assuming that sequences were linked by a “star” phylogeny, where lineages diverge simultaneously from a single root node; this removes the effect of phylogenetic history, including recombination events from the outcome. This analysis again indicated positive selection (p < 0. 0001 for M3 versus M2 and M8 versus M7), with higher ω values under the M3 and M8 models, and with the same sites usually in the positively selected class as in the original analysis (Table S6C). Next, a two-ratio LSM model, M2J (ωJ for Japanese foreground branch; ωR for all other branches (background lineages) ) was constructed, to test a priori whether ωJ was significantly different from ωR. M2J fit the data significantly better than M0 (p < 0. 0001; Table S6B) and suggested that divergence of Japanese hp0519 lineage was driven by positive selection (ωJ = 1. 6). In accord with this, the FR model, which assigned an independent ω for each lineage, also fits the data significantly better than M0 (p < 0. 0001; Table 6B). This suggested that hp0519 alleles had been subject to significantly different selective pressures in different lineages (Figure 3A). To identify rapidly evolving codons in the Japanese lineage, we constructed two LSSMs, M2JM2 and M2JM3, which assigned a different value to the Japanese lineage (ωJ) and compared them for fit against M1 and M3 SSMs, respectively. Both LSSMs confidently identified 26 sites that had been strongly selected (ωJ >> 20; Table S6B) in the ancestral Japanese lineage. This confirmed that divergence of Korean and Japanese hp0519 alleles was driven by strong positive selection that had favored specific adaptive changes in Japan. These differences are not explained by models invoking founder effects or random genetic drift alone. We suggest, rather, that this divergence reflects a condition unique to H. pylori in the Japanese islands at some recent evolutionary time. Possibilities include differences between the islands and the East Asian mainland in prevalence of other pathogens or parasites that affect host responses to H. pylori [55,56] and how H. pylori can best manage them; or host genotype, diet or nutrition, or sociocultural features that could also affect host responses to infection. The divergent Japanese-type hp0519 alleles might have existed at low frequency before being strongly selected in Japan. Alternatively, they might have arisen by more recent stepwise mutation and selection, perhaps only starting when rice-based agriculture was brought to Japan some 2,300 years ago, along with changes in diet, lifestyle, risk of infection, etc. To examine adaptive evolution in the Japanese hp0519 lineage in a protein structure–function context, we applied several methods for comparative secondary structure analysis and modeled Hp0519 three-dimensional structure on the experimentally determined crystal structures of its homologs HcpB (Hp0336) [57] and HcpC (Hp1098) [58]. Twenty four of 26 (92%) sites under positive selection were in regions near but not overlapping the predicted SLRs, with 19 of 26 (73%) between the second and third SLR. Two of these adaptive residues (44 and 216) were located within the Δ24 or Δ15 indels, suggesting that these regions were also important functionally, whereas the other adaptive residues were separate from these indels. Strong conservation of Hp0519 SLR motifs suggests that they are critical for protein function. Secondary structure comparisons revealed striking differences between Japanese-type Hp0519 proteins and all others in amino acid charge distribution, hydrophilicity, surface probability, and antigenic index (Figure 3B). Although modelling of Hp0519 structure was complicated by significantly lower sequence conservation and the presence of indels, threading analysis suggested that Hp0519 folds into a three-dimensional structure similar to that of HcpB and HcpC (z-score = 31. 19; Table S7). Secondary structure analysis localized most sites under positive selection to loop regions, which are generally surface-exposed and potentially positioned to contact cognate host proteins. Data supporting the computational prediction that Hp0519 protein is secreted was obtained by generating plasmids that encode recombinant Hp0519-FLAG fusion proteins with and without predicted signal peptides, and expressing them in BL21 DE3 Escherichia coli cells. Tests using the α-FLAG M2 antibody showed that most Hp0519 encoded by full-length hp0519 (predicted signal peptide intact) was secreted into the culture supernatant, whereas that encoded by an engineered hp0519 variant that lacked the signal peptide coding sequence remained with the cell pellet (Figure S3). Expecting equivalent signal peptide–dependent secretion of Hp0519 protein in H. pylori, we propose that many of the surface-exposed adaptive changes identified computationally may affect the strength or specificity of Hp0519′s interaction with cognate receptors or other host components. Other surface-exposed Hp0519 residues might have evolved under host immune selection, in particular if certain Hp0519-specific immune responses could inhibit H. pylori growth or persistence. In this last scenario, the directional nature of positive selection, specifically in the Japanese lineage, would suggest unique immunological pressures in Japan at some point in H. pylori' s evolution. Finally, some of the observed adaptive changes might have been context-dependent, compensating for deleterious mutations elsewhere in the same gene [59–61], which themselves could have accumulated by chance (drift), or been specifically selected at an earlier time, if different selective pressures operated then. To learn if selection for amino acid sequence change in different populations was common to the slr gene family, in general, versus specific to hp0519, we sequenced five additional slr family members present in nearly all H. pylori strains (hp0160, hp0211, hp0235, hp0628, and hp1117) from ≥32 isolates variously from East Asia (Japan, Korea), West Europe (Spain, England), and Africa (South Africa, The Gambia). The slr gene phylogenies revealed separate clustering of African, European, and East Asian alleles, as is typical of housekeeping genes. In striking contrast to hp0519, there was no distinction between Korean and Japanese alleles of these other slr genes (Figures S4A–S7A and S8). Analysis of selective pressures by SSMs suggested that each slr gene, except hp0628, that was analyzed here had experienced different selective pressures at different amino acid sites during their evolution (Figure 4A–4D; Tables S8–S12). Homology-based structural modeling suggested that most sites under positive selection in these slr genes also encoded surface-localized amino acids (Figure 4E– 4F). This outcome supports the idea of adaptive sites in SLR proteins potentially affecting their interactions with cognate host components. LSMs suggested that the different hp0160, hp0211, and hp1117 lineages also evolved at different rates (Figure S4B–S7B). For example, positive selection was more evident in the East Asian and The Gambian hp0160 lineages (ω = 2. 26 and 2. 32, respectively), than in South African and European lineages (ω = 0. 29 and 0. 001, respectively). Similar lineage-specific positive selection was also seen in phylogenies of hp1117 and of hp0211; Hp0211 (HcpA) protein induces a cytokine IL12–mediated pro-inflammatory response [30]. Although SSMs suggested heterogeneous selective pressures and rapid evolution of certain codons in hp0235, LSMs indicated that ω did not vary significantly among the lineages studied (Table S11); a sampling of additional populations, however, might identify hp0235 lineages that had evolved more rapidly. The hp0628 gene was exceptional, in that SSMs and LSMs indicated evolution dominated by purifying selection (no codon class under positive selection), suggesting that it is functionally constrained, although ω did vary among codon sites (Table S12). Taken together, these outcomes illustrate that H. pylori evolution has not been strictly neutral genome-wide: that amino acid level polymorphisms fixed by positive selection are particularly abundant in certain key genes, perhaps often selected by features of the host milieu that vary geographically. Different sets of selective pressures operated in different members of the slr gene family and only on particular sites in any given gene. The many decades during which H. pylori can persist in the gastric mucosa despite inflammation and other host defenses, the differences among individuals and geographic regions in the intensity and specificity of host responses, and the changes in responses with age and as infection progresses, all coupled with the possibility of H. pylori exploiting these responses while avoiding clearance by them [23], suggests a need for active response management by the bacterium. It is especially in this framework that we studied the evolutionary dynamics of the H. pylori slr gene family. The members of this family encode secreted proteins with homology to the Sel1 group of eukaryotic regulatory proteins that, through their interaction with other eukaryotic proteins, affect cell proliferation, apoptosis, immune response, and intracellular trafficking [62,63]. We found that positive selection played a dominant role in slr gene evolution: that different amino acids were selected at particular sites in a given protein in different geographic areas, and that effects were more extreme for some slr genes than for others in the populations examined. We suggest that these findings be interpreted in terms of within- and between-host pathogen dynamics: that differences among hosts in physiologic traits had selected for changes in cognate microbial proteins (here H. pylori SLR proteins). Also to be included in this category, we suggest, should be selection for altered interaction with or recognition by critical components of the host immune system, which can differ among humans genetically or physiologically, reflecting factors such as infectious disease history, nutrition, stress, etc. Successful adaptation to these forces should also contribute to the extraordinary chronicity of H. pylori infection. Geographic differences in predominant H. pylori–associated diseases noted above [13,14,64] are likely due to multiple factors that may include infections by other pathogens that affect responses to H. pylori infection, diet, and human genotype [3,15]. It is tempting to consider these trends also being affected by several aspects of H. pylori genotype, including predominant allele types of slr genes. H. pylori isolates were obtained from phylogenetically distinct West European (Spain, England), African (The Gambia, South Africa), and East Asian (Japan, Korea) populations and have been described earlier [8,52]. All isolates were obtained from patients with gastric complaints who had undergone diagnostic endoscopy with informed consent. Standard methods were used for H. pylori propagation and storage. Primers used for PCR and sequencing are listed in Table S13. Standard methods were used for genomic DNA preparations, PCRs, and sequencing. Nucleotide diversities within and between population, FST and permutation tests, and McDonald–Kreitman tests were done with DNASP version 4. 1 (http: //www. ub. es. /dnasp). Phylogenetic reconstruction using slr gene sequences was performed with the ML approach implemented in PAUP*4b10 (http: //paup. csit. fsu. edu/). An ML phylogeny was reconstructed under the best-fit model [determined with MODELTEST, version 3. 7 (http: //darwin. uvigo. es/software/modeltest. html) ] by using a combination of heuristic searches and branch swapping to further optimize the likelihood score and substitution parameters. The significance of observed phylogenetic groupings was assessed by a bootstrap analysis performed with 1,000 replicates under the distance optimality criterion, while incorporating ML-optimized model and parameters. Phylogenetic trees were visualized with TreeView version 1. 6. 6 (http: //taxonomy. zoology. gla. ac. uk/rod/treeview. html). The selective pressures operating on H. pylori slr genes were measured using an ML method that takes into account the sequence phylogeny and assesses the fit to the data of various models of codon evolution that differ in how ω varies across the sequence or across the phylogeny [33]. Three classes of codon-based analysis were used in this study: (a) SSM analysis, whereby models (M0, M1, M2, M3, M7, and M8) assume a single ω for all branches of the tree, but allow ω to vary among individual codon sites, thereby providing a measure of heterogeneity in selection pressures acting across the gene sequence [35]; M7 and M8 perform robustly even when recombination has occurred; (b) LSM analysis, wherein models (FR, two-ratio, etc.) assume that ω varies among individual branches of the phylogeny, but that all codon sites are under the same selective pressure, thereby providing a measure of selective pressures acting on the gene in different lineages [51]; and (c) LSSM analysis, which allows ω to vary simultaneously among sites and lineages [36]. Positive selection was inferred when codons with ω of >1 were identified and the likelihood score (−InL) of the codon substitution model in question was significantly higher than the likelihood of a nested model that did not take positive selection into account. The probability that a specific codon belonged to the neutral, negative, or positively selected class was calculated by using Bayesian methods implemented in PAML version 3. 14 (http: //abacus. gene. ucl. ac. uk/software/paml. html). Multiple runs, assuming different initial ω and κ values, and different models for estimating equilibrium codon frequencies (calculated from the average nucleotide frequencies at the three codon positions tables (F3X4) or used as free parameters) were analyzed for each gene to verify the convergence optima for each model. Homology modeling of Hp0160, Hp0211, Hp0235, Hp0519, and Hp1117 started with a multiple sequence alignment including the protein sequences of the template structures HcpB (1KLX) [57] and HcpC (1OUV) [58] using program CLUSTALW. Due to the modular architecture of Hcps, different superpositions of HcpB and HcpC are possible and meaningful. This molecular feature increased conformational space and allowed modeling of protein sequences that were significantly longer than the template structures. The resulting structure-based sequence alignment was merged with the multiple sequence alignment. The alignment was manually curated taking into account the predicted secondary structure [program JPRED (http: //www. compbio. dundee. ac. uk/∼www-jpred) ]. Homology models were generated using program MODELLER (http: //salilab. org/modeller/modeller. html). To identify distant structural homologues of Hp0519, its protein sequence was threaded against databases of protein structures using the programs FUGUE (http: //www-cryst. bioc. cam. ac. uk/fugue) and LOOPP (http: //cbsu. tc. cornell. edu/software/loopp/). Figures were generated with PYMOL (http: //www. pymol. org). GenBank (http: //www. ncbi. nlm. nih. gov/Genbank) accession numbers for sequences generated in this study are from EF372636 to EF372923.

Helicobacter pylori is a genetically diverse bacterial species that infects billions of people worldwide, typically for decades. Long-term infection is a major risk factor for stomach ulcers and cancer, although most infections are benign, and the risks of various disease outcomes vary markedly among human populations. Analyses of housekeeping genes, whose encoded proteins perform normal cellular metabolic functions, had established that H. pylori strains from different geographic regions differed in their DNA sequences. Here, we analyzed the H. pylori slr multigene family that encodes up to nine secreted proteins (called SLR proteins) quite similar to the human protein Sel1. We showed that most members of the H. pylori slr gene family evolved significantly more rapidly than normal housekeeping genes. Different amino acids were selected in different H. pylori lineages, often on exposed surfaces of SLR proteins where they were potentially positioned to interact with host components. We propose that these amino acid differences affect the SLR protein function, likely contributing to H. pylori' s adaptation to local differences in human stomach physiology. Further characterization of H. pylori proteins with lineage-specific differences in amino acids should improve understanding of geographic differences in H. pylori–host interactions and human disease, and of the interplay between different evolutionary forces in natural populations of any species.

Abstract Introduction Results/Discussion Materials and Methods Supporting Information

infectious diseases microbiology computational biology evolutionary biology molecular biology genetics and genomics eubacteria

2007

Helicobacter pylori Evolution: Lineage- Specific Adaptations in Homologs of Eukaryotic Sel1-Like Genes

Wolbachia bacteria are now being introduced into Aedes aegypti mosquito populations for dengue control. When Wolbachia infections are at a high frequency, they influence the local transmission of dengue by direct virus blocking as well as deleterious effects on vector mosquito populations. However, the effectiveness of this strategy could be influenced by environmental temperatures that decrease Wolbachia density, thereby reducing the ability of Wolbachia to invade and persist in the population and block viruses. We reared wMel-infected Ae. aegypti larvae in the field during the wet season in Cairns, North Queensland. Containers placed in the shade produced mosquitoes with a high Wolbachia density and little impact on cytoplasmic incompatibility. However, in 50% shade where temperatures reached 39°C during the day, wMel-infected males partially lost their ability to induce cytoplasmic incompatibility and females had greatly reduced egg hatch when crossed to infected males. In a second experiment under somewhat hotter conditions (>40°C in 50% shade), field-reared wMel-infected females had their egg hatch reduced to 25% when crossed to field-reared wMel-infected males. Wolbachia density was reduced in 50% shade for both sexes in both experiments, with some mosquitoes cleared of their Wolbachia infections entirely. To investigate the critical temperature range for the loss of Wolbachia infections, we held Ae. aegypti eggs in thermocyclers for one week at a range of cyclical temperatures. Adult wMel density declined when eggs were held at 26–36°C or above with complete loss at 30–40°C, while the density of wAlbB remained high until temperatures were lethal. These findings suggest that high temperature effects on Wolbachia are potentially substantial when breeding containers are exposed to partial sunlight but not shade. Heat stress could reduce the ability of Wolbachia infections to invade mosquito populations in some locations and may compromise the ability of Wolbachia to block virus transmission in the field. Temperature effects may also have an ecological impact on mosquito populations given that a proportion of the population becomes self-incompatible. Aedes aegypti mosquitoes are the principal vectors of dengue and are widespread in the tropics where they live near humans [1,2]. Chemical insecticides have historically been used to control Ae. aegypti populations during disease outbreaks, but this approach is unlikely to be sustainable as insecticide resistance is now widespread in many parts of the world [3,4]. There is increasing interest in ‘rear and release’ programs where mosquitoes modified with desirable traits are released into natural populations as an alternative approach to disease control [5]. At the forefront of these programs is the deployment of mosquitoes infected with Wolbachia bacteria. Wolbachia occur naturally in many insects but have been introduced experimentally into Ae. aegypti where they can interfere with the transmission of dengue and other pathogens [6–8]. Wolbachia are transmitted maternally and typically affect host reproduction [9] or provide other advantages [10] to facilitate their spread into populations. These phenotypes have been utilized in disease control programs where Wolbachia-infected mosquitoes have been deployed to replace natural populations [11–13] or suppress populations through the release of only males [14,15]. Both approaches rely on cytoplasmic incompatibility induced by Wolbachia, where uninfected females that mate with infected males do not produce viable offspring, but viability is restored if the female is also Wolbachia-infected [16,17]. Over ten Wolbachia strain associations have now been generated in Ae. aegypti and they exhibit a diverse range of phenotypes. Some Wolbachia strains are relatively benign and have little impact on host fitness or virus blockage such as the wRi strain [18]. Others impose large fitness costs but also strongly reduce virus transmission including wMelPop [17] and wAu [19] while others like wAlbB fall somewhere in between [20,21]. There are also superinfections where two or more Wolbachia strains infect the same host, which can have combined or unexpected effects [21,22]. Aedes aegypti infected with the wMel strain of Wolbachia have been or are now being released in over ten countries (https: //www. worldmosquitoprogram. org/) and have successfully established in suburban areas in Cairns and Townsville in Queensland, Australia [11–13] and in Brazil [23]. The wAlbB strain has also been deployed successfully in Malaysia for population replacement (http: //www. imr. gov. my/wolbachia/) and in several countries for population suppression, where the release of only infected males has reduced population sizes by more than 80% due to cytoplasmic incompatibility (https: //debug. com/; https: //www. nea. gov. sg/corporate-functions/resources/research/wolbachia-aedes-mosquito-suppression-strategy). Despite these successes, there are limitations of Wolbachia infections that may affect their utility as disease control agents (reviewed in Ritchie et al. [24]). The majority of Wolbachia infections in Ae. aegypti reduce mosquito fitness and these costs tend to be exacerbated in stressful environments such as when larvae are starved [25] or in quiescent eggs [18,19,26,27]. Fitness costs can have enormous effects on invasion success. For example, the wMelPop infection failed to persist in release zones in Australia and Vietnam despite reaching frequencies above 90%, likely due to the massive fitness costs of this strain [28]. Wolbachia infections that occur naturally in mosquitoes can interfere with patterns of cytoplasmic incompatibility and limit the potential for population replacement and suppression [29]. Density-dependent interactions [30] and spatially heterogeneous environments [31] can also slow the rate of invasion, as can pesticide susceptibility in released mosquitoes [23]. For population replacement programs to be successful, Wolbachia infections should persist at high frequencies in the environment and block virus transmission under field conditions for many years following deployment [24]. There is a risk that Wolbachia infections, viruses or mosquitoes will evolve following the establishment of Wolbachia in populations, leading to less effective virus protection in Wolbachia-infected mosquitoes in the long-term [32]. However, the wMel infection has remained stable so far in terms of virus blockage [33] and its effects on fitness [34]. After seven years in the field, wMel has retained a high titer and continues to induce complete cytoplasmic incompatibility in the laboratory [35], indicating that attenuation is unlikely for at least several years following deployment. While the wMel infection in northern Queensland Ae. aegypti populations does not appear to have changed phenotypically since release, environmental conditions in the field such as temperature can have transient effects on Wolbachia infections, influencing their ability to suppress virus transmission or establish in populations. This issue is particularly important as climate change is leading to warmer average conditions and higher temperature extremes, including in the tropics [36]. Wolbachia infections in Ae. aegypti are vulnerable to high temperatures; heat stress during larval development reduces Wolbachia density in adults [37] and decreases the fidelity of cytoplasmic incompatibility and maternal transmission [35,38]. The fidelity of cytoplasmic incompatibility and maternal transmission are two key parameters for Wolbachia spread [39] while Wolbachia density is positively associated with the strength of virus blockage in both Drosophila [40,41] and mosquitoes [42]. Wolbachia strains in Ae. aegypti differ in their response to heat stress; the wMel and wMelPop strains are relatively susceptible while wAlbA, wAlbB and wAu are more robust, retaining high densities when larvae are reared at cyclical temperatures of 26–37°C [19,38]. These laboratory studies demonstrate the potential for heat stress to affect the success of Wolbachia interventions, but conditions experienced by mosquitoes in field situations are more complex than in an incubator. To understand the effects of high temperatures under more natural conditions, we reared Ae. aegypti larvae infected with wMel in the field with varying levels of exposure to sunlight. We then performed crosses to test for effects on cytoplasmic incompatibility and measured Wolbachia density. Finally, we performed experiments with Ae. aegypti eggs in the laboratory to determine the range of temperatures that adversely affect Wolbachia infections. Blood feeding on human subjects was approved by the Human Research Ethics Committee, James Cook University (approval H4907). All adult subjects provided informed oral consent (no children were involved). Aedes aegypti mosquitoes infected with the wMel strain of Wolbachia were collected in 2013 from locations near Cairns, Australia where wMel had successfully established [11]. Aedes aegypti infected with wMelPop were collected from Cairns, Australia in 2012 following field releases and local field breeding [28]. Aedes aegypti infected with wAlbB were derived from laboratory colonies and are described in Xi et al. [16] and Axford et al. [20]. Uninfected Ae. aegypti were collected in 2016 from locations where Wolbachia-infected mosquitoes had not been released. Mosquitoes were maintained in an insectary at the University of Melbourne according to methods described by Ross et al. [43]. Females with each Wolbachia strain were crossed to males from the uninfected population for at least three generations prior to the start of experiments to ensure that genetic backgrounds between populations were similar [27]. We conducted two experiments during the wet season in Cairns, Australia in 2018 to test the stability of the wMel Wolbachia infection under field temperature conditions. Experiments took place at James Cook University under a protective awning with different levels of shade as described in Ritchie et al. [44]. The relative reduction in light intensity compared to direct sunlight was determined using an EasyView EA30 light meter (Extech Instruments Corporation, Waltham, MA 02451 U. S. A.); two shade levels were chosen for the experiments which reduced light intensity by 50% and 99% respectively. We then established water-filled containers of various sizes to simulate a range of field larval habitats. The wMel-infected eggs were hatched in a single tray in 99% shade, then approximately 100 1st instar larvae were placed into each container. Larvae were provided with TetraMin tropical fish food tablets (Tetra, Melle, Germany) ad libitum throughout their development. Water temperatures were recorded every 30 minutes for the duration of larval development with data loggers (Thermochron; 1-Wire, iButton. com, Dallas Semiconductors, Sunnyvale, CA, USA) placed in zip-lock bags and submerged in the centre of each container. Pupae were collected one week after hatching and were returned to the laboratory for adult emergence. In the first experiment, Ae. aegypti eggs infected with the wMel strain of Wolbachia were hatched on the 10th of January 2018 and pupae were collected on the 17th of January. We used containers of three types with a wide range of water volumes which were expected to experience a variety of temperature conditions. Black buckets (13 cm radius, 25 cm height) were filled with 8 L of tap water, plant pots (5 cm radius, 10 cm height) with 500 mL and cups (2 cm radius, 6 cm height) with 60 mL. Containers were covered with mesh or stockings to prevent wild mosquitoes from ovipositing, and experimental mosquitoes from escaping. This setup was repeated for both the 99% and 50% shade levels. wMel-infected larvae were also reared in a single bucket filled with 8 L of water and placed in direct sunlight. In the second experiment, we repeated this procedure but used two container types: black buckets filled with 8 L of water and small round clear plastic containers (10 cm radius, 7 cm height) filled with 400 mL of water, with each container replicated three times at 99% and 50% shade levels. Eggs were hatched on the 26th of January 2018 and pupae were returned to the laboratory on the 1st of February. Populations of wMel-infected and uninfected Ae. aegypti were also reared in the laboratory concurrently at 26°C ± 1°C according to Ross et al. [43] for experimental crosses. We tested the ability of wMel-infected males to induce cytoplasmic incompatibility and wMel-infected females to restore compatibility after being reared under field temperature conditions. Adults emerging from each container type and shade level were added to 15 cm3 cages (BugDorm-4M1515, Megaview Science Co. , Taichung, Taiwan) where sexes were maintained separately. Crosses were performed two to three days after adults emerged by aspirating approximately 50 females into cages with an equal number of males. Females were blood fed three days later and then isolated in plastic cups containing 15 mL of larval rearing water and a strip of sandpaper for oviposition. Eggs were collected four days after blood feeding, partially dried, and then hatched four days after collection. Eggs were then counted under a dissecting microscope and hatch rates were determined by counting the proportion of eggs that had a clearly detached cap. In the first experiment, we performed crosses with adults reared in buckets at either 99% or 50% shade. Field-reared wMel-infected males were crossed to uninfected females to determine their ability to induce cytoplasmic incompatibility. We also crossed field-reared wMel-infected females to either uninfected males or wMel-infected and laboratory-reared males to determine the ability of females to restore compatibility. Infected males and uninfected females both reared under laboratory conditions were crossed to each other to confirm that wMel induces complete cytoplasmic incompatibility at 26°C. Twenty females were isolated for oviposition in each of these crosses, but individuals that died or did not lay eggs were excluded from the analysis. In the second experiment we performed a similar set of crosses but only for adults emerging from buckets held in 50% shade. In crosses where both sexes were infected we used males and females from the same container rather than using males reared at 26°C. This was done to see if populations became self-incompatible when both sexes were reared at warmer temperatures. Thirty females were isolated for oviposition in each cross. Crosses between males and females reared under the same conditions were also performed with adults that were reared in buckets held in 99% shade and small containers held at 50% and 99% shade. We determined egg hatch proportions and Wolbachia densities of females in each of these crosses to see if there was a relationship between Wolbachia density and hatch rate. We performed two experiments using thermocyclers (Biometra, Göttingen, Germany) to test the thermal tolerance of Wolbachia-infected Ae. aegypti eggs and the density of Wolbachia under a range of temperature conditions. We followed methods described in Kong et al. [45] with some modifications. Eggs from uninfected, wMel, wAlbB and wMelPop colonies were collected on sandpaper strips which were then partially dried, wrapped in paper towel and held in sealed zip-lock bags. Four days after collection, eggs were brushed onto filter paper with a small paint brush and then tipped into 0. 2 mL PCR tubes using a funnel. Batches of 15 to 39 eggs (mean 25. 7) were added to each tube. Tubes were closed and then tapped on the bench to ensure that eggs sank to the bottom of the tube where temperature control in the thermocycler is the most accurate [45]. Tubes were then placed in heat blocks of Biometra TProfessional TRIO 48 thermocyclers with tubes from each population arranged randomly in each block. In both experiments we used three thermocyclers, each with three heated blocks that can run independently for a total of 9 temperature regimes. In the first experiment we chose a broad range of temperature cycles to cover the entire range of temperatures that Ae. aegypti may experience in the field (http: //www. bom. gov. au/climate/averages/tables/cw_031011_All. shtml). Each regime had a fluctuation of 10°C between the minimum and maximum temperature; the lowest being 8–18°C and the highest being 32–42°C, with a difference of 3°C between each regime (S3A Fig). In the second experiment we chose a narrower temperature range based on when egg hatch and Wolbachia density started to decline in the previous experiment. The lowest regime was set to 24–34°C and the highest was 32–42°C, with difference of 1°C between each regime (S3B Fig). In the first experiment there were six replicate tubes of eggs for each temperature cycle and Wolbachia infection type and the second experiment had 12 replicates. After all tubes were added to the thermocyclers we closed the lids and started programs simultaneously. Eggs in tubes were also maintained at 26°C in a controlled temperature room in both experiments. After one week, tubes were removed from the thermocyclers and eggs were hatched by holding PCR tubes sideways above 70 mL specimen cups and then pipetting water into the tubes so that eggs fell into the cup. Each cup was filled with 40 mL of water and provided with a small amount of TetraMin and a few grains of yeast. Two days after hatching we determined egg hatch proportions by dividing the number of larvae by the number of eggs. We counted larvae again every 2 days as some eggs were slow to hatch, allowing one week in total for larvae to appear before we ceased counting. All larvae that hatched were added to plastic containers filled with 500mL of RO water and reared to adulthood. Multiple replicate cups of larvae were combined into trays for rearing, but the larval density was controlled to 100 larvae per tray or fewer to account for effects of larval competition and development time on Wolbachia density [46]. All adults were stored in ethanol for Wolbachia density measurements. In each experiment, random subsets of adults were stored in ethanol within 24 hours of emergence for Wolbachia screening. For both field experiments we extracted DNA from 16 males and 16 females from each container type and shade level. For experiments with eggs held in thermocyclers we extracted DNA from up to 10 (first experiment) or 12 (second experiment) males and 10 or 12 females from each Wolbachia infection type and treatment. Some treatments had lower sample sizes due to low egg hatch proportions. DNA was extracted from whole adults with 150 μL of 5% Chelex 100 resin (Bio-Rad Laboratories, Hercules, CA) and 3 μL of Proteinase K. We then conducted qPCR to detect and estimate the density of Wolbachia in each whole adult using methods described previously [47]. Individuals were considered uninfected if the Ae. aegypti-specific marker amplified successfully (Cp value < 35) but the Wolbachia-specific marker did not (Cp value of 35 or no Cp value) in two independent runs. For individuals that were positive for Wolbachia, (Cp value < 35 for both markers), differences in Cp between the two markers were transformed by 2n to provide an estimate of Wolbachia density, averaged from at least two independent runs. For the second field experiment, we also estimated the Wolbachia density of females after they had laid eggs to see if there was a relationship between Wolbachia density and egg hatch rate when crossed to infected males reared under the same conditions. All data were analyzed using SPSS statistics version 24. 0 for Windows (SPSS Inc, Chicago, IL). Hatch proportion and Wolbachia density data were often not normally distributed, and we therefore compared treatments for these variables with Kruskal-Wallis and Mann-Whitney U tests. We also used Spearman’s rank-order correlation to test the relationship between egg hatch and female Wolbachia density. We monitored water temperatures experienced by larvae in each container type at different levels of shade. Maximum temperatures differed between container types at 99% shade, with cups having average maximum daily temperatures that were 2. 5°C higher than buckets, though average temperatures were similar because smaller containers reached cooler temperatures at night (S1 Fig). Temperature cycles were similar between containers in 50% shade, which was unexpected given the large differences in water volume. The level of shade affected temperature substantially, with buckets in direct sunlight experiencing average maximum temperatures of 38. 3°C, while containers in 50% shade (average minimum: 23. 7°C, average maximum: 35. 3°C) were much warmer than containers in 99% shade (23. 2–29. 6°C) (Fig 1A). We tested the ability of wMel-infected males reared under field temperature conditions to induce cytoplasmic incompatibility with uninfected females. In a control cross, wMel-infected males reared in the laboratory at 26°C caused complete cytoplasmic incompatibility (no eggs hatched) with uninfected females (Fig 1). wMel-infected males reared in buckets at 99% shade induced almost complete cytoplasmic incompatibility, though 2/18 females produced a single viable progeny each (Fig 1B). wMel-infected males reared in buckets at 50% shade induced weaker cytoplasmic incompatibility, with 9/16 females producing some viable progeny (Fig 1C). We also tested the ability of wMel-infected females to retain their compatibility with wMel-infected males reared in the laboratory. When females were reared in buckets at 99% shade, there was no difference in hatch rate between crosses with uninfected males and crosses with wMel-infected males (Mann-Whitney U: Z = 0. 348, P = 0. 726, Fig 1B). In contrast, wMel-infected females reared in buckets at 50% shade had a 47. 6% reduction in egg hatch rate when crossed to wMel-infected males (Z = 3. 612, P < 0. 001). This indicates partial incompatibility with Wolbachia-infected males, suggesting a substantial loss of Wolbachia infection. We estimated the Wolbachia density of a subset of adults from each container type and level of shade (Fig 2). Wolbachia density was not consistently affected by container type for both females (Kruskal-Wallis χ2 = 2. 598, df = 2, P = 0. 273) and males (χ2 = 4. 419, df = 2, P = 0. 110), likely because the container types experienced similar temperature fluctuations at 50% shade. Conversely, Wolbachia density was affected substantially by shade level for both females (χ2 = 71. 261, df = 1, P < 0. 001) and males (χ2 = 68. 563, df = 1, p < 0. 001). Females reared under 50% shade had a median Wolbachia density that was just 0. 32% of the laboratory control, while males had a density of 8. 09% of the control. This reduction likely reflects the substantially higher maximum daily temperatures experienced in containers at 50% shade. In contrast, the Wolbachia density of adults reared at 99% shade was not significantly different to laboratory-reared adults (females: χ2 = 0. 650, df = 1, P = 0. 420, males: χ2 = 0. 085, df = 1, P = 0. 771). All adults screened from containers in 99% shade, 50% shade and the laboratory were positive for Wolbachia. However, we were unable to detect any Wolbachia infection in a sample of 11 adults taken from a bucket placed in direct sunlight. This indicates a complete loss of infection which is likely due to the extreme temperatures experienced in that container (up to 43°C, Fig 1A). Though we did not score survival to adulthood in containers directly, the bucket placed in direct sunlight experienced high mortality since only 11 adults emerged out of the approximately 100 larvae added initially. We conducted a second experiment later in the month where we also tested cytoplasmic incompatibility and measured Wolbachia density in adults. Temperatures were affected substantially by the location of containers where average maximum temperatures were nearly 7°C warmer in 50% shade compared to 99% shade (Fig 3A). Maximum temperatures also differed between container types; at 50% shade small containers reached 39. 26°C on average while buckets reached 36. 54°C, but average temperatures did not differ much between containers at the same level of shade because of warmer minimum temperatures in buckets. We set up crosses with adults emerging from buckets held in 50% shade to test for any effects on cytoplasmic incompatibility. wMel-infected males induced strong but incomplete cytoplasmic incompatibility with uninfected females; 7/18 females produced some viable progeny, compared to 0/20 in the control (Fig 3B). wMel-infected adults reared in buckets at 50% shade that were crossed to each other experienced a 79. 9% reduction in egg hatch rate when crossed to each other relative to crosses with uninfected males (Mann-Whitney U: Z = 5. 615, P < 0. 001), suggesting a greatly reduced ability of females to restore compatibility under these rearing conditions. We estimated the Wolbachia density of adults that emerged from each treatment in the second experiment. We found that many individuals had lost their Wolbachia infection (no detectable infection) when reared in containers held in 50% shade (Fig 4), particularly in small containers where larvae experienced higher maximum daily temperatures (Fig 3A). Of the adults reared at 50% shade that were still infected with Wolbachia, their density had been reduced to 0. 19% and 0. 23% of the 26°C control in females and males respectively. In this experiment, Wolbachia density was also reduced at 99% shade relative to the 26°C control (females: χ2 = 14. 828, df = 1, P < 0. 001, males: χ2 = 16. 519, df = 1, P < 0. 001), with densities being approximately 50% the level of the control. wMel-infected adults emerging from the two container types and shade levels were returned to the laboratory and allowed to mate with individuals from the same container. We then scored egg hatch proportions of individual females and measured their Wolbachia density after oviposition to determine the relationship between Wolbachia density and egg hatch proportion. Females with high Wolbachia densities exhibited high hatch proportions while females with lower densities tended to have very low hatch proportions or produced no viable offspring, with the correlation between density and egg hatch being highly significant (Spearman’s rank-order correlation: ρ = 0. 899, P < 0. 001, n = 65, Fig 5). This indicates that females with low densities had partially or completely lost their ability to restore compatibility, but males reared under the same conditions had largely retained their ability to induce cytoplasmic incompatibility. The strong relationship between Wolbachia density and egg hatch indicates that a high density in females is important for restoring compatibility with infected males. We tested the tolerance of Wolbachia-infected and uninfected eggs to a broad range of temperature conditions. When eggs were held at 26°C for one week, Wolbachia-infected eggs did not differ from uninfected eggs in terms of hatch proportion (Mann-Whitney U: all P > 0. 05). At higher temperatures fitness costs of Wolbachia infections were evident; wMelPop-infected eggs had lower hatch proportions than uninfected eggs under temperature cycles of 26–36°C and 29–39°C (both Z = 2. 802, P = 0. 005, Fig 6A). wAlbB-infected eggs also had reduced hatch proportions relative to uninfected eggs at 29–39°C (Z = 2. 162, P = 0. 031) but wMel-infected eggs did not differ from uninfected eggs under any temperature cycle (all P > 0. 05). We reared larvae hatching from eggs held at each temperature cycle and measured Wolbachia density in adults. Wolbachia density did not differ between males and females across all temperature conditions (wMel: Kruskal-Wallis χ2 = 0. 271, df = 1, P = 0. 603, wAlbB: χ2 = 2. 398, df = 1, P = 0. 122) except for wMelPop, where density was higher in females than in males (χ2 = 14. 507, df = 1, P < 0. 001). When eggs were held at 26°C, wMelPop-infected adults had the highest density of Wolbachia while wAlbB had an intermediate density and wMel had the lowest density (Fig 6B and 6C), consistent with previous studies [20,38]. This pattern was consistent across the cooler temperature cycles (maximum daily temperatures of 18–33°C) where Wolbachia densities for wAlbB (χ2 = 6. 505, df = 5, P = 0. 260), wMelPop (χ2 = 9. 108, df = 5, P = 0. 105) and wMel males (χ2 = 2. 950, df = 5, P = 0. 708) were stable (Fig 6B and 6C). In contrast, wMel density in females declined with increasing maximum temperatures across this range (χ2 = 27. 190, df = 5, P < 0. 001). When eggs were held at 29–39°C, adult Wolbachia density declined steeply for both wMel and wMelPop infections (Fig 6B and 6C). Median wMel densities were reduced to only 0. 41% and 0. 14% of densities at 26°C in females and males, respectively. In wMelPop the relative loss was even steeper, with females reared from eggs held at 29–39°C having just 0. 02% the density of females at 26°C, while Wolbachia was not detected in males. wAlbB density also declined when eggs were held at 29–39°C in both females (Mann-Whitney U: Z = 2. 797, P = 0. 005) and males (Z = 2. 735, P = 0. 006) but the effect was much weaker than the other strains, with median densities of 34. 88% (in females) and 42. 04% (in males) of eggs held at 26°C. In a second experiment, we used a narrower temperature range to investigate egg thermal tolerance and the loss of Wolbachia infections on a finer scale and with greater replication. Egg hatch proportions declined for all Wolbachia infection types as maximum temperatures increased, with the effect being most severe for the wMelPop infection (Fig 7A). Egg hatch proportions of the wMel and wAlbB strains did not differ significantly from uninfected eggs at maximum temperatures of 37°C and below (Mann-Whitney U: all P > 0. 05). However, at maximum temperatures of 38–41°C both the wMel and wAlbB strains had lower hatch proportions than uninfected eggs held at the same temperature (all P ≤ 0. 026). This indicates that Wolbachia infections in Ae. aegypti lower the tolerance of eggs to high temperatures, particularly in the case of wMelPop. Consistent with the previous experiment, Wolbachia density declined as eggs were exposed to increasing maximum temperatures, beginning at 35°C for wMelPop and 36°C for wMel (Fig 7B and 7C). The wMel and wMelPop infections were lost from some individuals when eggs were exposed to 29–39°C and absent from all adults at 30–40°C (Fig 7B and 7C). In contrast, all wAlbB adults were infected across all temperature cycles, though density was reduced in both females (Mann-Whitney U: Z = 3. 204, P = 0. 001) and males (Z = 3. 897, P < 0. 001) at 30–40°C relative to 26°C. We tested the stability of the wMel Wolbachia infection in Ae. aegypti under field temperature conditions and performed laboratory experiments to determine the range of temperatures that affect different Wolbachia strains. Our experiments demonstrate three main outcomes of heat stress on Wolbachia-infected mosquitoes. Firstly, there are direct costs of Wolbachia infections on Ae. aegypti thermal tolerance, at least during the egg stage. Secondly, heat stress under partial shade conditions in the field reduces cytoplasmic incompatibility fidelity in wMel-infected males, while infected females become partially incompatible with infected males. Thirdly, heat stress reduces Wolbachia density and may impair the ability of Wolbachia to block virus transmission for a subsection of the mosquito population reared under specific field conditions. Heat stress could therefore adversely affect the success of disease control programs depending on the location and nature of the field breeding sites. There are relatively few examples of symbionts affecting the thermal tolerance of their hosts [48]. In Drosophila melanogaster, the wMelCS strain of Wolbachia increases the survival of adults under heat stress while the wMelPop infection decreases survival [49], though wMel appears to have no effect on high temperature tolerance [50]. Here we show that Wolbachia infection reduces the tolerance of Ae. aegypti eggs to high temperatures, with the severity of the effect depending on Wolbachia strain. In addition, we have determined the temperature range where deleterious effects on Wolbachia infections start to occur and where the infections are lost, at least during the egg stage. For the wMelPop and wMel infections, Wolbachia density declined beginning at temperatures of 25–35°C (30°C mean) and 26–36°C (31°C mean) respectively, while for wAlbB this occurred at a much higher temperature range (30–40°C, 35°C mean). The higher tolerance of wAlbB to heat stress is consistent with prior studies in Ae. aegypti larvae [19,35,38], but the increased resolution in this experiment provides a better estimate of the maximum daily temperatures that could affect Wolbachia interventions. In field situations the temperature ranges where Wolbachia infections are adversely affected will depend on the duration and timing of heat stress (see below). In our field experiments we found substantial effects of heat stress on cytoplasmic incompatibility that could limit the potential of wMel to invade natural populations during disease control programs and persist following releases. When reared in partial shade, wMel-infected males partially lost their ability to induce cytoplasmic incompatibility while females partially or completely lost their ability to restore compatibility when crossed to infected males reared in the lab. Infected females reared in partial shade had greatly reduced fertility in crosses with infected males from the same container. Female Wolbachia density was positively associated with egg hatch, consistent with a study in Drosophila [51]. High densities therefore appear needed for females to restore compatibility with infected males, but the density required for males to induce cytoplasmic incompatibility appears to be lower. Heat stress conditions in the field could greatly diminish or even reverse the reproductive advantage provided by Wolbachia, making invasion challenging, particularly when Wolbachia is at a low frequency, when its fitness relative to uninfected individuals is relatively lower and where it is susceptible to stochastic [52,53] and density related [30,54] effects. Where wMel has already established in a population, reduced egg hatch in wMel-infected mosquitoes that mate with each other could provide an opportunity for an increase in the frequency of uninfected mosquitoes although once wMel invaded areas of North Queensland it appears to have been stable [34]. Fitness costs and self-incompatibility between infected mosquitoes could also have unexpected ecological effects; a decline in the Ae. aegypti population could lead to shifts in species composition [24] which could be beneficial for disease control efforts. Though we attempted to rear mosquitoes under realistic temperature conditions, our field experiments will only be relevant to a subset of natural breeding sites. We provided abundant food to speed up and synchronise larval development to facilitate experimental crosses between strains. Larval development times in nature are variable and can exceed two months under competitive conditions [30]. Increasing the rate of larval development in this experiment likely underestimated the effect of heat stress; longer development times increase the chance that larvae will experience a heat wave and increased durations of heat stress may further reduce Wolbachia density, though density may also recover over time in the absence of heat stress [37]. wMel-infected larvae provided with a low level of food have a greatly reduced Wolbachia density when reared at 26–32°C compared to 26°C, indicating that even moderate temperatures can reduce Wolbachia density when combined with nutritional stress (see Figure S4 of Ross and Hoffmann [35]). The effects of heat stress on Wolbachia density can carry over into the next generation [55] which may lead to reduced virus blockage or cytoplasmic incompatibility across a generation after a heat wave. In our laboratory experiments eggs were maintained for one week before hatching, but in the field the egg stage can be shorter or much longer. During the dry season eggs can remain quiescent for months before hatching [56], increasing their potential exposure to high temperatures. Wolbachia infections reduce the viability of quiescent Ae. aegypti eggs [20,26,27] and under high temperatures these fitness costs will likely be exacerbated. A further limitation of our field experiment is that the containers used for larval rearing were not colonized naturally. Aedes aegypti seem to prefer laying eggs in shaded areas but will also utilize containers in sunlight [57–59]. Wolbachia infections may also affect thermal preference; adult Drosophila melanogaster infected with Wolbachia prefer cooler temperatures than uninfected flies [60,61]. Nevertheless, data from sentinel containers indicates that wMel-infected mosquitoes will lay eggs in containers placed in direct sunlight. Sentinel buckets and small containers placed within the wMel release zone in Cairns were all colonized by Ae. aegypti despite some of these experiencing similar temperatures to the experimental containers held at 50% shade (S2 Fig). Ae. aegypti tend to lay eggs during cooler parts of the day [62,63] and therefore may be unable to discriminate against habitats that reach high maximum temperatures later. Unlike adult mosquitoes, immature stages cannot easily escape heat stress as they are unable to move beyond the container. Since Wolbachia density and egg hatch in wMel-infected mosquitoes appears to depend strongly on the level of shade, temperature and productivity surveys of larval habitats could be conducted in release areas if there are concerns around heat stress impacts in a release area. Despite the substantial effects on Wolbachia density and fertility in our experiments, Ae. aegypti mosquitoes infected with wMel have successfully established in Cairns [11,12] and Townsville [13], Australia and in Brazil [23], with the infection persisting at a high frequency in most locations. In areas where the releases succeeded, the costs of heat stress observed here were clearly not prevalent or severe enough to prevent the establishment of wMel. Once a Wolbachia infection has attained a high frequency in a population it may stay high unless the fitness costs are extreme, as is the case for wMelPop [28]. Nevertheless, heat stress will likely slow the rate of Wolbachia invasion and spread, increasing the number of mosquitoes required for releases, and potentially creating an unstable situation around critical invasion points that must be exceeded for Wolbachia to invade [52]. Heat stress could partially explain why infection frequencies have persisted at intermediate levels in some suburbs [13] and may also contribute to the incomplete maternal transmission fidelity of wMel observed in Cairns [64] given that some individuals were cleared of their Wolbachia infections in our experiments. wMel-infected mosquito releases outside of Australia in locations where maximum daily temperatures are warmer may be more challenging. Reduced Wolbachia densities may also reduce virus protection provided by Wolbachia even if infection frequencies remain high in a population, though we do not demonstrate this effect directly. The wMel strain has retained its susceptibility to heat stress for seven years after field deployment in Australia [35], indicating that alternative strains may be needed in areas where wMel has difficulty establishing or where viral blockage is insufficient.

Aedes aegypti mosquitoes infected with Wolbachia symbionts are being deployed in the tropics as a way of reducing disease transmission. Some Wolbachia strains are vulnerable to high temperatures but these effects have not been evaluated outside of a laboratory setting. We reared Ae. aegypti infected with the wMel strain of Wolbachia in the field during the wet season in Cairns, Australia, where the first releases of Wolbachia-infected Ae. aegypti took place. wMel-infected mosquitoes became partially self-incompatible, with reduced egg hatch, when larvae were reared in partial shade where maximum daily temperatures exceeded 39°C. Under these conditions the amount of Wolbachia in adult mosquitoes was reduced to less than 1% of laboratory-reared mosquitoes on average, while some mosquitoes were cleared of Wolbachia entirely. In contrast, wMel was stable when mosquitoes were reared under cooler conditions in full shade. Field trials with the wMel strain are now underway in over 10 countries, but high temperatures in some locales may constrain the ability of Wolbachia to invade natural mosquito populations and block disease transmission.

Abstract Introduction Methods Results Discussion

invertebrates medicine and health sciences classical mechanics geographical locations australia mechanical stress light animals wolbachia electromagnetic radiation developmental biology infectious disease control insect vectors bacteria infectious diseases aedes aegypti thermal stresses life cycles disease vectors insects arthropoda physics people and places mosquitoes sunlight solar radiation eukaryota oceania biology and life sciences species interactions physical sciences larvae organisms

2019

Loss of cytoplasmic incompatibility in Wolbachia-infected Aedes aegypti under field conditions

UNC-6/Netrin is a conserved axon guidance cue that directs growth cone migrations in the dorsal-ventral axis of C. elegans and in the vertebrate spinal cord. UNC-6/Netrin is expressed in ventral cells, and growth cones migrate ventrally toward or dorsally away from UNC-6/Netrin. Recent studies of growth cone behavior during outgrowth in vivo in C. elegans have led to a polarity/protrusion model in directed growth cone migration away from UNC-6/Netrin. In this model, UNC-6/Netrin first polarizes the growth cone via the UNC-5 receptor, leading to dorsally biased protrusion and F-actin accumulation. UNC-6/Netrin then regulates protrusion based on this polarity. The receptor UNC-40/DCC drives protrusion dorsally, away from the UNC-6/Netrin source, and the UNC-5 receptor inhibits protrusion ventrally, near the UNC-6/Netrin source, resulting in dorsal migration. UNC-5 inhibits protrusion in part by excluding microtubules from the growth cone, which are pro-protrusive. Here we report that the RHO-1/RhoA GTPase and its activator GEF RHGF-1 inhibit growth cone protrusion and MT accumulation in growth cones, similar to UNC-5. However, growth cone polarity of protrusion and F-actin were unaffected by RHO-1 and RHGF-1. Thus, RHO-1 signaling acts specifically as a negative regulator of protrusion and MT accumulation, and not polarity. Genetic interactions are consistent with RHO-1 and RHGF-1 acting with UNC-5, as well as with a parallel pathway, to regulate protrusion. The cytoskeletal interacting molecule UNC-33/CRMP was required for RHO-1 activity to inhibit MT accumulation, suggesting that UNC-33/CRMP might act downstream of RHO-1. In sum, these studies describe a new role of RHO-1 and RHGF-1 in regulation of growth cone protrusion by UNC-6/Netrin. The connectivity of neuronal circuits is established through properly guided axons which form functional synaptic connections. The growing axon is guided to its target by the motile, actin-based growth cone at the tip of the growing neurite. Growth cone response to extracellular guidance cues allows the axon to extend, retract, turn and branch, regulated by the reorganization and dynamics of the actin and microtubule cytoskeletons of the growth cone [1]. In C. elegans and vertebrates, the conserved laminin-like UNC-6/Netrin guidance cue and its receptors UNC-40/DCC and UNC-5 direct dorsal-ventral axon outgrowth [2–10]. UNC-6 is secreted by cells in the ventral nerve cord [11], and growth cones grow toward UNC-6/Netrin (i. e. ventral migration; attraction) and away from UNC-6/Netrin (i. e. dorsal migration; repulsion). The prevailing model of UNC-6/Netrin-mediated axon guidance involves a ventral-to-dorsal chemotactic gradient of the molecule, which growth cones interpret by migrating up or down the gradient using the “attractive” receptor UNC-40/DCC or the “repulsive” receptor UNC-5, respectively [12,13]. However, this model has recently been challenged by studies in mouse spinal cord showing that floorplate Netrin is dispensable for commissural axon guidance, and that ventricular expression is important, possibly in a close-range, haptotactic event [14–17]. Experiments leading to the statistically-oriented asymmetric localization (SOAL) model in neurons with growth cones that grow ventrally toward UNC-6 were among the first studies to show that UNC-6/Netrin gradients were not required to explain directed outgrowth [18–20]. In the HSN neuron, which extends an axon ventrally, UNC-6/Netrin controls the biased ventral accumulation of the UNC-40 receptor in the HSN cell body, and UNC-5 acts to bias UNC-40/DCC ventrally, resulting in probabilistic bias of protrusion to the ventral surface [18–20]. Our previous work with the VD growth cones that migrate dorsally (repelled) suggests that UNC-6/Netrin first polarizes protrusion and F-actin to the dorsal side of the growth cone via the UNC-5 receptor, and then regulates protrusion based on this polarity (the polarity/protrusion model). UNC-5 inhibits protrusion ventrally, close to the UNC-6/Netrin source, and UNC-40 stimulates protrusion dorsally, away from the UNC-6/Netrin source, resulting in directed dorsal growth away from UNC-6/Netrin [21–23]. That polarity and protrusion are separable events was suggested previously in HSN by missense mutations in UNC-6 and UNC-40 that uncouple their roles in polarity and migration [24]. Neither the SOAL model in ventrally-growing axons or the polarity/protrusion model in dorsally growing axons rely on chemotactic gradients and instead involve growth cone asymmetries coupled with regulation of protrusive growth by these asymmetries. Chemotactic gradient models imply a tight coupling of growth cone polarity and protrusion (i. e. different concentrations of UNC-6/Netrin lead to different protrusive activities across the growth cone). While the SOAL model is based on asymmetry of axon initiation in the HSN cell body, and the polarity/protrusion model is based on analyzing growth cones during outgrowth, the idea of separability of polarity and protrusion in directed migration is similar in both models. Also similar in both models is that the UNC-5 receptor, considered the “repulsive” receptor in classical gradient models, acts in both growth toward and away from UNC-6/Netrin. UNC-40/DCC drives growth cone lamellipodial and filopodial protrusion via the small GTPases CDC-42, CED-10/Rac, and MIG-2/RhoG, the Rac-specific guanine nucleotide exchange factor (GEF) TIAM-1, and actin cytoskeletal regulators Arp2/3, UNC-34/Enabled and UNC-115/abLIM [25–29]. UNC-5 inhibits growth cone protrusion via the Rac GEF UNC-73/trio, CED-10/Rac and MIG-2/RhoG (also used to drive protrusion), the FMO flavin monooxygenases which might act via actin, and the actin and MT-interacting proteins UNC-33/CRMP and UNC-44/Ankyrin [22,23,30]. UNC-5 also restricts the accumulation of microtubule + ends in VD growth cones which have pro-protrusive effects [21]. Thus, in unc-5 mutants, VD growth cones are larger and more protrusive, display unpolarized protrusion including ventral protrusions, display unpolarized F-actin around the periphery of the growth cone, and have increased accumulation of MT+ ends [21,22]. This unregulated protrusion results in unfocused growth cones that fail to migrate dorsally away from UNC-6/Netrin, causing the severe VD axon guidance defects seen in unc-5 mutants. The Rho-family GTPases CED-10/Rac, MIG-2/RhoG, and CDC-42 control neuronal protrusion [23,26,28,31]. Here we dissect the role of RHO-1, the single RhoA molecule encoded in the C. elegans genome, in regulation of VD growth cone polarity and protrusion. rho-1 RNAi results in early embryonic arrest, with a failure in cytokinesis and severe morphological defects [32–35]. We used cell-specific expression of constitutively-active RHO-1 (G14V) and dominant-negative RHO-1 (T19N), and cell-specific RNAi of rho-1 and found that RHO-1 inhibited growth cone protrusion and MT+ end accumulation. RHO-1 did not, however, affect polarity of protrusion or F-actin, demonstrating that growth cone polarity can be separated from growth cone protrusion. We also found that the RHO-1 activator RHGF-1, a RHO-1 GTP exchange factor of the LARG family [36,37], was required to inhibit protrusion and MT+ end accumulation similar to RHO-1. Genetic interactions with UNC-5 signaling and UNC-33/CRMP suggest that RHGF-1 and RHO-1 might act downstream of UNC-5 and in parallel to other regulators of protrusion and MT+ end accumulation. These studies also revealed that RHO-1 requires UNC-33/CRMP to prevent MT+ end accumulation. In sum, results reported here show that RHGF-1 and RHO-1 are key inhibitors of growth cone protrusion and MT+ end accumulation and act with UNC-5 in protrusion, but not growth cone polarity. RHO-1 is the single RhoA homolog in C. elegans. Loss of rho-1 leads to embryonic lethality, with a failure in cytokinesis [38], and perturbation of RHO-1 signaling in adults results in dysfunction in numerous neuronal and non-neuronal functions leading to death [39]. To understand the role of RHO-1 in VD growth cone morphology, we constructed constitutively-active G14V and dominant-negative T19N versions of RHO-1, and expressed them in the VD/DD neurons using the unc-25 promoter. Constitutively-active rho-1 (G14V) expression significantly reduced the VD growth cone area and shortened filopodial protrusions as compared to wild-type (Fig 1A, 1B and 1D). In contrast, dominant-negative rho-1 (T19N) expression displayed significantly longer filopodial protrusions as compared to wild-type VD growth cones (Fig 1A, 1B and 1E). Growth cone area was increased, but not significantly so. These results indicate that RHO-1 activity inhibits growth cone protrusion. We used a transgenic RNAi approach to knock down rho-1 in the VD/DD motor neurons as previously described (see Materials and Methods) [40,41]. Plasmids were generated to drive expression of sense and antisense RNA fragments complementary to the rho-1 under the control of the unc-25 promoter. Animals were made transgenic with a mix of the sense and antisense plasmids, and the resulting transgenes were used in analysis. The average length of filopodial protrusions and growth cone area were significantly increased in rho-1 (RNAi) (Fig 1A, 1B and 1F). These data suggest that RHO-1 normally inhibits VD growth cone protrusion. The polarity of filopodial protrusions was not affected by rho-1 (DN) or rho-1 (RNAi), as protrusions still displayed a dorsal bias similar wild-type (Fig 1G–1I). Thus, despite showing increased protrusion, the polarity of growth cone protrusion was not affected by rho-1. rho-1 (G14V), rho-1 (T19N), and rho-1 (RNAi each resulted in low-penetrance but significant VD/DD axon guidance defects (Table 1), including wandering, branching, and failing to reach the dorsal nerve cord. This suggests that the effects of RHO-1 on the growth cone result in axon guidance defects. Previous studies indicate that in VD growth cones, F-actin accumulates at the dorsal, protrusive edge of the growth cone and acts as a polarity mark to specify protrusion in this region (Fig 2A and 2B) [21,22]. Furthermore, microtubule + ends are present in the growth cone and have a pro-protrusive role [21]. In wild-type, MT+ ends are rare in VD growth cones (~2 per growth cone) (Fig 2E and 2F) [21], and protrusion is tightly regulated and localized to the dorsal leading edge at the site of F-actin accumulation Fig 2) [21]. VD growth cone F-actin was monitored using the VAB-10ABD: : GFP reporter, and MT+ ends were monitored using EBP-2: : GFP as described previously [21,22]. Dominant-negative rho-1 (T19N) and rho-1 (RNAi) had no effect on dorsally-polarized F-actin accumulation (Fig 2A and 2D), consistent with no effects on growth cone polarity of protrusion (Fig 1). However, growth cone EBP-2: : GFP puncta number were significantly increased by dominant-negative rho-1 (T19N) and rho-1 (RNAi) (Fig 2E, 2G and 2H), consistent with increased protrusion in these backgrounds. Constitutively-active rho-1 (G14V) resulted in fewer EBP-2: : GFP puncta, consistent with reduced growth cone protrusion (Fig 2E). F-actin polarity was also abolished, with distribution along the periphery of the entire growth cone (Fig 2A and 2C). Possibly, constitutive activation reveals a role of RHO-1 in F-actin polarity that is not affected in reduction of function treatments. However, a similar effect on F-actin was observed with constitutively-active Rac GTPases MIG-2 and CED-10 [21]. Possibly, this effect on F-actin is a consequence of small growth cones with severely-restricted protrusion, and not a direct role in F-actin organization. In sum, these results suggest that RHO-1 normally restricts growth cone protrusion by preventing accumulation of growth cone MT+ ends. RHGF-1 is a PDZ RhoGEF with PDZ, RGS, C1, DH, and PH domains (Fig 3A). RHGF-1 is a RHO-1-specific GEF and acts with RHO-1 in neurotransmitter release and axonal regeneration [36,37,42–44]. rhgf-1 (ok880) is a 1170bp in frame deletion which removes a large part of the DH domain and is predicted to have no RhoGEF activity [24], rhgf-1 (gk217) is a 247bp in frame deletion which removes the C1 domain, and rhgf-1 (gk292502) produces a premature stop just before the C1 domain (Fig 3A). rhgf-1 mutants each displayed increased growth cone area and longer filopodial protrusions compared to wild-type (Fig 3B–3F). The dorsally-biased polarity of growth cone protrusion was not significantly affected by rhgf-1 mutation (Fig 3G–3I). These data indicate that RHGF-1 is normally required to limit the extent of growth cone protrusion, but does not regulate growth cone polarity, similar to rho-1. rhgf-1 mutants displayed low-penetrance but significant VD/DD axon guidance defects (Table 1), suggesting that the effects of rhgf-1 on the growth cone has ramifications on axon guidance. The Drosophila RHGF-1 homolog DRhoGEF2 is a key regulator of morphogenesis and associates with the tips of growing MTs and exhibits plus end tracking [45]. In C. elegans, RHGF-1 associates with MTs and initiates an axon regeneration pathway [37]. rhgf-1 mutant VD growth cones displayed significantly increased numbers of EBP-2: : GFP puncta (Fig 4A–4C), but caused no significant defects in F-actin organization, similar to rho-1 knockdown (Fig 4D–4F). These results indicate that RHGF-1 might act with RHO-1 to inhibit growth cone protrusion by excluding MT+ ends from entering the growth cone periphery. The results above indicate that the VD growth cones of activated rho-1 (G14V) displayed reduced protrusion, and that those of rhgf-1 loss of function were overly-protrusive. The VD growth cones of activated rho-1 (G14V) double mutants with rhgf-1 loss of function resembled the small, inhibited growth cones of rho-1 (G14V) alone (Fig 5A–5E), with a significant reduction in filopodial length and growth cone area as compared to wild-type and rhgf-1 mutants alone (Fig 5A–5E). Similarly, double mutants of rhgf-1 and rho-1 (G14V) showed a significant decrease in the average number of EBP-2 puncta in the growth cone similar to rho-1 (G14V) alone (Fig 6A–6D). VAB-10ABD: : GFP distribution in these double mutant growth cones also resembled activated rho-1 (G14V) with F-actin distributed randomly all across the growth cone (Fig 6E–6H). That activated RHO-1 (G14V) was epistatic to rhgf-1 loss of function is consistent with RHO-1 acting downstream of RHGF-1 in limiting growth protrusion and EBP-2 accumulation in VD growth cones. Double mutants of dominant-negative rho-1 (T19N) and rho-1 (RNAi) with rhgf-1 did not result in significant enhancement of growth cone protrusion compared to single mutants (Fig 5F and 5G). VD/DD axon guidance defects were also not enhanced, except in one case (Table 1). These results further support the idea that RHO-1 and RHGF-1 act in the same pathway in growth cone protrusion and axon guidance. Previous studies showed that UNC-6/Netrin signaling via the heterodimeric UNC-40/UNC-5 receptor is required for inhibition of growth cone protrusion in UNC-6/Netrin repulsive axon guidance [22,23]. Constitutive activation of UNC-40 and UNC-5 using myristoylated versions of the cytoplasmic domains of UNC-40 and UNC-5 (myr: : unc-40 and myr: : unc-5) in the VD neurons result in small growth cones with few or no filopodial protrusions [22,23,25]. Loss of rhgf-1 significantly suppressed inhibition of filopodial protrusion and growth cone size caused by myr: : unc-40 and myr: : unc-5 (Fig 7). myr: : unc-40 and myr: : unc-5 growth cones show a significant decrease in the average number of EBP-2: : GFP puncta in the VD growth cones as compared to wild-type (Fig 8A–8C) [21]. Double mutants of rhgf-1 with myr: : unc-40 and myr: : unc-5 resembled rhgf-1 mutants alone, with significant increases in protrusion and MT+ end accumulation (Fig 8A and 8D). Similar to activated Racs and RHO-1 (G14V), F-actin is distributed throughout the small growth cones in activated myr: : unc-5 and myr: : unc-40 (Fig 8E–8G). rhgf-1 mutation restored dorsal polarity of F-actin (Fig 8E and 8H). In sum, the growth cones of rhgf-1 double mutants with myr: : unc-5 and myr: : unc-40 displayed increased protrusion and EBP-2 puncta accumulation compared to myr: : unc-40 and myr: : unc-5, but normal dorsal F-actin polarity. These data indicate that RHGF-1 is required for the inhibitory effects of myr: : unc-40 and myr: : unc-5 on growth cone protrusion and EBP-2: : GFP puncta accumulation. unc-5 loss of function results in unpolarized, overly-protrusive VD growth cones. Excess MT+ ends accumulate in unc-5, and dorsal polarity of F-actin accumulation and thus protrusion is lost [21,22]. Activated rho-1 (G14V) expression did not suppress the large growth cone area and long filopodial protrusions seen in unc-5 mutants (i. e. double mutants resembled unc-5 alone) (Fig 9). Furthermore, we observed no significant change in EBP-2: : GFP and VAB-10ABD: : GFP distribution in the VD growth cones as compared to unc-5 mutants alone (Figs 10 and 11). This suggests that UNC-5 might have RHO-1-independent roles. Double mutants of unc-5 and dominant-negative rho-1 (T19N) and rho-1 (RNAi) showed significantly enhanced protrusion compared to single mutants, but did not exceed the additive effects of each (Fig 9F and 9G). This might reflect roles of these molecules that are independent of one another. Consistent with this notion, VD/DD lateral midline crossing axon guidance defects were significantly enhanced in unc-5 double mutants with rho-1 (T19N) and rho-1 (RNAi) (Table 1). The Collapsin-response mediator protein (CRMP) UNC-33 and the Ankyrin-like molecule UNC-44 are required for inhibition of growth cone protrusion of activated myr: : unc-40 and myr: : unc-5. Loss of unc-33 and unc-44 results in VD growth cones resembling unc-5 mutants, with increased protrusion, increased MT+ end accumulation, and loss of F-actin dorsal polarity [21,23]. Double mutants of unc-33 and rho-1 (G14V) resembled those of activated rho-1 (G14V) mutants alone, with a significant decrease in growth cone area and filopodial protrusions (Fig 11). Despite reduced protrusion and smaller growth cone size, EBP-2: : GFP puncta accumulation was increased in double mutants of unc-33 and rho-1 (G14V) (Fig 12). By contrast, double mutants of unc-44 with rho-1 (G14V) resembled unc-44 mutants, with excessive growth cone filopodial as evidenced with increased filopodial length and growth cone area, as well as an increase in EBP-2 puncta distribution (Figs 11 and 12). Double mutants of unc-33 and unc-44 with rho-1 (G14V) showed no significant change in F-actin distribution as compared to single mutants alone (Fig 12). These complex interactions reveal a differentiation of function between UNC-33/CRMP and UNC-44/Ankyrin in interaction with RHO-1 in growth cone morphology regulation. Expression of activated RHO-1 (G14V) resulted in VD growth cones with a marked decrease in growth cone protrusion and EBP-2 puncta distribution (Figs 1 and 2). Expression of the dominant negative form of RHO-1 (T19N) in the VD neurons and rho-1 (RNAi) resulted in increased protrusion and EBP-2: : GFP accumulation. MT+ ends in the growth cone periphery (Figs 1 and 2). Notably, neither dominant-negative RHO-1 (T19N) or rho-1 (RNAi) resulted in altered growth cone polarity and F-actin dorsal bias (Figs 1 and 2), suggesting that RHO-1 might specifically affect growth cone protrusion but not polarity. Previous work has identified roles of the Rho GTPases in regulation of both microtubules and actin [46]. RhoA has been shown to regulate formation of contractile actin structures such as stress fibers and promote stabilization of microtubules [47,48] through actomyosin contraction. In cultured growth cones, RhoA is involved in F-actin retrograde flow, wherein actin filaments in the periphery undergo constant retrograde transport to growth cone body [49–52]. RhoA activates RhoA kinase (ROCK), which activates contractility by phosphorylating the regulatory myosin light chain (MLC). This actin retrograde flow is thought to restrict MTs from the growth cone through physical association with these actin filaments undergoing retrograde flow, thereby reducing leading edge protrusion resulting in growth cone collapse and retraction [50,53]. Growth cone advance can occur when this actin-MT linkage is disrupted or when actin becomes attached to the substrate (the “clutch” hypothesis) [54] resulting in anterograde flow over the anchored actin filaments. One hypothesis explaining our results is that, in VD growth cones, RHO-1-mediated retrograde flow of actin restricts MT+ ends from the growth cones, and when RHO-1 activity is reduced, more MTs enter the growth cones resulting in increased growth cone protrusion. RHO-1 does not control growth cone polarity. We envision that it controls the general entry of pro-protrusive factors into the growth cone, possibly delivered to the growth cone by microtubules. The disposition of these pro-protrusive factors then depends on earlier growth cone polarity. In other words, where these pro-protrusive factors are active, at the dorsal leading edge, depends on growth cone polarity. When more pro-protrusive factors are delivered as a result of rho-1 loss, more protrusion occurs, but at the normal location. Loss of rhgf-1 resulted in increased growth cone protrusion and accumulation of EBP-2: : GFP, similar to but more pronounced than dominant-negative RHO-1 (T19N) and rho-1 (RNAi) (Figs 3 and 4). Furthermore, rhgf-1 mutants had no effect on growth cone polarity of protrusion or F-actin distribution (Fig 4). RHGF-1 might be an activator of RHO-1 to inhibit growth cone protrusion and MT accumulation. Consistent with this idea, activated rho-1 was epistatic to rhgf-1 loss-of-function (i. e. activating RHO-1 bypasses the need for RHGF-1). Growth cones in these double mutants displayed inhibited protrusion and reduction in MT distribution similar to activated rho-1 alone, suggesting that RHGF-1 acts as an upstream RHO-1 regulator in this process (Figs 5 and 6). Previous studies in Drosophila S2 cells have shown that the RHGF-1 homolog, DRhoGEF2, induces contractile cell shape changes by regulating myosin II dynamics via Rho1 pathway. Furthermore, DRhoGEF2 associates with tips of growing MTs and travels to the cell cortex [45]. In C. elegans, RHGF-1 functions through Rho and ROCK to activate the MAPKKK DLK-1 during MT disruption, triggering synaptic branch retraction and overgrowth of PLM neurites ultimately leading to neuronal remodeling [37]. Possibly, RHGF-1 activates RHO-1 to mediate a potential retrograde flow of F-actin to restrict MT accumulation in the growth cone. rhgf-1 loss-of-function suppressed the inhibitory effects of activated myr: : unc-40 and myr: : unc-5 on growth cones. Double mutant growth cones resembled those of rhgf-1 alone, with increased protrusion and EBP-2: : GFP puncta (Figs 7 and 8). That RHGF-1 is required for the effects of constitutively active MYR: : UNC-40 and MYR: : UNC-5 suggest that RHGF-1 acts downstream of MYR: : UNC-5 and MYR: : UNC-40. However, it is possible that RHGF-1 defines a parallel pathway. In any event, the inhibitory effects of MYR: : UNC-5 and MYR: : UNC-40 require functional RHGF-1. Receptors to several attractive or repulsive guidance cues signal through complex pathways through the Rho family of small GTPases to direct changes in growth cone cytoskeletal organization [55,56], and Rho activity is thought to be induced by “repulsive” cues [57]. Loss of the UNC-6/Netrin receptor unc-5 has been shown to cause excessively large VD growth cones with increased protrusion and excess EBP-2: : GFP accumulation [21,22]. If RHO-1 is activated by UNC-5, we expect that activated rho-1 (G14V) would be epistatic to unc-5 loss-of-function. This was not the case, as growth cones of rho-1 (G14V); unc-5 (lof) double mutants resembled those of unc-5 (lof) alone, with increased protrusiveness and EBP-2: : GFP accumulation (Figs 9 and 10). Possibly, loss of UNC-5 affects multiple parallel pathways, including RHO-1, and activation of the RHO-1 pathway alone cannot compensate for loss of UNC-5. Alternately, RHO-1 might act in parallel to UNC-5. That RHGF-1 function is required for the effects of activated MYR: : UNC-5 and MYR: : UNC-40 suggests that RHGF-1 (and by extension RHO-1) might, in part, act in the UNC-5 pathway directly. Previous studies have shown that the C. elegans UNC-33/CRMP is required in a pathway downstream with Rac GTPases for inhibition of growth cone protrusion in response to UNC-6/Netrin [23]. unc-33 loss-of-function mutants show large protrusive growth cones with excess EBP-2 accumulation in the growth cones, similar to unc-5. While activated RHO-1 (G14V) did not suppress the excessively-protrusive growth cones of unc-5 mutants, it did suppress those of unc-33 (Fig 11). Protrusion of growth cones of rho-1 (G14V); unc-33 double mutants resembled rho-1 (G14V) alone (i. e. protrusion was reduced and growth cones were small). Interestingly, despite their small size, inhibited unc-33; rho-1 (G14V) growth cones displayed increased EBP-2 puncta compared to wild-type animals, but significantly lower than unc-33 mutants alone (Fig 12). Thus, activated RHO-1 (G14V) can fully suppress excess protrusion, but not EBP-2: : GFP accumulation, of unc-33 mutants. Together, these results suggest that UNC-33 is required for activated RHO-1 (G14V) to restrict MTs from growth cones. They also suggest that RHO-1 has a role in protrusion that is independent of MT accumulation, as protrusion was reduced in rho-1 (G14V); unc-33 double mutants despite excess MT accumulation. UNC-44/Ankyrin is required to properly localize UNC-33/CRMP to the axons [58], and mutants are phenotypically indistinguishable in the VD growth cones (both are required to polarize protrusion and F-actin and to inhibit protrusion and EBP-2: : GFP accumulation) [21,23]. However, unc-44 loss was completely epistatic to activated RHO-1 (G14V), including both protrusion and EBP-2: : GFP accumulation. This suggests that UNC-44/Ankyrin has a role that is independent of UNC-33/CRMP involving non-MT-based regulation of protrusion. The FMO flavin monooxygenases inhibit growth cone protrusion with UNC-5 [30], possibly in an actin-based manner similar to MICAL [58,59]. Possibly, UNC-44/Ankyrin acts in this pathway or another independently from UNC-33/CRMP. Our results show that RHO-1 and the Rho activator GEF RHGF-1 are required to inhibit VD growth cone protrusion and to restrict EBP-2: : GFP puncta accumulation in growth cones, possibly downstream of the UNC-6/Netrin receptor UNC-5. One potential scenario for how these molecules interact is shown in Fig 13. UNC-5 might activate RHGF-1 and thus RHO-1, and UNC-33/CRMP might then be required to exclude MTs from growth cones in response to RHO-1 activation. In parallel, the Rac GTPases CED-10 and MIG-2 also act with UNC-33/CRMP to regulate MT exclusion [21]. CRMP interactions with Rho, actin, and microtubules have been documented in other systems. In cultured mammalian neurons, CRMP interacts with F-actin and with tubulin dimers to promote microtubule assembly [60,61], and expression of CRMP2 can alter Rho-GTPase-driven neurite morphology. Co-expression of Crmp-2 with activated Rho can promote cell spreading and neurite growth and this function of Crmp-2 is regulated by Rho Kinase [62]. Furthermore, CRMP-2 has been shown to be phosphorylated by Rho Kinase II [63,64] which disrupts the association of mature full-length CRMP-2 with tubulin heterodimers so that tubulin cannot be transported to the plus ends of microtubules for assembly [61] causing neurite retraction and growth cone collapse [65]. This reduced binding capacity to tubulin by phosphorylated CRMP-2, can be reversed by inhibiting RhoA activity [66]. Thus, RHO-1 may regulate growth cone protrusion and MT distribution through the phosphorylation activity of UNC-33/CRMP possibly through the same pathway or in parallel to it. If RHO-1 is indeed involved in F-actin retrograde flow, the role of UNC-33 might be to link F-actin to microtubules, such that in an unc-33 mutant, MTs are not excluded despite retrograde flow (including in the activated RHO-1 (G14V) background). RHO-1 might have an additional non-UNC-33 and non-MT-dependent role in inhibiting protrusion, along with UNC-44, possibly involving actin. RHO-1 is a key negative regulator of growth cone protrusion and MT accumulation that acts specifically in the protrusion aspect of the polarity/protrusion model of directed growth cone migration away from UNC-6/Netrin. The separability of growth cone polarity and protrusion indicate that these are controlled by distinct mechanisms. Possibly, short-range interactions with UNC-6/Netrin result in growth cone polarity, and longer-range interactions (e. g. diffusible UNC-6/Netrin) maintain polarity and regulate protrusion as the growth cone moves away from the UNC-5/Netrin source. In the SOAL and polarity/protrusion a models, chemotactic gradients are not rerquired to explain directed outgrowth. Experiments were performed at 20°C using standard C. elegans techniques [67]. Mutations used were LGIV: unc-5 (e53 and e152), unc-33 (e204), unc-44 (e362); lqIs128 [Punc-25: : myr: : unc-40] LGX: rhgf-1 (gk217, ok880 and gk292502), lqIs170 [rgef-1: : vab-10ABD: : gfp]. Chromosomal locations not determined: lqIs279 [Punc-25: : ebp-2: : gfp] by integration of lqEx809, lhIs6 [Punc-25: : mCherry], lqIs296 [Punc-25: : myr: : unc-5], lqIs312 [Punc-25: : rho-1 (G14V) ] by integration of lqEx1043, lqIs314 [Punc-25: : rho-1 (T19N) ] by integration of lqEx1070. Extrachromosomal arrays were generated using standard gonadal injection [69] and include: lqEx999 and lqEx1000 [Punc-25: : myr: : unc-40; Pgcy-32: : yfp], lqEx1131, lqEx1132, lqEx1133 and lqEx1134 [Punc-25: : rho-1 RNAi; Pgcy-32: : yfp], OX347 [Prgef-1: : vab-10ABD: : gfp; ttx-3: : rfp]. Multiple (≥3) extrachromosomal transgenic lines of Punc-25: : ebp-2: : gfp, Punc-25: : rho-1 (G14V) and Punc-25: : rho-1 (T19N) were analyzed with similar effect, and one was chosen for integration and further analysis. In wild-type, and average of 16 of the 19 commissures of the VD/DD axons are distinguishable, as commissural axons sometimes run together as a bundle and cannot be resolved. For these experiments, 100 animals were scored for an average total of 1600 axons. In Table 1, “% defective VD/DD axon guidance” includes axon wandering greater than 45 degrees laterally, axon branching, and premature axon termination. As axon guidance defects are nearly completely penetrant in unc-5 mutants, another guidance metric was used. “% failure to cross lateral midline” were axons that failed to extend dorsally past the lateral midline. Significance of difference was determined by Fisher’s Exact Test. VD growth cones were imaged and quantified as previously described [22]. Briefly, animals at ~16 h post-hatching at 20°C were placed on a 2% agarose pad and paralyzed with 5mM sodium azide in M9 buffer, which was allowed to evaporate for 4 min before placing a coverslip over the sample. Some genotypes were slower to develop than others, so the 16 h time point was adjusted for each genotype. Growth cones were imaged with a Qimaging Rolera mGi camera on a Leica DM5500 microscope. Images were analyzed in ImageJ, and statistical analyses done with Graphpad Prism software. As described in [22,23], growth cone area was determined by tracing the perimeter of the growth cone body, not including filopodia. Average filopodial length was determined using a line tool to trace the length of the filopodium. Unless otherwise indicated, ≥25 growth cones were analyzed for each genotype. These data were gathered in ImageJ and entered into Graphpad Prism for analysis. Analysis of Variance (ANOVA) was used to determine significance of difference between genotypes. Any of the VD growth cones visible at the time of imaging were scored (VD2-VD13), and we did not focus on any single VD growth cone for analysis. The F-actin binding domain of VAB-10/spectraplakin fused to GFP has been used to monitor F-actin in C. elegans [68,69]. We used it to image F-actin in the VD growth cones as previously described [22]. To control for variability in growth cone size and shape, and as a reference for asymmetric localization of VAB-10ABD: : GFP, a soluble mCherry volume marker was included in the strain. Growth cones images were captured as described above. ImageJ was used image analysis to determine asymmetric VAB-10ABD: : GFP localization. For each growth cone, five line scans were made from dorsal to ventral. For each line, pixel intensity was plotted as a function of distance from the dorsal leading edge of the growth cone. The average intensity (arbitrary units) and standard error for each growth cone was determined. For dorsal versus ventral comparisons, the pixel intensities for VAB-10ABD: : GFP were normalized to the volumetric mCherry fluorescence in line scans from the dorsal half and the ventral half of each growth cone. This normalized ratio was determined for multiple growth cones, and the average and standard error for multiple growth cones was determined. Statistical comparisons between genotypes were done using ANOVA on these average normalized ratios of multiple growth cones of each genotype. EBP-2: : GFP has previously been used to monitor microtubule plus ends in other C. elegans cells including neurons [70–72]. We constructed a transgene consisting of the unc-25 promoter driving expression of ebp-2: : gfp in the VD/DD neurons. In growth cones, a faint fluorescence was observed throughout the growth cone, resembling a soluble GFP and allowing for the growth cone perimeter to be defined. In addition to this faint, uniform fluorescence, brighter puncta of EBP-2: : GFP were observed that resembled the EBP-1: : GFP puncta described in other cells and neurons. For each growth cone, the perimeter and filopodia were defined, and the EBP-2: : GFP puncta in the growth cone were counted. For each genotype, the puncta number for many growth cones (≥25 unless otherwise noted) was determined. Puncta number displayed high variability within and between genotypes, so box-and-whiskers plots (Graphpad Prism) were used to accurately depict this variation. The grey boxes represent the upper and lower quartiles of the data set, and the “whiskers” represent the high and low values. Dots represent major outliers. Significance of difference was determined by ANOVA. We used a cell-specific transgenic RNAi approach as described previously [40]. Fragments of the rho-1 coding region was amplified by PCR and inserted behind the unc-25 promoter in a plasmid (primer and plasmid sequences available upon request). A “sense” and “antisense” orientation relative to the unc-25 promoter was isolated. An equimolar mixture of the sense and antisense plasmids was used to construct transgenic animals. These transgenic animals were predicted to express both sense and antisense RNAs driven by the unc-25 promoter in the VD/DD motor neurons, which was expected to trigger a double-stranded RNA response in these cells (RNAi).

Neural circuits are formed by precise connections between axons. During axon formation, the growth cone leads the axon to its proper target in a process called axon guidance. Growth cone outgrowth involves asymmetric protrusion driven by extracellular cues that stimulate and inhibit protrusion. How guidance cues regulate growth cone protrusion in neural circuit formation is incompletely understood. This work shows that the signaling molecule RHO-1 acts downstream of the UNC-6/Netrin guidance cue to inhibit growth cone protrusion in part by excluding microtubules from the growth cone, which are structural elements that drive protrusion.

Abstract Introduction Results Discussion Materials and methods

cell physiology invertebrates microtubules caenorhabditis enzymes enzymology neuroscience animals cell polarity animal models caenorhabditis elegans model organisms experimental organism systems genetic interactions cellular structures and organelles cytoskeleton research and analysis methods developmental neuroscience animal studies proteins animal cells guanosine triphosphatase biochemistry cellular neuroscience eukaryota hydrolases cell biology axon guidance phenotypes gene identification and analysis neurons genetics nematoda biology and life sciences cellular types organisms

2019

RHO-1 and the Rho GEF RHGF-1 interact with UNC-6/Netrin signaling to regulate growth cone protrusion and microtubule organization in Caenorhabditis elegans

Cytoplasmic transport of organelles, nucleic acids and proteins on microtubules is usually bidirectional with dynein and kinesin motors mediating the delivery of cargoes in the cytoplasm. Here we combine live cell microscopy, single virus tracking and trajectory segmentation to systematically identify the parameters of a stochastic computational model of cargo transport by molecular motors on microtubules. The model parameters are identified using an evolutionary optimization algorithm to minimize the Kullback-Leibler divergence between the in silico and the in vivo run length and velocity distributions of the viruses on microtubules. The present stochastic model suggests that bidirectional transport of human adenoviruses can be explained without explicit motor coordination. The model enables the prediction of the number of motors active on the viral cargo during microtubule-dependent motions as well as the number of motor binding sites, with the protein hexon as the binding site for the motors. The small number of motor proteins involved in microtubule transport implies a system where the fluctuations in the behavior of motors and the randomness of molecular reactions are essential characteristics [13] suggesting a stochastic modeling of the governing processes. Here we propose a stochastic representation of the main events involved in motor transport, namely stepping along microtubules and binding and unbinding of molecular motors to the cargo. The proposed model has six parameters, namely the binding, unbinding and stepping rates of plus-end and minus-end motors (herein presumed to be dynein and kinesin, respectively). The step sizes of the motors were set to −8/+8 nm for dynein/kinesin as suggested by the results of single molecule experiments [14], [15]. We note that we do not impose any geometrical information on the motors and their binding sites on the virus capsid. The motor protein binding sites on the adenovirus capsid are not known even though a recent cryo-EM image of the structure of the human adenovirus type 2 temperature sensitive mutant revealed the organization of the surface of the virus capsid [16]. The six model parameters are inferred through a system identification process using the velocity and displacement distributions of segmented trajectories as the cost function of our optimization. An evolutionary algorithm, capable of handling noisy cost functions, is used to obtain the rates that minimize the distance between the velocity and displacement distributions of the in silico and in vivo trajectories. The velocity distribution in virus trajectories has led to several suggestions regarding the cooperation or lack thereof between molecular motors. High velocities, in the order of a few microns per second, were observed for intracellular viruses (Fig. 2E) [17]. Similar high speeds have been observed for vesicles moving along microtubules such as peroxisomes [18] and endosomes [19]. These velocities are above the maximum velocities measured for single motors without load (3 µm/s for dynein, [14]; 0. 4 µm/s for kinesin-1, [20]; 3 µm/s for kinesin-1, [21]; 0. 8 µm/s for kinesin-1, [22]; 0. 8 µm/s kinesin-1 and 0. 5 µm/s kinesin-2, [23] in in vitro experiments. It has also been reported for drosophila lipid droplets, that multiple processive motors do not move cargoes faster [24]. It is likely that yet unknown mechanisms account for the high velocities measured in in vivo biological systems. These mechanisms may involve motors which are able to increase their velocities up to few microns per second or motors are able to act additively to achieve higher speeds. Both assumptions have not been experimentally validated or discarded in in vivo experiments. Additive behaviour of motors is an underlying assumption in our model (Fig. 2A). The additive behaviour is inherent to the Stochastic Simulation Algorithm [25] used herein to simulate the model, since the time step to the next event depends on the total propensity (numbers and event rates). The proposed stochastic model does not impose any explicit coordination between motor proteins, e. g. a switching mechanism that selects a certain motor protein type to be active. We emphasize that our model does not aim at a mechanistic description at the motor level. Forces are known to affect motor properties, but it is not clear how they are distributed among multiple motors [26]. Furthermore while it is possible to obtain data relating forces for certain motors in vitro, there is no such data for in vivo experiments. In the present model the forces between molecular motors and cargo are implicitly taken into account through the binding/unbinding/stepping rates of the stochastic models. The simulation of the stochastic model produces cargo trajectories that depend on the parameter settings. The model contains no a-priori assumptions on the existence of either a tug-of-war or coordination between molecular motors. In turn, the model parameters are systematically identified with a derandomized evolution strategy that minimizes the difference between the length and velocity distributions of directed motions (fast microtubule dependent runs [11]) produced by the model and those of fluorescently labelled human adenovirus type 2 (Ad2) as measured by confocal microscopy at 25 Hz temporal resolution. The two-dimensional virus trajectories are extracted by a single particle tracking algorithm [10] (Fig. 1A, B). Directed motions along microtubules are classified by trajectory segmentation [11] and the distance travelled along the microtubule is determined as a function of time (1D trajectory shown in Fig. 1C). The same analysis is applied to trajectories obtained by the simulation of our model using the Stochastic Simulation Algorithm (SSA) [27]. These trajectories are also subsequently segmented to classify directed motions [11]. In turn an optimization algorithm is employed to identify the parameters of the stochastic model [28]. Here the six model parameters (binding, unbinding and stepping for both kinesin and dynein, Fig. 2A) were identified by minimizing the Symmetric Kullback-Leibler divergence between the in silico and in vivo length and velocity distributions using an Evolution Strategy with Covariance Matrix Adaptation (CMA-ES) [29] (Fig. 2B, C). The proposed de-randomized optimization algorithm is an essential aspect of our method. CMA-ES samples the six-dimensional multivariate normal distribution involving the parameters of this problem at each iteration and it encodes relations between the parameters of the model and the objective that is being optimized without requiring explicitly the gradients of the cost function [29]. The CMA-ES is a method capable of optimizing noisy cost functions (such as those from the present stochastic model) and its efficiency, reliability and robustness were demonstrated over a number of benchmark problems [30], [31]. The CMA-ES is particularly suitable to this optimization problem as it is know to perform best [29] in problems that are low dimensional (here six parameters), inherently noisy (here a stochastic model), multimodal and computationally expensive (for each parameter set thousands of trajectories are generated and segmented to collect reliable statistics). The algorithm identifies an optimal set of parameters and at the same time provides a sensitivity analysis of the model. The standard deviations of the six principal axes are shown to converge (Fig. 3), thus yielding a minimum (Text S1). After the convergence of the optimization process (Fig. 3) we found that the directed motion length and velocity distributions of the in silico trajectories, under the optimal set of parameters, matched with high accuracy the experimental data (Fig. 2D, E). The maximum number of motors attached to the viral cargo is limited by the number of binding sites on the virus. The present model enables predictions on the number of motor binding sites on the viral capsid, a quantity that is difficult to determine experimentally but important for understanding the mechanisms of transport. We first estimated the number (between 2 and 20) of motor binding sites on the virus by an optimization procedure (Fig. 2F). In models with 6–16 binding sites, the cost function values were almost constant around the minimum value obtained for 14 binding sites (Text S1). For less than 6 motor binding sites, the optimization process did not converge to the experimentally observed directed motion length and velocity distributions. Above 16 binding sites, an unbalanced configuration of motors was feasible only at low binding and unbinding rates, and yielded largely unidirectional trajectories due to infrequent motor binding to the virus. We concluded that 14 common binding sites for dynein and kinesin correspond optimally to the experimental data. Since it is not possible to differentiate between common and separate binding sites, we additionally investigated the possibility that the experimental data support separate binding sites for the different motors. We optimized a model where dynein and kinesin have distinct binding sites, namely 4+4,5+5,6+6,7+7 binding sites, and various permutations thereof (Fig. 2F, Text S1), and found that an equal number of motor binding sites was optimal in all cases. This is consistent with the observation that center and periphery directed length and velocity distributions were almost symmetric (Fig. 2D, E). We note that the optimal number of binding sites, i. e. , 14, is the same for the models with common and separate binding sites (Fig. 2F, black curve). Molecular motors carrying cargo on microtubules operate as individuals or as an ensemble. We found that, on average, during virus directed motions, 1. 56±0. 56 dynein or kinesin (for minus-end and plus-end directed motions, respectively) motors, and 0. 15±0. 22 motors of opposite polarity were bound to the virus (Fig. 4A). The probability of binding more than four motors to one virus particle was below 10−3, and most often only one type of motor was bound (Fig. 4A, B). These data are in agreement with low number of motors estimated on vesicular cargo in squid axoplasm by cryo-EM [7]. For other organelles, the estimates for motor numbers range from a few to dozens based on immunological detections in chemically fixed cells. In order to quantify the correlation between the number of bound motors and the directed motion length, the Sample Pearson Product Moment correlation coefficient (with a range of 0 to 1, where 1 is maximal correlation) between motor numbers and directed motion length was computed to be 0. 51 for dynein and 0. 49 for kinesin for minus-end and plus-end directed motions, respectively. This implies a weak correlation between the number of bound motors and the directed motion length, showing that long runs do not necessarily require many motors, as two or three already account for lengths in the order of micrometers (Fig. 4B). This result is consistent with the recent in vitro observation that two motors are sufficient to move a cargo over several micrometers [32]. The velocities, derived from optimized stepping rates, for single dynein and kinesin motors were 866 nm/s and 833 nm/s, respectively, consistent with values reported for dynein and conventional kinesin-1 or kinesin-2 [21]–[23], [33]. Although kinesins are currently not known to be involved in cytoplasmic transport of adenovirus [1], the model makes a clear prediction for a plus end directed motor in cytoplasmic transport of adenovirus. Our findings indicate that microtubule-based motility of adenovirus requires a low number of bound motors compared to the number of binding sites on the capsid. This allows configurations where only one motor type is bound, and thereby produce directed motions. Low numbers of motors allow fast switches between directions and therefore, bidirectional motion. Importantly, the binding and unbinding rates were much smaller than the stepping rates, which is key for directed motion runs (Fig. 2C). Small perturbations of binding and unbinding rates greatly affect the model dynamics (Text S1). The susceptibility of motor based cargo transport to these parameters has been reported in other theoretical studies [26] and hints to a possible mechanism to regulate the run length of the motors [32]. The present results enabled an assessment on the virus binding sites for motor proteins. The outer surface of adenovirus particles is composed of five polypeptides, three of which are still present on cytosolic viruses that have undergone stepwise disassembly [34]. Cytosolic particles contain the major protein hexon, a large fraction of the pentameric penton base at the icosahedral vertex, and protein IX (pIX), which stabilizes hexon. By considering the size (90 nm in diameter) and icosahedral geometry of the virus and the cylindrical microtubule (25 nm in diameter), we can postulate that the maximum number of microtubule motor-capsid interactions occurs along the edge of a capsid facet, in this case on hexon (Fig. 5A, B). This arrangement implies that 9 hexon trimers are aligned with the microtubule, giving a maximum of 27 hexon binding sites for the motors. This is above the value of 14 binding sites predicted from the simulations. If we assume, however, that the motor protein binding sites are located at the interface of two trimeric hexons, one microtubule filament could cover 1–15 sites (Fig. 5B, red lines), which is within the predicted range of 6–16. In addition to hexon, 6 to 8 binding sites were available for pIX, and less than 5 for penton base which detaches to a significant extent from the incoming virions before reaching the cytosol [34]. We analyzed trajectories of pIX-deficient adenoviruses to distinguish between hexon and pIX binding sites for motor proteins [35]. The directed motion length and velocity distributions of pIX-deleted adenovirus were similar to those from wild-type viruses without significant deviations or asymmetries, indicating that pIX may not provide a binding site for microtubule dependent motors during cytoplasmic transport (Fig. 5C, D). Therefore, we predict that hexon harbours the binding sites for dynein and kinesin motors. In this study, we use in vivo imaging to identify a stochastic model of cargo transport by molecular motors on microtubules. The model parameters were systematically identified using live imaging data of virus trajectories and a de-randomized optimization algorithm to minimize the Kullback-Leibler divergence between the length and velocity distributions of adenovirus directed motions on microtubules with the in silico trajectories produced by the model. The model accounts for directed motions at µm/s speeds, processive stepping over hundreds of nanometres, and periods of stationary behaviour. The results show that the stochastic model can result in bidirectional support without an explicit coordination mechanism. In our work kinetic rates of a stochastic model are determined via an evolutionary optimization approach using experimental data. The identified model enables a number of predictions. First, it shows that one to four motors are active on virus particles during microtubule-dependent motions, although the number of motor binding sites is estimated to be 6–16. The observation that the cost function value is constant within this range suggests that the virus may align with the microtubule in different orientations (Fig. 5B) and still preserve its motility. This range is consistent with the maximum of 15 hexon trimer-trimer interfaces along the edge of a capsid facet. The low number of motors involved in directed motions supports an emerging concept from wet lab experiments and in silico simulations, that key events of cell functions are in many cases executed by only a few polypeptides [36]. Second, if equal numbers of opposite motors are attached, the cargo oscillates and eventually stops, or remains confined to small areas. This may be an important mechanism for fine-tuning the subcellular velocity to achieve localized delivery of the cargo. We anticipate that viral transport is tuned by the binding and unbinding rates of motors to microtubules or the cargo, rather than by additional regulatory factors. Such tuning could be cell-type specific [17], and could control the number of engaged motors and motor configuration, and also provide specific segregation or guidance cues for traffic. In support of this, it has been suggested that the microtubule binding protein Tau can fine-tune the distance that the cargo travels by reducing microtubule binding of kinesin in distal parts of neuronal axons [7], [37]. In addition, motor properties can be tuned by post-translational modifications, such as phosphorylation of dynein or kinesin binding partners, which could affect their enzymatic functions and hence their stepping rates [7]. We close by noting that besides the results on motor transport on microtubules, the algorithm taken here is in line with reverse engineering and systems identification approaches [28], [38]–[40] which are gaining significance as discovery and model validation tools in systems biology. The CMA-ES algorithm is capable of handling noisy and multimodal cost functions that are inherently associated with stochastic models. The CMA-ES optimization algorithm along with image analysis of in vivo systems can be a robust process to help identify parameters of stochastic models employed in several areas of systems biology. HeLa cells were grown to 30% confluency on 18 mm glass cover slips (Menzel Glaser) and kept in Hanks buffered salt solution containing 0. 5% BSA (Sigma) and 1 mg/ml ascorbic acid (Sigma). Adenovirus serotype 2 and protein IX deficient adenoviruses were grown, isolated, and labeled with atto565 (Atto-tec, Germany) as described by Nakano and Greber in [41] and Suomalainen et al. in [17]. HeLa cells were infected with Ad-atto565 and imaged between 30 and 90 minutes post infection at 25 Hz. Flat regions of the cell were chosen for imaging in order to minimize the cytoplasmic volume above the imaging plane. The center of the cell was detected by differential interference contrast imaging to assign directionality to the virus motions. Images were recorded using a spinning disc confocal microscope (Olympus IX81) fitted with an UplanApo100x objective of N. A. 1. 35 on a back-illuminated monochrome Cascade 512 EM-CCD camera (Photometrics) containing a 512×512 pixel chip (with 16×16 micrometer large pixels). For the computational methods see Text S1.

Molecular motors, due to their transportation function, are essential to the cell, but they are often hijacked by viruses to reach their replication site. Imaging of virus trajectories provides information about the patterns of virus transport in the cytoplasm, leading to improved understanding of the underlying mechanisms. In turn improved understanding may suggest actions that can be taken to interfere with the transport of pathogens in the cell. In this work we use in vivo imaging of virus trajectories to develop a computational model of virus transport in the cell. The model parameters are identified by an optimization procedure to minimize the discrepancy between in vivo and in silico trajectories. The model explains the in vivo trajectories as the result of a stochastic interaction between motors. Furthermore it enables predictions on the number of motors and binding sites on pathogens, quantities that are difficult to obtain experimentally. Beyond the understanding of mechanisms involved in pathogen transport, the present paper introduces a systematic parameter identification algorithm for stochastic models using in vivo imaging. The discrete and noisy characteristics of biological systems have led to increased attention in stochastic models and this work provides a methodology for their systematic development.

Abstract Introduction Results Discussion Materials and Methods

cell biology virology computational biology/systems biology computational biology

2009

A Stochastic Model for Microtubule Motors Describes the In Vivo Cytoplasmic Transport of Human Adenovirus

Dengue is a serious vector-borne disease, and incidence rates have significantly increased during the past few years, particularly in 2014 in Guangzhou. The current situation is more complicated, due to various factors such as climate warming, urbanization, population increase, and human mobility. The purpose of this study is to detect dengue transmission patterns and identify the disease dispersion dynamics in Guangzhou, China. We conducted surveys in 12 districts of Guangzhou, and collected daily data of Breteau index (BI) and reported cases between September and November 2014 from the public health authority reports. Based on the available data and the Ross-Macdonald theory, we propose a dengue transmission model that systematically integrates entomologic, demographic, and environmental information. In this model, we use (1) BI data and geographic variables to evaluate the spatial heterogeneities of Aedes mosquitoes, (2) a radiation model to simulate the daily mobility of humans, and (3) a Markov chain Monte Carlo (MCMC) method to estimate the model parameters. By implementing our proposed model, we can (1) estimate the incidence rates of dengue, and trace the infection time and locations, (2) assess risk factors and evaluate the infection threat in a city, and (3) evaluate the primary diffusion process in different districts. From the results, we can see that dengue infections exhibited a spatial and temporal variation during 2014 in Guangzhou. We find that urbanization, vector activities, and human behavior play significant roles in shaping the dengue outbreak and the patterns of its spread. This study offers useful information on dengue dynamics, which can help policy makers improve control and prevention measures. Dengue is a mosquito-borne disease caused by one of the four dengue virus serotypes (DENV 1–4), and is primarily transmitted by Aedes aegypti and Aedes albopictus [1,2]. The virus and its vectors are now widely distributed throughout tropical and subtropical regions, resulting in about half the world’s population being at risk of infection [1]. The World Health Organization (WHO) has estimated that 50–100 million infections occur annually in over 100 endemic countries [1,3]. More recently, Bhatt et al. took into account the inapparent infections and found that the global burden is probably much higher, at about 390 million infections per year [4]. The problem of dengue epidemics in China has intensified over the past two decades [5], and between 1991 and 2013 about 21,532 dengue cases and 620 deaths were reported. In 2014, the incidence reached a peak, with 46,864 reported cases, 80% of which were infected in Guangzhou. Dengue is not endemic in China [5], but the current situation has become more complicated, and the exact causes of the increase in incidences and the detailed transmission characteristics are unclear [5]. In this study, we aim to identify the spatio-temporal transmission patterns of dengue epidemics in Guangzhou in 2014 by addressing the following questions: How can we estimate the temporal and geographical distributions of infection cases? How can we evaluate the infection risk and the control effects? How can we assess the interactions of the factors involved among different geographical locations and thus infer the diffusion process from one location to another? The answers to these questions will be influential in epidemiological inference and public health planning. Clear information about disease burden and infection risk can help in correctly evaluating the effects of the factors involved and the correct allocation of resources for intervention [4]. An accurate reflection of the transmission dynamics and diffusion process of dengue can help us more fully understand and further predict the prevalence of epidemic propagation. There are, however, many challenges to be addressed, such as misreported surveillance data, obscure vector indices, the hidden effects of hosts and vectors, and the heterogeneous infection processes. These challenges and existing works related to our questions are discussed in more detail below. First, disease surveillance data is usually the baseline for estimating the infection burden, but it does not directly reflect the full extent of infection, for the following reasons: (1) multiple factors, such as inapparent infections, under-reporting, and misdiagnosis can lead to the misreporting of infected cases [4,6]; (2) the incubation period of the dengue virus can create a delay between infection and reporting; (3) human mobility can lead to the mis-registration of infection locations [7]. Feasible techniques to handle these issues have been proposed, such as analyzing index clusters and serologic testing to evaluate inapparent dengue infections [8], using epidemiological models with exposed states (e. g. , SEIR model) to account for the incubation period [9], and incorporating human mobility into the transmission model to examine its effect on dengue infection [10]. Epidemiological information can be extracted from hidden infections and unclear data through statistical and stochastic methods [11]. Ster et al. deployed reversible jump MCMC methods to reveal hidden infections and inferred the infectivity profile of the U. K. 2001 foot and mouth epidemic [12]. By fitting the partially observed data sequences of hospital infection, Cooper et al. estimated key epidemiological parameters using a structured hidden Markov model [13]. We take the distribution of reported cases as the essential data, and also incorporate the factors of incubation time lag, host mobility and reported rate, to estimate the 2014 actual dengue burden in Guangzhou. Second, dengue infection risk is in reality primarily evaluated by vector indicators, such as the house index (HI), the container index (CI), the Breteau index (BI), the pupa index (PI), and the adult productivity index [2]. However, the traditional vector indices (e. g. , HI, CI, and BI) have been shown to be poor proxies for measuring adult mosquito abundance and dengue risk, possibly due to the inadequate quality of the vector and incidence data, diversity of vector indices and adult vector densities, or to geographic/temporal mismatches of infection sites and index records [14–16]. Most vector indices only reflect vector prevalence rather than abundance [17], as they do not take into account the container type productivity. In view of this, other vector indicators have been suggested, such as sampling adult mosquitoes [14,15,18], or integrating vector indices and other information (e. g. , combing demographics and indices [18]. Beyond the vector indices, standard notions have been proposed to assess infection potentials, such as the basic reproduction number [19,20], the vectorial capacity (VCAP) [21–23], and the entomological incubation rate (EIR) [22,24]. In this study, we systematically integrate environmental and ecological factors and BI data to evaluate the adult vector densities. Third, the large-scale spread of dengue viruses is often caused by the spatial and temporal dynamics of vectors and hosts [25–27]. A combination of elements, including dense populations, frequent human-vector contact, rural-urban migration, serotype circulation, and inadequate infrastructure, can lead to dengue infection and mosquito breeding opportunities [16,28,29]. These factors are significant in shaping the spatial and temporal transmission of dengue epidemics. Two types of studies have been performed to reveal transmission process and assess the relevant factors [30,31]. First, mapping techniques and statistical methods can be used to process various data, such as geographic information systems method [32], time-series Poisson regression [29], Moran’s I statistic [33,34], and spatial scan statistics [35]. These methods can identify the hot spots, evaluate the relationship between different factors (e. g. , climate, imported cases and urbanization) and dengue incidence, and estimate the dispersion process [29,32–35]. The second type of analysis methods is primarily based on mathematical or computational models, and focus more on the intrinsic biting-based transmission process and the interaction between hosts, vectors, and viruses. These include differential equations [19,20], spatially agent-based transmission models [36], and metapopulation models [10]. These methods are able to estimate the infection capacity and simulate the time evolution and spatial diffusion of dengue epidemics [10,19,20,36]. However, it has been suggested that existing models less take into account the heterogeneity of mosquito densities/behaviors and mosquito-host encounters [31,37], so may not effectively reflect the spatial heterogeneity and temporal variation in the transmission [25,31]. In this study, we use mathematical models and computational methods to tackle the aforementioned problems. First, based on the Ross-Macdonald theory [9,21–23], a transmission model is defined to simulate the spatial diffusion of the dengue virus. Second, large-scale and consecutive vector indices, together with environmental and ecological information, are integrated to estimate the vector quantities. Third, human mobility as a spatio-temporal driver of dengue spreading dynamics [36], is estimated based on the radiation model proposed by Simini et al. [38]. Underlying transmission parameters are quantified by fitting the model to real-world observations using machine learning methods, such as the Markov chain Monte Carlo (MCMC) method [39]. The issue of incomplete surveillance data is addressed by comparing the estimated incidence rates and surveillance data with a reported rate. The research framework is shown in Fig 1. We conduct an empirical study in Guangzhou, where serious dengue epidemics have recently been experienced, particularly in 2014. Guangzhou is an international metropolis located in the tropical/subtropical region, and its climate and geography are ideal for vector growth and virus survival. In Guangzhou, the population is densely concentrated in the urban areas, with frequent movement between districts. By implementing the proposed model, we aim to (1) estimate the actual incidences and reveal the effects of environment and urbanization on vector activities; (2) evaluate the infection risk in terms of the basic reproduction number and explain the temporal pattern of infection potential in each district; (3) identify the underlying transmission process and mechanisms of dengue in Guangzhou. The city of Guangzhou (112°57′E to 114°3′E and 22°26′N to 23°56′N) is the capital of Guangdong province in South China, and has an area of 7434 square kilometers and about 12. 93 million residents. The climate is humid and subtropical, with high temperatures and humidity in summer, and comparatively mild and dry in winter. The annual mean temperature is 22°C and the annual accumulate precipitation is 1,800 mm. Guangzhou is an international port and an important foreign trade gateway into China. The above information is based on the Guangzhou government site (http: //www. gz. gov. cn/gzgov/s2289/zjgz. shtml). Guangzhou consists of 12 districts and is divided into three areas: urban, suburban, and exurban, which follows the city’s overall planning (2010–2015) and its Five-Year Plan (2011–2015) that take into account the urbanization, population density, and green coverage. The urban areas are Liwan, Yuexiu, Haizhu, Tianhe, and Baiyun (south of Liuxihe and north second ring), which account for 3. 8% of the city area, 46. 6% of the total population, and about 56% of transportation. The suburban areas are Panyu, Huangpu, Luogang, Huadu, and Baiyun (outside the central area), and the exurban areas are Nansha, Zengcheng, and Conghua. A detailed map is shown in Fig 2. To identify the underlying transmission patterns of dengue in 12 Guangzhou districts, the following data are collected. The parameters are set as follows. In this section, we propose a mathematical model integrating the geographic, transportation, demographic, environmental, and surveillance information in dengue transmission. The relationship between incomplete surveillance data and the estimated number of incidences is discussed below. Surveillance systems usually record disease incidences in different locations as a set of time series, so if we have observed the incidences of H locations during time period t = 1,2, ⋯, T, and the spatio-temporal surveillance data at time t are denoted by a vector Γt = (γ1t, γ2t, ⋯, γHt) T, then the correspondence of incidences between reporting and modeling can be quantified as follows: Γ t = δ ρ t + ε t, ε t ∼ N (0, Σ), (1) where δ is the reported rate, ρt = (ρ1t, ρ2t, ⋯, ρHt) T is the numbers of incidences derived from the modeling approach presented below, and ε is the error term, which follows normal distribution with variations Σ = diag (σ 1 2, σ 2 2, …, σ H 2). To specify and locate the infection events, the time step is set to be one day, and each is divided into daytime and nighttime, to take daily commuting and biting difference into account. For simplicity, the notations of subscript and superscript correspond to the district number and vector age, and the hat and check correspond to daytime and nighttime, respectively. According to the proposed model and the MCMC algorithm, the underlying model parameters are estimated, and the values are presented in Table 2. Based on the scale factors between BI and mosquito density (i. e. , K1, K2 and K3), it is observed that the aquatic habitats contain the highest concentration of larval mosquitoes in urban areas and the lowest concentration in exurban areas, but the difference between the exurban and suburban areas is not significant. This finding is consistent with the results in [44], where the authors have found that in urban areas of Guangzhou, the larvae and pupae of Aedes albopictus are more abundant in container habitats. The possible reason for the disparity of K is that the temperature, food sources, and types of aquatic habitats and containers vary between the urban, suburban, and rural areas [44]. The estimated number of bites per person per 12 hours by each female Aedes mosquito is shown in Table 2, presented as ai in district i. This rate is equal to the product of the human blood index (i. e. , the proportion of blood meals of mosquitoes taken from humans) and the mosquito feeding frequency, which is possibly associated with the status of demography, temperature, geography and environment [42]. In Guangzhou, the terrain slopes downward from the north to the south. The temperature usually falls 1–2 degree from the south to the north, but the urban heat island effect results in over 1. 7 higher degrees in the center areas in 2014. Low latitude and high temperature can cause frequent mosquito feeding. Specifically, our results indicate that urban areas possess high values of human blood index or mosquito feeding frequency. This is perhaps due to the dense population and the urban heat island effect. Tianhe is the most prosperous district (with the highest GDP) and one of the densely populated areas. Mosquitoes there prefer to bite humans frequently. In the suburban and exurban areas, however, probably due to various blood sources (e. g. , chicken, dogs, and cattle), cool temperature, high altitude, and the sparse population, the biting rate on humans is relatively low. The infection risk is usually evaluated by the basic reproduction number R0, defined as the expected number of secondary infections averagely generated by one case in a completely susceptible population [9,19,22]. R0 is widely used as an invasion threshold: if R0 is less than one, then the disease will become extinct; otherwise, there will exist an endemic state. In epidemiology, R0 reflects the biology of the transmission dynamics and quantifies the transmission potential of an epidemic. For vector-borne diseases, the basic reproduction number was first derived by Macdonald (1957) and Ross (1911) [9], based on which, we present the following formula: R 0 = 4 b c ϑ ∑ j = σ h - k V j, (13) where ϑ = ∑l lPI (l) = 5 days [48] is the average duration of human infection. Eq (13) is an evolutionary form of the basic reproduction number with time-varied VCAP. Averaging over the vector densities through the study period and inserting it into Eq (13), we obtain the mean value of the basic reproduction number R0. The evolutionary and average values of R0 in each district of Guangzhou are presented in Fig 3. It can be observed that R0 decreases from the largest value, 4. 4 (in Tianhe), in late September to less than 1 in early November, and the average value is between 1. 81–2. 59. We find that R0 is much higher but decreases more quickly in urban areas, which implies that the infection capacity is at first greater in urban areas, and the following intervention measures are effective there. Two peaks of R0 are observed in Huadu and Conghua, due to the increase in mosquito numbers. It should be noted that a larger R0 does not determine a higher incidence, and a rapid decrease of R0 does not means a rapid reduction in incidences, as the incidence rate is also dependent on the infection sources. Applying the estimated parameters to the proposed model, we obtain the following estimation: (1) The longitudinal number of infections in each district with the infection difference between daytime and night is as shown in Fig 4, in which the time corresponds to when people are bitten and get infected; (2) The longitudinal numbers of the incidences are as shown in Fig 5, in which the reported cases are demonstrated as a part of them; (3) The levels of the remote and local infections are as summarized in Table 3, in which the living areas and bit locations of the patients are estimated; (4) The spatio-temporal incidence rates are as shown in Fig 6. From the above Figures and Table, we observe heterogeneous and interesting patterns in dengue transmission in these districts, which are specified from the following aspects. First, the number of infections is estimated at about 113,108 cases from September 24 to November 9,2014, some of which are recorded in the surveillance system. Most of the unreported cases are inapparent. We can classify the 12 districts into 3 classes. Second, the patterns of infection differences between daytime and nighttime are shown in Fig 4. A slightly higher infection rate can be observed at night for residents in the urban center, which may be due to people returning to their living districts in late evening, and the infections risk is relatively high in urban areas. However, in most suburban and exurban areas, the lower infection risk creates a lower probability of getting infected at night. If the number of people leaving a certain district each day is larger than the number moving in (i. e. , in Baiyun, Panyun, Huadu, Nansha, Zengcheng, and Conghua), then the incidence rate in the daytime could be higher due to remote infections. Third, typical temporal patterns of infection are summarized as follows: Based on the surveillance data, the weight of human mobility, the quantity of remote incidences, and the arriving time of incidence peak, particularly the spatio-temporal incidence rates, we are able to identify the primary route of dengue diffusion in Guangzhou. It should be noted that the effects of human mobility are not just reflected in the remote infections, and by introducing infectivity to local Aedes mosquitoes, human mobility can lead to large number of autochthonous dengue cases. In this study, we have developed an inference technique to identify dengue transmission patterns and applied it to the 2014 dengue outbreak in Guangzhou, China. From this approach, we can improve our understanding of the dengue burden, infection risk, and the transmission dynamics in Guangzhou. Our results can help policy makers formulate effective measures to control and prevent dengue transmission. Our model is based on the Ross-Macdonald theory, which can be viewed as an epidemiological compartment model with an SEI infection process. The model closely combines four key sub-models necessary for describing the integrated dynamics of the system, namely, those representing mosquito population dynamics, human movement, virus transmission, and parameters estimation. First, we present a novel method to estimate the quantities of adult female mosquitoes from the BI data and environmental information. This approach differs from other studies that directly use vector indices and involved factors (e. g. , climate, sociodemographic indicators, and land-cover types) to estimate the potential dengue risk from a statistical perspective [14,16,26]. Second, based on the available transportation data, we use a standard radiation model to approximate the human mobility pattern [38]. As an inevitable component in dengue spatial transmission, human mobility can also be tracked by many other methods, such as GPS data [27], agent-based models, [25], and metapopulation models [10]. Next, we integrate well-recognized formulas (e. g. , the vectorial capacity [21–23] and the entomological incubation rate [22,24]) to elucidate the transmission process. We take into account the spatial heterogeneity of vector-host interactions, and the corresponding biting rates are estimated by MCMC methods. Our model further explore the real dynamics of disease transmission behind the observed incidences. The framework can also be applied to the space-time analysis of other vector-borne diseases. Based on our empirical study in Guangzhou, we find that the spatio-temporal distribution of incidences is extremely heterogeneous, with 81. 6% infections occurring in urban centers with different shapes of peak existing in mid-October. By considering the underlying dynamics, we observe temporal and spatial disagreement between infection cases and reported cases. The rank of disease burden in 12 districts is also inconsistent with the surveillance results. Further, We find that, in Guangzhou, the basic reproduction number R0 as an indicator of the infection risk decreases from the peak (3. 45) on September 22 to the trough (0. 73) on November 9,2014, with a mean value of 2. 24. This estimated R0 can be applied to quantify the infectivity in 12 districts and measure the effectiveness of the control strategies. From September, the Guangzhou government began to adopt various measures to control dengue transmission, such as disseminating knowledge about dengue through different media, asking every family to clean and clear their water containers, and organizing a sweep each Friday and regular spring-cleaning throughout the city. Consequently, we find that R0 begins to decrease from late September in most areas, particularly urban regions. However, due to a large number of infectious sources, the incidences decreased in about mid-October. This indicates that to control dengue transmission, intervention measures must be taken in a timely fashion. Due to the availability and validity of current surveillance data, the proposed models have certain limitations, which are worthy of further improvement and discussion: (1) People in Guangzhou are assumed to be without immunity against dengue; (2) The biological parameters are extracted from the literature (see Table 1), which may show geographical disparities; (3) The reported rate is used as the the proportion of symptomatic infections from [4]. Further experiments and survey are necessary to validate these parameters.

Dengue transmission is a spatio-temporal process with interactions between hosts, vectors, and viruses. Its transmission also involves multiple complex or even hidden factors, such as climate, social environment, vector ecology, and host mobility. These complexities make the underlying process of dengue transmission difficult to clarify. We address how the patterns of dengue transmission can be inferred by investigating the 2014 dengue outbreak in the city of Guangzhou, China, taking the available surveillance data and applying mathematical models and computational methods. We can then estimate the distribution of dengue infections and identify the transmission mechanisms. In our study, we systematically investigate the critical factors, enabling us to estimate the real patterns and dynamics of dengue transmission beyond the surveillance data.

Abstract Introduction Materials and Methods Results Discussion

invertebrates medicine and health sciences infectious disease epidemiology china vector-borne diseases geographical locations social sciences spatial epidemiology human mobility animals infectious disease control insect vectors human geography infectious diseases geography epidemiology disease vectors insects arthropoda people and places infectious disease surveillance mosquitoes asia urban areas earth sciences geographic areas disease surveillance biology and life sciences organisms

2016

Inferring the Spatio-temporal Patterns of Dengue Transmission from Surveillance Data in Guangzhou, China

Neutrophils and dendritic cells (DCs) converge at localized sites of acute inflammation in the skin following pathogen deposition by the bites of arthropod vectors or by needle injection. Prior studies in mice have shown that neutrophils are the predominant recruited and infected cells during the earliest stage of Leishmania major infection in the skin, and that neutrophil depletion promotes host resistance to sand fly transmitted infection. How the massive influx of neutrophils aimed at wound repair and sterilization might modulate the function of DCs in the skin has not been previously addressed. The infected neutrophils recovered from the skin expressed elevated apoptotic markers compared to uninfected neutrophils, and were preferentially captured by dermal DCs when injected back into the mouse ear dermis. Following challenge with L. major directly, the majority of the infected DCs recovered from the skin at 24 hr stained positive for neutrophil markers, indicating that they acquired their parasites via uptake of infected neutrophils. When infected, dermal DCs were recovered from neutrophil depleted mice, their expression of activation markers was markedly enhanced, as was their capacity to present Leishmania antigens ex vivo. Neutrophil depletion also enhanced the priming of L. major specific CD4+ T cells in vivo. The findings suggest that following their rapid uptake by neutrophils in the skin, L. major exploits the immunosuppressive effects associated with the apoptotic cell clearance function of DCs to inhibit the development of acquired resistance until the acute neutrophilic response is resolved. Leishmaniasis is a vector-borne disease initiated by the bite of an infected sand fly. Based on exhaustive findings in the murine model of cutaneous leishmaniasis due to Leishmania major, the clinical course of disease is thought to depend on the balance of activating cytokines, produced largely by Th1 cells, and deactivating cytokines, produced largely by Th2 cells and subsets of regulatory T cells [1]. Even in genetically resistant C57BL/6 mice, however, that develop self-limiting lesions due to a strongly polarized Th1 response, the early growth of the parasite is unrestrained, suggesting that innate killing mechanisms and the development of acquired resistance are avoided or delayed [2]. There is evidence that the acute neutrophilic response is itself critical to the early establishment of infection in the skin [3], [4]. Inoculation of L. major by the bite of a sand fly, or by needle injection, induces an intense infiltration of neutrophils that phagocytose the majority of parasites but fails to kill them, and neutrophil depletion prior to sand fly challenge leads to more rapid parasite clearance [5]. The manner in which the acute neutrophilic response inhibits the development of immunity to L. major infection is not understood. Neutrophils and DCs are normally located in distinct anatomical compartments, but converge at sites of inflammation in response to infection or tissue injury. The essential function of neutrophils in phagocytosis and killing of bacteria and in tissue repair is well described [6], [7]. Their additional role in modulating the adaptive response is suggested by their ability to release chemokines, cytokines, and anti-microbial peptides, [8], [9], and by more recent findings suggesting that activated neutrophils can deliver both activation signals and microbial antigens to DCs [10], [11]. By contrast, engulfment of apoptotic cells, including neutrophils, by DCs under steady state conditions has been shown to suppress DC maturation and is thought critical to the maintenance of peripheral tolerance [12]–[14]. Thus, the immunologic outcome of neutrophil - DC interactions may vary depending on the activation state of the neutrophils, their type of cell death, and the presence or absence of additional danger signals in the microenvironment in which these encounters occur. Importantly, the cross-talk between neutrophils and DCs has not been investigated in the context of any vector borne pathogen for which the co-localization of these cells at the site of transmission by bite or injection by needle in the skin is apt to be especially pronounced. In the present studies, we have monitored the sequence of inflammatory events following infection with L. major in the mouse ear dermis. We provide clear evidence that dermal DCs are preferentially infected via their capture of parasitized neutrophils in the skin, and that the Leishmania specific CD4+ T cell response is compromised until the acute neutrophilic response is resolved. We investigated the sequence of local inflammatory responses and identified the cells harboring L. major following injection of Lm-RFP metacyclic promastigotes (2×105) in the ear dermis of C57BL/6 mice. Myeloid populations were identified as CD11b+ cells, and further classified based on their expression of additional markers (Figure 1A) as follows: neutrophils (Ly6CintLy6G+, region 1); inflammatory monocytes (Ly6ChiLy6G−CD11c−MHCII−, region 2); monocytes/macrophages (Ly6ChiLy6G−CD11c−MHCII+, region 3); monocyte-derived DCs (Ly6ChiLy6G−CD11c+MHCII+, region 4); dermal DCs (Ly6C−Ly6G−CD11c+MHCII+, region 6); and dermal macrophages (Ly6C−Ly6G−CD11c−MHCII+, region 5). The cells in region 5 were uniformly F4/80+ (data not shown). The CD11b+ cells recovered from naïve ears included few neutrophils and inflammatory monocytes, and relatively greater numbers of dermal DCs and macrophages. The total number of CD11b+ cells recovered from the infected ears increased slowly over the first week, and expanded dramatically over the second week (Figure 1B). A prominent and transient neutrophil infiltrate accounted for the earliest increase in myeloid cells in the site, beginning at 1 hr, peaking at 12 hr, and dropping markedly between 1- 4 days (Figure 1C). Interestingly, neutrophils were found infiltrating the site again at day 7, and by day 14 their numbers exceeded the peak numbers observed during the first wave of the neutrophilic response. Comparison of L. major infected and sham injected mice demonstrated that at 1 hr the initial neutrophil infiltrate was induced, at least in part, by the tissue injury associated with the needle injection. At subsequent time points, however, the recruitment was dependent on the infectious status of the inoculum (Figure 1I). The increase in the number of inflammatory monocytes (Figure 1D) lagged slightly behind the neutrophil response, beginning at 12 hr and peaking at 24 hr. Similarly to the neutrophils, their numbers dropped markedly by 4 days but began to increase again by day 7. Very few of the CD11b+Ly6ChiLy6G− cells recovered from the site during the first week of infection were MHCII+ or CD11c+ (Figure 1F), of note because of recent findings implicating monocyte-derived DCs formed at the infection site as crucial to the induction of protective immunity during the active stage of disease [15]. The number of macrophages and DCs remained relatively unchanged from steady state conditions until 7 days post-infection, marking the onset of their massive accumulation in the site (Figure 1G and 1H). By analyzing the total population of RFP+ gated cells, we could follow the subsets of infected cells in the injection site over time (Figure 2A–C). Regions 1–6 define to the same subsets of myeloid cells as the corresponding regions in figure 1A, and in each case their CD11b expression was confirmed (data not shown). By contrast, many of the infected cells in region 7 were CD11b−, and their identity was not established using additional markers. Considering the total population of RFP+ cells (Figure 2D), low numbers were recovered at 1 and 4 hr which significantly increased between 4–12 hr and dramatically increased between 7–14 days (Figure 2D). In Figures 2E–L, the infected subsets are expressed both as a percentage of the total infected cells and their absolute numbers recovered from the ear dermis at each time point. Neutrophils were the predominant infected cells during the first 1–12 hrs (Figure 2E). At 12 hrs, 72% of the infected cells were neutrophils, with the remainder inflammatory monocytes, macrophages, DCs and other populations of CD11b+ and CD11b− cells. At 24 hr, neutrophils still represented approximately 32% of the total RFP+ cells. By day 4, the percentage of neutrophils in the RFP+ gate had dropped to fewer than 1%. Interestingly, their numbers began to increase again by day 7, and by day 14, the absolute number of infected neutrophils in the site exceeded the peak numbers observed during the first wave, although they remained <5% of the total population of infected cells. The inflammatory monocytes in the RFP+ gate also demonstrated two phases of recruitment, the first peaking at day 1 when they represented 22% of the total RFP+ cells, and the second at day 7 (Figure 2F). Their absolute numbers were greatest at day 14, again reflecting the massive expansion in the total number of infected cells at this time point. Very few of the RFP+ inflammatory monocytes recovered during the first 4 days were MHCII+ or CDllc+, while at 7 and 14 days, the majority of the RFP+ Ly6ChiLy6G− cells were MHCII+ and CD11c− (Figure 2G), reflecting the early stage of their differentiation to macrophages in the site. By day 14, appreciable numbers of infected monocyte-derived DCs (Figure 2H) were recovered, though they still represented only around 4% of the total RFP+ cells. By contrast, the infected macrophages (Figure 2I), expressed both as a percentage and absolute number of infected cells, started to increase at 4 days, and accounted for up to 20% of the total RFP+ cells at 14 days. The RFP+ dermal DCs remained few in number and <5% of the RFP+ cells over the first 24 hr, and while they remained a low percentage of the RFP+ cells at later time points, their absolute numbers increased markedly at 7 days and especially at 14 days post-infection (Figure 2J). In summary, our detailed analysis of infected cells in the L. major loaded dermis confirmed that neutrophils rapidly infiltrating the site represent the vast majority of infected cells over the first 12 hr, with the infections transitioning to inflammatory monocytes, and finally to monocyte derived macrophages and DCs during the active stage of disease. Given their predominance both as the earliest infiltrating and parasitized cells in the injection site, we investigated the influence of neutrophils on the subsequent program of infection and immune response. In vitro studies have suggested that macrophages can acquire L. major by phagocytosing infected, apoptotic neutrophils [16], [17]. To investigate the fate of infected neutrophils and their internalized parasites, Lm-RFP metacyclic promastigotes were injected into the ears of LYS-eGFP mice [18], in which neutrophils (CD11bhiGr-1hiF4/80−MHCII−), including those recovered from the skin, are eGFPhi [5]. eGFPhiRFP+ infected neutrophils were purified by cell sorting (Figure 3A) and injected into the ears of C57BL/6 mice. Analysis of a stained, cytospin preparation of the sorted cells just prior to injection indicated that approximately 30% of the parasites had already been released from the neutrophils during the 4–5 hr collection. Four hours after injection, the vast majority of the RFP+ cells recovered from the ear (90%) were found in an eGFP− population (Figure 3B), suggesting that in addition to the free parasites present in the inoculum, most of the remaining parasites were released from the infected neutrophils and available to be taken up by host cells in the skin. These cells were CD11clo, and their F4/80 and CD11b expression indicated that they were endogenous macrophages/monocytes or neutrophils (Figure 3D). Of the RFP+ cells that retained their eGFP fluorescence, approximately half appeared to be the injected population of intact, infected neutrophils (Figure 3C). The remaining RFP+eGFP+ cells were CD11c+, suggesting that the capture of infected neutrophils in the skin was largely accomplished by DCs. Of the total number of CD11c+RFP+ cells, 68% were eGFP+ (Figure 3E), suggesting that most of the DCs acquired their parasites via uptake of infected neutrophils. Of note, the eGFP fluorescence in these cells was reduced relative to the starting population of infected neutrophils. To rule out the possibility that the eGFPhiRFP+ infected neutrophils could have differentiated into CD11c+ cells, or that a small contaminating population of CD11c+ cells in the purified eGFPhiRFP+ infected neutrophils was responsible for the RFP+eGFP+CD11c+ cells observed in figure 3B, we sorted eGFPhiRFP+ infected neutrophils (donor, CD45. 2), and injected them into the ears of B6SJL mice (host, CD45. 1). Analysis of CD45. 1 expression on the subpopulations of RFP+ cells indicated that virtually all of the RFP+eGFP+CD11c+ cells were CD45. 1+, ruling out their donor origin (Figure S1). To investigate whether DCs might favor engulfment of infected neutrophils over uninfected neutrophils in the skin, equal numbers of eGFPhiRFP− uninfected and eGFPhiRFP+ infected neutrophils (Figure 3A) were co-injected into the ears of C57BL/6 mice. The analysis of eGFP+ gated cells recovered four hours later confirmed that DCs are able to take up neutrophils in vivo, representing approximately 11% of the eGFP+ cells (Figure 3F). Importantly, and despite their exposure to equivalent numbers of infected and uninfected neutrophils, an average of 66% of the CD11c+eGFP+ cells were RFP+ (Figure 3F and G), indicating that the dermal DCs favored the uptake of the infected neutrophils. The capture of mammalian cells by DCs and others phagocytes is a common event during tissue remodeling and at infection sites, where cells dying by apoptosis expose signals that are recognized for engulfment. The best studied ‘eat me’ signal on apoptotic cells is phosphatidylserine (PtdSer) whose outer membrane exposure can be quantified by staining with Annexin V. When neutrophils were recovered from the ear dermis of LYS-eGFP mice 12 hr after infection, 53% of the infected neutrophils compared to 19% of the uninfected neutrophils were Annexin V+ (Figure 4A and B). TUNEL staining confirmed a higher degree of apoptosis in the infected population of neutrophils recovered from the injection site (Figure 4C and D). These findings suggest that the uptake of L. major leads to accelerated apoptosis and earlier exposure of PtdSer on neutrophils infiltrating the injection site, which may favor their recognition and capture by DCs in the skin. To investigate neutrophil - DCs interactions following injection of the parasite directly, we evaluated dermal DCs recovered from C57BL/6 mice 24 hr after infection with Lm-RFP parasites, and stained for neutrophil derived-myeloperoxidase (MPO) and elastase (NE). In addition, mice were treated with two neutrophil-depleting antibodies: the GR-1 specific antibody RB6-8C5, which recognizes an epitope shared by Ly6G and Ly6C, and the Ly6G specific antibody, 1A8. Administration of 1A8 one day before infection depleted 85% of the CD11b+GR1hiLy6Cint neutrophils present in the ear dermis 24 hr after infection (Figure 5A and C). The remaining neutrophils showed lower GR1 staining, likely due to competition with the surface bound 1A8 antibody. The CD11b+GR1intLy6Chi population was unaffected. By contrast, and consistent with the prior reports [19], [20], the RB6-8C5 antibody depleted both neutrophils and a population of inflammatory monocytes (Figure 5B and C). Furthermore, the neutrophil depletion achieved using RB6-8C5 was virtually complete (99%). Neither reagent affected the total number of DCs recovered from the ear at 24 hr, or the number RFP+ DCs as a percentage of the total population of RFP+ cells (Figure 5D and E). Gating on RFP+ or RFP− dermal DCs (Figure 5F), MPO staining on cells recovered from the control treated mice was observed in an average of the 58% of the RFP+ DCs, suggesting that the majority of the infected DCs acquired their parasites via uptake of infected neutrophils (Figure 5G and H). By contrast, only a low proportion (<5%) of the RFP−DCs were MPO+, although because far more RFP−DCs were recovered from the site compared to RFP+DCs (Figure 5F), the percentage of RFP−DCs staining for MPO was on average 60% of the total population of MPO+ DCs (data not shown). The intracellular MPO staining in the majority of the infected DCs was comparable to the MPO staining observed in the neutrophils themselves, and greater than the MPO staining observed in the inflammatory monocytes recovered from the site (Figure S2), reinforcing the conclusion that the acquisition of the MPO marker by infected DCs was due to their uptake of infected neutrophils. Importantly, the RFP+ DCs recovered from the RB6-8C5 treated mice were virtually all MPO−, confirming that the uptake of parasites in the absence of neutrophils or inflammatory monocytes does not upregulate the expression of MPO in the DCs. The number of RFP+ DCs recovered from the 1A8 treated mice that stained for MPO was also significantly reduced, though an average of 24% of the cells were still MPO+ cells, consistent with the incomplete neutrophil depletion using this antibody (Figure 5G and H). Staining for NE, while relatively weak compared with MPO, reinforced the MPO result in that the majority of the RFP+ DCs recovered from the non-depleted mice stained positive for NE (Figure S3). Finally, RFP+ DCs recovered from control treated mice 14 days after infection were mainly MPO− (Figure 5I), suggesting that following the resolution of the acute neutrophilic response, infected neutrophils were no longer the main source of parasite delivery for DCs in the skin. We further characterized the possible subsets of the RFP+ DCs recovered from the site based on their expression of Langerin and CD103. As reviewed [21], and confirmed in our analysis of the DCs recovered from the ear dermis 24 hr after infection, the DC subsets include Langerhans cells (LC) and migratory LC (CD11c+MHCII+Lang+CD103−), Langerin+ DC (CD11c+MHCII+Lang+CD103+), and Langerin− DC (CD11c+MHCII+Lang−CD103−) (Figure S4). The RFP signal was associated exclusively with the Langerin− DCs. To address whether neutrophils might modulate the antigen presentation functions of DCs during the early stages of infection, the expression of activation markers on infected DCs recovered from the ear dermis 3 days after infection in neutrophil-depleted (RB6-8C5) or control treated C57BL/6 mice was compared (Figure 6A–C). Expression of MHC class II, CD86 and CD40, but not CD80, was increased on RFP+ DCs recovered from the neutrophil depleted mice (Figure 6B and C). Functional studies involving these infected DCs required pooling of dermal cells from 10 mice (20 ears) for each treatment group in order to obtain a sufficient source of antigen and antigen presenting cells for the co-culture assays. Using CD11c+ RFP+ cells that were normalized for their RFP signals by cell sorting (Figure 6D), the infected DCs from neutrophil depleted mice were more efficient than the infected DCs from the control treated mice in activating Leishmania-primed T cells from healed mice to secrete IFN-γ, observed in two independent experiments (Figure 6E). To evaluate the influence of neutrophils on CD4+ T cell priming to L. major- derived antigen in vivo, B6. SJL congenic mice were depleted of neutrophils 24 hr prior to infection with L. major SP-OVA or control 3′NT transgenic parasites in the ear. CFSE-labeled, naïve OT-II CD4+ T cells specific for OVA were adoptively transferred into the same recipients. Draining lymph nodes were harvested on day 6 and dilution of CFSE fluorescence was determined on CD45. 2+ and CD4+ gated cells. Infection of control treated mice with Lm SP-OVA failed to induce OT-II proliferation above the background levels (6–7%) observed in control treated or neutrophil depleted mice infected with Lm 3′NT (Figure 6E). By contrast, mice treated with 1A8 and RB6-8C5 had an average of 20% and 34% of the gated cells in division, respectively (Figure 6F). We also assessed the ability of CD45. 2+ OT-II CD4+ cells to produce IFNγ, IL-10 and IL-17, following ex-vivo restimulation with PMA/ionomycin for 4 hours in the presence of brefeldin-A. A percentage of proliferating CD45. 2+ OT-II CD4+ cells from 1A8 and RB6-8C5 treated mice (29% and 25%, respectively) produced IFNγ (Figure 6G). Neither IL-10- nor IL-17A-producing T cells were detected (data not shown). The influence of early neutrophil depletion on CD4 priming was no longer apparent when OT-II cells were transferred 14 days post-infection with Lm SP-OVA (Figure 6H), at a time when infected DCs no longer harbored neutrophil markers (Figure 5I). Taken together, these findings suggest that the favored uptake of infected neutrophils by dermal DCs effectively prevents the activation of Leishmania-specific CD4+ T cells until the acute neutrophilic response is resolved. We have recently described the efficient capture of L. major metacyclic promastigotes by neutrophils at the site of needle inoculation or infected sand fly bite, and the powerful effects of early neutrophil depletion in promoting rather than compromising host resistance to sand fly transmitted infection [5], [22]. The current studies provide an underlying mechanism to explain the immunomodulatory role of neutrophils in the L. major loaded dermis. Under steady state conditions, DCs are strategically positioned in peripheral and lymphoid tissues to sense microorganisms and endogenous stress signals, including apoptotic cells. Neutrophils, by contrast, are present mainly within the blood, and circulate in a non-activated state with a half-life of 6–7 hrs. Following inoculation of L. major into the skin by needle or by the bite of an infected sand fly, the parasites are taken up by neutrophils that are rapidly recruited to and accumulate with DCs at the injured site. We observed that phagocytosis of L. major significantly accelerated the rate of neutrophil apoptosis, which was associated with the favored uptake of infected over uninfected neutrophils by DCs in the skin. More importantly, for the majority of infected DCs in the skin their initial encounter with the parasite occurred via capture of infected neutrophils, with a negative impact on CD4+ T cell priming. These studies confirm the previous findings in L. major [5], [23], recently extended to L. infantum [24], that neutrophils are rapidly recruited to and accumulate in the inoculation site, and represent the predominant parasitized cell during the first 1–12 hours of infection in the skin. The inflammatory and infectious process induced by L. major in the skin may be regulated in a tissue specific manner, since recent observation by Gonçalves et al. [25] and confirmed by our own studies (data not shown) have revealed that when L. major metacyclics are introduced into the peritoneal cavity, neutrophils are neither the first infiltrating nor predominant infected cells. Our kinetic analysis of the L. major loaded dermis revealed that the rapid neutrophilic response is initiated in part by signals generated by the tissue injury produced by the needle injection itself, since a transient recruitment was observed in sham injected mice, and amplified by more durable signals derived from the parasite and/or from infected cells [26]. The fate of the infected neutrophils was followed by transfer of eGFPhiRFP+ cells into the ear dermis of C57BL/6 mice. By 4 hr, the majority of RFP+ cells recovered from the site were endogenous neutrophils and monocytes/macrophages that were eGFP−, consistent with our prior in vivo imaging results that readily captured infected neutrophils undergoing apoptosis and releasing viable parasites for subsequent uptake by other cells in the skin [5]. Thus, the ‘Trojan Horse’ hypothesis as originally proposed [17], in which neutrophils serve as a vector for silent entry of Leishmania into macrophages, has not been directly substantiated in these studies. We cannot, however, dismiss the possibility that phagosomal degradation of the eGFP signal occurred rapidly following engulfment of the infected neutrophils by macrophages. It is also possible that clearance of neutrophil-derived, apoptotic bodies by infected macrophages would still contribute to their deactivation and promote the intracellular survival and growth of the parasite, as proposed. By contrast to macrophages, the evidence for the uptake of L. major infected neutrophils by DCs in the skin seems clear. Firstly, CD11c+ cells were the only endogenous cells associated with both the RFP and eGFP signals. Secondly, when the infections were initiated by RFP L. major metacyclics, the majority of the RFP+ DCs recovered from the injection site at 24 hr also stained positive for neutrophil-derived MPO and elastase. In studies by Ng et al. [27], two-photon imaging captured dermal DCs but not Langerhans cells taking up Leishmania promastigotes in the skin. We also found Langerin− dermal DCs as the major infected DC subset in the skin, but conclude based on their staining for neutrophil markers, and the absence of these markers in DCs that have taken up parasites in the absence of neutrophils, that the majority of the infected DCs acquired their parasites via engulfment of infected neutrophils. Favored uptake of infected over uninfected neutrophils was also observed, correlated with their accelerated expression of apoptotic markers that may have targeted them for more efficient recognition and clearance by DCs. Neutrophil ingestion of other microbial pathogens, notably E. coli [28], Str. pneumoniae [29], [30], C. albicans [31], Sta. aureus [32], and M. tuberculosis [33], has also been found to accelerate their apoptotic program. The findings involving Leishmania are inconsistent on this point, with delayed or enhanced expression of PtdSer observed on neutrophils obtained from human blood or the mouse peritoneal cavity and exposed to Leishmania in vitro [34]–[37]. The current studies are the first to compare the apoptotic profile of tissue infiltrated neutrophils that have taken up parasites, or not, in the inflamed dermis. Apoptosis is an active process to regulate cellular homeostasis. Efferocytosis refers to the capture of apoptotic cells by phagocytes, primarily macrophages and immature DCs (iDC), and is itself thought to be a homeostatic mechanism to resolve inflammation and to maintain peripheral tolerance [13]. Recognition and engulfment of apoptotic cells, including apoptotic neutrophils, by DC is known to inhibit their production of pro-inflammatory cytokines, expression of costimulatory molecules, and their ability to stimulate T-cell proliferation [14], [38], [39]. The exploitation of these inhibitory signals by microbial pathogens is suggested by in vitro studies showing that M. tuberculosis-induced activation of human iDC can be inhibited by their co-culture with apoptotic neutrophils [40], and that Plasmodium falciparum-infected erythrocytes can inhibit the maturation of mouse DCs by binding to CD36, a known recognition receptor for apoptotic cells [41]. The present studies are the first to demonstrate efferocytosis involving neutrophils and DCs in an infection driven inflammatory setting in vivo. The sequestration of Leishmania antigens within apoptotic neutrophils would seem an especially efficient process to exploit the immunosuppressive signals conferred by the clearance of dying cells by DCs. Removing host neutrophils as a source of apoptotic cells was sufficient to reconstitute the immune function of infected DCs. It should be noted that in contrast to recent studies [42], we did not observe a reduction in either the total number of DCs nor infected DCs recovered from the ear following neutrophil depletion (Figure 5D and E). We would offer that while the prior study was confined to cells migrating out of the ear dermis ex vivo, our analysis was based on the greater recovery of cells following enzymatic digestion of the tissue. By comparing the ex vivo APC function of infected DCs recovered from the skin of mice depleted or not of neutrophils, and normalized for their RFP signals, the inhibitory effects of neutrophil uptake on DC maturation and Leishmania specific T cell activation could be formally demonstrated. The consequence of this inhibition in effectively delaying the onset of Leishmania specific T cell priming in vivo was directly supported by the enhanced, early OT-II priming to Lm-derived OVA in the neutrophil depleted mice. The neutrophil - DC interactions that inhibit T cell priming following needle challenge with L. major might be relevant to more general vaccination protocols in which an acute neutrophilic infiltrate accumulates at the site of antigen deposition. A recent report by Yang et al. [43] described the negative influence of neutrophils on the T and B cell responses to protein antigens administered by needle in the footpad. It is clear, however, that apoptotic neutrophils can also provide a source of immunogenic molecules to DC, especially for cross-priming, and especially if accompanied by extrinsic maturation signals [44], [45]. The relative paucity of activation signals associated with the phagocytosis of Leishmania promastigotes by neutrophils is suggested by the fact that the parasite traffics to a non-lytic compartment, avoids activation of the NADPH oxidase, and survives capture by these cells [5], [37]. It should be noted that PtdSer exposure on the parasites themselves has been suggested to facilitate their silent entry into macrophages, [46], [47], and may be especially relevant to their initial survival in neutrophils. Following neutrophil depletion, or the resolution of the first wave of neutrophils in the site, the majority of the infected DCs recovered from the skin lacked neutrophil markers, and are presumed to have taken up the parasite directly. By contrast to the absence of activation signals associated with the direct uptake of L. major metacyclic promastigotes by macrophages, the activation of human and mouse DCs following their phagocytosis of these organisms in vitro is well described [48]. Direct uptake might allow for parasite antigens to be more accessible to the MHC class I and II processing machinery, for parasite encoded TLR agonists to more efficiently engage their respective receptors, and for activation pathways to proceed in the absence of the inhibitory signals induced by apoptotic cell clearance. By two weeks, the priming conditions had clearly improved, and neutrophil depletion did not further enhance the CD4+ T cell response, despite the reappearance of neutrophils in the site. In contrast to the initial wave, however, the infected neutrophils recovered at two weeks represented a small percentage of the total population of infected cells, and the majority of infected DCs no longer harbored neutrophil markers. It is likely that the conditions of neutrophil recruitment to and activation in the skin during the active stage of disease, possibly Th17 driven at this later time, are distinct from those associated with the acute infiltrate, and that the influence of these respective neutrophil populations on the anti-leishmanial response will also be distinct. In the current studies, there was a significant difference in the effects of the neutrophil depleting antibodies, 1A8 and RB6-8C5, in potentiating the early OT-II response to infection with Lm SP-OVA in the skin. The 1A8 treatment critically confines the enhanced priming observed to specific depletion of Ly6G+ neutrophils. The more powerful effects observed with the RB6-8C5 antibody is consistent with the more efficient depletion of neutrophils that was achieved, although the removal of an additional population of GR-1+ myeloid cells with suppressor activity [49], [50] cannot be discounted. While our studies have employed a relatively high dose, needle challenge in order to recover a sufficient number of infected cells from the ear dermis for analysis, it should be emphasized that the initial wave of neutrophil recruitment to the infected sand fly bite site is more massive, localized, and sustained compared to the needle injection site [22]. This may explain why the ablation of the early neutrophilic response had such a strong effect in promoting protection against sand fly transmitted infection as compared to needle challenge [5], [51]–[53]. Thus, the impact of the early neutrophil - DC interactions described in these studies may be especially relevant to the inflammatory conditions elicited by natural sand fly transmission, as well as to that of other vector borne pathogens, in promoting the early establishment of infection and the progression of disease. This study was carried out in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. The protocol was approved by the Animal Care and Use Committee of the NIAID, NIH (protocol number LPD 68E). All mice were maintained at the NIAID animal care facility under specific pathogen-free conditions. Female C57BL/6 and B6. SJL congenic mice, and RAG1-deficient OT-II CD4+ TCR transgenic mice were purchased from Taconic Laboratories. C57BL/6 LYS-eGFP knock-in mice [18] were a gift from T. Graf (Albert Einstein University, NY) and were bred at Taconic Laboratories. Experiments were carried out using different lines of L. major: L. major Friedlin strain FV1 (MHOM/IL/80/FN); a stable transfected line of L. major FV1 promastigotes expressing a red fluorescent protein (Lm-RFP), L. major FV1 promastigotes expressing a portion of the ovalbumin gene encoding amino acids 139 to 386 containing the class II restricted epitope recognized by OT-II TCR transgenic CD4+ T cells (Lm- SP-OVA), and L. major V1- transfected with the control plasmid expressing Leishmania donovani 3′ nucleotidase-nuclease (Lm-NT). Transfected lines were generated as described previously [54]–[55]. Parasites were grown at 26°C in medium 199 supplemented with 20% heat-inactivated FCS (Gemini Bio-Products), 100 U/ml penicillin, 100 µg/ml streptomycin, 2 mM L-glutamine, 40 mM Hepes, 0. 1 mM adenine (in 50 mM Hepes), 5 mg/ml hemin (in 50% triethanolamine), 1 mg/ml 6-biotin (M199/S), and 50 µg/ml of Geneticin (Gibco). Infective-stage, metacyclic promastigotes of L. major were isolated from stationary cultures (4–5 days old) by negative selection using peanut agglutinin (PNA, Vector Laboratories Inc). For flow cytometric studies of dermal and draining lymph node cells, mice were infected with the specified number of metacyclic promastigotes in the ear dermis by i. d. injection in a volume of 10 µl. In parallel, sham mice received i. d. injection of DMEM in a volume of 10 µl. To obtain chronically infected mice, animals were infected 16–20 weeks previously with 104 L. major FV1 metacyclic promastigotes in the left hind footpad. Ear tissue was prepared as previously described [2]. Briefly, the two sheets of infected ear dermis were separated, deposited in DMEM containing 100 U/ml penicillin, 100 µg/ml streptomycin, and 0. 2 mg/ml Liberase CI purified enzyme blend (Roche Diagnostics Corp.), and incubated for 1 h and 30 min at 37°C. Digested tissue was placed in a grinder and processed in a tissue homogenizer (Medimachine; Becton Dickenson). Retromaxillary (ear) lymph nodes were removed, and mechanically dissociated using tweezers and a syringe plunger. Tissue homogenates were filtered through a 70 µm cell strainer (Falcon Products). Single-cell suspensions were incubated with an anti-Fc-γ III/II (CD16/32) receptor Ab (2. 4G2, BD Biosciences) in RPMI without phenol red (Gibco) containing 1% FCS and stained with fluorochrome-conjugated antibodies. The following antibodies were used: APC- anti-mouse CD11c (HL3, BD Biosciences), PE-Cy7- anti-mouse CD11c (N418, eBioscience), PerCP-Cy5. 5 or PE-Cy7- anti-mouse CD11b (M1/70, eBioscience); PerCP-Cy5. 5- anti-mouse Ly6C (HK1. 4, eBioscience); FITC- anti-mouse Ly6G (1A8, eBioscience); FITC- anti-mouse GR-1 (RB6-8C5, BD Biosciences); eFluor anti-mouse F4/80 (BM8, eBioscience), Alexafluor-700 anti-mouse MHC II (M5/114. 15. 2, eBioscience), APC- anti-mouse CD103 (M290, eBioscience), A488- anti-mouse Langerin (929F3. 01, Dendritics), APC- anti-mouse CD40 (1C10, eBioscience), FITC- anti-mouse CD80 (16-10A1, eBioscience), PerCP-Cy5. 5- anti-mouse CD86 (GL-1, BioLegend), APC- anti-mouse CD4 (RM4–5, eBioscience), PerCP-Cy5. 5- anti-mouse CD45. 2 (104, eBioscience); APC-eFluor 780 anti-mouse CD45. 1 (A20, eBioscience), FITC- anti-mouse myeloperoxidase (MPO) (8F4, Hycult), anti-human neutrophil elastase (NE) (H-57, Santa Cruz), FITC conjugated using and amine reactive probe (Sigma-Aldrich). The isotype controls used (all obtained from BD Biosciences) were rat IgG1 (R3–34) and rat IgG2b (A95-1). The staining of surface and intracytoplasmic markers was performed sequentially: the cells were stained first for their surface markers, followed by a permeabilization step with BD Cytofix/Cytoperm (BD Biosciences) and staining for Langerin, MPO or NE. For intracellular detection of cytokines, cells were first stimulated with Leukocyte Activation Cocktail, plus GolgiPlug (BD Biosciences) according the manufacturers' instructions for 4 h in vitro. Following surface staining and permeabilization, cells were then stained with a combination of anti-mouse antibodies: PerCP-Cy5. 5 anti-IL17A (eBio17B7, eBioscience) APC anti-IFN-g (XMG1. 2, eBioscience), PE anti-IL-10 (JES5-16E3, BD Bioscience) in Perm/Wash buffer (BD Bioscience). Intracellular staining was carried out for 30 minutes on ice. The data were collected and analyzed using CELLQuest software and a FACScalibur or FacsDIVA software and a FacsCANTO flow cytometer (BD Biosciences). Neutrophils, dendritic cells, macrophages and monocytes from the ear dermis were identified based on size (forward scatter) and granularity (side scatter) and by surface phenotype as indicated in the text and figure legends. Infected DCs (CD11c+RFP+) were purified using a FACSVantage or a FACsAria (BD Biosciences) cell sorter on cells recovered from the ear dermis 3 days after infection with 2×106 L. major-RFP. For the analysis of the capacity of infected, dermal DCs to induce the secretion of IFNγ by L. major specific T cells, 4×104 (Exp. 1) or 4. 5×104 (Exp. 2) infected dermal DCs pooled from 10 mice (20 ears) for each treatment group were co-cultured with 1×105 T cells purified by negative selection (Miltenyi Biotec) from draining lymph nodes (dLNs) of B6 mice with a healed, primary infection with L. major FV1. After 3 days, culture supernatants were analyzed for IFN-γ production by ELISA (eBioscience). For adoptive transfer experiments, CD4+ T cells were purified from spleens and lymph nodes of RAG1-deficient OT-II CD4+ TCR transgenic mice by negative selection (Miltenyi Biotec). Purified CD4+ T cells were incubated at 2. 5–5×107 cells/ml in PBS with 0. 5 µM CFSE (Invitrogen) for 10 min at 37°C. The reaction was stopped with 10% normal mouse serum, and the cells were washed twice with cold PBS/0. 1% BSA. B6. SJL congenic mice received intravenously (i. v.) 2–5×105 CFSE-labeled, purified CD4+ OT-II T cells either the same day or 14 days after challenge in the ear dermis with 105 metacyclic promastigotes. Six days after adoptive transfer, the dLNs were removed and analyzed by flow cytometry. To obtain neutrophils recruited to the site of infection in the skin, LYS-eGFP mice were inoculated in the ear dermis with 2×106 Lm-RFP. Twelve hours later the ear tissue was prepared as described above and infected (RFP+eGFPhi) and uninfected (RFP−eGFPhi) neutrophil populations were sorted from dermal tissue using a FACSVantage or a FACsAria (BD Biosciences) cell sorter. Sorted populations were washed once and immediately analyzed for apoptosis or injected into the ear dermis of C57BL/6 and B6SJL recipient mice in a volume of 10 ul. Sorted, infected (RFP+eGFPhi) and uninfected (RFP−eGFPhi) neutrophil populations were stained with Annexin-V-APC and 7-AAD (BD Biosciences) as recommended by the manufacturer. For TUNEL assays, neutrophil populations were fixed in 4% paraformaldehyde, and then labeled with the Beckman Coulter Mebstain Apoptosis kit using biotinylated dUTP. Cells were then incubated with streptavidin-conjugated APC (BD Pharmingen) for 30 min at room temperature. Cells were analyzed by flow cytometry. Neutrophils were depleted employing a single i. p. injection of 0. 5 mg RB6-8C5 (anti-Gr-1), or 1 mg of 1A8 (anti-Ly6G, BioXCell), or GL113 (control IgG, BioXCell), 1 d prior to parasite injection. The efficiency and specificity of the depletions were evaluated on dermal cell preparations, and on heparinized whole blood. Statistical significance between groups was determined by the unpaired, two-tailed student' s t test using Prism software (GraphPad).

Prior studies in mice have shown that the inoculation of Leishmania major into the skin by sand fly bite or by needle provokes a massive recruitment of neutrophils that take up the parasite, and that this response somehow suppresses immunity since neutrophil depletion results in better control of the infection. We investigated how neutrophils recruited to the injection site might interact with and suppress the function of dendritic cells (DCs) in the skin. Infected neutrophils recovered from the skin expressed increased levels of apoptotic markers compared to uninfected neutrophils, and were efficiently taken up by dermal DCs when injected back into the skin. When dermal DCs were permitted to take up parasites in the absence of neutrophils, their expression of activation markers and their ability to present Leishmania antigens were enhanced. Neutrophil depletion also enhanced the activation of Leishmania specific CD4+ T cells in vivo. The results suggest that for insect borne pathogens like Leishmania that provoke a strong inflammatory response at the site of infection, the immunosuppressive effects associated with the apoptotic cell clearance function of DCs will inhibit the early development of immunity.

Abstract Introduction Results Discussion Materials and Methods

medicine infectious diseases immunology biology

2012

Efficient Capture of Infected Neutrophils by Dendritic Cells in the Skin Inhibits the Early Anti-Leishmania Response

YAP1 is a major effector of the Hippo pathway and a well-established oncogene. Elevated YAP1 activity due to mutations in Hippo pathway components or YAP1 amplification is observed in several types of human cancers. Here we investigated its genomic binding landscape in YAP1-activated cancer cells, as well as in non-transformed cells. We demonstrate that TEAD transcription factors mediate YAP1 chromatin-binding genome-wide, further explaining their dominant role as primary mediators of YAP1-transcriptional activity. Moreover, we show that YAP1 largely exerts its transcriptional control via distal enhancers that are marked by H3K27 acetylation and that YAP1 is necessary for this chromatin mark at bound enhancers and the activity of the associated genes. This work establishes YAP1-mediated transcriptional regulation at distal enhancers and provides an expanded set of target genes resulting in a fundamental source to study YAP1 function in a normal and cancer setting. YAP1 (Yes-associated protein 1) is a major transcriptional effector of the evolutionary and functionally conserved Hippo pathway, which is a crucial regulator of organ size, proliferation but also tumor growth [1–3]. Activation of the Hippo pathway leads to phosphorylation and inactivation of the transcriptional co-activator YAP1 by cytoplasmic retention or enhanced degradation [4–8]. YAP1 has a potent growth promoting activity and the YAP1/Hippo pathway has been tightly linked to cancer [8–11]. Loss of Hippo signaling by mutations or down-regulation of core pathway components is associated with cancer development, while YAP1 is reported as a potent oncogene that can promote tumorigenesis in a wide range of tissues [2,12,13]. Elevated expression or activity of YAP1 occurs through multiple mechanisms. YAP1 gene amplification and mutations in upstream pathway regulators, such as NF2, have been described in various human tumors [2,14–20]. YAP1 lacks an intrinsic DNA-binding domain and is thought to exert its co-activator function through binding to promoter sequences via interaction with transcription factors (TF), such as TEAD1/-2/-3/-4, Smads, Runx1/-2, p73, ErbB4, Pax3, AP-1, or TBX5 [12,21]. Among these the TEAD TF family members play a dominant role as primary mediators of YAP1-dependent gene regulation and YAP1 growth-promoting activity [22–28]. Although the tumor-promoting function of YAP1 and TEAD by controlling a remarkable range of cellular processes is undisputed [1,13,27], the comprehensive ensemble of direct downstream target genes and the underlying mechanisms of target gene regulation remain poorly understood. In the past decade, gene expression studies have identified several YAP1-responsive genes [22,29–31]. In contrast, the number of validated direct target genes remains small. Besides validating YAP1 binding to proximal promoter regions of individual genes using ChIP-qPCR [22,29,31–38], a ChIP-on-chip approach using a microarray consisting of promoter regions has been conducted to identify direct YAP1-target genes in MCF10A mammary epithelial cells [22]. While focusing on YAP1-binding to promoter proximal regions a substantial set of functional YAP1 genomic binding sites might have been missed given the importance of distal regulatory elements in establishing a precise pattern of gene expression [39–43]. Here, we comprehensively mapped YAP1 chromatin binding genome-wide, independent of gene location, using ChIP-seq in two human cancer cell lines from different lineages with elevated YAP1 activity (SF268 and NCI-H2052) as well as in non-transformed cells (IMR90) enabling an unbiased identification of YAP1 binding sites and their dependence on cellular context. We demonstrate that YAP1 chromatin recruitment is primarily mediated by binding of TEAD1 to single as well as double TEAD motifs with 3bp spacer at distal enhancers. Aside from presenting a global view of YAP1 and TEAD1 binding in a cancer context, our study also provides novel mechanistic insights into YAP1 transcriptional co-activation of TEAD TFs. We show that YAP1-dependent enhancer activation entails characteristic chromatin changes at lysine 27 of histone H3 and activation of associated genes. Finally we identify a set of YAP1 targets genes by expression profiling following YAP1 knockdown representing a gene signature that can predict YAP1 activity in tumor samples. To gain insight into YAP1 genomic recruitment in a YAP1-relevant cancer context, we used SF268 glioblastoma cells, previously demonstrated to have elevated YAP1 activity due to a 13-fold genomic amplification of the YAP1 locus [44]. Accordingly, YAP1 mRNA and protein levels are increased in SF268 cells as compared to LN229 glioblastoma cells that do not harbor any genetic aberrations of YAP1/Hippo pathway components (Fig 1A and S1 Fig). As a consequence, YAP1 transcriptional activity appears significantly elevated, as suggested by an increased expression of known YAP1 target genes ANKRD1, CYR61, and NPPB but not of unrelated genes FAM171A1 and HAX1 (Fig 1A). To identify YAP1 binding sites genome-wide we performed chromatin immunoprecipitation with a YAP1-specific antibody followed by high-throughput sequencing (ChIP-seq). The chosen antibody proved to be highly specific and sensitive as measured by western blot analysis as well as immunoprecipitation (S2 Fig). We observed high reproducibility between two independent biological ChIP-seq replicates with a Pearson correlation coefficient (PCC) of 0. 95 (Fig 1B). We identified 2,498 binding sites enriched over matching input using the ChIP-seq peak-finder peakzilla [45] (S1 and S2 Tables). To further benchmark our approach we have analyzed the dataset for the presence of peak regions in the most commonly described YAP1 target genes. As anticipated, peaks were identified in the vicinity of published YAP1 target genes, such as CTGF [22], CYR61 [6], NPPB [32], CCND1 [31], AXL [36], DKK1 [33], ITGB2 [22], WWC1 [35], and ANKRD1 (Fig 1C and S3 Fig). Although ANKRD1 expression is commonly used to monitor YAP1 transcriptional activity, to our knowledge, it has not formerly been proven as a direct YAP1 target gene. Our data proofs direct YAP1 binding to the promoter of ANKRD1 (Fig 1C and 1D). ChIP-qPCR validation for several randomly selected loci confirmed YAP1 occupancy at those sites (Fig 1D), further supporting the specificity of binding and overall reliability of the dataset. YAP1 does not contain a DNA-binding domain and thus, relies on interactions with other TFs for recruitment to chromatin. To investigate which TFs mediate binding in SF268 cells we searched the YAP1 peak regions for motifs de novo using MEME [46]. This identified CATTCC, the known consensus motif for TEAD, as the predominant hit (Fig 2A). When allowing for 1 base pair (bp) mismatches to the TEAD consensus motif (S3 Table) we observed that more than 86% of all YAP1 peak regions contained at least one TEAD binding site. This represents a 2. 3-fold enrichment over random control regions (hypergeometric P < 10−288) and provides evidence that TEADs are the predominant co-factors facilitating YAP1 association with chromatin in YAP1-amplified glioblastoma cancer cells. To ask whether additional TFs might recruit YAP1, we searched YAP1 peak regions for enrichment of other known TF motifs. Besides the TEAD consensus motif, we identified only the AP-1/JDP2 motif TGACTCA to be significantly enriched (Fig 2B and S4 and S5 Tables). AP-1 is a heterodimeric protein complex composed of c-Fos and c-Jun, both highly expressed in SF268 cells (S5 Table). Cooperative binding of AP-1 with other TFs has been previously reported as a mechanism of context specific gene regulation [47]. Therefore YAP1/TEAD might act cooperatively with AP-1 in a stimulation-dependent manner or dependent on the pathway genetic context of the analyzed cell type to regulate context-specific gene expression programs. In support of this, c-Fos has recently been described to regulate YAP1 transcriptional activity in the context of KRAS-driven cancers [30] and AP-1/TEAD were found to act as regulators of the invasive gene network in melanoma [48]. When allowing for 1bp mismatches we identified TGACTCA motifs in 60% of YAP1 peak regions that do not contain a TEAD motif but observed the motif as well in 45% of peak regions with a TEAD binding motif. Furthermore we observe that peak regions containing a TEAD binding motif have significantly higher YAP1 ChIP occupancy (as defined by the peakzilla peak score), while the presence of an AP-1 motif does not significantly increase YAP1 occupancy (Fig 2C). Our genome-wide binding data therefore do not provide convincing evidence that AP-1 might serve as an alternative factor for the recruitment of YAP1 to chromatin. However we cannot exclude that AP-1 might serve as a co-factor for YAP1/TEAD under specific experimental conditions. Taken together, our genome-wide binding data support the notion that TEADs account for the vast majority of YAP1 binding to chromatin. We noted that 52% of peak regions contained more than one TEAD binding motif (Fig 2D) with two consecutive sites (double motif) being particularly prevalent. Binding of TEADs and other TFs to double motifs has been recently shown in vitro using high-throughput SELEX [49]. Indeed, we found a specific enrichment of double motifs oriented in the same direction separated by a 3bp spacer (18%) as compared to other spacing or random control regions (hypergeometric P <10−145 vs. control and P < 10−40 vs. other spacer lengths) (Fig 2E). This is consistent with a cooperative mechanism of TEAD1 binding to DNA that has previously been suggested based on structural analyses [50] and in vitro binding experiments [51,52]. We observed that peak regions are significantly conserved as compared to random control regions especially at their peak summit (Fig 2F and 2G). Further, in contrast to peaks without a TEAD motif, peaks with single or double motifs had significantly higher ChIP occupancies (Fig 2H). To directly investigate the functionality of the TEAD double motif we utilized a luciferase reporter gene assay. Double motifs from two independent peak regions (CATTCC-NNN-CATTCC) were cloned upstream of a luciferase reporter. Both constructs caused an increase in luciferase reporter expression as compared to control regions. Importantly, mutations in either one or both of the double motif sites reduced reporter gene expression to the levels of control regions (Fig 2I, red line), indicating that both sites of the double motif are required to enhance transcription. We conclude that, at a subset of binding sites, TEAD binds homotypic clusters of motifs as previously shown for other human TFs [53]. The four different TEAD proteins display distinct expression patterns in cultured cell lines even though they have been suggested to be functionally redundant [54]. To establish which TEADs are essential for YAP1-mediated transcriptional activity in SF268 cells, we assessed the expression of the TEAD-dependent MCAT-luciferase reporter upon siRNA-mediated depletion of individual TEADs. This revealed that the depletion of TEAD1 had a potent effect on reporter gene activity, while knockdown of TEAD4 had only marginal effects (Fig 3A). As TEAD1 appears to be the primary transcriptionally active TEAD family member in SF268 cells, we next mapped its genome-wide binding profile by ChIP-seq (S4 Fig). This led to the identification of 2,652 TEAD1 binding sites based on two independent, but highly reproducible biological replicates and matching input (S1 Table). We first noted a high similarity between TEAD1 and YAP1 ChIP samples, which is reflected in a high positive correlation (PCC = 0. 71) (Fig 3B) and a remarkable overlap of 90% with YAP1 peaks regions (Fig 3C and S3 Fig). siRNA-mediated depletion of TEADs strongly reduced TEAD1 binding to all tested loci, thereby confirming the specific binding of TEAD1 to the identified peak regions (Fig 3D and S5 Fig). Reduction of TEADs also reduced YAP1 levels at all tested sites (Fig 3E). This further argues that YAP1 association with chromatin is mainly mediated via TEAD TFs and specifically by TEAD1 in the tested glioblastoma setting. Reciprocally, we observed that the majority of TEAD1 peaks overlap with YAP1 peaks arguing that all TEAD1 binding sites recruit YAP1. Previous studies focused primarily on the association of YAP1 with proximal promoters [22,31,33,34,36,38,55–57]. It is therefore not surprising that the majority of target genes described up to now contain a YAP1/TEAD1 peak in their promoter region. However, less than 4% of the YAP1/TEAD1 peaks identified in our study are actually located within 2Kb of a gene TSS, and only 15% are located in the 5’UTR of known genes (Fig 4A and S6 Fig). Thus, the majority of YAP1/TEAD1 binding sites reside distal to gene TSSs (Fig 4B, top) providing evidence that YAP1 acts at distal enhancers, which account for a large fraction of regulatory regions [58]. To evaluate whether these distal binding sites occur indeed within functional regions such as enhancers, we took advantage of the fact that acetylation at lysine 27 of histone H3 (H3K27ac) can serve as a signature mark of active enhancers [59,60]. We performed ChIP-seq in SF268 cells using an H3K27ac-specific antibody in two independent biological replicates and matched input (S1 Table). This identified 38,331 H3K27ac positive regions both proximal and distal to gene TSSs (Fig 4B, bottom). Intersecting this dataset with YAP1 and TEAD1 reveals that 95% of the YAP1/TEAD1 peaks overlap with H3K27ac in particular on nucleosomes flanking YAP1/TEAD1 peaks (Fig 4C and 4D and S3 Fig). Thus, most of YAP1/TEAD1 binding appear to occur within active enhancers and likely represent functional binding events. To test this hypothesis, we inserted several YAP1/TEAD1 occupied putative enhancer regions into reporter plasmids. In this experiment indeed five out of six tested elements were able to activate the transcription of a luciferase reporter. Notably, siRNA-mediated depletion of YAP1 or TEADs blunted their enhancer activity demonstrating their necessity for proper enhancer function (Fig 4E). These results provide experimental evidence that YAP1/TEAD1 bind primarily at active distal regulatory regions, contributing to enhancer activity. To gain further mechanistic insight into YAP1/TEAD1 transcriptional regulation, we assessed the impact of YAP1 inactivation on TEAD1 chromatin recruitment, target gene expression, and the H3K27ac enhancer chromatin mark. More specifically we took advantage of contact inhibition as a physiological impetus to control YAP1 activity [7]. Although SF268 cells overexpress YAP1, they are nevertheless fully responsive to contact inhibition. When cultivated at high density, YAP1 translocates to the cytoplasm and is degraded as reflected by decreased protein levels (Fig 5A and 5B). This results in reduced target gene expression (Fig 5C) and coincides with diminished YAP1 recruitment (Fig 5D). Interestingly, inactivation of YAP1 also leads to a reduction of TEAD1 expression (Fig 5C), which resulted in reduced cellular TEAD1 protein levels (Fig 5B) and subsequently diminished chromatin occupancy (Fig 5E). Importantly, YAP1 nuclear depletion also decreases H3K27ac at YAP1/TEAD1 peaks (Fig 5F). This observation appears highly specific since global H3K27ac levels were not affected and regions not bound by YAP1/TEAD1 showed no reduction (Fig 5B and 5F). To gain further mechanistic insight into how YAP1 affects H3K27ac we performed ChIP-qPCR for p300, the major histone acetyltransferase that has been linked to enhancers [61,62]. This reveals that p300 indeed binds to YAP1 positive H3K27 acetylated sites (Fig 5G). Next we asked if p300 recruitment to these sites is YAP1-dependent by testing p300 occupancy upon YAP1 inactivation under high cell density conditions. This revealed reduced p300 levels mirroring the reduction in H3K27 acetylation. To independently test the link between YAP1 activity and enhancer chromatin, we furthermore depleted YAP1 using siRNA, which similarly led to reduced chromatin binding (S7 Fig). In agreement with YAP1 inactivation by high cell density, siRNA-mediated depletion of YAP1 resulted in diminished H3K27ac levels, p300 occupancy, and reduced TEAD1 expression and chromatin occupancy probably through disruption of a positive feedback loop (S3 and S7 Figs). Together, these data confirm the link between YAP1 chromatin binding and transcriptional activation of target genes and establish a requirement for YAP1 for proper chromatin structure at enhancers. Next, we asked if the observed YAP1 binding to distal enhancers is specific for cellular situations with extensive YAP1 amplification such as in SF268 cells. Towards this goal we investigated YAP1 binding in NCI-H2052 malignant mesothelioma cells, a cell line of different lineage and with a different mechanism of YAP1 activation (NF2 mutation, LATS2 deletion) [63]. ChIP-seq analysis of YAP1 in two independent biological replicates and matching input identified 16,470 binding sites (S1 Table). This larger number of peak regions is due to many weak peaks that were not detected in SF268 cells. However, YAP1 binding is well conserved between SF268 and NCI-H2052 cells particularly at strong peaks. This is evident in a global positive correlation (PCC = 0. 32) but also at the level of individual loci (Fig 6A and 6B). Indeed 82% of the YAP1 peaks identified in SF268 overlap with peaks in NCI-H2052 cells. As expected, we also identified 1,142 SF268- and 2,510 NCI-H2052-specific YAP1 binding sites using stringent thresholds but only 48% and 36%, respectively, were assigned to genes not targeted by shared peaks (S8 Fig) suggesting a common function for YAP1 in cancer cells. We also found that YAP1 binds mainly to distal regulatory regions in NCI-H2052 cells (S8 Fig) and that occupied sites were also enriched in TEAD single and double motifs as well as AP-1 motifs (S8 Fig). Similarly to SF268 cells, genetic knockdown of TEADs resulted in reduced YAP1 chromatin binding to NCI-H2052-specific and shared loci with SF268 cells, supporting that TEADs are the main mediators of YAP1 binding also in NCI-H2052 cells (S9 Fig). Cell-type and context-dependent binding of TFs involves chromatin architecture and epigenetic modifications which are often altered during tumor development [64]. Thus, YAP1 binding in cancer cells might differ from non-transformed cells. To investigate whether YAP1 also binds primarily at TEAD-mediated enhancers in non-transformed cells, we investigated its binding profile in non-transformed lung fibroblast cells (IMR90) as primary cells for which many genomic datasets exist [65,66]. ChIP-seq analysis of YAP1 in two independent biological replicates and matching input identified 1,111 binding sites (S1 Table). Notably, we found no significant global correlation (PCC = 0. 002) of YAP1 binding profiles between SF268 and IMR90 (Fig 6B and 6C). Indeed only 42% of YAP1 peaks in SF268 overlapped peaks in IMR90. This difference in binding also holds true at the gene level (S8 Fig). Despite these differences binding nevertheless takes place predominantly at distal regulatory regions (S8 Fig). Furthermore, cell type-specific binding generates cell type-specific presence of the H3K27ac mark suggesting that those sites are functional (Fig 6D). Finally, YAP1 binding sites in IMR90 are also predominantly enriched in TEAD motifs. Importantly, depletion of TEADs using siRNAs resulted in reduced YAP1 occupancy at cell type-specific and shared loci confirming the general observations made in the cancer cell lines (S9 Fig). Different from the cancer cell lines the consensus motif for forkhead TFs (FOX) is significantly enriched as a secondary motif at IMR90-specific sites (S8D Fig). This might indicate that FOX factors act as cell type-specific contributors to YAP1/TEAD transcriptional regulation. This is compatible with a recent publication that shows a functional interaction of YAP1 and FoxO1 in cardiomyocytes [67]. This however remains challenging to test experimentally due to the fact that more than 20 FOX TFs are expressed in these particular cells that are all predicted to bind to this consensus motif. While it remains to be determined if FOX TFs contribute to cell type-specific TEAD binding our data clearly reveal that also YAP1 binding in non-transformed cells is mainly mediated by TEAD (S8 Fig). Taken together, these findings indicate that YAP1 binding to enhancers, as well as the presence of double TEAD motifs with a 3bp spacer, are general features of YAP1-mediated transcription in YAP1-activated cancer as well as non-transformed cells even though targeted enhancers can largely differ. The high occurrences of YAP1 binding sites that we identify at distal enhancers suggest that the number of direct YAP1 target genes is much larger than previously anticipated based on studies that focused on promoter regions. While it is undisputed that enhancers are highly relevant for gene activation it remains challenging to correctly assign their target genes due to the fact that enhancers can regulate genes over long distances [68]. Nevertheless, assigning enhancers to the gene in their nearest vicinity is a useful approximation that is correct in the majority of cases [62,69]. Based on this observation we assigned each YAP1/TEAD1 peak to its nearest gene TSS, yielding 1,738 genes in SF268 cells (S2 Table). In agreement with the notion that YAP1 mainly functions as a transcriptional co-activator this gene set was expressed at a significantly higher level as compared to a random control set (Fig 7A) indicating that the peak-to-gene assignment is overall accurate. Distal YAP1/TEAD1 peaks (over 2Kb away from gene TSS) were assigned to 1580 genes, 52 of which also had a proximal peak. Importantly, the 1528 genes with only distal YAP1/TEAD1 peaks were also more highly expressed than a random control set (Fig 7A). To further, determine the accuracy of the peak to gene assignment we directly tested the expression of 19 randomly selected genes upon YAP1 or TEAD siRNA-mediated depletion (S10 Fig). Gene expression levels were affected in all tested cases arguing for a direct link between binding events and target gene expression. Gene Ontology (GO) analysis of this gene set showed enrichment for previously reported YAP1 functions such as regulation of cell migration (hypergeometric P < 10−17), extracellular matrix organization (P < 10−15), actin cytoskeleton organization (P < 10−07), and regulation of epithelial cell proliferation (P < 10−6) (Fig 7B and S6 Table). In agreement with recent studies demonstrating a complex interaction network between the YAP1/Hippo and others signaling pathways such as WNT, BMP, TGF-β and PI3K-mTOR [70], our analysis reveals an enrichment of terms associated with signaling (Fig 7B and 7C and S6 Table). We noted that this set of genes contains a number of core components and downstream targets of diverse signaling pathways (S11 Fig). In addition, a number of YAP1/Hippo pathway components including WWC1, LATS, NF2, and AMOT are bound by YAP1/TEAD1. This suggests an extensive feedback mechanism in vertebrates and confirms previous reports in Drosophila [71,72]. Next we set out to determine which genes are transcriptionally activated by YAP1 and performed RNA-seq profiling following YAP1 siRNA-mediated depletion in SF268 cells (Fig 7D and S12 Fig). This identified 219 and 360 genes that were down- or up-regulated by at least 2-fold, respectively, upon YAP1 knockdown compared to control siRNA-treated cells. Among the down-regulated genes, 70 (32%) contained a YAP1/TEAD1 peak assignment in SF268 cells (Fig 7D). To evaluate the physiological relevance of these YAP1-activated target genes, we sought to predict YAP1 expression in tumor samples using expression data for 528 primary glioblastoma and 279 head and neck squamous cell tumor samples [73]. For each indication, we labeled the samples as YAP1 “high” or “low” expression and divided the datasets into training and test datasets (2/3 and 1/3 of the samples respectively) over 1000 randomized iterations. Using a naïve Bayes classifier this allowed to predict YAP1 expression level with high accuracy (an area under the receiver operating characteristics curve (AUC) = 0. 83 for glioblastoma samples and AUC = 0. 78 for head and neck squamous tumor samples) (S13 Fig). Feature selection allowed reducing the full YAP1 gene signature to ten genes without losing prediction performance. Hierarchical clustering of the samples shows consistent patterns of expression depending on YAP1 “high” or “low” expression (Fig 7D). This result supports the use of the acquired gene signature to identify YAP1-activated cancers. By providing a comprehensive account of YAP1 genomic binding and its impact on transcription this study establishes that transcriptional regulation of YAP1 target genes is predominantly mediated by TEAD binding to distal enhancers. In addition to demonstrating this mode of regulation we show that this activity entails the establishment of chromatin marks typical to enhancers linking YAP1 activity to H3K27ac. The identified distal regions enabled us to largely expand the set of YAP1 target genes, which we foresee to be a valuable source for functional studies and which we show to have predictive power to identify YAP1-activated cancers. Due to lack of a DNA-binding domain, YAP1 requires TFs for genomic recruitment. TEAD family members are considered the main TFs for YAP1-mediated regulation of gene expression [22,23,25]. In support of a dominant function of TEADs in cancer cells, overexpression of an artificial TEAD2-VP16 construct in NIH3T3 cells was reported to mimic the effects of YAP1 overexpression at the transcriptional level and lead to cell transformation [23]. Furthermore, mutations in YAP1 that prevent binding to TEAD were shown to abolish YAP1-induced transcription and cell transformation in NIH3T3 and MCF10A cells [22]. Here, we comprehensively mapped YAP1 chromatin binding genome-wide in two different cancer cell lines and in non-transformed cells, enabling an unbiased assessment of the sequence features that direct YAP1-mediated regulation. Our genome-wide map of TEAD1 binding sites revealed that the vast majority of YAP1 binding sites were co-occupied by TEAD1 confirming the dominant role of TEAD factors in the control of YAP1 transcriptional activity. To the best of our knowledge this is the first study demonstrating a genome-wide co-occupancy of both factors in cancer cells. Our data extends the results from a previous ChIP-on-chip study that used a promoter specific microarray and demonstrated a comparable overlap of >80% for YAP1 and TEAD1 binding around start sites in MCF10A mammary epithelial cells [22]. Despite the major role of TEADs to mediate YAP1 co-activator activity, additional TFs are described to interact with YAP1 (reviewed in [12]). Our data, however, do not provide evidence for the importance of additional TFs in targeting YAP1 chromatin binding. We demonstrate that this finding is not just limited to a cellular situation where YAP1 is amplified since we observe a similar predominant enrichment of TEAD motifs in YAP1 peak regions in a YAP1-activated NCI-H2052 cancer cell line as well as in non-transformed IMR90 cells. TEAD1 also referred to as TEF-1, for transcription enhancer factor 1, was first cloned in HeLa cells as an activator of the simian virus 40 (SV40) “enhancer”, which is a short 72bp sequence element that is a component of the viral early promoter [54,74,75]. So far, however, binding and function at endogenous elements that by the current definition of an enhancer act distal to promoters had not been investigated. Our genome-wide analysis of YAP1/TEAD1 binding indicates that the vast majority of endogenous sites, in cancer and non-cancer cells (SF268, NCI-H2052 and IMR90), are actually located within distal regulatory regions representing enhancer elements. This mimics the typical distribution of sequence-specific TFs and is in line with the concept that distal TF and co-activator binding are key determinants of enhancer activity and in turn cell-type specific gene expression patterns [39,68,76]. Recent efforts in mapping enhancers in different tissues revealed that the human genome contains up to several hundred thousand distal regulatory regions, most of which are cell-type specific [77]. Their misregulation can be highly disease relevant since mutations in these regions have extensively been associated with disease susceptibility [78]. Distal binding has not been reported for Yki (Yorki) the Drosophila homolog of YAP1 and it remains open if this reflects a functional difference or the organization of the smaller and gene denser fly genome [57,79]. Similarly, enhancer binding of Yap1 has not been reported in mouse embryonic stem cells [55]. However, when reanalyzing an available list of Yap1 peak regions from Lin et al. , we observed that a large fraction of these Yap1 binding sites are located in regions distal from promoters. In support of our data two recent reports demonstrated that YAP1/TEAD regulate transcription by binding to distal enhancers [67,80]. Together this argues that YAP1 distal binding is a general feature of YAP1/TEAD-driven transcription activation also in non-transformed cells and is not an acquired feature of cancer cells. Distinct chromatin modifications are associated with various aspects of gene expression. In particular H3K27ac was found to be an effective means to determine enhancer activity [59,81]. Our data show that the vast majority of YAP1 binding sites overlap H3K27ac positive regions and that cell-type specific YAP1 sites match cell type specific H3K27ac regions. Interestingly we show that, in SF268 cells, YAP1 chromatin association is a prerequisite for the deposition of H3K27ac supporting the fact that YAP1 binding sites represent functional enhancers. Interestingly, genome-wide binding analysis in Drosophila revealed a correlation between Yki chromatin binding and trimethylation of H3K4 (H3K4me3) [57]. Consistent with this finding, nuclear receptor coactivator 6 (Ncoa6), a subunit of the Trithorax-related H3K4 methyltransferase complex, has been identified as a Yki binding protein that is required for transcriptional regulation [79]. Importantly, besides the H3K4 methyltransferases, the mammalian Ncoa6 has been reported to enhance the activity of TFs by interacting with histone acetyltransferases CBP/p300 [82]. Whether this NCOA6 function possibly facilitates YAP1-dependent acetylation of H3K27 (H3K27ac) and which additional cofactors are recruited to trigger transcriptional activity warrants further investigations. The determination of genes targeted by specific enhancers remains a challenge. We observed that only a minority of genes nearest to binding sites was transcriptionally affected by depletion of YAP1. Notably however, only 10–25% of TF binding events in higher eukaryotes contribute to the expression of the closest proximal gene in any given cell type. This is likely to be an underestimate given the nature of enhancers and the complexity of transcription regulatory networks [76]. Besides the uncertainty of assigning binding sites to target genes, enhancers function in a modular manner such that they contribute additively and redundantly to the expression of their target genes [76,83]. Therefore YAP1/TEAD contribution to transcriptional activity might not be apparent at many target genes. In addition to shedding light on basic principles of YAP1 transcriptional regulation, the identification of distal regulation as the primary means of YAP1 transcriptional control enabled us to identify an extended list of target genes based on both YAP1 chromatin binding and gene expression changes. This novel YAP1 signature from YAP1-amplified glioblastoma cells should have predictive potential for the identification of YAP1-dependent tumors. SF268 cells (NCI DCTD tumor/cell line repository) were maintained in RPMI 1640 Medium, GlutaMAX supplement, 25 mM HEPES, 10% (v/v) fetal calf serum and 1 mM sodium pyruvate. NCI-H2052 cells (ATCC) were maintained in RPMI, 10% (v/v) fetal calf serum, 1% (v/v) non-essential amino acids, 1 mM sodium pyruvate. IMR90 cells (ATCC) were maintained in EMEM (Sigma M 4655) supplemented with 1% non-essential amino acids (NEAA) and 10% FBS. All media and supplements were from Life Technologies. To obtain low density (LD) or high density (HD) cultures, SF268 cells were plated at 10. 000 cells/cm2 or 100. 000 cells/cm2, respectively and harvested 48h or 96 hours after seeding. Transient transfections of SF268, IMR90 and NCI-H2052 cells with siRNA (final concentration: 25 nM) were performed using Lipofectamine RNAiMAX (Life Technologies). Transient transfection of SF268 cells with plasmid DNA was performed using Cell Avalanche Transfection Reagent (EZ Biosystems). Cells were harvested at 48 hours or 72 hours post-transfection. The following antibodies were used for western blot, immunoprecipitation and chromatin immunoprecipitation (ChIP): anti-YAP1 [EP1674Y] (ab52771), anti-KAT3B/p300 (ab14984), and anti-H3 (ab1791) from Abcam; anti-TEAD1 (610922) from BD Transduction Laboratories; anti-TEAD4 (ARP33426_P050) from Aviva Biosystems; anti-β-Actin (A2066) from Sigma-Aldrich; anti-H3K27ac (AM 39133) from Active Motif. YAP1 and TEAD1 antibodies for ChIP-seq were characterized in western blot, immunoprecipitation and ChIP-qPCR (S2 and S4 Figs). ChIP was essentially carried out as previously described [84], with slight modifications. Chromatin was sonicated for 14 minutes using a Covaris E210 (Settings: 5% duty cycle, intensity 4). 60μg of chromatin were incubated over night at 4°C with 5μg of the corresponding antibody and for 2 hours with preblocked (tRNA, BSA) Dynabeads protein G. DNA was purified using the Minielute PCR purification kit (Qiagen). Quantitative PCR was performed using Maxima SYBR Green / ROX qPCR Master Mix (Thermo Scientific) and the ViiA 7 Real-Time PCR System (Life Technologies) and 1/80th of the ChIP sample or 0. 01% of input chromatin per PCR, respectively. Amplifications were performed in triplicate, and mean values were expressed as percentage input. Standard deviation was calculated from the triplicates, and error bars are indicated accordingly. Primers are listed in Table 1 and Table 2. YAP1, TEAD1 and H3K27ac ChIPs from SF268 cells and YAP1 ChIPs from NCI-H2052 and IMR90 cells were subjected to high-throughput sequencing on a 356R Illumina HiSeq 2500 sequencer using standard NEB library preparation kits and protocols. Additional ChIP-seq dataset for H3K27ac in IMR90 cells was obtained from the Gene Expression Omnibus under the accession number GSM469967. We mapped the ChIP-seq sequencing reads (single-end, 50bp) to the human reference genome (hg19 only chromosomes 1 to 22, X, Y and M) using bowtie [85] version 1. 0. 0 with parameters-v 3-m 1—best—strata. We extended the reads to 150bp (average estimated fragment length) and calculated for each genomic position the read density normalized to one million reads in the library to generate wiggle files. Genome screenshots were taken using the UCSC genome browser [86]. We identified peaks in YAP1 and TEAD1 ChIP samples compared to the corresponding input samples using peakzilla [45] and in H3K27ac ChIP using MACS [87] version 1. 4. 2. The strategy used to define and overlap peak regions is described in [88] and S1 Table. We defined control peak regions by shuffling the peaks randomly within the same chromosome. We calculated the Pearson correlation coefficient (PCC) and plotted scatterplots between two samples using the mean fragment density of each peak region from all samples. Differentially bound regions were identified with the R package DESeq [89] using an adjusted p-value threshold of 10−5 and a 2-fold enrichment with enrichment in the reference sample below 100 normalized reads per kilobase. We searched for motif de novo using MEME [46] within 31bp around peak summits and for occurrences of the known motifs from Jaspar [90], and [49] using MAST [91] (from the MEME suite programs version 4. 1. 1) with a P-value of 10−3 in an area of 151bp (average genomic fragment length) around each peak summit. We assigned each peak to its closest gene transcriptional start site (TSS) using the reference transcriptome (GRCh37. 71). For each gene ontology biological processes [92] and WikiPathways [93], we calculated the enrichment and associated hypergeometric P-values of genes in each class compared to all genes. We calculated the conservation rate of regions using the PhastCons 46 way placental mammals [94]. RNA was isolated using RNeasy Mini Kit (Qiagen) and cDNA synthesis was performed using the High-Capacity RNA-to-cDNA Kit (Applied Biosystems). cDNA was subjected to quantitative PCR (qPCR) analysis in triplicates with gene-specific primers (see Table 3) using Maxima SYBR Green / ROX qPCR Master Mix (Thermo Scientific) and the ViiA 7 Real-Time PCR System (Life Technologies). Total RNA of three biological replicates was extracted from SF268 cells 48h after transfection with two individual siRNAs targeting YAP1 and unspecific control siRNAs using the Total RNA purification kit from Norgen Biotek. RNAseq libraries were prepared using the Illumina TruSeq RNA Sample Prep kit v2 and sequenced using the Illumina HiSeq2500 platform (76-bp paired-end reads). Additional RNA-seq datasets for SF268 and LN229 cells were obtained from the Cancer Genomics Hub (https: //browser. cghub. ucsc. edu). We mapped the RNA-seq sequencing reads (paired-end, 100bp) to the human reference transcriptome (GRCh37. 71) using tophat [95] version 1. 3. 1 with parameter—no-novel-juncs. We calculated genes FPKMs (fragments per kilobase of transcript per million mapped reads) using cufflinks [96] version 2. 0. 2 with parameter-G using the reference transcriptome (GRCh37. 71). Differentially expressed genes in YAP1 knockdown were identified with the R package DESeq [89]. Genes either down- or up-regulated were selected using an adjusted p-value threshold of 10−5 in all four pairwise comparisons of YAP1 and control siRNA treated samples and an at least 2-fold enrichment in one comparison and at least 1. 2-fold in the other three. GeneArt Strings DNA fragments encompassing approximately 200bp of six distal enhancers bound by YAP1/TEAD1 (see Table 4) and two negative regions carrying BglII restriction sites were cloned into pGL3 promoter vector (E1761, Promega) upstream of the luciferase gene with SV40 minimal promoter. For two regions mutations were introduced in either one or both motif sites of the double TEAD motif with 3bp spacer (see Table 5). One day prior transfection SF268 cells were plated on 384-well plates (1800 cells/well). Cells were co-transfected with 28. 5 ng of the respective reporter constructs and 1. 5 ng pRenilla. For luciferase assays in YAP1 and TEADs-depleted cells, SF268 cells were transfected with the indicated siRNAs (see Table 6) at the day of seeding (1800 cells/well) in 384-well plates. The next day, the medium was changed and cells were transfected with DNA (pGL3 reporter constructs and pRenilla). Firefly and Renilla luminescence signals were measured at 24 hours after DNA transfection using Dual-Glo luciferase assay system (Promega). Firefly luminescence signals were normalized according to their corresponding Renilla signals resulting in relative luciferase activity. Each sample was transfected in triplicate, and each experiment was repeated independently at least three times. SF268 cells stably expressing the MCAT-Luc YAP1/TEAD responsive reporter [44] were transfected with siRNAs targeting YAP1, TEAD1, TEAD2, TEAD3, and TEAD4 with 8 siRNAs per gene (see Table 7). At 72 hours after transfection medium was aspirated and cells were incubated with fresh medium containing 1. 4μM resazurin (SIGMA; MO, USA) for 2 hours before measuring fluorescence (Ex: 540 nm, Em: 590 nm) as a read-out for cell viability. Subsequently the cells were lysed in fresh medium containing 1: 10 (v/v) Steady-Glo luciferase assay reagent (Promega; WI, USA). Luciferase measurements were taken according to the manufacturer' s protocol. Fold change in MCAT-Luc reporter activity was calculated by normalizing luminescence signal to resazurin and to negative control siRNA. Each experiment was carried out in triplicate. SF268 cells were lysed in FT lysis buffer (20 mM Tris / HCl at pH 7. 8,600 mM NaCl, 20% glycerol, proteinase inhibitor), and proteins including histones were extracted by repeated freeze-thaw cycles followed by Benzonase (Novagen) treatment. Lysates were separated using Novex NuPAGE SDS-PAGE gel system transferred to Immobilon-P membranes (Millipore) and subjected to immunoblotting. Cells were fixed with 4% PFA (Paraformaldehyde 20% solution, EM grade #15713-S) for 15 minutes at room temperature. Subsequently cells were washed 1x PBS and permeabilized in PBS / 0. 1% Triton X-100 at room temperature for 10 minutes. Cells were rinsed in PBS and incubated with anti-YAP1 diluted 1: 300 in PBS / 1. 5% BSA over-night at 4°C. Cells were washed with PBS and incubated with the secondary antibody, anti-rabbit Alexa647 (1: 1000, Life Technologies) and Hoechst (1: 10. 000) for DNA staining for 2 hours at room temperature. After washing with PBS, staining was analyzed by fluorescence microscopy (Operetta, Perkin Elmer), 20x objective. We used as gene signature 70 genes that were 2-fold down-regulated in YAP1 depleted SF268 cells and had a YAP1/TEAD1 binding peak in their vicinity. Expression data were collected from cBioPortal [73,97] for 528 primary glioblastoma and 279 Head and neck squamous cell tumor samples generated by the TCGA Research Network. We used either all or the top and bottom 10% of samples according to the ranksum for YAP1 expression and copy number. Over 1000 iterations we randomly divided the datasets into training and test subsets (2/3 and 1/3 respectively) and used a naïve Bayes predictor from the Bioconductor package e1071 to predict the YAP1 expression level (“high” or “low”). Prediction accuracy was measured using recall statistics and receiver-operating-characteristic (ROC) curves and performance statistics were generated using the ROCR package [98]. Raw and processed ChIP-seq data are deposited in the Gene Expression Omnibus (GEO) under the accession number GSE61852. The raw RNA-seq reads are available in the NCBI Short Read Archive under the accession number SRP056665.

The YAP1/Hippo signaling pathway is a key regulator of organ size and tissue homeostasis, and its dysregulation is linked to cancer development. Elevated activity of YAP1, a transcriptional coactivator and well-established oncogene has been reported to occur in human cancers. Comprehensive identification of YAP1 regulated genes and its mode of action will be of high importance to uncover YAP1 biology that could be exploited for a therapeutic intervention. To this end, we performed genome-wide analyses to identify YAP1 occupied sites in cancer cell lines representing different YAP1/Hippo pathway tumor etiologies and in non-transformed fibroblasts. Our data demonstrate that YAP1 activity is mediated predominantly via TEAD transcription factors supporting the importance of TEADs as main mediators of YAP1-coactivator activity. We further show that YAP1 and TEAD1 exert their transcriptional control via binding to enhancers, leading to characteristic chromatin changes and distal activation of genes. By linking enhancers to genes, we provide a list of novel YAP1 target genes in an oncogenic setting that we show can readily be exploited in tumor classification and provides a foundation for further investigations.

Abstract Introduction Results Discussion Materials and Methods

2015

YAP1 Exerts Its Transcriptional Control via TEAD-Mediated Activation of Enhancers

Kaposi' s sarcoma-associated herpesvirus (KSHV) is an oncogenic herpesvirus associated with multiple AIDS-related malignancies. Like other herpesviruses, KSHV has a biphasic life cycle and both the lytic and latent phases are required for tumorigenesis. Evidence suggests that KSHV lytic replication can cause genome instability in KSHV-infected cells, although no mechanism has thus far been described. A surprising link has recently been suggested between mRNA export, genome instability and cancer development. Notably, aberrations in the cellular transcription and export complex (hTREX) proteins have been identified in high-grade tumours and these defects contribute to genome instability. We have previously shown that the lytically expressed KSHV ORF57 protein interacts with the complete hTREX complex; therefore, we investigated the possible intriguing link between ORF57, hTREX and KSHV-induced genome instability. Herein, we show that lytically active KSHV infected cells induce a DNA damage response and, importantly, we demonstrate directly that this is due to DNA strand breaks. Furthermore, we show that sequestration of the hTREX complex by the KSHV ORF57 protein leads to this double strand break response and significant DNA damage. Moreover, we describe a novel mechanism showing that the genetic instability observed is a consequence of R-loop formation. Importantly, the link between hTREX sequestration and DNA damage may be a common feature in herpesvirus infection, as a similar phenotype was observed with the herpes simplex virus 1 (HSV-1) ICP27 protein. Our data provide a model of R-loop induced DNA damage in KSHV infected cells and describes a novel system for studying genome instability caused by aberrant hTREX. Genome instability, an enabling characteristic of the hallmarks of cancer, has long been established as a major contributing factor to cancer formation and progression [1], [2]. However, our understanding of the underlying molecular causes is still in its relative infancy. Contributing factors of genome instability are wide ranging and incorporate those from exogenous sources, such as ionising radiation, endogenous sources such as reactive oxygen species (ROS) and reactive nitrogen species (RNS), as well as mutations incorporated into the genome during cell replication, including DNA replication errors and error prone DNA repair [3], [4]. Cells have evolved to deal with this onslaught of damage through several DNA repair pathways, each specific to certain types of damage [4]. The most severe types of DNA damage result in double strand breaks (DSB) that can be repaired primarily through error-free homologous recombination (HR) [5], [6], or error-prone non-homologous end-joining (NHEJ) [6]. DSBs are closely associated with cancer progression and can include severe chromosome pulverisation and chromothripsis [7]–[9] leading to major chromosome rearrangements, as well as smaller mutations. As such, DSBs are known to be an integral part of many cancers, for example, breast cancers, Burkitt' s lymphoma, and multiple leukaemia' s [10]–[12]. The Kaposi' s sarcoma-associated herpesvirus (KSHV) is an important oncogenic virus associated with multiple AIDS-associated malignancies including Kaposi' s sarcoma (KS), primary effusion lymphoma (PEL) and multicentric Castleman' s disease (MCD) [13], [14]. Like all herpesviruses, KSHV has a bi-phasic lifecycle incorporating latency and lytic replication [15]. During latency the virus expresses a small subset of genes that allows it to persist in the host cell, while reactivation to the lytic cycle results in expression of the full viral genome and production of infectious virus progeny. KSHV differs from the other oncogenic herpesviruses as both the latent and lytic cycles are required for tumorigenesis [13], [16]. Interestingly, KSHV has recently been shown to cause DNA damage in infected cells [17]. Specifically, lytic infection of cells has been shown to directly induce DNA double-strand breaks, a severe form of genome instability [17]. Moreover, several reports have identified chromosomal instability in KSHV infected cells, which have been suggested to contribute to the neoplastic process of KSHV [18], [19]. Specifically, loss of chromosomes 14 and 21. and non-random translocations and deletions in chromosome 3 [19]–[21]. Furthermore, loss of the Y chromosome in early tumour stages, recurrent gains in chromosome 11 and further chromosomal stages during late KSHV-associated tumour stages have been reported [22]–[24]. Importantly however, there is currently no known mechanism that describes how KSHV lytic replication can cause DNA double-strand breaks. Recent evidence has suggested there may be a surprising link between mRNA export, genome instability, and cancer development [25]. The Transcription and Export complex (TREX) [26], has multiple roles throughout mRNA processing including recruitment of the nuclear export receptor, TAP, to initiate efficient bulk mRNA export, as well as stabilisation of mRNA during transcription [27]. Human TREX (hTREX) comprises the DEAD-box helicase UAP56, the export adapters Aly and UIF, the recently discovered CIP29, PDIP3, Chtop and ZC11A proteins, and the multi-protein THO complex (summarised in Schumann et al [28]). Aberrations that affect hTREX protein expression and function have been implicated in human cancer [25], [29]. For example, the export adapter protein Aly, which functions to recruit the hTREX complex to the mRNA, as well as the THO component THOC1, are known to be deregulated in multiple cancers [25], [29], [30]. Strikingly, there appears to be disparity not only between different cancer types, but also between low- and high-grade tumours. THOC1 is overexpressed in ovarian, colon and lung cancers, but there is a loss of expression in testicular and skin cancer [30]. Moreover, while Aly is highly expressed in several low-grade lesions, it is undetectable in multiple high-grade tumours of the colon, stomach, thyroid, testis and skin [30]. Furthermore, UAP56 has been identified as having a possible role in causing chromosome instability during mitosis, where knock-down of UAP56 leads to premature sister chromatid separation [31]. Together, these observations highlight the importance of the hTREX complex, not only in exporting mRNAs from the nucleus, but also in maintaining genome integrity during transcription. Thus, aberrations in components of hTREX can have devastating consequences on genome stability [31]. Importantly, a body of work, primarily performed in yeast, has highlighted a potential mechanism for how defects in mRNA export may contribute to genome instability [32]. Yeast THO mutants show impaired transcription elongation and RNA export defects, as well as a high level of transcription-associated recombination. It is thought that an absence of THO at the 5′ end of a nascent mRNA during transcription leads to the loss of mRNA stability and the formation of abnormal RNA: DNA hybrids known as R-loops. These R-loops may form as the newly transcribed mRNA anneals to the template strand of the DNA, and although a mechanism has been proposed [33] our understanding of the processes involved is still lacking. The KSHV open reading frame 57 (ORF57) protein is a multi-functional protein that, like its homologues in other herpesviruses, facilitates all stages of viral mRNA processing throughout lytic replication [28], [34]–[37]. Through an interaction with the KSHV transactivator protein, Rta, ORF57 has been shown to function co-transcriptionally promoting expression of KSHV genes [38], [39]. ORF57 is also known to enhance the splicing of several viral transcripts [40]. Moreover, through an interaction with the cellular protein PYM, ORF57 is able to facilitate the recruitment of the cellular pre-initiation complex and enhance the pioneer round of viral translation [41], [42]. The majority of work on ORF57 has focused on its roles in RNA stability and mRNA export [43]. Several recent studies have shown the importance of ORF57 in stabilising the KSHV PAN RNA [44]–[46]. Importantly, ORF57 also acts as a viral mRNA export factor and recruits the entire hTREX complex to viral mRNA through a direct interaction with the export adapter proteins Aly and UIF [47]–[49]. The interaction with these two separate cellular export adapter proteins allows for redundancy in the ORF57-mediated viral mRNA export. Significantly, ORF57 has been shown to be essential for KSHV replication [50], and ORF57-mediated recruitment of hTREX to viral mRNA is essential for efficient KSHV lytic replication [47]. The link between mRNA export and genome instability is intriguing. We therefore set out to investigate whether the known interaction between ORF57 and hTREX could have implications for genome instability in KSHV lytically infected cells. Herein, we demonstrate that sequestration of hTREX by the KSHV ORF57 protein is sufficient to induce genome instability, due to a consequence of the formation of R-loops. This work highlights the importance of viral models for our understanding of cellular processes, and also demonstrates and confirms a novel link between mRNA export and genome instability, a major driving force behind tumorigenesis. Moreover, it describes a novel mechanism to account for the DNA damage observed in lytically active KSHV infected cells. KSHV has been shown to induce genome instability either through chromosome instability [18] or, more recently, through the observation that lytic KSHV infection in BCBL-1 cells reactivated with TPA and sodium butyrate induces the phosphorylation of the double strand break marker, γH2A. x [17]. To confirm this data in a well characterised KSHV infected cell line, we utilised the Tet-regulated expression system T-REx with the TREx BCBL1-Rta cell line [51]; a KSHV latently infected cell line containing a Myc-tagged version of the viral transcriptional activator, Rta, under the control of a doxycycline-inducible promoter. We first used confocal fluorescence microscopy to assess the phosphorylation of the H2A histone variant, H2A. x, as a marker for DNA strand breaks [52]. After 24 hours, doxycycline induced TREx BCBL1-Rta cells showed significant levels of the phosphorylated form of H2A. x, γH2A. x compared to uninduced cells (Figure 1A). Furthermore, we induced TREx BCBL1-Rta cells over a 24 hour time-course and analysed cell lysates by western blotting. The KSHV early gene product, ORF57, can be seen at 8 hours post-induction with increasing expression at 16 and 24 hours (Figure 1B). ORF57 is observed as a double band to differing degrees in different cell lines due to a known caspase-7 cleavage event [53]. Interestingly, γH2A. x increases through the same time-course concurrent with ORF57, while total levels of H2A. x are slightly decreased. It is important to note that an interesting recent study highlighted the importance of γH2A. x for KSHV episome persistence [54]. This study demonstrated that H2A. x is phosphorylated during KSHV latency, confirmed by our western blot analysis (Figure 1B), but our data convincingly show that this level is dramatically increased during the KSHV lytic replication cycle. To analyse DNA damage directly in KSHV lytically infected cells we performed comet assays on uninduced versus induced TREx BCBL1-Rta cells. Cells were first induced for 24 hours allowing sufficient ORF57 expression (Figure 1C) before alkaline comet assays were performed to assess the total level of single and double strand DNA breaks (Figure 1D and 1E). Tail moments were scored giving values of 3. 93 for uninduced TREx BCBL1-Rta cells versus 9. 87 for induced cells indicating a significant level of DNA strand breaks. Data were analysed using an unpaired 1-tailed T-test to verify the statistical significance (P = 4. 92×10−5). Together, these data show the relevance of KSHV as a model of genome instability and, using an inducible KSHV system, confirm previous observations that KSHV lytic replication induces DNA double-strand breaks. To assess the effect of ORF57 expression on the cellular proteome, and in particular on DNA repair pathways, we undertook a stable isotope labelling by amino acids in cell culture (SILAC) based quantitative proteomics approach [55], [56]. The Flp-In T-REx-293 system was used to create a stable KSHV ORF57 inducible cell line, iORF57-293, with ORF57 under the control of a tetracycline/doxycycline-inducible promoter. Uninduced cells were grown in heavy isotope labelled DMEM, R6K4, while the sample to be induced was grown in label-free DMEM, R0K0, for 6 passages to allow incorporation of the respective isotopes. Cells grown in DMEM R0K0 were then induced for 24 hours to allow for expression of ORF57 before being fractionated into cytoplasmic, nuclear and nucleolar fractions to reduce sample complexity. The quality of the cellular fractions was confirmed using specific markers to cellular proteins (Figure S1). Protein samples were separated by SDS-PAGE and stained with colloidal blue stain. Each protein gel lane was excised in 10 fragments, digested with trypsin and analysed by LC-MS mass spectrometry (LTQ-Orbitrap Velos, service provided by the University of Dundee). Quantification of peptide changes was performed using MaxQuant [57], [58] and expressed as a fold change of endogenous proteins in induced cells compared to uninduced cells. A 2. 0-fold cutoff was chosen as a basis for investigating potential proteome changes between data sets. Bioinformatical analysis was performed using IPA analysis software (Ingenuity systems) and multiple proteins were found to be enriched upon ORF57 expression, including many associated with DNA repair (Table S1). Of particular interest were proteins involved in the error-prone double-strand repair pathway, NHEJ. Analysis of cellular pathways highlighted that the majority of the proteins involved in this repair pathway are enriched upon ORF57 expression (Table 1). Indeed, Ku70 and Ku80 are enriched by 3. 2-fold and 2. 9-fold, respectively, DNA-PK by 2. 1-fold and Rad50 by 2. 4-fold. This enrichment suggests a possible activation of DSB repair upon ORF57 expression. To demonstrate that ORF57 expression alone is sufficient to induce chromosome instability, as has previously been reported for KSHV infection [18], we undertook confocal fluorescence microscopy of mitotic cells comparing uninduced versus induced ORF57-expressing cells to look for characteristic markers of genome instability. In the first instance we looked for the presence or absence of chromosomal anomalies in the form of lagging chromosomes during mitosis where chromosomes do not separate correctly into daughter cells, an aberration that has been shown to have links to defective hTREX [31]. This can lead to the formation of micronuclei, a common occurrence in many tumour cells [59]. iORF57-293 induced cells showed numerous chromosome abnormalities in the form of chromosome lagging (Figure 2A) compared to uninduced cells, implying that ORF57 expression alone has a significant effect on genome maintenance. This observation was confirmed in HEK 293T cells either mock transfected, transfected with an EGFP or an EGFP tagged form of ORF57, EGFP-ORF57. Only the cells expressing EGFP-ORF57 were observed to contain chromosome lagging (Figure 2B). Of the total cells examined, only 1 out of 22 showed chromosome lagging in cells not expressing ORF57 compared to 6 out of 14 in cells expressing ORF57. The importance of this observation is still to be fully characterised, but it demonstrates the ability of ORF57 to induce genome instability in cells. Moreover, there is an established link between hTREX siRNA knockout and mitotic anomalies [31] that suggests interfering with hTREX function, for example with the over-expression of ORF57, could lead to the formation of mitotic anomalies such as chromosome lagging. These data together suggest that ORF57 expression is sufficient to induce a DSB response and increase the levels of chromosome instability in cells. To further confirm that ORF57 expression alone is required for a DSB response we performed confocal fluorescence microscopy on uninduced versus induced iORF57-293 cells and assayed γH2A. x levels (Figure 3A). Uninduced cells showed minimal γH2A. x staining, as would be expected of a healthy cell population. As a positive control the topoisomerase II inhibitor etoposide was used to treat the cells for 30 minutes at a concentration of 50 µM leading to increased levels of γH2A. x. Importantly, ORF57 induction also led to an increased level of γH2A. x, confirming that ORF57 expression promotes the DSB response. Moreover, western blot analysis of iORF57-293 cell lysates shows an increase in γH2A. x upon ORF57 expression when compared to uninduced cells (Figure 3B). Additional controls confirmed that the parental cell line, 293 Flp-In, shows low levels of γH2A. x (Figure S2). Further to this we performed alkaline comet assays on iORF57 293 cells that had been left uninduced or induced to express ORF57 for 16 hours (Figure 3C and 3D). Uninduced cells had a tail moment of 1. 68 compared to 3. 38 for induced cells (P = 0. 009) demonstrating that ORF57 expression can lead directly to DNA strand breaks. In addition, we performed neutral comet assays to demonstrate the presence of double strand DNA breaks in an over-expression system using mCherry-ORF57, mCherry as a negative control and etoposide treatment as a positive control (50 µM for 15 mintues) (Figure 4). As the assay is reliant on transfection, the transfection efficiency was confirmed by immunofluorescence microscopy, and protein expression levels determined by western blot (Figure 4A and 4B). mCherry-ORF57 expressing cells showed significant DNA damage with a tail moment of 7. 59 compared to 1. 32 for mCherry expressing cells (P = 1. 53×10−7) (Figure 4C and 4D). These data strongly support the conclusion that ORF57 expression leads to DNA damage. Furthermore, western blot analysis was performed on the total protein levels of ORF57 and Aly. iORF57-293 cells either left uninduced or induced for 16 hours to express ORF57,293T cells either mock transfected or transfected for 24 hours with EGFP-ORF57 and a HEK 293T based cell line containing the entire KSHV genome, termed 293T rKSHV. 219 [60], that was reactivated with 20 ng/ml TPA and 1. 5 mM sodium butyrate for 36 hours showed that the relative levels of Aly remain relatively unchanged upon ORF57 expression (Figure S3). There is a slight increase in Aly expression in iORF57-293 cells upon induction of ORF57 which could be explained by a compensatory effect by the cells due to the sequestration of hTREX away from cellular transcription. Moreover, we endeavoured to investigate the effect that ORF57 over-expression has on cellular nuclear export of polyadanylated mRNA as it seems likely that ORF57 recruitment of hTREX would mirror a hTREX knockdown (Figure S4). To this end we performed fluorescence in situ hybridisation (FISH) experiments on 293T cells either mock transfected, transfected with EGFP or transfected with EGFP-ORF57. Importantly, no effect was observed on polyadenylated mRNA in mock transfected or EGFP transfected cells. However, EGFP-ORF57 over-expression had a marked impact on the subcellular localisation of polyadenlyated mRNA with a large proportion retained in the nucleus. Interestingly, the retained polyadenylated mRNA does not co-localise completely with ORF57, suggesting that it is not ORF57 directly that is retaining the cellular mRNA. This data shows convincingly that ORF57 binding to hTREX mimics a hTREX knockdown and causes a block to bulk cellular mRNA export. We have previously shown that ORF57 recruits the entire hTREX complex [4], [47], [61]. Taking into account the link between hTREX aberrations and genome instability, we hypothesised that the DSB response observed upon ORF57 expression could be due to the interaction between ORF57 and hTREX. To test this hypothesis we undertook a series of comet assays in HEK 293T cells expressing ORF57. Initially, we tested whether ORF57 expression alone led to an increase in single and double strand breaks using alkaline comet assays. Cells were either transfected with a construct expressing mCherry, or an mCherry tagged ORF57 (mCherry-ORF57), as well as untransfected cells treated with etoposide (50 µM for 15 minutes) as a positive control. Western blot analysis shows exogenous protein expression (Figure 5A) and fluorescence microscopy images are provided to show the high level of transfection efficiency (Figure 5B (i) ). Alkaline comet assays were performed to determine the level of total single and double strand breaks (Figure 5B). Cells transfected with mCherry showed minimal levels of DNA damage with a tail moment of 1. 78, compared to 33. 27 for etoposide treated cells (P = 3. 81×10−37, unpaired 1-tailed T-tests were used for this and all subsequent statistical analyses). Notably, cells expressing mCherry-ORF57 also showed significant DNA damage with a tail moment of 10. 35 (P = 2. 37×10−7 when compared to mCherry transfected cells). These findings are consistent with our previous observations that ORF57 expression induces a DSB response and causes DNA damage and genome instability. The loss of hTREX function has previously been shown to lead to the formation of R-loops in both yeast and siRNA knockdown in human cells [33], [62]. We hypothesise that ORF57 sequestration of hTREX away from cellular transcription to sites of viral transcription could lead to a loss of cellular mRNA stability and the formation of R-loops. To determine this, we utilised a technique shown previously to confirm the presence of R-loops that involves the over-expression of RNaseH and activation-induced cytidine deaminase (AID). The principle being that an R-loop consists of an RNA: DNA hybrid and that overexpression of an RNaseH enzyme in a system containing R-loops will lead to the degradation of the RNA component allowing resolution of the R-loop and a reduction in DNA damage [63]. Alternatively, overexpression of AID can lead to a state of hypermutation [64] whereby the single-stranded DNA present in an R-loop is subjected to increased deamination that would lead to an increase in the level of observed DNA damage. HEK 293T cells were again transfected with mCherry-ORF57 along with either EGFP-RNaseH1 or pMSCVgfp: : AID (Figure 5C (i) ). The cells expressing mCherry-ORF57 and pMSCVgfp: : AID showed significant increase in DNA damage with a tail moment of 14. 72 (P = 2. 72×10−13 compared to mCherry expressing cells and P = 0. 019 compared to mCherry-ORF57 expressing cells) (Figure 5C). This significant increase demonstrates that in cells expressing ORF57 there is single stranded DNA present, indicative of R-loops. Moreover, cells expressing mCherry-ORF57 and EGFP-RNaseH1 had a tail moment of 5. 81 and showed a significant decrease in the level of DNA damage when compared to mCherry-ORF57 expression alone (P = 0. 006) (Figure 5C), indicating the presence of RNA: DNA hybrids. Importantly, together these data confirm that ORF57 induces the formation of R-loops. Finally, to confirm that R-loop formation in the presence of ORF57 is a consequence of ORF57 sequestering hTREX away from cellular transcription and to sites of viral transcription, we first transfected HEK 293T cells with an ORF57 mutant, ORF57pmut, which has previously been shown to be unable to interact with Aly and recruit the remainder of the hTREX complex (Figure 5D (i) ) [49]. If indeed ORF57 does induce R-loops and the subsequent genome instability is due to sequestration of hTREX this mutant should show a reduction in DNA damage. Expression of ORF57pmut produced a tail moment of 5. 46, significantly lower than expression of mCherry-ORF57 (P = 0. 002) (Figure 5D). It should be noted that although this mutant is unable to recruit hTREX via Aly the inherent redundancy within hTREX means that UIF could be recruited, albeit at a slightly lower level. This could explain the observation that there is still a level of DNA damage in ORF57pmut expressing cells. To further investigate this sequestration mechanism, we co-transfected HEK 293T cells with constructs expressing mCherry-ORF57 as well as a Myc-tagged Aly construct (Figure 5D (i) ). Myc-Aly in this system should function as a molecular sponge and prevent ORF57 from sequestering all of the endogenous hTREX proteins, thereby reducing the level of DNA damage observed. Strikingly, Myc-Aly overexpression did reduce the tail moment to 4. 31 (P = 2. 26×10−4). Further controls were performed on cells transfected with pMSCVgfp: : AID alone, or transfected with EGFP-RNaseH1 alone, and show no significant increase in tail moment compared to mock transfected cells (Figure S5). Taken together, these results suggest that sequestration of hTREX by ORF57 results in the observed genome instability due to of R-loop formation. To examine whether our hypothesis of hTREX sequestration by ORF57 away from cellular transcription to sites of viral transcription leads to R-loop formation during KSHV infection we utilised the HEK 293T based cell line containing the entire KSHV genome, 293T rKSHV. 219. We performed a series of alkaline comet assay experiments to determine the levels of DNA strand breaks comparing KSHV latent and reactivated cells (Figure 6). Unreactivated cells showed a tail moment of 0. 43, whilst after being reactivated using 20 ng/ml TPA and 1. 5 mM sodium butyrate for 36 hours cells showed a significant increase in the tail moment to 5. 49 (P = 6. 18×10−9) demonstrating a level of DNA strand breaks. This confirms our previous observations in TREx BCBL-Rta cells showing that KSHV lytic replication directly causes DNA damage. To determine whether the observed DNA damage could be attributed, at least in part, to R-loop formation we again utilised a construct expressing EGFP-RNaseH1 and transfected it into the 293T rKSHV. 219 cells 8 hours prior to reactivation. Importantly, there was a significant decrease in the tail moment when compared to the reactivated cells without EGFP-RNaseH1 over-expression down to 1. 42 (P = 4. 31×10−6). This demonstrates that the DNA strand breaks in KSHV lytically replicating cells are as a result of the formation of R-loops. Moreover, to demonstrate that the R-loop formation is as a result of ORF57 sequestering hTREX away from cellular transcription to sites of viral transcription we transfected Myc-Aly into the 293T rKSHV. 219 cells. As was observed in the ORF57 over-expression system this again dramatically and significantly reduced the tail moment to 2. 06 (P = 1. 24×10−4 when compared to reactivated 293T rKSHV. 219 cells) indicating that hTREX sequestration by ORF57 is the cause of R-loop formation and DNA strand breaks in lytically active KSHV infected cells. It follows that if ORF57 is capable of recruiting hTREX in such a way as to mimic hTREX knockdown leading to R-loops and genome instability, that other herpesvirus homologues of ORF57 that interact with hTREX could function in a similar way. We therefore decided to investigate whether the ORF57 homologue of HSV-1, ICP27, was sufficient to elicit a similar response. We chose ICP27 as a DNA damage response has been observed in HSV-1 infected cells and ICP27 is known to interact with hTREX via a direct interaction with Aly [65], [66]. Moreover, mutant proteins are available which abolish the ICP27-Aly interaction [67]. Alkaline comet assays were again performed on 293T cells transfected with EGFP-ICP27 (Figure 7). Cells expressing EGFP-ICP27 showed a tail moment of 10. 71 compared to 3. 50 for EGFP transfected cells (P = 3. 32×10−6), demonstrating that ICP27 is indeed sufficient to induce a significant amount of DNA strand breaks. To assess whether the observed DNA damage is a consequence of R-loop formation cells were co-transfected with EGFP-ICP27 along with either pMSCVgfp: : AID or EGFP-RNaseH1. AID expression led to an increase in the tail moment to 13. 62, although not significant (P = 0. 259) whereas RHaseN1 expression significantly decreased the tail moment to 2. 45 (P = 5. 42×10−8), indicating that, similar to KSHV ORF57, HSV-1 ICP27 causes DNA damage via R-loop formation. To further characterise the mechanism we either co-transfected EGFP-ICP27 along with Myc-Aly or transfected a mutant of ICP27 which fails to interact with Aly, known as EGFP-ICP27WRL [65], [67]. Again, as expected, both showed reduced tail moments when compared to EGFP-ICP27 alone (4. 03 and 3. 90 respectively, with P-values of 1. 74×10−5 and 9. 17×10−6) suggesting that the R-loop formation is as a result of ICP27 recruiting hTREX away from cellular transcription. The role of hTREX in genome stability and cancer is an important emerging field [25], [29]. As we have previously shown, the KSHV ORF57 protein interacts directly with hTREX [47], [68]. Moreover, KSHV is known to induce DNA double-strand breaks during lytic replication, although no mechanism has been previously described [17], [49]. Therefore, we set out to investigate what effect sequestration of hTREX by KSHV ORF57 would have upon R-loop formation and genome instability, and whether this would describe a mechanism by which KSHV could induce DNA strand-breaks. The data presented here convincingly show that using the KSHV mRNA export factor, ORF57, we can accurately replicate a system deficient in hTREX and that this lack of hTREX does lead to R-loop-mediated genome instability in KSHV infected cells. Our model proposes that aberrant hTREX levels, sequestered by ORF57, leads to the formation of R-loops, which in turn increases the number of DNA DSBs and the rate of mutation. This model explains for the first time a possible novel mechanism of lytic KSHV induced DNA double-strand breaks (Figure 8). We suggest that this model could also describe a possible driving force behind KSHV tumorigenesis as several studies have shown significant defects in proteins involved in mRNA processing in numerous cancers [25], [29], [30]. Importantly, the observed decrease in DNA damage in 293T rKSHV. 219 cells when either Aly or RNaseH1 are over-expressed demonstrates that this model is viable in the context of a KSHV infection. The complex nature of lytic replication in KSHV makes it difficult to confirm that all observed DNA damage is as a result of ORF57-induced R-loops as other studies have shown a DNA damage response caused by other KSHV proteins, for example v-cyclin [69] and LANA [54], [70]. However, our data present compelling evidence that R-loop formation as a result of ORF57 sequestration of hTREX away from cellular transcription to sites of viral transcription could have a significant impact on genome instability in KSHV infected cells. An additional consideration is how the induction of DNA damage during the early stages of KSHV lytic replication could lead to mutations in progeny cells during cancer development. It is established that early events in primary KSHV infection of target cells involves the expression of a subset of lytic and latent viral genes prior to the establishment of latency [71]. This includes genes that require the activity of ORF57 for their expression, including K8, K8. 1 and ORF59 [72], [73]. It seems likely then that expression of ORF57 during de novo infection prior to the onset of latency could give rise to a background level of genome instability in infected cells, as has been suggested previously [17]. Interestingly, recent data suggests that KSHV lytic replication leads to genome instability, although no underlying mechanism was described [17]. This is in addition to observations of chromosomal abnormalities in KSHV infected cells and KSHV-associated tumours [18]–[24]. Furthermore, it has been suggested that the DNA damage checkpoint response could function as an anticancer barrier in KSHV infected cells, as KSHV v-cyclin expression leads to a DNA damage response [69]. Importantly, other herpesviruses are also known to lead to a DSB response [74]–[76]. This could have significant implications, as all herpesviruses encode an ORF57 homologue that interacts with the cellular mRNA export machinery [65], [67], [77]–[79]. However, not all herpesviruses are associated with cancer development. Therefore, the R-loop induced DNA damage could be a side-effect of virus replication, although with persistent infections where lytic replication plays a role in tumourigenesis, as is the case with KSHV, this could be an important driving force behind mutation and cancer development. Alternatively, the DNA damage observed in this study caused by ICP27 during HSV-1 lytic replication may have less impact during the viral cycle where persistent infection is not linked to tumourigenesis. Interestingly, the genome instability caused by the formation of R-loops by ORF57-mediated sequestration of hTREX does not fully explain the observation that ORF57 can induce chromosome instability in the form of chromosome lagging. This observation is, however, intimately linked to the sequestration of hTREX. Depletion of UAP56, as well as other components of hTREX, by siRNA knockdown has been shown to lead to a large increase in chromosomal instability [31]. Loss of UAP56 is shown to lead to premature sister chromatid separation which is known to be a cause of micronuclei formation [80]. This could have significant implications for cells deficient in UAP56 and hTREX because of the strong link between micronuclei formation, genome instability and cancer [7], [81]. Importantly, both chromosome lagging and micronuclei formation have been reported previously in KSHV infected cells [18]. Therefore, loss of functional hTREX due to sequestration by KSHV ORF57 may cause genome instability through multiple mechanisms including the induction of R-loops as presented here, but also through aberrations in mitosis that can lead to micronuclei formation, work that is currently being investigated in our laboratory. In summary, our data sheds new light on the mechanism of R-loop induced genome instability in the oncogenic herpesvirus, KSHV. We have demonstrated that sequestration of hTREX to sites of viral transcription by the KSHV ORF57 protein leads to severe DNA damage, and that R-loop formation is the cause of this genetic instability. We have shown the importance of the hTREX complex in maintaining genome instability in KSHV infected cells, and that loss of function could be a significant cause of double strand breaks. Importantly, when combined with data that shows loss of hTREX components in multiple cancers this model could be an important mechanism of DNA damage during tumorigenesis. Our work highlights a novel mechanism by which KSHV can induce genome instability in infected cells, an enabling characteristic of the hallmarks of cancer. It also demonstrates the importance of further understanding the complex links between mRNA processing and genome instability and the roles that mRNA processing factors have in cancer formation and progression. pmCherry-N1, pEGFP-C1 and pEGFP-N1 were obtained from Clontech. pmCherry-ORF57 was cloned into pmCherry-N1, and pEGFP-ORF57 and pORF57pmut were cloned into pEGFP-N1, all described previously [47], [49], [61]. pMSCVgfp: : AID contains full-length human AID and was obtained from Addgene (Addgene plasmid 15925) [64]. pEGFP-RNH contains full-length human RNAse H1 gene and was a kind gift from Dr Anneloor ten Asbroek [82]. EGFP-ICP27 contains the ICP27 gene from HSV-1 in the pEGFP-N1 vector, and the EGFP-ICP27 WRL mutant contains ICP27 with the point mutations W105A R107A L108A [67]. Etoposide was purchased from Cambridge Bioscience and cell treatments were as described in the text. For the inducible iORF57-293 cell line, ORF57 was cloned into pcDNA5/FRT/TO and produced using the Flp-In T-REx system (Life Technologies) following the manufacturer' s instructions and as previously described [83]. Both the Flp-In T-REx parental cell line and the iORF57-293 cell line were grown in DMEM supplemented with 10% v/v foetal bovine serum and penicillin/streptomycin (Lonza). The TREx BCBL Rta cell line was provided by Professor Jae Jung [51] and grown in RPMI supplemented with 10% foetal bovine serum and penicillin/streptomycin. Both cell lines were kept under hygromycin B selection at a concentration of 100 µg/ml, and inductions of both cell lines were performed using 2 µg/ml doxycyclin. The HEK-293T cell line was obtained from the Health Protection Agency Culture Collection and grown in DMEM supplemented with 10% foetal bovine serum and penicillin/streptomycin. The 293 rKSHV. 219 cells are a 293T cell line containing a recombinant bacterial artificial chromosome harbouring the KSHV genome [60]. Cells were frown in DMEM supplemented with 10% foetal bovine serum and penicillin/streptomycin. Transfections were performed using Lipofectamine 2000 (Life Technologies), as previously described [84]. The monoclonal antibodies to GAPDH and fibrillarin and the polyclonal antibodies to RFP and lamin B1 were purchased from Abcam and used at 1∶5000,1∶1000,1∶1000 and 1∶2500, respectively. The monoclonal antibodies to Myc and Flag and the polyclonal antibody to Flag were from Sigma and were used at 1∶2500,1∶1000 and 1∶1000, respectively for western blotting. Polyclonal Flag was used at 1∶250 for immunofluorescence. Monoclonal ORF57 antibody was obtained from Santa Cruz and used at 1∶1000 for western blotting and 1∶100 for immunofluorescence, and the monoclonal antibody to γH2A. x and polyclonal antibody to H2A. x were from BioLegend and used at 1∶2500 and 1∶1000, respectively for western blotting. Monoclonal γH2A. x was used at 1∶100 for immunofluorescence. Secondary antibodies for western blotting were HRP-conjugated polyclonal goat anti-mouse and polyclonal goat anti-rabbit supplied by Dako. Fluorescently-conjugated secondary antibodies were all obtained from Life Technologies and used at 1∶500: monoclonal Alexa Fluor 488, monoclonal Alexa Fluor 546, polyclonal Alexa Fluor 546 and polyclonal Alexa Fluor 546. Western blots were performed as previously described [85]. Briefly, protein samples were run on 10–12% polyacrylamide gels and transferred to nitrocellulose membranes via semi-dry blotting. Membranes were blocked with TBS + 0. 1% v/v Tween 20 and 5% w/v dried skimmed milk powder. Membranes were probed with relevant primary and secondary antibodies, treated with EZ-ECL (Geneflow), and exposed to Amersham hyperfilm ECL (GE Healthcare). Full details of the SILAC proteomics methods can be found in supplementary information (Text S1). The pathway analysis was performed with the Ingenuity Systems software packet, IPA 9. 0 (Ingenuity Systems, Inc). Comet assays were performed using the CometAssay ES II system (Trevigen) and performed to the manufacturer' s instructions. In brief, cells were transfected as necessary and grown for 48 hours before being harvested and kept on ice in PBS. Cells were counted and diluted to 1×105/ml in PBS and mixed with low melting point agarose at 37°C at a ratio of 10∶1, agarose: cells. 50 µl of the cell/agarose suspension was spread onto a 2-well CometSlide and allowed to set before being placed in lysis solution at 4°C for 1–2 hours in the dark. For the alkaline assay, slides were then immersed in alkaline unwinding solution for 20 minutes at room temperature. Electrophoresis was performed in a CometAssay ES II unit in alkaline electrophoresis buffer for 30 minutes at 21 V and subsequently washed twice in water and once in ethanol for 5 minutes each before being dried at 37°C. For the neutral assay, after lysis slides were incubated in neutral electrophoresis buffer for 30 minutes at 4°C. Electrophoresis was performed in a CometAssay ES II unit in neutral electrophoresis buffer for 45 minutes at 21 V. Slides were then incubated in DNA precipitation buffer for 30 minutes at room temperature followed by 30 minutes in 70% ethanol at room temperature before being dried at 37°C. For both neutral and alkaline assays, slides were then stained with SYBR Gold for 30 minutes and subsequently imaged on a Zeiss LSM 700 laser scanning confocal microscope. Images were exported from Zen 2011 and comets were scored using TriTek CometScore. Cell fixation and staining was performed as previously described [86], [87]. Briefly, cells were grown on sterilised glass coverslips and either transfected or induced. After a specified time cells were washed in PBS and fixed in PBS containing 4% v/v paraformaldehyde for 10 minutes, washed further in PBS and permeabilised using PBS containing 1% Triton X-100 for 10 minutes. Coverslips were then incubated with appropriate primary and secondary antibodies for 1 hour each at 37°C before being mounted onto microscope slides using Vectashield with DAPI. Slides were visualised on a Zeiss LSM 700 laser scanning confocal microscope and images analysed using Zen 2011 (Zeiss). For analysis of cell transfection by fluorescence microscopy, cells were visualised using a Leica DC 300F fluorescence microscope. Cells were either untreated or treated with 50 µM etoposide for 30 minutes at 37°C immediately prior to fixation. Cells were fixes in PBS containing 1% paraformaldehyde for 5 minutes at room temperature and washed twice in PBS. Cells (2×106 cells/ml) were permeabilised in 1× permeabilisation buffer (PBS with 1. 25 mM EDTA, 2% foetal calf serum, 0. 5% Triton X-100) for 20 minutes on ice before being incubated on ice for 1 hour with primary antibody. Cells were washed in PBS for 10 minutes before being incubated in secondary antibody for 1 hour on ice and washed again for 10 minutes in PBS. Fluorescence was measured using the 288 nm laser of a Becton Dickson BD-LSRFortessa flow cytometer and the proportion of fluorescent cells was calculated using the DiVa 6 software, as previously described [88]. To detect polyadenylated RNA, an HPLC-purified oligo dT (70) probe labelled at the 5′ end with Alexa Fluor 546 NHS Ester was used. 293T cells were attached to coverslips coated with poly-L-lysine and transfected for 24 hours before being washed once with PBS and fixed for 10 minutes in PBS with 4% paraformaldehyde, washed three times in PBS, permeabilised in PBS with 0. 5% Triton X-100 for 5 minutes, washed twice more with PBS and once with 2× SSC for 10 minutes, all at room temperature. The oligo dT (70) probe was then added at 1 ng/µl in ULTRAbyb-oligo hybridisation buffer (Ambion) for 16 hours at 42°C. Cells were washed for 15 minutes each, twice with 2× SSC, once with 1× SSC and once with PBS all at room temperature. Coverslips were mounted onto microscope slides using Vectashield with DAPI. Slides were visualised on a Zeiss LSM 700 laser scanning confocal microscope and images analysed using Zen 2011 (Zeiss).

The hallmarks of cancer comprise the essential elements that permit the formation and development of human tumours. Genome instability is an enabling characteristic that allows the progression of tumorigenesis through genetic mutation and therefore, understanding the molecular causes of genome instability in all cancers is essential for development of therapeutics. The Kaposi' s sarcoma-associated herpesvirus (KSHV) is an important human pathogen that causes multiple AIDS-related cancers. Recent studies have shown that during KSHV infection, cells show an increase in a double-strand DNA break marker, signifying a severe form of genome instability. Herein, we show that KSHV infection does cause DNA strand breaks. Moreover, we describe a novel molecular mechanism for genome instability involving the KSHV ORF57 protein interacting with the mRNA export complex, hTREX. We demonstrate that over-expression of ORF57 results in the formation of RNA: DNA hybrids, or R-loops, that lead to an increase in genome instability. DNA strand breaks have been previously reported in herpes simplex, cytomegalovirus and Epstein-Barr virus infected cells. Therefore, as this work describes for the first time the mechanism of R-loop induced genome instability involving a conserved herpesvirus protein, it may have far-reaching implications for other viral RNA export factors.

Abstract Introduction Results Discussion Materials and Methods

biochemistry rna medicine and health sciences rna transport cell biology nucleic acids pathology and laboratory medicine host-pathogen interactions virology viruses and cancer biology and life sciences microbiology molecular cell biology pathogenesis

2014

A Novel Mechanism Inducing Genome Instability in Kaposi's Sarcoma-Associated Herpesvirus Infected Cells

The fixation of into living matter sustains all life on Earth, and embeds the biosphere within geochemistry. The six known chemical pathways used by extant organisms for this function are recognized to have overlaps, but their evolution is incompletely understood. Here we reconstruct the complete early evolutionary history of biological carbon-fixation, relating all modern pathways to a single ancestral form. We find that innovations in carbon-fixation were the foundation for most major early divergences in the tree of life. These findings are based on a novel method that fully integrates metabolic and phylogenetic constraints. Comparing gene-profiles across the metabolic cores of deep-branching organisms and requiring that they are capable of synthesizing all their biomass components leads to the surprising conclusion that the most common form for deep-branching autotrophic carbon-fixation combines two disconnected sub-networks, each supplying carbon to distinct biomass components. One of these is a linear folate-based pathway of reduction previously only recognized as a fixation route in the complete Wood-Ljungdahl pathway, but which more generally may exclude the final step of synthesizing acetyl-CoA. Using metabolic constraints we then reconstruct a “phylometabolic” tree with a high degree of parsimony that traces the evolution of complete carbon-fixation pathways, and has a clear structure down to the root. This tree requires few instances of lateral gene transfer or convergence, and instead suggests a simple evolutionary dynamic in which all divergences have primary environmental causes. Energy optimization and oxygen toxicity are the two strongest forces of selection. The root of this tree combines the reductive citric acid cycle and the Wood-Ljungdahl pathway into a single connected network. This linked network lacks the selective optimization of modern fixation pathways but its redundancy leads to a more robust topology, making it more plausible than any modern pathway as a primitive universal ancestral form. Six different autotrophic carbon-fixation pathways have been identified across the tree of life [1]–[3]. It has been recognized that some of these pathways share reaction sequences, but a comprehensive framework does not yet exist to interpret the relatedness among all these extant phenotypes, or to judge which if any is the best candidate for a preserved ancestral form [3]–[6]. Phenotypes exclusive to both the bacterial and archaeal domains have been found, but a full explanation for the patterns of exclusivity has not yet been given [1], [2]. Discussions of the ancestral mode of carbon fixation have focussed primarily on the Wood-Ljungdahl (WL) pathway [7] – as part of the reductive acetyl-CoA pathways – and on the reductive citric acid cycle (rTCA) [8], but diverse observations of metabolic universality and simplicity, network topology, and phylogenetic distribution have not yet been given a single compatible interpretation [3]–[6]. As will be elaborated upon below, throughout this work we use the concepts of carbon-fixation ‘pathways’ and ‘phenotypes’ interchangeably as we focus only on the initial biochemical sequences of how enters living cells. While consensus remains elusive on which pathway (if any) represents the ancestral form of carbon fixation, it has become increasingly accepted over the last 30 years that the first forms of life were likely chemoautotrophic, deriving all biomass components from, and all energy from inorganic redox couples in the environment [3]–[6], [9]. (Photoautotrophs, in contrast, derive energy from sunlight, while heterotrophs derive energy and biomass from organic sources of carbon.) Most discussions of autotrophy in the origin of life are complicated because they combine chemical requirements for carbon and energy uptake with considerations of whether organisms or syntrophic ecosystems were required to complete the required pathways, and the ways the resulting population processes would have related to genomics on one hand and to metabolic regulations on the other. The disentangling of these issues is addressed in [10], and we will not revisit them here. Instead we focus on the purely biochemical requirements for autotrophy, and adopt the working hypothesis that at the ecosystem level the biosphere has been autotrophic since its inception. Autotrophy provides a simple yet powerful constraint from which we derive a coherent framework for the evolution of all extant fixation pathways. The modern biosphere may be described, most fundamentally, as implementing a biological carbon cycle based on, in which carbon fixation is the metabolic anchor embedding life within geochemistry. If the earliest ecosystems were also autotrophic, then a carbon cycle based on must have existed continuously to have supported biosynthesis. Any local ecosystems that could no longer fix either would have become appendages to neighboring autotrophic ecosystems, or would have dwindled and been recolonized by such ecosystems. Under either scenario it must be possible, relative to appropriately defined ecosystem boundaries, to trace all extant carbon-fixation pathways through fully autotrophic sequences to the earliest forms. The question for historical reconstruction is then whether these sequences are sufficiently simple that their innovations could have occurred independently, or whether complex, cross-species and interdependent sequences of innovations sustained continuous carbon-fixation. In the latter case ecosystems become the relevant units of phylogenetic comparison, and clear lineages in carbon-fixation may be difficult, if not impossible, to reconstruct. A history of independent innovations, in contrast, allows us to be indifferent to the distinction of ecosystems and species, and should make clear carbon-fixation lineages distinguishable. In this paper we reconstruct the evolutionary sequences that relate modern carbon-fixation pathways to each other and to a single ancestral form, showing that the history of carbon-fixation can indeed be described as simple sequences of independent innovations in autotrophy. To define the constraint of autotrophy we will use metabolic-flux balance analysis, and because we do not use it to model cellular-level mechanisms of either regulation or heredity, it does not distinguish among strictly defined autotrophic species, populations of diverse cells (or pre-cells) tightly linked by transfer of genes and metabolites [11], [12], or syntrophic ecosystems. Therefore the ‘carbon-fixation phenotypes’ that we analyze may, but need not, have corresponded to phenotypes integrated within single species. The intensity of gene transfer and the integrity of species lineages thus become moot points, unless they lead to signatures of complexity and non-independence in the sequence of carbon-fixation phenotypes. Moreover, we focus here on the integrated pathways of carbon fixation alone, requiring only that in the bottom-up construction of biomass all initial metabolic branching points be directly accessible from, and do not extend to the more complicated reconstruction of full intermediary metabolism. The remarkble modularity and redundancy of carbon fixation pathways, and the small number of metabolites through which they are connected to the rest of metabolism, make this separation feasible for autotrophy. We return to the complexities that arise upon consideration of heterotrophic organisms and larger components of intermediary metabolism in a later section. We thus describe a diversification of the of input channels of carbon into the biosphere, from which downstream anabolic pathways may diverge in different organisms. (Anabolism is the process by which life constructs larger organic metabolites from smaller ones, while catabolism is the converse breakdown of larger into smaller molecules.) Distinctions caused by the participation of heterotrophs in complex ecosystems, as well as those among organisms that share carbon-fixation pathways but use these to feed diversified intermediary metabolism, all arise outside the networks we model here. The ancestral pathways we reconstruct are therefore best understood as divergences in the ecosystem-level metabolic foundation of an emerging biosphere. Our evolutionary reconstructions are based on a novel integrated metabolomic/phylogenomics approach, whose basic principles we outline next. The two main statistical tools that exist to probe genetic information and study the early metabolic evolution of life are phylogenetics [13]–[15] and metabolic flux-balance analysis (FBA) [16], [17]. Whole-genome FBA models have become a widely used and successful method to study the metabolism of individual organisms [16], [18]. One can use this approach and target deep-branching organisms in an attempt to study conserved metabolic features near the base of the tree of life. However, due to the inherent ambiguities and errors in using gene sequence comparisons to determine the presence of enzymes [19], [20] and therefore reactions, heuristics are needed to fill the “gaps” in the initial network that is derived from the genome to produce a viable organism model [16], [18]. These techniques work remarkably well in predicting overall growth rate dynamics under various conditions, especially for well-characterized model organisms [16], [18]. It is less clear, however, what confidence to assign to such heuristic rules when targeting individual pathways of evolutionary interest, especially for deep-branching organisms that lack extensive laboratory characterization and that have significant gene divergences from well-characterized model organisms. Similarly, phylogenetics based on gene presence/absence and sequence similarity, without other prior constraints, has given significant insights into the relatedness and historical divergences among organisms [13], [14]. However, branching relationships become more ambiguous at greater phylogenetic depths, as tree-like descriptions fail for whole organisms due to extensive lateral gene transfer (LGT) near the root [12], [14], [21]. This can make the phylogenetic position (and evolutionary divergence) of metabolic pathways uncertain, as has been the case for carbon-fixation pathways [1], [2]. Phylogenetics and flux-balance analyses, if used together, have complementary strengths that may ameliorate some of these problems. We refer to the joint use of the two methods as “phylometabolic” analysis, and illustrate its main features in Fig. 1. Instead of using heuristics to fill gaps in order to complete an initial network derived from genome annotation (as for example in Fig. 1A), we compare the gene profile for a metabolic pathway in a focal organism to those in related organisms both within and across neighboring clades as shown in Fig. 1B. By focusing at the pathway level, the comparison may reveal variations in multi-genic functional units, providing context for the completion of individual organism networks, while also restricting the plausible sequences for divergence in the evolution of metabolic structure. Especially in cases where individual reactions or growth conditions, rather than complete pathways, have been chemically characterized in different organisms, comparison of functional units can pool evidence that would not be restrictive in isolation. Conversely, as shown in Fig. 1C, placing hypothesized constraints on functional units, such as requiring the continuous production of essential metabolites, can lead to specific conclusions about uncertain clade-level branches in phylogenetic trees. This may be understood as adding a semantic dimension for the contribution of genes to phenotypes, which provides a different form of disambiguation, when phylogenies based on gene presence/absence alone yield poorly defined deep branching relationships as summary statistics for reticulated networks of single-gene histories [14], [21]. The reconstructed “phylometabolic tree”, shown in Fig. 1D, by including multiple complete pathways to common essential metabolites, then suggests which evolutionary substitutions are allowed (at either organism or ecosystem levels) among these pathways by the constraint that essential metabolites are continuously produced. As we will show for the evolution of carbon-fixation, complete pathway reconstructions, combined with characterization of reaction mechanisms and enzymes, often suggests the causes for reconstructed divergences. In an analysis of large or highly ambiguous networks, statistical methods already employed for gene phylogenies, or those used to suggest enzyme functions in automated FBA modeling, could be combined into joint-maximum-likelihood or Bayesian Markov Chain Monte Carlo (MCMC) reconstruction algorithms. The small size and high parsimony of the network we will consider permitted manual reconstruction. Future work will extend these methods to the reconstruction of full models of intermediary metabolism – in a separate manuscript we will describe the metabolic reconstruction of the deep-branching hyperthermophilic chemoautotroph Aquifex aeolicus. For the carbon-fixation networks considered here, important interactions of phylogenetic and flux-balance constraints occur at two distinct points in the network. The first, of the kind shown in Fig. 1B, determines our reconstruction of the ancestral synthesis route to glycine and serine, which from the perspective of input form a unique set among the monomer precursors to biomass (e. g. amino acids, nucleotides). It has been observed [22] that all anabolic pathways originate from five universal precursors: acetyl-CoA, pyruvate, oxaloacetate, succinyl-CoA and -ketoglutarate, and that all of these are intermediates in the citric acid (TCA) cycle. Even in organisms whose carbon-fixation pathways do not pass through this cycle, TCA arcs are used as supplementary pathways to connect primary carbon inputs to these standard precursors. The sole exceptions are glycine and serine (and a few higher order derivatives of these), which, as shown in Fig. 2, can in some organisms be produced directly from, bypassing both the TCA intermediates and all of the recognized complete carbon-fixation pathways. Operation in the fully reductive direction of the folate-based chemistry that forms the core of this pathway has only been observed within the context of the complete Wood-Ljungdahl pathway, in which acetyl-CoA is synthesized as the final step. However, all steps in this sequence to glycine and serine are fully reversible [23], [24], and so there is no reason a priori to exclude the possibility of much wider use of this pathway in the context of carbon-fixation. In the next section we will present evidence strongly suggesting that this pathway is in fact widely present across the tree of life without the final step to acetyl-CoA in organisms that lack alternate routes to glycine and serine, and that it forms the most likely ancestral pathway to these amino acids. This is surprising, not only because carbon fixation on folates had previously only been recognized in anaerobic organisms using it as the sole route to fix, but also because our results suggests that organisms commonly use two disjoint parallel pathways to fix carbon. The presence of two parallel carbon-fixation pathways in an autotroph was only recently noted in a single, late branching -Proteobacterium [25] (begging the question how common this might be [2]) but even in that case the two pathways are connected through metabolic intermediates. We will show evidence that direct reductive synthesis of glycine and serine combines with all other carbon-fixation pathways, and that in many of these cases the two pathways supply distinct, disconnected components of biomass. Perhaps the most important consequence of these findings is that it significantly increases the similarity among all carbon-fixation phenotypes. In particular, the most similar carbon-fixation phenotypes are now the deep branching form of rTCA, which combines a complete rTCA cycle with an incomplete WL pathway to synthesize glycine and serine, and Wood-Ljungdahl, which combines a complete direct reductive sequence from to acetyl-CoA with partial TCA cycle sequences to complete the set of universal anabolic precursors. Moreover, among all carbon-fixation machinery, the folate-based direct reduction of and partial TCA sequences appear to have the most universal distributions across the tree of life, supporting an early appearance for both. Together these various observations suggest general principles underlying the evolutionary diversification of carbon-fixation, as well as an avenue to the reconciliation of phylogenetic and metabolic observations that previously appeared incompatible. The elaboration of these results below will lead to us to a second junction at which we will invoke phylometabolic constraints, this time of the type in Fig. 1 C/D, in reconstructing the complete evolutionary sequences that connect all carbon-fixation phenotypes to a single ancestral form. As we go through these analyses we also make observations on the nature of the associated reactions and enzymes of key steps in the different forms of carbon-fixation, from which we identify plausible ecological and evolutionary explanations for the divergences at different stages. In particular we include a section immediately following the analysis of glycine/serine pathways that describes in detail how one of the major driving forces - energy optimization - is inferred from a wide range of evidence. The discussion of this driving force, and its interaction with others, will then set the stage for interpreting the full evolutionary history of carbon-fixation. We begin, however, by describing direct reduction of one-carbon () units in more detail, which is not only an essential source of core metabolites in all WL organisms, but occurs ubiquitously as a module for carbon fixation in interaction with other autotrophic pathways. Direct reduction follows the ‘central superhighway’ of tetrahydrofolate (THF) metabolism (some archaea use tetrahydromethanopterin () and related compounds, all analogues of THF [24]), which links the synthesis of purines, thymidilate, formyl-tRNA, serine, and methyl-group chemistry mediated by S-adenosyl methionine (SAM) [24]. The core reactions of this pathway, summarized in Fig. 2, are widespread in both oxidative and reductive form throughout the tree of life [24], [26]. In their reductive form these reactions, followed by the acetyl-CoA synthase reaction – catalyzed by one of the most oxygen-sensitive enzymes in our biosphere [27] – make up the WL pathway, which is the principal carbon fixation route in a variety of bacteria and archaea, including acetogens and methanogens, sulfate reducers, and possibly anaerobic ammonium oxidizers [2], [28]. In autotrophs, the WL pathway couples in a variety of ways to an incomplete rTCA cycle forming what are collectively known as reductive acetyl-CoA pathways. We will present evidence from genome comparisons that direct reduction is not only a carbon fixation route in WL organisms, but that it is ubiquitous, and was actually the ancestral route to glycine and serine, which took on diversified roles independent from the complete WL pathway very early in the rise of oxygen. The extant glycine and serine synthetic pathways provide the key constraints in reconstructing ancestral carbon fixation, so we describe these next. From their phylogenetic distribution, and their energetics in the context of fully autotrophic networks, we then reconstruct a sequence for the major innovations in carbon fixation. Three main synthetic pathways to glycine and serine are found in modern organisms and provide evidence about ancestral forms. In eukaryotes and most late-branching bacteria, serine is derived from 3-phosphoglycerate (3-PG) as a branch from either glycolysis or gluconeogenesis [26], [29], through reactions of Fig. 2. Serine is subsequently cleaved to glycine, donating a methylene group to THF, and glycine can be further broken down in what is known as the glycine cleavage system to and, donating a second methylene group to THF [23]. We will refer to this route as the “oxidative” pathway because the first step from 3-PG involves the oxidation of the alpha-hydroxyl to a carbonyl group, and parts of the THF pathway operate in the oxidative direction. In the alternative, direct synthesis of glycine from units (reactions in Fig. 2), the THF-mediated reactions proceed exclusively in the reductive direction, and we refer to this route as the “reductive” pathway. The third major route to glycine is via reductive transamination of glyoxylate [29] (reaction 12 in Fig. 2), which is important in cyanobacteria and plants undergoing photorespiration [30]. Following its synthesis from glyoxylate, one molecule of glycine may be cleaved to and, donating a methylene group to THF, which is then combined with a second glycine to produce serine. In photosynthetic organisms, glyoxylate arises from the Calvin-Benson-Bassham (CBB) cycle upon exposure to oxygen [30], but in other organisms it can arise from other sources such as cleavage of isocitrate in the glyoxylate shunt [29]. We track the distribution of glyoxylate transaminase in our gene profile comparisons as the key reaction in all these pathways, and refer to them collectively as the “glyoxylate” pathways. The synthesis of glycine through cleavage of Threonine has been shown to play a role in some organisms (notably Saccharomyces cerevisiae [31], [32]), but its physiological importance is generally not well understood [33], [34]. We interpret this route as a salvage pathway and do not consider it further here. To understand the distribution and reconstruct the history of glycine and serine synthesis, we acquired gene profiles for all three pathways from the UNIPROT database [35], for all strains in a wide range of deep-branching organisms (see Methods for details). UNIPROT derives from the manually reviewed SWISSPROT database, which has one of the lowest rates of mis-annotation among public databases [20]. We find the complete gene complement for the reductive pathway widely distributed among both bacteria and archaea, as shown in Fig. 3, including many non-WL organisms that lack the genes for alternative routes. This latter group includes members of clades that use the rTCA cycle (the Nitrospirae) [36] or the 3-Hydroxypropionate (3-HP) bicycle (the Chloroflexi) [37], anaerobic and aerobic heterotrophs (the Thermotogae and Isosphaera pallida of the Planctomycetes) [38], [39], and also several archaea (all listed in Table S1 in Text S1). The very wide distribution of direct reduction on THF suggests that hybrid carbon fixation is much more common than has been recognized, and that the reductive pathway is not limited to WL organisms in which it is the sole pathway. Indeed, this pathway appears to be more widely distributed than any single primary fixation pathway. A detailed distribution of gene profiles for the three pathways is shown in Table 1. Although the reductive pathway requires 7 or 8 steps (heterotrophic from formate or autotrophic from, respectively) to reach glycine, versus 3 and 1 in the alternative pathways, its frequency in the sample is nearly double the combined frequencies of the oxidative and glyoxylate pathways. The reductive pathway from appears in two common forms. The full pathway (“Red” in the tables) comprises 8 reactions. In the alternative form (denoted “Red (-2) ”), reaction 2, attaching formate to THF, does not appear in genome searches based on sequence similarity, but all 7 other reactions are present. We will argue from detailed analysis of gene frequencies and growth conditions, that the 7-reaction form is active and is in fact a carbon fixation pathway, suggesting that formate incorporation is catalyzed by an unrecognized protein or an unidentified function among the known THF-interconversion enzymes. We suspect that an alternate route involving incorporation through rather than of THF may be active in these cases (see Text S1 for further details). In Table 1 pathways from (autotrophic form) and formate (heterotrophic form) are combined under columns Red and Red (-2). For full 8-step autotrophic gene profile see e. g. organisms MTA, CAG and NDE in Table S1 in Text S1. For full alternate 7-reaction autotrophic gene profiles see e. g. organisms AAE, CCH or NPU in Table S1 in Text S1. Full organism names are also found in Text S1. The most informative distributions come from three clades that are consistently placed among the deepest-branching bacteria: the Firmicutes, Thermotogales and Aquificales [40]–[42]. Thermotogales, Aquificales, and several groups of Firmicutes are among the most thermophilic bacteria and are generally restricted to hydrothermal vents [7], [38], [42], [43]. These environments are among the least changed from early-Earth conditions, and clades apparently restricted to them throughout history may well be the most conservative of metabolic features from the base of the tree [44]. Among the Firmicutes, a remarkably diversified clade, the reductive pathway is the common form, the oxidative pathway is less common, and glyoxylate pathways are very rarely found. The only systematic exception to the common pattern among Firmicutes is found in the Lactobacillales, where the ‘glycine cycle’ (reactions in Fig. 2) is completely absent, apparently having been replaced by the oxidative pathway. Indeed, this group shows the most complete such replacement found among deep-branching bacterial clades. The Lactobacillales, however, are mesophiles, highly adapted to environments rich in organic carbon, and are known to have a high degree of associated gene loss and acquisition through LGT [45]. Therefore we conclude that the reductive pathway is the ancestral route to glycine and serine in the Firmicutes. The Thermotogales and Aquificales are much less diverse than the Firmicutes, comprising almost exclusively hydrothermal vent/spring organisms. Metagenomic evidence suggests the existence of specialized mesophilic Thermotogales [46], [47], but the amino acid composition of reconstructed ancestral states of a large number of gene families supports a highly thermophilic origin for this clade as a whole [42]. Mesophilic Thermotogales have not yet been cultured and our genomic sample includes thermophiles only. As can be seen, all Thermotogales and Aquificales lack two, or even all three, of the genes to synthesize serine from 3-PG, and none has the gene for glyoxylate transaminase. As the alternative, all Thermotogales have a complete reductive pathway, while several Aquificales show the 7-reaction sequence missing only the ATP-dependent formyl-THF synthase (reaction 2). Evidence that formate is, nevertheless, taken up in the reductive pathway by Aquificales comes from the presence of a formate dehydrogenase (reaction 1) that, in obligate autotrophic members of this clade, has been shown not to operate in the oxidative direction [48], [49]. Experiments in [48] followed the protocol of [49] that test only for the oxidative direction and found no activity in obligate autotrophs. Hence, this enzyme likely functions as a reductase rather than a dehydrogenase, despite the gene nomenclature. In the remaining deep bacterial lineages, oxidative and reductive pathways are co-present, although the reductive pathways remain more common. The glyoxylate pathway is common only in cyanobacteria (for reasons explained below), and otherwise only in the Halobacteria, a late-branching mesophilic archaeal clade restricted to hypersaline environments. Therefore all distributional evidence suggests that not only is the (either latent or active) reductive pathway common across the bacterial domain, but also that it represents the ancestral pathway to glycine and serine. This leads to the general conclusion that the ancestral function of THF-based chemistry was completely reductive starting from, until alternative routes to glycine and serine became available and parts of this chemistry could reverse direction. The pattern of pathways for glycine synthesis in archaea is more complex than in bacteria, because pterin diversity is greater (see next section and Fig. 4), and their functions and associated enzymes have been characterized in much less detail [24], [50], [51]. However, pathway distributions, in particular the widespread presence of the glycine cycle, continue to suggest the reductive pathway as the archaeal ancestral form. In bacteria this cycle appears to be a good indicator for the presence of the reductive pathway, with nearly all (213 out 217) strains that have a complete glycine cycle showing either the complete (Red) or alternate (Red (-2) ) forms of the direct reductive pathway. Among non-methanogenic archaea a majority has a complete or nearly complete cycle. Of these, some show a complete reductive pathway, while most lack only the genes specific to bonding at of THF (reactions in Fig. 2). As the syntheses of archaeal (non-THF) pterins and of THF use different enzymes starting from the first commited step from GTP [50], [51], suggesting a deep evolutionary divergence of these molecules, it does not seem completely surprising that enzymes performing the equivalent pterin- interconversions would not show up in homology searches against the THF- interconversion enzymes, even if this pathway is in reality present and active. This underscores the importance of further characterization of archaeal pterin- chemistry (see also Text S1). Among archaea a complete oxidative pathway occurs only in the absence of a glycine cycle (a stricter version of a similar alternation in bacteria), and the majority of these cases are found among methanogens whose -based pathways play fewer biosynthetic roles than their THF homologues [24]. In the following section we combine the distribution signature with an energy analysis of the different pathways, from which it can be seen that the loss of the glycine cycle in methanogens was probably a result of pterin optimization within this derived subgroup of Euryarcheota. Genes for the -methanofuran system that interconverts oxidation states of one-carbon units have been found in both archaeal and bacterial clades, and this observation has been the basis for some hypotheses that they were present in the LUCA [52]–[54]. Several lines of evidence, however, argue against this conclusion. The most direct counterevidence comes from the synthesis and structural variation within the folate family, to which both THF and belong. Both structures are derived from GTP, but their synthetic pathways diverge as shown in the left panel of Fig. 4 [55]. Whereas in the synthesis of THF the final steps are simply the addition of an aminobenzoate and one or more glutamates, followed by the reduction of a double bond, in the case of the aminobenzoate is first combined with a phosphoribosyl pyrophosphate (PRPP) before being incorporated at the homologous step [56]. This is then followed by the addition of a second PRPP, an -ketoglutarate [56], two methyl groups derived from SAM [57] and finally the reduction of the same double bond, leading to the same chirality [58], as in THF. The synthesis of is thus an elaborated version of the synthesis of THF, which has led and its structural variants (shown in the right panel of Fig. 4) [59]–[64] to be termed ‘modified folates’ [65]. The variation within this group is based around two central structural features that likely affect chemistry. Combining aminobenzoate with PRPP eliminates the electron-withdrawing carbonyl group of aminobenzoate (indicated in red in Fig. 4), raising the electron density at in the final pterin structure and lowering the free energies of the bound units in the direct reduction of [24]. The addition of structural methyl groups (indicated in purple in Fig. 4) results in steric hindrance and a more rigid pterin structure for than for THF, in turn lowering the conformational entropies of pterins containing single-bound units [24]. The order of these modification in the synthesis of and the nature of the observed variation within the family of modified folates – elimination of the carbonyl group is universal outside of THF; the number of methyl group varies from zero to two in the same order in which they are added in – suggests a step-wise exploration of folate modification. Further variation outside of these two aspects mostly occurs within a molecular appendage which is far removed from the active site and is thus unlikely to affect local chemistry. Within these general outlines, a wide range of structural variants is observed within this class of molecules (see Fig. 4). Of these, only THF and are observed within the bacteria (the latter only in a few clades), while all are observed within the archaea. The general premise that diversity remains highest in the domain of origin, and the fact that is an apparent end-point in the step-wise modification within this class of molecules suggests already that the exploration in folate modifications occurred within the archaea, and that genes for were subsequently laterally transfered to bacteria. This hypothesis is further strengthened if we consider the ecological roles of the metabolic machinery, which are very different in the two domains. Methanogenic archaea use this machinery in both the oxidative and reductive direction, with autotrophs among this group using it exclusively in the fully reductive direction starting from, which as we previously showed was most likely the ancestral function of folate chemistry. In bacteria, by contrast, the methanopterin system is used exclusively in the oxidative direction, either as part of a methylotrophic metabolism or possibly to serve in formaldehyde detoxification. Even methylotrophic bacteria that have been classified as autotrophs first oxidize reduced forms of carbon before feeding it into traditionally recognized -fixing pathways such as the Calvin-Benson Cycle [66]. To summarize: not only do we find a much greater structural diversity within the modified folates (including a range of apparent structural intermediates between THF and) present in the archaeal domain, we also find the likely ancestral function of this whole class of molecules preserved in exclusively within this domain. Variations in the synthesis, structures and ecology of the folate family are all thus explained consistently by an emergence within the archaea and subsequent transfer to the bacteria, as they would not be by the reverse transfer from bacteria to archaea, or by the presence of in the LUCA. Finally we briefly discuss phylogenetic studies of the distribution of genes, some of which have been used to argue against the transfer of genes from archaea to bacteria. While several phylogenetic studies [53], [67] led to the assumption that the most probable scenario was a transfer from archaea to bacteria [53], [67]–[69], one study reached a different conclusion [54]. In this work, unrooted trees were built for genes found within the Planctomycetes, proteobacteria, and methanogenic archaea. It was then argued that because all three clades separate in most such trees, transfer from methanogens to bacterial clades (including sequential transfers) could be eliminated, leading to the conclusion that these genes were most likely present in the LUCA [54]. We first note that this study does not address variations in the structure and synthesis of the modified folates or their ecological roles as just discussed. In addition, in any tree with just three clades, the topology automatically renders any two clades monophyletic relative to the third, which means that transfer from methanogens to the ancestor of proteobacteria and Planctomycetes cannot in fact be eliminated on these topologies alone. There are several trees in [54] that show non-methanogenic archaea branching between methanogens and bacterial clades, but these refer to genes for the biosynthesis of, which because modified folates are common across archaea are less useful in distinguishing these scenarios. Finally, transfer from archaea to bacteria explains the pattern of absence of genes in most bacterial clades, which in the case of a presence in the LUCA would require an unexplained mechanism for massive differential loss within the bacteria [68]. Thus we conclude from a wide range of evidence that incorporates phylogenetics and the synthesis, structural diversity, and ecology of the folate family, that emerged within the archaea, and that this particular modified folate (but not others) was subsequently transferred to bacteria, where in a few clades it was adopted in a later derived functional role. If the folate family diversified within the archaea as we have proposed, through step-wise modification to the synthesis of THF, what then drove these modifications? As alluded to in the description of the individual key structural changes, we argue here that it was an energetic optimization of the fixation of through direct reduction. We calculated the biosynthetic cost of glycine and serine in units of ATP and reductant (equivalents) for both the reductive and oxidative pathways, shown in Table 2. Since the cost of the oxidative pathway rises with general cost of fixing carbon for different primary fixation pathways, while the reductive pathway has a fixed cost, Table 2 only shows cost for the oxidative pathway as part of the two most energy efficient autotrophic strategies, the rTCA cycle and the reductive acetyl-CoA pathways [1], [70]. It can be seen that the only autotrophic context in which the oxidative pathway is more efficient than the reductive pathway is methanogenesis. The combined effect of higher electron density on of attained by fusing aminobenzoate with PRPP before incorporation, the lowering of the entropy of single-bound -folate structures through the addition of structural methyl groups – both of which may have on average conveyed an energetic advantage in isolation – and the usage of the methanofuran thus resulted in a reduction of the number of ATP' s required for the uptake of through this pathway. The higher electron density at of that facilitates the uptake of without ATP hydrolysis increased the stability of the C- bond, but thereby sacrificed the capacity to donate the methylene group and synthesize glycine directly, explaining the absence of the glycine cycle in methanogens. (We have noted that, in contrast, most non-methanogens show all genes for this cycle.) In addition, the lowering of the entropies of single-bound - molecules may contribute to the smaller free energy differences between these and closed ring (methenyl- and methylene-) - structures, allowing easier transitions between oxidation states and thus facilitating reversal of pterin chemistry from the reductive to the oxidative direction [24]. This robust change in functionality would also explain why only among all the modified folates was transfered from archaea to bacteria. Finally, as we shall see below, energy optimization through elimination of redundant ATP-consuming reactions was likely a more general selective force that also explains the initial emergence of WL phenotypes. We interpret the pathway distributions as showing, generally, that energy optimization is a secondary selection force to oxygen toxicity, when the latter is present. For example, once acetyl-CoA synthase was eliminated in the direct reduction of upon oxygen exposure in deep-branching autotrophs, modification within the folate family as just described would no longer have been advantageous as the oxidative synthesis of glycine and serine could no longer be connected to folate-mediated direct reduction of, eliminating this route to lowering ATP cost. Another example is the cyanobacteria, which use the glyoxylate pathways to synthesize glycine and serine even though energy accounting suggests that a hybrid strategy involving the reductive pathway would be more economical than using glyoxylate emerging out of the CBB cycle, which has one of the highest ATP costs of carbon-fixation [1], [70]. However, it is known that 2-phosphoglycolate, the precursor to glyoxylate in these organisms, is formed when replaces in the CBB Rubisco reaction, and subsequently inhibits the cycle [30]. In this case the adoption of the glyoxylate pathway thus furnishes a mechanism to remove -induced growth inhibition and to recycle 2-phosphoglycolate through anabolism. We predict that the reductive pathway retains a role in cyanobacteria living under anoxic [71] or high- [30] conditions, where the CBB cycle does not produce (significant) 2-phosphoglycolate, or in mutants where the glyoxylate pathways have been deactivated [30]. Having established that direct reduction of is the most likely ancestral metabolic pathway to glycine and serine, as well as the ancestral function of THF- chemistry in general, we next consider a full evolutionary reconstruction of carbon-fixation. We first note, as briefly mentioned in the introduction, that with these new results the deep-branching rTCA and Wood-Ljungdahl carbon-fixation phenotypes show a high degree of similarity. For deep-branching rTCA phenotypes we find parallel use of an incomplete WL pathway that lacks only the final synthesis of acetyl-CoA, which as noted previously is catalyzed by one of the most oxygen sensitive enzymes in the biosphere. Anaerobic WL phenotypes, in contrast, do possess this complete sequence to acetyl-CoA, and they then complete the set of universal anabolic precursors through a variety of incomplete rTCA cycles. Closer inspection of these incomplete cycles shows that while connection to the universal anabolic precursors is always maintained, in all cases the usage of one of the redundant ATP-dependent steps involved in thioester bond formation is eliminated: the synthesis of citryl-CoA from citrate (which is subsequently cleaved to acetyl-CoA and oxaloacetate), or the synthesis of succinyl-CoA from succinate [72], [73]. The incomplete WL pathway as part of the hybridized deep-branching rTCA cycle is thus associated with oxygen sensitivity, while the incomplete rTCA cycles as part of deep-branching WL are associated with ATP economy. The template that underlies both is a fully connected rTCA-WL network. The modular role and ancestral status of direct reduction thus anchors the most fundamental division in carbon fixation strategies, between the autocatalytic rTCA loop and the non-autocatalytic reductive acetyl-CoA pathways, and suggests that the linked rTCA-WL network preceded both. From the linked rTCA-WL phenotype, oxygen toxicity to the acetyl-CoA synthase causes divergence of Aquifex-like rTCA phenotypes, while energy optimization through elimination of redundant ATP-dependent citryl-CoA or succinyl-CoA synthetases leads to the divergence of WL phenotypes. Other fixation pathways may be derived from these by loss or addition of only a few key reactions, linked again to ATP, oxygen sensitivity, or in some cases alkalinity or redox states. Within our basic assumption that the biosphere has always been autotrophic from, we may therefore ask: 1) whether all carbon-fixation phenotypes may be connected while maintaining uninterrupted access to all initial anabolic branching points from; 2) whether these connections can be made with no repeated innovations; 3) which networks that are unobserved in extant biology must be posited to connect all networks that are observed. These questions are answered by arranging the known phenotypes, and the new hybrid forms revealed here, as nodes on a tree according to parsimony, as outlined in Fig. 1D, where links represent evolutionary innovations. A maximum-parsimony tree connecting all known autotrophic pathways, in which all nodes represent viable carbon-fixation phenotypes, is shown in Fig. 5. We note first the position of the linked rTCA-WL network between rTCA and WL phenotypes. Fully connecting the deep tree of life while maintaining autotrophy requires connecting rTCA and WL phenotypes. No single change can connect them while maintaining uninterrupted access to all essential branching points from, and the only sequence of two changes that does maintain autotrophy passes through the linked rTCA-WL network. The ancestral state required to connect the network is thus consistent with the alternative selective filters from oxygen and ATP use discussed above. From this inserted node, we may then connect all other observed carbon-fixation phenotypes, with no further insertions needed, through sequences of single changes for which we can invoke plausible ecological causes. If we assume that once the complex, ATP-dependent, citryl-CoA or succinyl-CoA synthetase enzymes were lost they could not easily be recovered, we can then explain the absence of the rTCA cycle in the archaea, together with the curious, chimeric Dicarboxylic/4-Hydroxybutyrate (DC-4HB) and 3-Hydroxypropionate/4-Hydroxybutyrate (3HP-4HB) cycles as the only autocatalytic loop pathways in this domain [1], [2]. Once autocatalytic rTCA cycling has been lost (as in the emergence of WL phenotypes), organisms can only survive subsequent loss of the acetyl-CoA synthase due to oxygen exposure if they either can survive as heterotrophs, or else possess a latent cycle that can be activated. Indeed, in the Euryarcheota all non-methanogens are heterotrophs lacking acetyl-CoA synthase. In the Crenarcheota, in turn, the DC-4HB cycle shares the first arc of the rTCA cycle (acetyl-CoAsuccinyl-CoA) but not the second (succinyl-CoAacetyl-CoA), and it has a significantly higher ATP cost per carbon fixed than does rTCA [1], [70]. Therefore, an ancestral status for DC-4HB cannot be motivated as requiring fewer chemical innovations, nor could this pathway have competed with rTCA energetically. If, however, a subgroup of WL organisms that possessed all the steps of the second 4HB arc for other purposes were exposed to oxygen, the activation of the cycle and the transition to this carbon-fixation phenotype would be enforced to maintain autotrophy from. The second arc of DC-4HB originates from acetate along the isoprene synthesis pathway essential to archaeal lipids, and from succinate along a fermentative pathway using the key 4-hydroxybutyryl-CoA dehydratase [74]–[76]. Both of these sequences have been found in the Clostridial clade of the Firmicutes [75], [76], which also contains many WL organisms, supporting this scenario. The most elaborate case is that of Clostridium kluyveri, which contains pathway segments to reach all DC-4HB intermediates, though these are not used as part of an autocatalytic cycle, instead supporting a fermentative heterotrophic metabolism [76]. In a similar fashion, the 3HP-4HB phenotype then branches from the DC-4HB phenotype through replacement of the first rTCA arc (acetyl-CoAsuccinyl-CoA), which was still preserved in DC-4HB. The central difference, from the perspective of carbon-fixation, between the rTCA arc and the 3HP arc that replaces it is that the latter requires only uptake of bicarbonate whereas the former takes in as well. The transition to 3HP-4HB thus appears to be driven by changes in the alkalinity in the environment, with an equilibrium shift from dissolved toward bicarbonate favoring the emergence of the 3HP pathway. The subsitution of the second rTCA arc by the 4HB arc within the crenarcheota had already removed all reactions involving in the second half of the pathway, thus imposing no further barriers to the transition from DC-4HB to 3HP-4HB in response to alkalinity. The archaeal case contrasts with the emergence of the 3HP pathway in bacteria, as part of the 3-HP bi-cycle, which similarly uses only bicarbonate uptake and has thus also been recognized as an adaptation to alkaline conditions. In the latter case, both (rTCA) arcs in the initial phenotype utilized uptake reaction, and adaptations to avoid these resulted in a more complex pathway structure. In addition to the appearance of the 3HP arc this involves the reversal of part of the first rTCA arc and the appearance of an alternate completion of the 3HP arc through combining with a glyoxylate that is one of the products of the reversed (first) rTCA arc. We note, however, that these complex pathways result from the relatively simple and uniform chemistry of aldol reactions, and that part of the 3HP-bicycle overlaps with the glyoxylate shunt, which is a similar bicycle. The emergence of the 3HP pathway in both archaea and bacteria as just discussed is one of the two main parsimony violations in Fig. 5. The 3HP pathway uses two key biotin-dependent carboxylation reactions (to malonyl-CoA and methylmalonyl-CoA), which suggests a bacterial origin as this enzyme class features prominently in the biosynthesis of the fatty acids that make up bacterial membranes, in contrast to archaeal lipids, which are based on isoprenoid backbones [77]. However, a comparison of enzymes for thioesterification of propionate to propionyl-CoA has been interpreted as implying convergent evolution for these essential steps in the pathway [78]. Since the appearance of the 3HP pathway is recognized as an adaptation to alkaline conditions in both bacteria and archaea, either convergence or LGT is plausible due to restricted common environments. The other main parsimony violation in Fig. 5 is the parallel emergence of the oxidative route to serine. Oxidative synthesis of serine involves three of the most ubiquitous reaction types/enzymes within metabolism, the dehydrogenation of an alcohol to a carbonyl (reaction 11), a transamination of that carbonyl to an amine group (reaction 10), and a dephosphorylation (reaction 9). The recurrent emergence of this pathway through either duplication or promiscuity of common enzymes is therefore not likely to be a low-probability event, and thus not a significant violation of the assumptions behind a maximum-parsimony reconstruction. The tree of Fig. 5 separates the Firmicutes together with all archaea as the branch from the linked rTCA-WL network originally driven by energy optimization in the absence of oxygen. All other bacterial lineages separate by an early loss of acetyl-CoA synthase. This basic division – with Firmicutes separate from all other bacterial clades – is supported by phylogenetic studies that focus on directed insertion-deletions in paralogous informational and operational genes [79], and on universal orthologous genes [41]. The fact that all archaea separate as a monophyletic branch from a pool of ancestral bacteria may be associated with the more tree-like archaeal phylogeny, compared to the more web-like phylogeny of bacteria, which suggests higher rates of LGT in the latter [21]. The co-evolution of the archaeal DNA-replication system together with less permeable isoprenoid membranes [77], [80], resulting in lower LGT in this domain, are possible mechanisms for isolation. The carbon-fixation phenotypes at all internal nodes of our reconstructed phylometabolic tree are found preserved in extant organisms, except for the linked rTCA-WL network at the root node. While many of these derived nodes incorporate parallel fixation pathways, in all such cases the parallel pathways are disjoint and supply carbon to distinct portions of biomass. The linked rTCA-WL network is qualitatively different, having two separate input channels that supply acetyl-CoA, resulting in a kind of metabolic redundancy. We must ask, does our choice to insert an unobserved node, in order to connect the tree while preserving autotrophy, justify the reconstruction of an ancestor with a form of redundancy not found in any of its descendants, and if so, what does this reconstruction tell us about the evolution of earliest life? We address both questions by considering the relation of network topology and redundancy to self-amplification. The capacity of life for exponential growth, resulting from proportional self-amplification by metabolic and other networks with “autocatalytic” topology [81], [82], is essential to self-repair and robustness in the face of perturbations. At the small-molecule substrate level the rTCA cycle is network-autocatalytic, and thus capable of exponential growth above a threshold rate of production of acetyl-CoA, but also fragile against collapse if production falls below this threshold [83]. The threshold fragility, which may have been a more serious problem in an era of primitive catalysts or regulation, is removed while preserving autocatalysis if acetyl-CoA is independently supplied by WL. Conversely, WL in modern organisms may be considered network autocatalytic at the level of whole-cell physiology including enzymes and cofactors, but it relies on the integrity of synthetic pathways for these molecules which are more complex even than the small-molecule substrate. rTCA, which provides an independent channel for carbon fixation and synthesis of precursors, would thus reciprocally support the robustness of WL in the earliest era of organic (versus proposed earlier mineral [69], [84]) cofactors. The difference in network topology between our reconstructed root, and its derived descendants, would then reflect a shift in the character of natural selection acting on the earliest versus later cells. In early cells (or pre-cells) with imprecise or unreliable enzyme function, consequent leaky pathways and fluctuating cofactor concentrations, or unreliable regulation of anabolism, robustness inherent in the topology of the substrate network would have carried a selective advantage. (Note that anabolism is a form of “parasitic side-reaction” from rTCA cycling, and that an inadequately regulated anabolism can as readily carry carbon fixation below the threshold for self-maintenance, as external factors can). As each of these cellular-level mechanisms was refined, redundant self-amplification of small-molecule substrates would have become unnecessary, and conservation of ATP or adaptation to oxidizing environments would have become more advantageous. These observations also motivate our reference to the linked rTCA-WL network as a “root” phenotype in a maximum-parsimony tree that, by its construction method, is otherwise unrooted. All other phenotypes in the tree may be explained as evolutionary diversifications away from the linked rTCA-WL network, which is both a template for these divergences, and because of its redundancy, a more plausible candidate for a primitive ancestral form than any modern phenotype. The good overlap of our carbon-fixation tree with the later branchings of bacteria and archaea indicates the fundamental role of autotrophy in shaping the deep evolution of the biosphere, and suggests that the later nodes describe not only cellular life, but emerging well-resolved clades. The less-clear separation of bacteria and archaea near the root in Fig. 5, the correspondence of these branches to the reticulated domain of gene phylogenies [12], [14], [19], the need and character of the inserted root node, and the flexible interpretation of our carbon-fixation phenotypes as species or consortia, leave open the possibility that the earliest branches were stages of chemical evolution that preceded modern life [3]–[6]. Our phylometabolic reconstruction, and the surprising tree that it yields, has focused on the particular function of carbon fixation. However, the same methods could be extended to a fuller description of core metabolism, and from the phenotypes we have already shown, we may anticipate certain specific complications that will be introduced with a wider reconstruction. These reflect the changing nature of ecological interactions with increasing oxygenation, and they give added insight into the interpretation of the high degree of parsimony we have shown for innovations in carbon-fixation, most of which took place in anaerobic or micro-aerobic conditions. For any phylogenetic reconstruction, it is important to remember that the nodes and links on a tree are summary statistics for relatedness of samples taken from a population process that may have been very complex. A high degree of parsimony in a tree does not indicate the absence of complex structure in populations, constraints on innovation, or ecological interactions; at most it indicates a lack of specific evidence that innovation required anything more than rare variations and environmental selection in vertically transmitted phenotypes. The cases in which violations of parsimony are inescapable provide evidence that multiple levels of organization were causally essential to the course of innovation, whether these were latent constraints causing some innovations to recur (leading to evolutionary convergence), or ecological interactions leading to gene or pathway transfer. The structure of parsimony violation then indicates what forms of multilevel interaction must be deduced to explain evolutionary causation. Two illustrative cases of parsimony violation that we have elaborated are the transfer (or combined transfer/convergence) of the 3HP pathway, and the gradual elaboration of pterin cofactors followed (as we argue) by the late transfer specifically of from archaea to bacteria. The parallel innovation of the 3HP pathway in Chloroflexi and in the Chrenarchaeota entails duplication of an entire (and rather elaborate) pathway segment, and not merely a single key gene. It is favored in specialized environments which we would expect to create long-term association between inhabiting species, these environments contain a stressor (alkalinity) which we expect to induce gene transfer, and relative to the very ancient divergences in our tree, the innovations of 3HP occur late, at an era when we expect organism lineages to have evolved refractoriness to many forms of gene transfer [85]–[87], along with more integrated control of chromosomes. Similar long-term associations in anaerobic environments such as coastal muds are believed to have led to the (otherwise uncommon) aggregate transfer of a large complement of operational genes from -proteobacteria to Aquificales [88]. The transfer of reflects an even more fundamental link between oxygen and ecosystem structure. As it occurs in methanotrophs and methylotrophs, is used for oxidation of methane and other reduced species, the most extensive form of heterotrophy of reduced carbon, which relies on environmental oxidants to link methane producers and consumers. This function, driven by trophic interactions, is layered over the foundation of reduction on folates which we reconstruct as ancestral in the clades harboring methanotrophs and methylotrophs. The complexity of methylotrophy [89] anticipates the enormous diversification of catabolic pathways that becomes available with oxidation of reduced biotic carbon [17], but which depends on details of ecological provision and accessibility of carbon sources. Finally, it is interesting that many of these complexities would fall cladistically near the tips of the tree in Fig. 5, suggesting that innovation in carbon fixation ceased, to be replaced by innovations in carbon exchange through ecosystems, on a horizon coinciding with the rise of oxygen. This pattern brings into sharper relief the striking lack of multilevel dynamics that would distinguish organism and ecosystem roles in the reconstruction we have shown. In this paper we have demonstrated a novel method that integrates constraints from FBA and from phylogenetics. Individually, metabolic and phylogenetic reconstructions are both subject to ambiguities, especially for deep-branching lineages with significant gene divergence from well-characterized model organisms, and extensive LGT near the root [12], [14], [19]. Integrating the two can resolve ambiguities inherent in each, as well as providing new ways to tailor questions to specific features of early evolution. Here, as a proof of principle, we have presented a coarse-grained reconstruction of the input channels of carbon into the biosphere only up to the initial anabolic branching points. We can relate all modern forms of carbon-fixation to a single ancestral form, and we find that innovations in carbon-fixation were the foundation for most major early divergences in the tree of life. We have also proposed specific causes for the major divergences, and argued for a very small role for lateral gene transfer or convergent evolution. The absence of any phylogenetic signature of important ecological co-evolution for this function, combined with selective forces that originate in energetics or inorganic chemistry, offers additional specific links between genome evolution and the geological record [15]. The consistency of the reconstruction demonstrates, with specific examples, how the fine structure of organic chemistry and geochemistry can enter as detailed constraints on long-range evolutionary dynamics. The hypotheses required by the reconstruction further imply several sets of specific experimental predictions, which are outlined in the Text S1. By limiting our reconstruction to the networks of core carbon fixation – and by virtue of the modular interface of this function with later stages of biosynthesis – we have selected a problem for which many distinctions between organism and ecosystem (any that do not leave phylogenetic signatures) may be passed over, which falls prior to most ambiguities from reticulated gene phylogenies, and which leads to a high-parsimony tree that can be identified manually. The extension of these methods to larger networks with more ambiguity, and to historical reconstructions for which trees provide less adequate representations of complex population processes, will require formal probability models, which may be implemented in joint-maximum-likelihood or Bayesian MCMC phylometabolic algorithms. The current work suggests ways to increase the number of dimensions of “meaning” that can be used to define such probabilistic models, by placing genes in both physiological and ecological context. In this way we may reconstruct trees of life that reflect more of the multi-level character of evolution and development than is suggested by gene counts, and that capture constraints which may have acted continuously since its emergence. For a description of the basic principles of phylometabolic analysis, see the introduction and Fig. 1. This method rests in part on metabolic flux-balance analysis (FBA), which has been described in detail elsewhere [90], [91]. Briefly, the core of FBA consists of the following three equations: (1) (2) (3) where is the concentration of metabolite n, is the stoichiometry of metabolite n in reaction m, and is the flux of that reaction. The nm matrix S and the m-dimensional vector are the the stoichiometries and fluxes of the total metabolic network. Under steady state growth, the principle of mass balancing can then be expressed as equation (2). Finally, Z is an objective function that is selected for optimization, and consists of a linear combination of individual individual fluxes weighted by proportionality constants. Z is often chosen to be the biomass composition of an organism, and maximizing its output thus maximizes growth. The full metabolic network of an organism is reconstructed from its annotated genome. While optimization and matching to laboratory growth data of specific organisms requires detailed analysis of the constraints of individual reactions and careful computational modeling, the initial reconstruction of the network requires only that the network represents a viable metabolism capable of supporting growth (Z0). Our analysis is restricted to the access (from) of the universal anabolic precursors that represent the initial branching points in anabolism, requiring as in the initial reconstruction of an organism only that they are produced (Z0). These anabolic branching points are acetyl-CoA, pyruvate, oxaloacetate, succinyl-CoA and -ketoglutarate, which is a very small number of intermediates to consider, and because studies of carbon-fixation phenotypes have already shown that they are in all cases reached through reuse of partial TCA sequences in fact no computational analysis is required. As explained in the main text, the only intermediates accessible through pathways that circumvent all the universal anabolic precursors are glycine and serine, so our analysis of metabolic genes focusses on the pathways to these intermediates. Three main pathways to glycine and serine are known (see Fig. 2 and the Results and Discussions section), and we analyze the gene profiles for a large number of species for the presence of the necessary enzymes for each of these three pathways. The gene profiles for all strains used in this study were obtained from the Uniprot database [35]. For all complete annotated genomes within each clade, gene profiles in the three pathways were obtained by searching for enzyme classes (through EC numbers) and names. The searches were done in a redundant manner to ensure that all naming variations in the annotation of a given gene/enzyme were included in the final result. While in general we used the annotations as given in the database, in a few cases a (Uniprot built-in) BLASTp search was done to confirm the absence or presence of an enzyme if the profile of a particular strain seemed to contradict the pattern of the clade overall. These are highlighted in Table 1 and Table S1 in Text S1. If within a particular strain all enzymes of a given pathway showed up in our database search the pathway (either active or latent) was counted as present in that strain. The EC numbers and one common variation of the full names of the numbered reactions as shown in the various Figures, and throughout the text are as follows: 1 = formate dehydrogenase (EC: 1. 2. 1. 2); 2 = -formyl-THF synthase (EC: 6. 3. 4. 3); 2A = -formyl-THF cycloligase (EC: 6. 3. 3. 2); 3 = methenyl-THF cyclohydrolase (EC: 3. 5. 4. 9); 4 = methylene-THF dehydrogenase (EC: 1. 5. 1. 5); 5 = dihydrolipoamide dehydrogenase (EC: 1. 8. 1. 4); 6 = aminomethyltransferase (EC: 2. 1. 2. 10); 7 = glycine dehydrogenase (EC: 1. 4. 4. 2); 8 = serine hydroxymethyltransferase (EC: 2. 1. 2. 1); 9 = phosposerine phosphotase (EC: 3. 1. 3. 3); 10 = phosphoserine aminotransferase (EC: 2. 6. 1. 52); 11 = 3-phospho-glycerate dehydrogenase (EC: 1. 1. 1. 95); 12 = alanine-glyoxylate transaminase (EC: 2. 6. 1. 44).

The existence of the biosphere today depends on its capacity to fix inorganic into living matter. A wide range of evidence also suggests that the earliest life forms on Earth likewise derived their carbon from. From these two observations one can assume that the global biological carbon cycle has always been based on, and we show here that this assumption can be used as a powerful constraint to help organize and explain the deep evolution of life on Earth. Using a novel method that fully integrates aspects of metabolic and phylogenetic analysis, we are able to reconstruct the complete early evolutionary history of biological carbon-fixation, relating all ways in which life today performs this function to a single ancestral form. The diversification in carbon-fixation appears to underpin most of the deepest branches in the tree of life, and this early metabolic diversification – reaching back to the first cells – appears to have been driven not by the contingencies of history, but by direct links to the physical-chemical environment. The ancestral carbon-fixation pathway that we identify is different from any modern form, but better suited to the capabilities of the earliest primitive cells.

Abstract Introduction Results/Discussion Methods

systems biology biochemistry microbial metabolism evolutionary biology microbial evolution origin of life biology computational biology metabolic networks metabolism microbiology genetics and genomics

2012

The Emergence and Early Evolution of Biological Carbon-Fixation

Leprosy Type 1 (T1R) reactions are immune-mediated events leading to nerve damage and preventable disability affecting hands, feet and eyes. Type 1 Reactions are treated with oral corticosteroids. There is little evidence on alternative treatments for patients who do not respond to steroids or experience steroid adverse effects. We report the results of a randomized controlled trial testing the efficacy and adverse effect profile of ciclosporin and prednisolone (CnP) in comparison to prednisolone only (P) in patients with new T1R in Ethiopia. Ciclosporin is a potent immunosuppressant. Outcomes were measured using a clinical severity score, recurrence rate, adverse events and quality of life. Seventy three patients with new T1R were randomized to receive CnP or P for 20 weeks. Recovery rates in skin signs was similar in both groups (91% vs 88%). Improvements in nerve function both, new and old, sensory (66% vs 49%) and motor (75% vs 74%) loss were higher (but not significantly so) in the patients on CnP. Recurrences rates of T1R (85%) were high in both groups, and recurrences occurred significantly earlier (8 weeks) in patients CnP, who needed 10% more additional prednisolone. Serious major and minor adverse events rates were similar in patients in the two treatment arms of the study. Both groups had a significant improvement in their quality of life after the study, measured by the SF-36. This is the first double-blind RCT assessing ciclosporin, in the management of T1R in Africa. Ciclosporin could be a safe alternative second-line drug for patients with T1R who are not improving with prednisolone or are experiencing adverse events related to prednisolone. This study illustrates the difficulty in switching off leprosy inflammation. Better treatment agents for leprosy patients with reactions and nerve damage are needed. Leprosy is a chronic granulomatous infection principally affecting the skin and peripheral nerves caused by the obligate intracellular organism Mycobacterium Leprae [1]. In 2014, the WHO reported 213 899 new cases globally [2]. Multi-drug therapy (MDT) cures the infection by Mycobacterium leprae. Although the bacteria may be eliminated, the damage done to nerves by the bacteria and by consequent immunological reactions leads to very visible and stigmatizing disabilities and deformities. Type 1 reactions (T1R) affect up to 30% of patients with borderline leprosy [3]. Although T1Rs can occur at any time, the frequency is higher in the first six months of MDT treatment [4]. T1R manifest clinically with erythema and oedema of skin lesions and tender peripheral nerves with loss of nerve function. Skin lesions become acutely inflamed and oedematous. Inflammation is usually in pre-existing lesions, but not all the lesions may be involved. Oedema of the hands, feet and face can also be a feature of a reaction but systemic symptoms are unusual. Nerves can become swollen, painful and tender. Acute neuritis may also occur without evidence of skin inflammation. The inflammatory process in leprosy reactions leads to nerve function impairment (NFI) which if not treated rapidly leads to permanent loss of nerve function causing peripheral sensory and motor neuropathy. Recurrent T1Rs can lead to further nerve damage [5]. Progressive NFI can also occur in the absence of a reactional state, so the history of timing of symptoms aids to differentiate from NFI due to a reaction. T1Rs are the result of spontaneous enhancement of cellular immunity and delayed hypersensitivity reactions to M. leprae antigens presented by macrophages and dendritic cells in the skin and by Schwann cells on nerves [6,7]. Immuno-suppression is required to control the symptoms and signs of T1R, but this management remains challenging. Oral prednisolone, the drug of choice, has frequent side effects and approximately 40% of individuals with T1R do not show clinical improvement [8,9]. There is a lack of evidence for efficacious and safe second line treatments for T1R. Ciclosporin is a potent immuno-suppressant that has been widely and successfully used as a treatment for psoriasis, Behcet’s disease, rheumatoid arthritis, inflammatory bowel disease and transplantation. Given that ciclosporin selectively inhibits the activation of CD4 T cells and the expression of cytokines such as IL-2 and TNF-α [10], it was thought to be useful in the treatment of T1R. Three case studies have been published [11,12], showing good response to ciclosporin and delayed recurrence of T1R. An uncontrolled pilot study was carried out assessing the efficacy of ciclosporin in severe T1R in Ethiopian and Nepali patients [13]. In the Ethiopian part of the study, performed in ALERT Hospital, Addis Ababa, ciclosporin was given to 33 patients with T1R for three months in a dose range of 5–7. 5mg/kg/day. This led to improvements in skin lesions in 85% of patients and 45% of patients had improvement in nerve pain and tenderness. Sensory nerve impairment improved in 45% of Ethiopian patients and motor function impairment in 53% of patients. Almost 88% of Ethiopian patients needed the higher dose of ciclosporin to show improvement partly because of the severity of the reaction. The study showed that in those patients treated with high-dose ciclosporin, 53% of patients with sensory impairment and 60% with motor impairment improved. A few Ethiopian patients with NFI of greater than 6 months duration, responded to ciclosporin. This was an encouraging result as in many leprosy endemic countries patients present late with chronic NFI. Almost 70% percent of Ethiopian patients developed new signs of reaction after stopping treatment, suggesting that they would benefit from a treatment period longer than three months. In the Nepali study, ten patients treated with ciclosporin were compared to a similar group of patients treated with prednisolone. Improvement in skin lesion was at 87. 5% in the ciclosporin group compared to 74% in the prednisolone group. Similarly the ciclosporin group showed 83% improvement in sensory testing compared to 22% in the prednisolone group. Few ciclosporin side effects were seen in the two clinical trials conducted in T1R. Of the 33 Ethiopian patients, three developed hypertension; of the ten Nepali patients one developed jaundice (possibly dapsone related), two developed raised serum creatinine levels and two other patients developed mild side effects (loss of appetite and indigestion controlled with antacids). The results of the above studies were encouraging as it appeared that ciclosporin monotherapy may be an effective alternative treatment in prednisolone-resistant or prednisolone-dependent cases of T1R. The study recommended using higher doses of ciclosporin (7. 5mg/kg/day) in future studies, longer periods of treatment, as well as tapering the drug slowly or adding low dose prednisolone to prevent relapse. We tested our hypothesis that ciclosporin would be as effective as prednisolone in the treatment of patients with leprosy reactions and nerve function impairment and that patients treated with ciclosporin would have fewer side effects than patients treated with prednisolone. A randomized controlled trial comparing ciclosporin and prednisolone in the treatment of leprosy T1R was designed and conducted. A double-blind controlled trial was conducted randomizing patients with new and recent onset T1R to treatment with either ciclosporin and prednisolone or prednisolone alone. Patients were recruited at the ALERT hospital leprosy clinic, in Addis Ababa, Ethiopia. Type 1 Reaction (T1R) was diagnosed when a patient with leprosy had erythema and oedema in skin lesions and/or neuritis. A patient could have skin reaction only, a nerve reaction only or a skin and nerve reaction. Neuritis was diagnosed when a leprosy patient had any of the following on history or examination: Nerve function impairment (NFI) was defined as clinically detectable impairment of sensory or motor nerve function using the definitions below [14]. New NFI was defined as less than six months duration of reduction in sensory or motor function on history or examination. Motor loss was defined by a decrease in voluntary muscle testing (VMT) score, by 1 point or more from the normal score of 5, using the modified MRC scale. Sensory loss was defined by a decrease in sensation as measured by Semmes Weinstein monofilament testing. In the hands, this was defined as not being able to perceive the 0. 2gm monofilament at 2 points out of 3 in each nerve of the hand. In the feet, this was defined as not being able to perceive the 2gm monofilament at 3 out of 4 sites of the foot. Silent neuropathy (SN): A patient had silent neuropathy when he/she had sensory and/or motor impairment of recent onset (less than six months duration) in an area innervated by one or more nerve without signs of a reaction (RR or ENL) or nerve pain with or without tenderness. T1R recurrence or flare-up was defined as an increase in skin severity score to 4 or more out of 9 AND/OR an increase in NFI defined as worsening of VMT by one point in two or more muscles, or by 2 points in one muscle and/or worsening of ST: decreased sensation in at least two out of 3 points per nerve on the hand and/or 3 or more points on the feet. NB: nerve tenderness was not part of the definition for T1R recurrence. NFI outcomes were defined clinically as (based on Marlowe study [13]: Clinical Severity Score: used to assess T1R severity (21 items; range of 0–63). The maximum score possible for skin (A), sensation (B) and motor function (C) are 9,24 and 30 respectively. Mild T1R is characterised by a score of 4 or less; moderate T1R by a score between 4. 5 and 8. 5 and severe T1R is a score of 9 or more. This Severity Scale for T1R based, on the INFIR clinical severity scoring system, was developed and prospectively validated in Bangladesh and Brazil [15]. It has so far been used in clinical trials on intravenous methylprednisolone [16], on azathioprine [17] and in the on-going TENLEP studies [18]. Participants (aged between 18 and 65 years and weighing more than 30 kg) were recruited from the leprosy clinic in ALERT Hospital, Addis Ababa, Ethiopia. Individuals with newly diagnosed T1R or neuritis were eligible for entry in the trial. The following individuals were excluded: those unwilling to give consent or return for follow up; those with severe active infections such as tuberculosis or severe inter-current disease; HIV positive individuals; pregnant or breastfeeding women. Women of reproductive age not willing to use contraception for the duration of the study were also excluded. The participants were randomly allocated to receive the standard ALERT hospital prednisolone regimen for T1R or the ciclosporin (Cn) arm (Table 1). We theorized that given the slow onset of action of ciclosporin compared to prednisolone and high relapse rate of T1R, the most effective regimen in leprosy reaction would be an initial ciclosporin dose of 7. 5mg/kg/day, divided in two doses, gradually tapered down over a total period of 20 weeks and adding prednisolone cover for the first four weeks of treatment. The TIR patients on the prednisolone arm (P) would get 20 weeks of a gradually reducing course of prednisolone only. Weight adjusted medication cards for each treatment arm were designed for the pharmacist, using a 10 kilogram range in patient weight. A double placebo system was used because of the different formulation of prednisolone (pink tablets) and ciclosporin (brown capsules). Each placebo was identical to its active counterpart and each participant took a combination of brown capsules and pink tablets as an essential way of blinding both patients and study physicians. Prednisolone tablets and prednisolone placebo (PP) tablets were produced by Ethiopian Pharmaceuticals Manufacturing Factory (EPHARM), in Addis Ababa, Ethiopia. Both were analysed for active ingredient by Dr Harparkash Kaur at LSHTM. Ciclosporin capsules (Panimune Bioral) and ciclosporin placebo (PC) capsules were produced by Panacea-Biotec Ltd, Solan, India and were provided with a certificate of analysis. A full history was taken and clinical examination performed. Nerve function was assessed at each visit, by one of three trained physiotherapists. Sensory testing was performed with five Semmes-Weinstein monofilaments at designated test sites on hands and feet. Voluntary muscle power was graded using the modified Medical Research Council scale. The results of the examination findings were recorded and a Clinical Severity Score calculated using the severity scale. Severity of reaction was also recorded as mild, moderate or severe by a second physician’s opinion blinded to the Clinical Severity Score. Laboratory investigations consisted of the following: slit skin smears for bacterial index, full blood count, HIV test, renal function, liver function tests, glucose, erythrocyte sedimentation rate (ESR), urinalysis and a stool specimen examined for ova, cysts and parasites. A skin biopsy was performed for Ridley-Jopling classification. Symptomatic screening for TB was carried out by chest x-ray and sputum samples for acid fast bacilli as necessary. All individuals received three days of albendazole 400mg daily to reduce the risk of hyper-infection with Strongyloides stercoralis at enrolment. Women of reproductive age were tested for pregnancy and contraception was prescribed, usually the oral contraceptive pill and condoms. Assessments were carried out at weeks 2,4, 6,8, 12,16,20,24,28, and 32 from enrolment. Assessment consisted of focussed questions about specific symptoms and adverse effects. The clinical examination including weight and blood pressure was repeated. Blood tests (full blood count, renal function and liver function), and urinalysis were carried out at each visit. Quality of life was assessed with a validated Amharic translation of the SF-36 health-related quality of life assessment tool [19] at recruitment and at week 28. Each patient’s quality of life is graded with two scores: a physical score (PCS) and a mental score (MCS), which in turn are composed of four subscales each. The primary outcome measure was the change in Clinical Severity Score and in clinical nerve function impairment and at week 4,20, and 28 for patients in each treatment arm. Secondary outcomes were: Criteria for using additional prednisolone were defined as sustained deterioration in nerve function nerve pain unresponsive to analgesics for a period of at least two weeks; new erythematous and raised skin patches; deterioration in nerve function which the study doctors believe requires immediate additional prednisolone and ENL flare-up with the appearance of new subcutaneous nodules. As the study was double-blinded, regimen for additional prednisolone depended on the time at which the reaction flare-up occurred. If the reaction recurrence was within the first ten weeks of treatment or there was facial involvement, extra prednisolone was added to make up a total of 40mg (with the pharmacist deciding on the exact additional dose of prednisolone required) and then tapered according to the original regimen. If T1R recurrence was after the first ten weeks of treatment, then prednisolone 20mg was added and tapered down according to the original regimen. The physician could prescribe more additional prednisolone if the reaction was severe. Adverse events were enquired about at each visit using a standardized form with anticipated adverse events attributable to prednisolone and ciclosporin. Any other adverse events reported by the participant or identified by the physicians were also recorded. Major adverse events were defined as any event leading to admission or prolonged admission, study un-blinding or death. Amongst these were included psychosis, severe infection including tuberculosis, peptic ulcer, glaucoma, cataract, diabetes mellitus, severe hypertension and haematological abnormalities. Minor adverse events were defined as moon face, acne, hirsutism, gum hyperplasia, fungal infections, gastric pain requiring antacids or any other minor adverse event not requiring admission to hospital or un-blinding. Patients who experienced blurred vision were referred to the ophthalmologist for ophthalmic review and had their serum glucose checked. Three study physicians (blinded to each other’s decision) reviewed each adverse event and decided whether it was linked to prednisolone or ciclosporin. Adverse events were also graded by severity, using the Common Terminology Criteria for Adverse Events [20] grading system. Eligible individuals were recruited consecutively and randomly assigned in 1: 1 ratio (block size of four), with a computer-generated randomisation list, to one of the two treatment arms. A standard envelope system was used for allocation concealment. The envelopes were prepared by an individual who had no other involvement in the study. The allocation procedure was done by the pharmacist who had no clinic contact and was the only individual aware of the treatment allocation. All study participants, physicians, nurses, ward staff, laboratory staff and the physiotherapists were blinded to the allocation. The allocation code was revealed to the researchers once the study was completed, except in the case of a serious adverse event necessitating un-blinding. The sample size, based on the Hypothesis of Non-Inferiority, was calculated with the study statistician, in consultation with ALERT hospital physicians. Prednisolone is known to show an improvement of 60% in nerve function in new T1R. Given that the true mean cure rates of the treatment agents and the active control are θ1 = θ2 = 60%, the non-inferiority margin was selected to be δ = 0. 25. The sample size was calculated using a power of β = 80% and significance of α = 0. 05, giving us a sample of n = 48 in each arm respectively. The data was entered in Access database and analysed using the Statistical Package for the Social Sciences (SPSS version 20. SPSS Inc. , Chicago, Illinois). An intention to treat analysis (ITT) was used for calculating the effects of treatment on individuals in each group and t tests and ANOVA (analysis of variance) were used as appropriate. The Mann-Whitney U test was used for all statistical tests of continuous variables and Fisher’s exact test was used to compare dichotomous variables. The studies were performed according to the Helsinki Declaration (2008 revision) and approved by the Ethics Committee of the London School of Hygiene and Tropical Medicine (5376), the ALERT and AHRI Ethical Review Committee (AA/ht/248/09), the National Ethics Review Committee of Ethiopia (RDHE/34-90/2009), and the Drug Administration and Control Authority of Ethiopia (02/12/70/926). All staff involved underwent Good Clinical Practice training and an independent Data and Safety Monitoring Board reviewed the study design and the safety and efficacy data. The study is registered with ClinicalTrials. gov: NCT00919815. Written informed consent was obtained in Amharic or if the patient spoke a different Ethiopian language, then the information and consent forms were translated verbally into the appropriate language before signing the consent form. The two groups did not differ significantly in respect of reaction type, or mean number of enlarged and tender nerves per patient (Table 3). There was a significant difference in the duration of NFI between patients recruited to the two groups (Chi Square, p = 0. 039). Twice as many patients in the ciclosporin arm reported isolated new NFI but in the prednisolone arm, there were more patients reporting combination of old and new NFI. Patients in the two treatment arms had similar duration of reported T1R symptoms prior to presenting at the clinic (p = 0. 2). Severity of T1R, assessed both by specialist opinion and by the Clinical Severity Score, was not significantly different between the two groups (Table 4). The 73 patients recruited had a total of 876 peripheral nerves examined. Nerve function impairment of less than 6 months duration (new NFI) was reported for 308 nerves (35%). A further 24% of nerves were reported to have been impaired for longer than 6 months (old NFI). In both old and new NFI, sensory loss was more frequent than motor loss or mixed loss. Of the nerves examined, 72% were enlarged, and 34% of nerves were tender on palpation. A larger proportion of nerves were impaired in the ciclosporin group patients (68% vs. 52%) and this group had significantly higher proportion of purely sensory and mixed sensory/motor types of new NFI (p = 0. 0387) The ulnar nerves were found to be both the most frequently enlarged and tender nerves, followed by the lateral popliteal, radial cutaneous and posterior tibial nerves. Nerve tenderness was present in 300 nerves and was more common in the ciclosporin group (40% vs. 29%). Apart from a higher number of affected sensory nerves in the ciclosporin group, there was no major significant difference between the two groups of patients with newly diagnosed T1R, recruited to the study. The change in group mean Clinical Severity Score over time for patients in each arm of the trial is shown in Fig 2. Changes in the three sub-scores are also shown. Variation in group mean T1R severity scores during the 32 weeks and between the two treatment arms, was assessed by ANOVA. Patients in both treatment arms had large and statistically significant improvement with time in all four scores (p<0. 000). This is consistent with a good clinical response with both treatments. There was no significant difference in all four severity scores between the two treatment arms over the 32 weeks (Score A, p = 0. 241; Score B, p = 0. 664, Score C, p = 0. 749 and Clinical Severity Score, p = 0. 531). In the ANOVA week by week breakdown, patients on the ciclosporin arm showed significantly higher skin score (A), at weeks 6 and 8 (p<0. 000). This was probably due to a greater number of patients in the ciclosporin arm experiencing a flare-up in skin reaction at this time. The difference between the two treatment groups in median improvement of Clinical Severity Scores were compared at week 0,4, 6,20, and 28 as shown in Fig 3. These time periods were deemed important, as at week 4, the prednisolone in the ciclosporin arm is stopped; at week 6 is the steroid free period for those on the ciclosporin arm; at week 20 the intervention period ends, and week 28 represents the end of the study. All four components of the severity scores show a downward trend, suggesting improvement in both groups of patients. The largest and sustained decrease in score occurs in the skin (A). At week 6, the difference in skin score between the two treatment arms is evident with the patients in the ciclosporin arm having a wider range in score despite a similar median score. Throughout the 32 weeks in the study, the median sensory score (B) does not reach the score of 0, which represents intact sensation. Analysis by patient and by nerves were also done to assess the improvement in T1R in patients treated with either ciclosporin or prednisolone. The general outcome for patients (Table 5) was decided by study physician assessment on review of patient notes and taking into account the changes in skin as well as nerves between week 0 and week 20, the end of the intervention period. There is no significant difference in all six clinical outcomes listed in Table 5 between the patients in the two treatment arms. Clinical outcomes in the follow-up period were recorded as those that maintained improvement and those that relapsed at the end of treatment. A larger proportion of patients appears to be maintaining improvement after the end of the intervention period in the ciclosporin arm (67% vs. 39%, p = 0. 044). The change in motor function, between baseline and the end of intervention, in nerves with reported weakness of less than six months duration is not significantly different between the two study arms (p = 0. 085). Fig 4 illustrates that motor function in both treatment arms recovered or improved in a large proportion of nerves (74% in the ciclosporin arm and 68% in the prednisolone arm (one tailed t test: p = 0. 043). 70% of nerves with sensory loss reported as being of less than six months duration in the ciclosporin arm and 56% in the prednisolone arm improved or recovered (Fig 5). Patients who received ciclosporin and prednisolone had better improvement in nerve function impairment than those who received prednisolone only (one tailed t test: p = 0. 038). Patients in both treatment arms had their nerves assessed three months after the end of the intervention and improvement in nerve function was maintained in the majority of patients. Motor function remained stable in 88% (Cn arm) and 76% (P arm), and sensory function in 78% (Cn arm) and 79% (P arm). Nerves reported to have been impaired for longer than six months also showed improvement. Of the nerves with old motor function 37% in the Cn arm and 39% in the P arm recovered or improved. Of the nerves with old loss of sensation, 46% in the Cn arm and 36% in the P arm recovered or improved. This study is the first double-blind RCT assessing ciclosporin, a potent immunosuppressant in the management of T1R. All the patients with T1R treated with ciclosporin and prednisolone or with prednisolone alone improved in all three Clinical Severity Score components. There was no statistically significant differences between the two study arms, suggesting that treatment of T1R with ciclosporin and prednisolone in non-inferior to prednisolone alone. Recurrences of T1R were equally frequent in both treatment arms. These recurrences were treated with additional prednisolone. The patients on the ciclosporin arm of the study received 10% less steroids than those on the prednisolone only arm during the 32 weeks of study. This study has shown that the steroid-sparing effect of ciclosporin is limited. The pilot study done by Marlowe in 2007 suggested that ciclosporin may be as efficient as prednisolone in the treatment of T1R. The study designs are different and no additional prednisolone was given in the Marlowe study for T1R or NFI flare-up; the dose of ciclosporin was increased in such cases. In view of the fewer side effects of ciclosporin compared to prednisolone, ciclosporin could be a useful safe alternative second-line drug for patients with T1R in whom prednisolone is not effective, or is causing adverse events. We would recommend that ciclosporin be prescribed in conjunction with oral prednisolone (unless the latter is contraindicated) for the initial eight weeks. Presently a 20 week course of ciclosporin for a patient in the weight range of 40–49kg, would cost USD 820, compared to a course of prednisolone costing USD 10. This study has highlighted that corticosteroid treatment for T1R and NFI is sub-optimal even when given in a standard reducing course over 20 weeks. The TENLEP multi-centre RCTs are comparing a 32-week vs 20-week course of prednisolone for NFI [18]. This would mean a cumulative dose of prednisolone greater than 5grams compared to 3. 5grams over 20 weeks recommended by Rao [24]. The development of more prolonged treatment protocols would require careful monitoring of adverse events and in particular the long term sequelae of corticosteroid therapy. This study illustrates the difficulty in switching off leprosy inflammation. Better treatment agents for reactions and nerve damage are needed. Clinical studies in T1R should be accompanied by laboratory based research to investigate the mechanisms of inflammation in T1R, identify patients at risk of recurrences and possibly identify a better agent for the treatment of T1R.

Leprosy infection is cured with multi-drug therapy (MDT), but patients may develop immune mediated skin and nerve lesions. These immunological reactions lead to disability and deformity secondary to neuropathy. Prednisolone is the main drug used to treat reactions but is only partially effective and patients have a high rate of side effects. Identifying better agents for treating leprosy reactions is an important clinical goal. We tested the safety and efficacy of ciclosporin, an immunosuppressant used in many inflammatory conditions, in Type 1 reactions (T1R) in leprosy patients in Ethiopia. A double-blind randomized controlled clinical trial comparing the efficacy and adverse event profiles of ciclosporin and prednisolone was conducted in patients presenting with acute T1R. Patients on ciclosporin and prednisolone had similar improvements in clinical outcomes which were measured as skin and nerve function improvement. Both groups had a high rate of T1R recurrence (85%) and the patients on ciclosporin required more additional prednisolone to treat recurrences. We assessed patient quality of life and this was significantly improved with both treatments. This is the first assessment of patient quality of life in a leprosy patients trial. Ciclosporin may be a useful alternative in the treatment of T1R, but that the need for additional steroids decreases its value.

Abstract Introduction Methods Results Discussion

medicine and health sciences physicians medical doctors pathology and laboratory medicine clinical research design nervous system tropical diseases social sciences neuroscience health care research design bacterial diseases health care providers signs and symptoms neglected tropical diseases pharmacology research and analysis methods infectious diseases lesions adverse reactions adverse events people and places professions psychology quality of life diagnostic medicine anatomy nerves leprosy biology and life sciences population groupings sensory perception

2016

A Randomized Controlled Double Blind Trial of Ciclosporin versus Prednisolone in the Management of Leprosy Patients with New Type 1 Reaction, in Ethiopia

The Gram-negative bacterium, Legionella pneumophila, is a protozoan parasite and accidental intracellular pathogen of humans. We propose a model in which cycling through multiple protozoan hosts in the environment holds L. pneumophila in a state of evolutionary stasis as a broad host-range pathogen. Using an experimental evolution approach, we tested this hypothesis by restricting L. pneumophila to growth within mouse macrophages for hundreds of generations. Whole-genome resequencing and high-throughput genotyping identified several parallel adaptive mutations and population dynamics that led to improved replication within macrophages. Based on these results, we provide a detailed view of the population dynamics of an experimentally evolving bacterial population, punctuated by frequent instances of transient clonal interference and selective sweeps. Non-synonymous point mutations in the flagellar regulator, fleN, resulted in increased uptake and broadly increased replication in both macrophages and amoebae. Mutations in multiple steps of the lysine biosynthesis pathway were also independently isolated, resulting in lysine auxotrophy and reduced replication in amoebae. These results demonstrate that under laboratory conditions, host restriction is sufficient to rapidly modify L. pneumophila fitness and host range. We hypothesize that, in the environment, host cycling prevents L. pneumophila host-specialization by maintaining pathways that are deleterious for growth in macrophages and other hosts. L. pneumophila is a Gram-negative intracellular pathogen with a broad host range that extends from unicellular protozoa to alveolar macrophages of the human lung [1]. L. pneumophila is an accidental pathogen: responsible for severe, sporadic disease in humans [2]–[4], but ubiquitous in nature [5]–[7]. Natural and man-made freshwater sources serve as the primary environmental reservoirs of L. pneumophila, with bacterial replication occurring within a diverse set of protozoan species within the aquatic environment. L. pneumophila has been shown to replicate in over 15 species of protozoa [8]–[10], consistent with the bacterium being a generalist in that it shows little evidence of species specificity. After uptake by these natural protozoan hosts, the L. pneumophila type IVB Dot/Icm translocation system translocates a large cadre of proteins across host membranes [11]–[13], remodeling the Legionella-containing vacuole (LCV) into a non-acidified compartment supportive of intracellular replication [14]–[16]. Over 300 bacterial proteins are thought to be substrates of this translocation system [6], [7]. Recent evidence supports a model in which the large repertoire of Dot/Icm translocated substrates is essential to the broad host range of L. pneumophila, with different subsets of these proteins contributing to optimal replication in distinct protozoan hosts [17]. We hypothesize that host cycling in the environment maintains L. pneumophila as a generalist, presumably through purifying selection against mutations that diminish fitness in any of several naturally encountered protozoan hosts. Environmental replication of L. pneumophila within man-made water sources frequently leads to human exposure to the bacteria, through inhalation of contaminated aerosols [18]. L. pneumophila is the causative agent of Legionnaires' disease, a severe, often-fatal pneumonia [19] and Pontiac Fever, a less severe, self-limiting disease [20]. Once inside the human lung, L. pneumophila bacteria are able to replicate within alveolar macrophages in a process that appears broadly similar to that which occurs within amoebae in the natural environment [21]. As in amoebae, the establishment by L. pneumophila of a non-acidified, replicative vacuole in macrophages is critically dependent on components of the Dot/Icm translocation system [22], [23]. During the evolutionary history of L. pneumophila, encounters between these bacteria and mammalian host cells are likely to be quite rare relative to their persistent encounters with protozoan hosts. No environmental mammalian reservoirs of L. pneumophila have been identified and, while L. pneumophila bacteria are capable of causing severe disease within humans, there is no evidence of human-to-human transmission of the bacteria [24]. For instance, when infected individuals returned home from the 1976 Philadelphia outbreak of Legionnaires' disease, none of the 193 surveyed contacts of these individuals developed symptoms of the disease [24]. This suggests that the interaction between L. pneumophila and mammalian host cells is likely an infrequent event of limited duration and may represent an evolutionary dead-end from the perspective of the pathogen. Therefore, during outbreaks of human disease, it is likely that the macrophages encountered by L. pneumophila represent a novel host environment for which each inoculating bacterium is suboptimally adapted. While many similarities exist between the shared intracellular survival strategies employed by L. pneumophila in both amoebal and mammalian host cells, little is known about the differences between these two intracellular environments. The adaptation of L. pneumophila to the intracellular niche is thought to have occurred within a diverse range of protozoan hosts in the natural environment [8]–[10], resulting in a pathogen with a remarkably broad host range. Conservation between protozoa and humans of several L. pneumophila targets, such as key host components of intracellular vesicular trafficking [25], is presumably responsible for the accidental pathogenesis of L. pneumophila in human hosts [9]. The same selective pressures are thought to have selected for a pathogen exhibiting significant genetic redundancy [17], making L. pneumophila recalcitrant to many forward genetic screens aimed at uncovering the function of individual virulence factors [26], [27]. We sought to directly test whether unique selective pressures are placed upon bacteria replicating within mammalian cells by restricting L. pneumophila to growth within mouse macrophages for hundreds of generations. Multiple mutations were identified that improved fitness in macrophages, either in isolation or synergistically with other mutations. Many of these mutations altered bacterial replication in natural amoebal hosts, suggesting fitness trade-offs between natural and accidental hosts. These data represent the first directed adaptation of L. pneumophila host range, a powerful experimental approach to understanding the evolution of host-pathogen interactions within specific host cell-types. L. pneumophila Philadelphia-1, strain LP01 [22], was modified to contain an integrated lux operon [28] and used to inoculate four independent cultures of 1×107 primary A/J bone marrow-derived mouse macrophages at a multiplicity of infection (MOI) of 0. 05 bacteria per host cell (Figure 1A). Under these culture conditions, L. pneumophila was unable to replicate outside of host cells (data not shown), but could replicate intracellularly for approximately 3 days before exhaustion of the macrophage culture. After 2–3 days, any remaining host cells were lysed, bacterial numbers were estimated using luminescence, diluted, and used to inoculate new macrophages, again at a low MOI (estimated 0. 05, see Materials and Methods). Samples of viable bacteria were taken every 10–20 days to allow for the study of intermediate time-points. In this manner, four independent lineages of L. pneumophila were confined to intracellular replication within mouse macrophages for several months. To determine if extended replication within macrophages would improve the fitness of L. pneumophila within this environment, competition experiments were performed [29]–[31] (Figure 1B). Macrophages were incubated for 3 days with both non-luminescent L. pneumophila and clones isolated from each of the four lineages at various points during passage. Bacteria were then plated on solid agar and the ratios of luminescent to non-luminescent colony forming units were determined through imaging. After each of these competitions, the relative frequency of the adapted strains was greater than that of the progenitor. Within each lineage, clonal isolates from later time-points uniformly displayed greater growth advantages than did those from earlier time-points (Figure 1B). We next identified the mutations that arose during passage in macrophages. Clones from each lineage were used to generate libraries for whole-genome sequencing (Materials and Methods). Reference and de novo genome assembly software [32] was used to identify differences between the passaged and progenitor strains (Table S1). These analyses identified several point mutations, small insertions, small deletions, and the precise start and stop points of one large (45. 5 kb) deletion known to exist in the LP01 laboratory strain [33]. Unlike other in vitro bacterial evolution studies [34], we did not observe differences in insertion sequence number or location, though it is possible that the strategies we used were not optimal for detecting alterations in insertion sequence copy numbers (see Materials and Methods). Several independent mutations in the flagellar regulator, fleN [35], [36], components of the lysine biosynthesis pathway, and the Dot/Icm translocated substrate, sdbA [37], were identified. Two clones from independent lineages contained an identical single nucleotide deletion, consistent with a mutational hotspot between pacS, a putative cation transporter, and lphB, a gene adjacent to several Dot/Icm components [23]. Of the 27 mutations identified across all lineages, 5 are predicted to result in frameshifts in open reading frames, 4 are nonsense mutations, 13 are non-synonymous missense mutations, 3 are intergenic, and 2 are predicted to be synonymous. To quantify the frequency of each mutation across the duration of passage, we next used population genotyping (qEGAN analysis [38], Materials and Methods) to compare populations from each intermediate time-point to known mixtures of wild type and mutant genomic template (Figure 2). By providing an unprecedented level of genotypic detail across the entire duration of extended passage, these data uncovered population dynamics consistent with selective sweeps, such as fleN (D75Y) becoming fixed in lineage A by day 118. In addition, there were several instances of clonal interference, in which subpopulations transiently increased in frequency. For example in lineage D, a strain carrying the fleN (V168del), which was the predominant clone at day 45, was ultimately replaced by bacteria harboring the lysC/A metG double mutation. This analysis also uncovered population changes that were not readily explained by the mutations identified in the initial set of clones chosen for sequencing. The proportion of mutants in lineage A having the sdbA (G157opal) lesion increased from days 133 to 175, but reduced dramatically at day 200 (Figure 2). The frequency of sdbA (G157opal) in the population then increased to apparent fixation concurrently with the acquisition of a mutation between lphB and pacS. As this population behavior was consistent with transient clonal interference from a subpopulation of bacteria, we sequenced a sdbA+ strain isolated from day 200 for further analysis, and identified two additional mutations, one in gidB and the other in what is annotated as an intergenic mutation between smpA and cdgS7 (Table S1, Figure 2). We performed TBLAST analysis [39] on the smpA and cdgS7 intergenic region, and identified a conserved intact open reading frame with an alternate GTG start codon corresponding to the ferric uptake regulation (fur) gene [40]. This open reading frame is missing from the published L. pneumophila Philadelphia-1 genome annotation [41], perhaps due to its alternate GTG start codon. As predicted, subsequent population genotyping analysis demonstrated that the frequency of the gidB and fur mutations in lineage A was inversely correlated to the frequency of sdbA from day 175 to 245. Using a similar approach, we identified clpA and dapE mutations in the C lineage from targeted resequencing of a fleN+ isolate at day 180 (Table S1, Figure 2). The population dynamics observed after these iterative rounds of sequencing is consistent with the majority of the significant mutations within each lineage being identified by this approach. Mutations in fleN were identified in three independent lineages, suggesting that modulation of this locus might confer a growth advantage in macrophages. The fleN locus is located in a dense neighborhood of flagellar regulation genes in L. pneumophila, which may restrict the types of mutations that might otherwise be isolated from this locus. We introduced the fleN (D75Y) mutation into wild type bacteria and measured the fitness of this strain in macrophages through competition against the luminescent progenitor strain (Figure 3A). The introduction of this single amino-acid change was enough to recapitulate the competitive advantage of an A lineage clone isolated at 133 days. Other fleN mutations (identified in the C and D lineages) when placed into the wild type background conferred similar growth advantages (Figure 3B). FleN is known to regulate flagellar number in other bacteria [36], [42], but little is known about its function in L. pneumophila. While both the fleN (D75Y) mutant and an in-frame deletion strain were motile and maintained a single polar flagellum by transmission electron microscopy (data not shown), we investigated whether known consequences of altered flagellar function could be observed. L. pneumophila with misregulated flagellar assembly often show lowered association with host cells [35], so we determined if the fleN non-synonymous mutations resulted in modified cell association. Macrophages were inoculated with equal ratios of luminescent and non-luminescent strains for one hour, incubated with gentamicin for one additional hour to kill extracellular bacteria, and then output to input ratios between strains were compared (Figure 3C). In these assays, there was enhanced internalization by macrophages of the fleN (D75Y) strain compared to wild type one hour after infection. To determine whether these fleN non-synonymous point mutations phenocopied a complete loss-of-function mutation, we next constructed an in-frame deletion of fleN in the wild-type LP01 background. In uptake assays, the ΔfleN strain, like fleN (D75Y), showed improved uptake into primary mouse macrophages 1 hour after infection (Figure 3C). We next measured intracellular replication of wild-type LP01, the fleN (D75Y) point mutant, and two independently-derived ΔfleN strain, each carrying an integrated lux operon. Macrophages were challenged with each strain and luminescence was monitored over 72 hours of incubation to determine the growth dynamics of each strain. The fleN (D75Y) non-synonymous point mutant showed improved growth relative to the wild-type strain (Figure 3D). In contrast, the growth of the ΔfleN clone was indistinguishable or slightly reduced from wild-type in these hosts (Figure 3D). Two independently-derived ΔfleN clones displayed phenotypes that were indistinguishable from one another (data not shown). Therefore, the specific fleN alleles isolated in this study were missense mutations that were selected because they resulted in proteins that had altered activities. These data are consistent with fleN serving multiple regulatory roles during infection, a hypothesis that is supported by recent data showing that fleN also influences cell division in Campylobacter jejuni [43]. The mutations identified in lineage B were similarly analyzed by moving individual changes into the wild type parent. Multiple mutations were introduced into the LP01 parent both in isolation and in the order in which they emerged within the adapted population. The effect of these mutations was determined by competing these newly constructed strains against the LP01: : lux strain (Figure 4A). In contrast to fleN mutations, multiple mutations were required to recapitulate the growth advantage of the lineage B isolate from day 133. Notably, the lqsS (N336Y), argD (frameshift), and sdbA (frameshift) mutations identified in lineage B did not detectably improve fitness in macrophages when placed in isolation, but introducing the lqsS argD (double) and lqsS argD sdbA (triple) mutations into LP01 resulted in improved fitness for both strains. Mutations were also observed within genes predicted to participate in the lysine biosynthesis pathway. To determine the impact of these mutations on the nutritional requirements of L. pneumophila, we introduced each mutation into LP01 and then measured the in vitro growth of each within defined broth media [44] in the presence or absence of added lysine (Figure 4B). Strains harboring the argD, dapE, and lysC/A mutations (from the B, C, and D lineages respectively), all grew in medium with lysine but, unlike LP01, were severely defective for growth in its absence. The addition of exogenous meso-diaminopimelic acid (meso-DAP), an intermediate metabolite directly converted into lysine by LysA in the last step of the pathway [45] partially rescued the growth of argD and dapE mutants in the absence of lysine, but failed to rescue the lysC/A mutant (data not shown). Models for the pathogenesis of L. pneumophila posit that the primary selective pressure for environmental maintenance of intracellular growth is the ability to replicate within amoebae [9], [46], [47]. Therefore, we tested if the adaptive changes affected growth in Acanthamoeba castellanii, one of several natural amoebal hosts of L. pneumophila, to determine if increased fitness in macrophages represents broadening of host range or causes a switch in host range specificity. As before, macrophages and amoebae were challenged with each strain and luminescence was monitored over 48–72 hours of incubation to determine the growth dynamics in both host cell types. The passaged strains replicated more efficiently in macrophages than did the progenitor strain (Figure 5A). In contrast, the lineage B and D strains displayed diminished growth in A. castellanii relative to the progenitor (Figure 5B). The two lysine biosynthesis mutations present in these strains, argD and lysC/A, were sufficient to generate these phenotypes in A. castellanii (Figure 5D) and in another amoebal host, Hartmannella vermiformis (Figure 5E). The lysC/A (G813R) also displayed a severe growth disadvantage during competitions with the wild type progenitor in A. castellanii (Figure 5F). The frequent observation of growth defects of lysine auxotrophs in A. castellanii and H. vermiformis is consistent with purifying selection during growth in environmental amoebae selecting for the maintenance of the wild type alleles of these genes, because the mutations that we analyzed resulted in a costly host-range specificity switch. In general, the results of growth curves and competition experiments were highly reproducible. The one exception was the lineage C clone, d180C′, which contains a dapE mutation as well as two other mutations (Figure 2). We saw variable levels of growth defects for this strain during incubations with A. castellanii and H. vermiformis, indicating that this genotype selects for frequent suppressors or else this strain background is extremely sensitive to small variations in bacterial or host cell culture conditions. In contrast to the lysine auxotrophs from the B and D lineage, the A and C lineage strains had equal or better fitness within amoeba as compared to the progenitor (Figure 5B). The A lineage mutation, fleN (D75Y), recapitulated this growth advantage in A. castellanii (Figure 5F), suggesting that some macrophage-driven adaptive changes can confer broadly increased replication in both macrophages and A. castellanii. None of the adapted strains displayed growth advantages during in vitro replication in rich media (data not shown), indicating that these phenotypic differences did not likely stem from the accumulation of mutations that broadly improved bacterial replication. Analysis of bacterial experimental evolution has generally focused on identifying mutations that arise during growth in chemically defined media [48]–[52]. One of the signatures of the selected strains is that they contain mutations that give insight into critical features of the growth conditions used. For instance, continuous passage of Escherichia coli in minimal media containing glycerol as the sole carbon source results in selection of glycerol kinase mutations [49]. A parallel can be found in our studies with the isolation of lysine auxotrophs during bacterial passage in macrophages. This provides information about the metabolic requirements for growth in both the natural and mammalian host. First, this is consistent with lysine being delivered efficiently to the Legionella-containing vacuole (LCV) in mouse macrophages (Figure 2, Figure 4B). Secondly, as many of these auxotrophs grow worse in amoebae, the natural host appears to be inefficient at delivering lysine to the LCV (Figure 5D–F). Therefore, continuous passage represents a powerful approach that allows identification of the restrictive features of a poorly defined growth environment. Experimental evolution of L. pneumophila within macrophages generated populations of mutants that had three features seen previously in experiments designed to model microbial adaptation in the laboratory as there was: 1) parallel evolution, with mutations being isolated in an overlapping set of loci across independent lineages; 2) frequent clonal interference between transient subpopulations of bacteria; and 3) genetic interactions, with the acquisition of successive mutations within a specific genetic background resulting in non-additive increases in the fitness levels of a strain. The independent isolation of mutations in multiple parallel lineages that alter a single loci or biochemical pathway, known as parallel evolution, is a frequent result of experimental evolution enrichments [34], [49], [51], [53]–[55]. In multiple macrophage-adapted lineages of L. pneumophila, we also observed several instances of parallel evolution: non-synonymous mutations in the flagellar regulator, fleN in three lineages (with a fleQ mutation isolated in the fourth), nonsense mutations isolated in three lineages in the Dot/Icm translocated substrate, sdbA, three independent lysine biosynthesis mutations (argD, dapE, and lysC/A), nonsense mutations in 2-oxoisovalerate dehydrogenase (oivA) found in two lineages, and the same exact mutation in between a hypothetical protein, lphB and a putative copper transporter pacS in two lineages, suggesting a hotspot for this mutation. By measuring the relative abundance of mutations in each population over time, all four of the macrophage-adapted lineages showed hallmarks of clonal interference, a property of asexual populations in which different beneficial mutations emerge in clones that subsequently compete with each other. Both experimental and mathematical modeling studies indicate that clonal interference is common in asexual populations and that these events should be positively correlated with both population size and mutation rate [56]–[58]. In our population genotyping data, we could readily identify time frames in which two mutations displayed inverse relative abundance, with only one reaching fixation in the population. In each case, we observed that the ultimate fixation of one of these mutations in a population was accompanied by the acquisition of one or more additional mutations. Our data cannot distinguish whether the acquisition of these additional mutations was stochastic or whether one of the competing genotypes was more or less compatible with additional beneficial mutations [59]. These two possibilities could be distinguished by replaying these events several times between the naturally competing clones of each lineage to determine whether one specific clone is more frequently fixed in each competition [59]. As organisms adapt to new environments, increased fitness under these conditions often correlates with reduced fitness in other environments [30], [60]. This relationship has been pursued using a number of model systems, including phage and other viruses adapted to novel host backgrounds [61]–[66], bacteria passaged under specific nutritional requirements [67], and light/dark cycling of the unicellular green algae, Chlamydomonas [68]. During extended passage through mouse macrophages, three out of four of the L. pneumophila lineages acquired mutations in lysine biosynthesis that resulted in lysine auxotrophy (Figure 2, Figure 4B). The most severe of these auxotrophic strains was the lysC/A mutation identified in lineage D. This gene encodes a bifunctional fusion protein predicted to catalyze both the first and last steps of lysine biosynthesis in L. pneumophila [69]–[72] (Figure 6), with the substitution occurring in the C-terminal domain that is similar to LysA, the DAP-decarboxylase responsible for converting meso-diaminopimelate (meso-DAP) to lysine. Consistent with this function, growth of this mutant in the absence of lysine cannot be rescued by addition of meso-DAP, whereas meso-DAP was able to partially rescue the growth of mutations in argD and dapE that are predicted to interfere with steps in the pathway upstream of meso-DAP (data not shown). Selection for lysine biosynthesis defects could have arisen in two ways: 1) the macrophage may supply sufficient lysine to the replication vacuole, allowing accumulation of lysine biosynthesis mutations due to neutral selection for pathway maintenance; or 2) loss of lysine biosynthesis confers a fitness advantage within macrophages but with a corresponding fitness cost in other hosts [67]. Our data support the second model, known as antagonistic pleiotropy, because we observe parallel selection for these mutations in independent lineages (Figure 2), the acquisition of these mutations is rapid, and the fixation of each mutation appears to occur as part of selective sweeps through the population. If neutral selection were a significant contributor to the isolation of mutations in this study, we would expect to have identified lesions in several Dot/Icm translocated substrates that have individually and collectively been shown to be dispensable for growth in macrophages [1], [17]. The Dot/Icm translocated substrates represent over 10% of the entire L. pneumophila genome, yet the only mutations that we identified were in sdbA and sidF, which when mutated may provide a selective advantage [11]–[13]. As nutrient availability is known to regulate the intracellular differentiation of L. pneumophila from replicative to transmissive phase [73], we hypothesize that modulation of lysine biosynthesis may serve to influence the activation of transmissive traits during late stages of infection. Within this context, the potential for crosstalk between the lysine and arginine biosynthetic pathways in L. pneumophila should be explored given the inclusion of argD in both pathways and a previously identified role for arginine levels in broadly regulating transcription [74]. Our data are consistent with a model in which the availability of lysine (or intermediate metabolites in the lysine/DAP biosynthesis pathway) is lower in at least some amoebal hosts than in bone marrow-derived macrophages, which explains why this biosynthesis pathway has been retained in L. pneumophila. The macrophage-selected lysC/A (DAP-decarboxylase) and argD mutants are defective for intracellular growth in A. castellanii and H. vermiformis (Figure 5D–F), consistent with the model that available lysine pools are lower in some amoebae than in other cells. Based on genome annotation, Coxiella burnetii [75], an obligate mammalian pathogen and close relative of L. pneumophila, lacks a functional lysA DAP-decarboxylase, showing a connection between pathogen adaptation to mammalian host cells and selection for loss of a complete lysine biosynthesis pathway. C. burnetii survival, but not replication, has been demonstrated in amoebal hosts [76]. We propose that lysine availability within amoebae limits the environmental dissemination of mammalian host-adapted pathogens and prevents specialization of L. pneumophila for growth within a mammalian niche. A recent outbreak of Q-fever was linked to the presence of C. burnetii in an air-conditioning unit, where it was proposed that amoebal hosts supported persistence of the pathogen in the aquatic environment [77]. Strain-to-strain variation of the lysine biosynthesis pathway in these clinical strains may explain the dissemination of C. burnetii during this outbreak. In light of these results, a systematic examination of the nutritional requirements of L. pneumophila within diverse host backgrounds that was initiated several years ago should also be revisited in this new context [78]. In conclusion, we have shown rapid parallel evolution of L. pneumophila during conditions of host restriction in mouse macrophages. Some of the adaptations resulted in fitness costs in chemically defined media and in amoebal hosts, whereas others provide broadly improved intracellular replication. These data are consistent with the model that cycling through diverse protozoan hosts maintains L. pneumophila in “evolutionary stasis” as a generalist [17]. The extension of our experimental evolution approaches to conditions of host restriction in different protozoan species will be critical to determining whether genome reduction and/or specialization can occur under these conditions as well. In vivo experiments with West Nile Virus, indicate that having relaxed purifying selection within some hosts can also influence pathogen evolution during host cycling [61]. As additional natural protozoan hosts of L. pneumophila are identified, it will be important to determine whether similar processes also influence the evolution of L. pneumophila during its passage through diverse natural hosts. Intracellular growth in isolated mouse macrophages can never capture the complexity of infection conditions in human hosts. While our adapted strains display improved replication in cultured mouse cells, both known deficits in the progenitor strain [79] and evolution towards reduced fitness in natural hosts would likely reduce their transmission in the wild. Remarkably little is understood about the transmission of L. pneumophila between its natural protozoan hosts and the human lung [80], the relationship between infectious dose and severity of disease, or genotypic diversity and selection during individual cases of disease. Just as access to low-cost sequencing has transformed the field of experimental evolution, the application of these technologies to environmental and clinical strains of bacteria has already started to define the selective pressures that influence the ultimate outcome of disease [81]–[86]. Applying these approaches to understanding the epidemiology of human outbreaks of L. pneumophila will be particularly critical to identifying the evolutionary pressures that shape these events. The L. pneumophila LP01 strain used in these studies is derived from a clinical isolate of Legionella pneumophila strain Philadelphia-1 [22] and is virulent in guinea pigs (data not shown). LP01, rather than a related thymidine auxotroph, LP02, was selected for these studies in order to facilitate amoebal challenge, as thymidine auxotrophy severely limits intracellular replication in these hosts. pSR47-ahpC: : lux, a plasmid containing the luxCDABE operon of Photorhabdus luminescens downstream of the L. pneumophila ahpC promoter (a kind gift from J. Coers and R. Vance) [28], was used in triparental matings with the E. coli Tra+ helper strain RK600 [87] to integrate the lux operon onto the chromosome of L. pneumophila strain LP01. Individual mutations were introduced to LP01 through triparental matings with pSR47S-derived plasmids containing 2 kb PCR products generated from amplification of regions flanking identified mutations using genomic DNA of each passaged strain as template. An in-frame deletion of the fleN open reading frame in LP01 was generated by first making a pSR47S-derived plasmid containing 4. 9 kb of sequence surrounding the fleN locus. Inverse PCR was then used to eliminate residues 27–284 of the 295 amino acid protein within this plasmid, using lpg1783ko_invF (5′-TCAAGGCAGATCTTTTCTTTTTGGAGCGTTTGG-3′) and lpg1783ko_invR (5′-TCAAGGCAGATCTCGGGACAAATTTCTAAGACCA-3′), followed by BglII/DpnI digestion and subsequent intra-molecular ligation with T4 DNA ligase. Primary bone marrow-derived macrophages from female A/J mice were isolated as described previously [88], frozen in 10% DMSO/90% fetal bovine serum, and thawed prior to use. Challenge of both primary A/J mouse bone marrow-derived macrophages and Acanthamoeba castellanii with L. pneumophila was performed as previously described [89], using bacteria grown overnight to post-exponential phase that were predominantly motile (A600 = 3. 7–4. 5). For growth curve analysis, 1×105 macrophages per well were plated in 240 µl RPMI1640 without Phenol Red+glutamine+10% heat-inactivated FBS in 96-well white tissue culture treated plates (Greiner). A. castellanii (ATCC30234; American Type Culture Collection) were plated at a density of 2. 5×105 amoebae per well in 240 µl Ac buffer [89]. H. vermiformis (ATCC50237; American Type Culture Collection) were plated at a density of 1×105 amoebae per well in 240 µl H. vermiformis medium (modified PYNFH medium, ATCC medium 1034). Macrophages and A. castellanii were challenged with bacteria at an MOI of 0. 05; H. vermiformis were challenged at an MOI of 0. 5. The plates were incubated at 37°C (in 5% CO2 for macrophages) in a Tecan Infinite M200 Pro with Luminescent and CO2 Gas Modules. Luminescence was measured for 20 seconds per well every 20 minutes. Growth experiments were performed multiple times, and in each case the data shown are from one representative experiment with 3 or more inoculations each experiment. An overnight culture of LP01: : lux bacteria was grown to post-exponential phase, as described above. 5×105 bacteria were used to inoculate each of 4 independent cultures of 1×107 freshly thawed A/J primary bone marrow-derived macrophages in 12 ml of RPMI1640+glutamine+10% heat-inactivated FBS+100 µg/ml streptomycin in 10 cm tissue culture treated Petri dishes. After inoculation, each dish was centrifuged for 5 minutes at 400× g and incubated at 37°C, 5% CO2 for 2–3 days. After incubation, supernatants from each culture were collected. Remaining host cells were lysed by adding 8 ml of sterile ultrapure water (Invitrogen) and incubating for 15 minutes at room temperature. After pipetting up and down, these lysates were combined with the supernatants. An estimate of bacterial density was determined by pelleting 1. 5 ml of each collection, resuspending in 100 µl PBS, and measuring luminescence in a 96 well plate in a Molecular Devices Spectramax M5 plate reader. These estimates were empirically determined by using as standards the luminescence of single passages of wild type bacteria harvested in this manner after 3 days in culture. An amount of culture equal to approximately 1×10−3 of the growth over these 3 days was used to inoculate new cultures of 1×107 macrophages thawed 1 day prior to this re-infection. Periodically, dilutions of these lysates were plated on solid CYE agar plates in order to ensure that luminescence continued to approximate the CFUs in each culture. Every 9–25 days, the remaining lysates were centrifuged for 5 minutes at 400× g to remove cellular debris and intact host cells. The supernatants were then centrifuged at 7000× g for 15 minutes and the resulting bacterial pellets were resuspended in AYE+20% glycerol and stored at −80°C for future analysis. For competition assays, host cells were challenged at total MOI = 0. 05, consisting of equal mixtures of two strains, one carrying the luxCDABE operon and one without, centrifuged for 5 minutes at 400× g and incubated at 37°C, 5% CO2. These assays were performed in 96 well plates, using either 1×105 primary A/J bone marrow-derived macrophages or 5×105 A. castellanii cells. A. castellanii challenge was performed without centrifugation. 3 days after inoculation, remaining host cells were lysed with 0. 05% saponin for 5–10 minutes as described previously [89]. Dilutions of each lysate were plated on CYE solid agar and colonies were visualized with and without epi-illumination using the Biorad Chemidoc XRS system. The competitive index (C. I.) was determined as described previously [31], C. I. = (mutant/wild type output ratio) / (mutant/wild type input ratio), normalized to the results from competitions between two differentially marked progenitor strains, and plotted on a logarithmic scale. P-values were determined in a two-tailed, unpaired Student' s t-test of the logarithmic-transformed normalized C. I. values, comparing each competition to a wild type/wild type control competition. A P-value of less than or equal to 0. 05 was considered a significant difference from this control competition. Uptake competition assays were performed with 1×105 primary A/J bone marrow-derived macrophages added to each well of a 96 well tissue culture plate. After a 4 hr incubation at 37°C, 5% CO2, host cells were challenged with a total MOI = 1. 0, consisting of equal mixtures of two strains, centrifuged for 5 minutes at 400× g and incubated at 37°C, 5% CO2 for 1 hr. 1 hr after inoculation, gentamicin was added to each well at the final concentration of 50 µg/ml and the cultures were incubated for an additional hr to kill extracellular bacteria. Each well was then washed 5 times with 200 µl phosphate-buffered saline, after which macrophages were lysed by incubating each well in 200 µl ultrapure water for 10–15 minutes at room temperature. Relative ratios of colony forming units (CFUs) for each strain in each well were calculated as above. Control wells of bacterial inoculations, in which no macrophages were present, were also performed under these conditions in order to confirm the anti-bacterial activity of gentamicin against extracellular bacteria, as indicated by the absence of CFUs from these wells. A competitive index of uptake was determined for each challenge, as described above. Bacteria were recovered from glycerol stocks by streaking onto CYE agar plates. Genomic DNA was isolated from individual clones grown to post-exponential phase, using the Qiagen DNeasy kit including the optional RNase digestion (Ambion RNase cocktail). 5 µg of genomic DNA was sheared by nebulization for 6 minutes at 35 psi (Invitrogen). Sheared DNA was purified on QIAquick spin columns (Qiagen), then treated with the End-IT DNA Repair kit (Epicentre). After spin-column purification, 3′ A-tailing was performed by incubating for 1 hour at room temperature with Exo-minus Klenow (New England Biolabs) and dATP. Samples were again purified using QIAquick columns, and custom, 4 nt 5′ barcoded, adapter sequences were ligated to each sample using the Fast-link ligation kit (Epicentre). Libraries were size-selected on 2% agarose gels, and fragments 400–450 nucleotides in length were purified using QIAquick columns. To enrich for properly ligated samples, approximately 4% of each library was amplified for 16 cycles using common primers, QIAquick purified, and then quantified using a Nanodrop spectrophotometer. Libraries were mixed at equal ratios and sequenced on an Illumina Genome Analyzer II. LP01: : lux, d133A, d200A, d314A, d133B, d314B, d90C, d180C, d180C′, d45D, d90D, and d180D libraries were sequenced using single-end, 40 nt sequencing reactions. d346A and d346B libraries were sequenced using paired-end, 2×40 nt reactions. Multiplexed sequence data was sorted by 5′ barcode identity into individual libraries. De novo assembly was performed with Velvet [90]. CLC Genomics Workbench 3 and Maq were used to generate reference assemblies comparing each strain against the published L. pneumophila Philadelphia-1 (Genbank accession: NC_002942) and the LP01: : lux assembly. Raw Illumina reads from each sequenced strain were deposited as Sanger-formatted FASTQ files in the Dryad Digital Repository (doi: 10. 5061/dryad. 95mt02sb). Chemically defined, Modified Ristroph media (MRM) [44], [73] supporting L. pneumophila in vitro growth was used to determine the growth requirements of strains harboring lysine-biosynthetic mutations. Bacteria were grown overnight with shaking at 37°C to an A600 = 2. 0 in AYE medium. For each strain, 2 tubes containing bacteria equivalent to 1 ml of A600 = 2. 0 were pelleted at 14,000× g, washed once in 1 ml of MRM (−) lysine, and resuspended in 10 ml of either MRM (−) lysine or MRM (+) lysine. 240 µl of each of these suspensions was aliquoted into each of 3 wells of a 96 well plate. The OD of each well was measured every 12 minutes during incubation at 37°C with shaking in a Biotek plate reader. Sterile pipette tips were used to scrape each of the frozen glycerol stocks collected during the experiment. These samples were resuspended in ultrapure water, pelleted in a microcentrifuge, and cleared supernatants were used as templates for analysis. The procedure used for determining the proportion of each allele is Quantitative Exon Grouping Analysis (qEGAN). This is an extension of the procedure used in the laboratory for high throughput sequence analysis of exonic regions that has been used extensively for the analysis of BRCA1 and BRCA2 [38]. qEGAN is a heteroduplex technology based on conformation-specific gel electrophoresis (CSGE) [91], [92] and conformation-specific capillary electrophoresis (CSCE) [93], [94]. Briefly, an approximately 200 nt region containing each polymorphism was PCR amplified using flanking primers containing universal 5′ sequence tails. These products were fluorescently labeled during a secondary round of PCR amplification, using FAM-tagged universal primers. These fluorescently labeled amplicons were directly analyzed on an Applied Biosystems 3730xl capillary sequencer with a 50 cm Capillary Array, using non-denaturing POP Conformational Analysis Polymer and ROX labeled size standard (Life Technologies, Carlsbad, CA). The curve height of fluorescent signal from each reaction was normalized and the relative patterns analyzed using DAx Data Acquisition and Data Analysis software (Van Mierlo Software Consultancy, Eindhoven, NL). In parallel, mixtures of genomic DNA from LP01 and each adapted strain were also used as templates for this analysis. By comparing the curves generated from each known mixture of wild type∶mutant genomic DNA, the ratio of each mutation was determined within the time-course glycerol stocks. These results were confirmed by phenotypic quantification and results from allele-specific quantitative PCR (data not shown).

Legionella pneumophila is an accidental pathogen of humans, responsible for the severe, often-fatal pneumonia known as Legionnaires' disease. In the environment, L. pneumophila survives and replicates within protozoa by co-opting the intracellular machinery of these microbial predators. These freshwater encounters between bacteria and protozoa likely provided L. pneumophila with the selective pressures required to evolve into an intracellular pathogen. Many of the host pathways that L. pneumophila manipulates during infection are highly conserved and this is presumably what allows L. pneumophila to infect human cells. It is likely that L. pneumophila is suboptimally adapted to replication within mammalian cells, however, as replication within human cells is thought to be an evolutionary dead end. In this study, we developed an experimental evolution approach to determine what unique selective pressures might be present within mammalian hosts and how these pressures might modify this pathogen. We subjected L. pneumophila to continuous passage within mouse macrophages for several months, selecting for spontaneous mutations that resulted in improved fitness within these cells. We sequenced the genomes of each of the adapted strains, measured the population dynamics of each evolving population, and identified mutations that improve replication in mammalian cells and alter bacterial fitness in amoebae.

Abstract Introduction Results Discussion Materials and Methods

medicine infectious diseases biology genomics evolutionary biology microbiology genetics and genomics

2012

Experimental Evolution of Legionella pneumophila in Mouse Macrophages Leads to Strains with Altered Determinants of Environmental Survival

The intuitive response to an invading pathogen is to start disease management as rapidly as possible, since this would be expected to minimise the future impacts of disease. However, since more spread data become available as an outbreak unfolds, processes underpinning pathogen transmission can almost always be characterised more precisely later in epidemics. This allows the future progression of any outbreak to be forecast more accurately, and so enables control interventions to be targeted more precisely. There is also the chance that the outbreak might die out without any intervention whatsoever, making prophylactic control unnecessary. Optimal decision-making involves continuously balancing these potential benefits of waiting against the possible costs of further spread. We introduce a generic, extensible data-driven algorithm based on parameter estimation and outbreak simulation for making decisions in real-time concerning when and how to control an invading pathogen. The Control Smart Algorithm (CSA) resolves the trade-off between the competing advantages of controlling as soon as possible and controlling later when more information has become available. We show–using a generic mathematical model representing the transmission of a pathogen of agricultural animals or plants through a population of farms or fields–how the CSA allows the timing and level of deployment of vaccination or chemical control to be optimised. In particular, the algorithm outperforms simpler strategies such as intervening when the outbreak size reaches a pre-specified threshold, or controlling when the outbreak has persisted for a threshold length of time. This remains the case even if the simpler methods are fully optimised in advance. Our work highlights the potential benefits of giving careful consideration to the question of when to start disease management during emerging outbreaks, and provides a concrete framework to allow policy-makers to make this decision. Effective management of infectious disease remains a significant challenge [1–3]. Mathematical modelling is a powerful tool for integrating over uncertainty in predicting the future spread of outbreaks [4,5], and provides a rational methodology for assessing the likely performance of proposed interventions. Policy-makers can then base decisions concerning when, where and how to control disease outbreaks on quantitative scientific evidence [6–9]. Where and how epidemics should be managed has received significant attention, and a range of complex, often spatially-heterogeneous, control strategies have been proposed. These often target individuals at high risk by virtue of location [10–13] or contact patterns [14,15]. The question of when to introduce interventions appears much simpler. The difficulty of controlling an outbreak depends upon its size, and is expected to be easier and/or less expensive if treatment starts when the outbreak is small [9,16]. Early action also potentially avoids the feedback by which the net growth rate of an epidemic increases as the outbreak gets larger [17]. Intuition therefore suggests that control should be performed as soon as possible, a result echoed in a number of modelling studies [12,18–20]. However, intuition is seldom infallible, and optimising the timing of disease management can be more nuanced. Early control is only guaranteed to be effective if there is absolutely no uncertainty surrounding the future spread of disease. Demographic stochasticity can lead to an outbreak dying out without any management before a major epidemic occurs [21,22]. For certain plant and animal diseases it might therefore be optimal simply to wait and see whether or not the outbreak takes off, potentially avoiding wasted prophylactic control [23]. Withholding or delaying treatment creates ethical problems over each individual’s right to treatment, particularly for human diseases. Nevertheless for emerging outbreaks of humans as well as of crops and livestock, control interventions can almost always be targeted more effectively later in epidemics, when more is known about the dynamics of disease spread. The optimal management strategy is typically conditioned on the values of parameters controlling pathogen transmission [24,25], often quite sensitively [13,26,27]. Epidemiological parameters can be characterised increasingly accurately as an outbreak progresses, since more data are then available [28–30]. This sets up a trade-off, in which the additional accuracy with which control could be targeted if management was to be delayed could potentially outweigh the increased cost of controlling a larger epidemic. This trade-off raises a question: when precisely is it optimal to treat? To work in a concrete but still general setting, we consider a large-scale epidemic affecting agricultural animals, which we model using a stochastic, compartmental model of pathogen transmission in which the host units represent individual farms. This broad class of model–or elaborations to include a spatial component or other additional population heterogeneities–is routinely used for a number of animal diseases including classical swine fever [31,32] and bovine tuberculosis [11,33], as well as for pathogens of plants [3,9, 34] and humans [7,35,36]. Estimated values of transmission parameters become more precise as the outbreak unfolds and more data to fit the model become available. Control involves prophylactic vaccination of all animals on a fraction of susceptible farms. Vaccination protects farms for the duration of the outbreak, and, for simplicity, we assume vaccination can be done only once. The overall aim is to minimise the total cost of the outbreak, which consists of the cost of herds lost to disease, together with a smaller cost assigned to each farm on which animals are vaccinated. While we refer to a disease of livestock and a control strategy based on vaccination here, this is only to provide a simple scenario that is straightforward to describe and to visualise. The main concept that we investigate–i. e. whether or not an initial period of observation and learning might be a sensible consideration in an outbreak response setting–is likely in fact to be applicable to a large number of pathogens and controls. We introduce the extensible Control Smart Algorithm (CSA) to determine when and how to control in real-time. The CSA integrates disease spread data, parameter estimation and outbreak simulation to determine whether to control now, or whether instead management should be delayed to allow the policy-maker to learn more about transmission and the likely severity of the outbreak. Crucially, the CSA takes account of how rapidly any imprecision is likely to be resolved, comparing this timescale with the time at which the epidemic is expected to become too large to control effectively. The CSA is used repeatedly at successive times at which control could potentially be introduced, leading to a sequence of decisions on whether or not to intervene, as well as a specification of the control to perform when management is finally to be attempted. We address the following questions. At any time during an emerging outbreak, the optimal number of farms on which to deploy vaccination depends on the current outbreak size and the values of parameters governing transmission. We showed how this optimal number can be estimated using the stochastic SIR model and how this number depends on the model parameters (S1 Text; S1 Fig). If a policy-maker is presented with an outbreak for which the infection statuses of every farm and the values of transmission parameters are all known, rather than simply the infection statuses as we assumed here, then optimising control is very straightforward. However, the estimated values of parameters change as an epidemic unfolds, and during this time the current outbreak size also changes. To understand time-dependence of these factors that drive the optimal amount of control to deploy, we simulated a very large number of outbreaks, quantifying temporal changes in the precision of parameter estimates (Fig 1A), the probability of the outbreak dying out (Fig 1B), and the size of the outbreak (Fig 1C). We then tested the effect of introducing control on each day in these simulated epidemics, using parameter estimates from data up to and including the day in question to determine the most efficient management at that time. To optimise the extent of control deployment, we used the Control Amount Optimisation Algorithm–CAOA, shown in S1 Algorithm and S2 Fig. The CAOA averages over uncertainty in the effect of management as well as uncertainty in the transmission parameter values via repeated forward simulation. The average cost over all simulated outbreaks was minimised with control deployed after 40 days (dotted black line in Fig 1D). Since delaying control was initially optimal, the trade-off between the additional accuracy with which control can be targeted and the increased size of the epidemic led to a non-zero delay before intervening became optimal. To separate the effects of increasingly precise parameter estimates from those caused by outbreaks potentially dying out naturally, we also examined only the subset of simulated outbreaks that persisted for more than 50 days in the absence of control (blue line in Fig 1D). The response of the average outbreak cost to time of treatment again had a minimum, here at t = 28 days, meaning that uncertainty in parameters alone favoured delayed control. When uncertainty about model parameters was removed instead, so that the exact value R0 = 1. 5 was used to optimise the amount of control deployed, we found that it was still optimal to wait before implementing control (the minimum was now at t = 30 days, red line in Fig 1D). The chance of natural disease fade-out therefore also suggested an optimal strategy of waiting before starting treatment. The optimal time to control any individual ongoing outbreak will usually differ from that dictated by the average behaviour over a large ensemble of outbreaks. We therefore developed the Control Smart Algorithm (CSA), a dynamic approach for deciding when to implement control in real-time during a single ongoing outbreak (see Fig 2 and Methods). We first showed how the CSA performs on a synthetic dataset from a single simulated outbreak, allowing for decisions concerning whether or not to introduce control on a weekly basis (Fig 3). For the example epidemic that we initially considered, there were a number of decisions to wait, followed by the decision to control after T = 29 days (Fig 3A). The estimated values of transmission parameters became increasingly precise during the waiting phase (Fig 3B). Control was introduced at the first possible intervention time at which the expected cost of the outbreak when controlling at the current time was less than the equivalent expected cost at all possible future control times (i. e. the first time at which the curve in Fig 3C was monotonic increasing). Of course even after control was introduced, demographic stochasticity meant that a range of outcomes remained possible, even though R0 was then fixed at the true value (Fig 3D). By repeating this analysis for a large number of different simulated epidemics for a range of values of R0, we compared the performance of the CSA with those of simpler methods for deciding when and how to control (Fig 4). In particular, we tested the CSA against alternative strategies in which control was deployed as soon as thresholds in outbreak severity or duration were reached [37]. By optimising the threshold-based methods in advance (S2 Text), we identified the following thresholds: duration of the epidemic (TT strategy; t = t* = 15 days; Fig 4A); number of farms that have ever been infected (TIR strategy; (I+R) = (I+R) * = 8; Fig 4B); and number of farms currently infected (TI strategy; I = I* = 7; Fig 4C). The CSA reduced the mean cost averaged across all outbreaks compared with these simpler methods (Fig 4D–4E). Although the single goal of the CSA was to minimise the mean outbreak cost, using this method also reduced the cost of those outbreaks that did invade (Fig 4F) and the probability of a major epidemic (Fig 4G). Our objective here was to introduce and test the CSA in the simplest possible setting, in which the infection rate was the only parameter requiring estimation from disease spread data and in which there were no constraints on control deployment. However, we additionally verified that the improved performance of the CSA in comparison to the simpler methods of control was robust to informed prior knowledge of the parameters (S4 Fig), estimation of multiple parameters (S5 Fig) and logistic constraints in control (S6 Fig). We also tested robustness to the accuracy of prior knowledge concerning epidemiological parameters (S7 Fig), as well as to systematic bias in this prior knowledge (S8 and S9 Figs). These elaborations to the CSA are described in S3 Text. For computational tractability given the extremely large number of simulated outbreak datasets we considered in these extensive sensitivity analyses, we ran the underlying simulations in a smaller population of N = 100 host farms. In all cases the CSA outperformed the other simpler threshold-based strategies for determining when and how to control. Delaying management to allow more information to be accrued has been proposed only rarely for infectious diseases [38,39], despite work focusing on a similar idea for managing invasive species [40–43]. Here we have shown–using a generic stochastic model capturing the key features of an emerging disease outbreak–that it can be beneficial to delay control of potential epidemics (Fig 1). This allows outbreaks that would die out naturally to do so at no additional cost. It also allows disease spread parameters to be estimated more accurately, which in turn allows management to be targeted more precisely. Having verified that deferring control can be optimal via a post-hoc analysis in which the costs of numerous simulated outbreaks were averaged, we introduced the Control Smart Algorithm (CSA, Fig 2). This algorithm could be extended for use in real-time to inform decision-making concerning when to control an ongoing outbreak. It makes use of stochastic compartmental models and rigorous parameter estimation techniques of the type routinely employed for real-time infectious disease epidemic forecasting. The CSA uses forward simulation to optimise the timing and amount of control to introduce. The algorithm scans over possible control interventions, and all possible values of the parameters governing pathogen transmission. By averaging according to entire posterior distributions for the values of these parameters, a pre-specified objective function derived from the distribution of costs of forward simulations can be minimised. Here, we used the mean cost of outbreaks as this objective function, and the expected cost of the outbreak with control now and the equivalent expected cost with control in future were calculated exactly. For this reason, the expected cost of outbreaks with control according to the CSA as presented here will always be lower than with any other possible control strategy, given a posterior estimate of the parameters. This is the case even if these other strategies are fully optimised in advance of the outbreak (Fig 4). Of course, if the posteriors of the parameters are incorrect, then the CSA will be badly informed, and so in that case it might be possible for other control strategies to outperform the CSA. In reality there is often prior information that leads to systematic bias in the decision-making, for example observations of previous epidemics of the same pathogen but in different environments. However, a strategy of waiting before deploying control has the additional benefit that evidence to overcome this initial bias can be accrued. We showed the robustness of our method to the choice of prior (S4–S9 Figs), including situations with systematic biases in the decision-making (S8 and S9 Figs). The benefit of waiting to learn about disease spread, and the rate at which this uncertainty can be resolved compared to the speed of the outbreak, are likely to impact on whether or not waiting before deploying control is a sensible strategy. As we have described, the CSA performs this assessment rigorously given outbreak data from up to the current time and priors on the values of transmission parameters. In cases where uncertainty can only be reduced slowly throughout an outbreak, then early control will be advantageous. Early control will avoid the outbreak size becoming large before the most efficient control strategy can be deduced. The recommendation of the CSA in this case would most likely be to control immediately. The concept presented here, that waiting to introduce management might outperform a strategy of controlling immediately, was inspired by the similar idea of “wait and see” approaches to invasive species management [40,41,44,45]. In the pest management setting, it has been shown that it can be most effective to survey sites intensively to learn where the pest is most likely to be found before introducing a mixed strategy of searching and removal [40]. The idea of investing in collecting data has also been considered for protecting habitat critical for species survival, by asking when interventions should be introduced to avoid an unacceptable risk of species extinction [43]. The optimal times at which to change management strategies have also been considered in the endangered species management setting [42,46]. Of course, while delaying control can be effective for invasive species, waiting too long is also ineffective [47,48]. We have shown here that this principle also applies to controlling invading pathogens. While studies about when to introduce interventions are rare for infectious diseases, certain outbreaks are thought to have perhaps been over-controlled. For example, costly widespread management interventions were enacted in 2009 to counter the perceived potential pandemic of H1N1 influenza [49]. However, the outbreak was far less serious than experts had forecast, leading to speculation that control deployment was disproportionate to the level of threat [50,51]. There are also a number of outbreaks, such as the 2007 United Kingdom foot-and-mouth outbreak, in which disease fade out has been used to justify the preceding aggressive interventions [52]. However, despite the small size of these outbreaks, control can be expensive [53], and outbreaks might be minor in the absence of attempted eradication strategies [22]. Immediate intervention might therefore be unnecessary and an avoidable cost. Such high-profile examples of the potentially sub-optimal effects of very early management have motivated the invention of various metrics for guiding whether or not to act. One example is the first 14 days incidence, FFI, which has been proposed by Hutber et al. [54] as a parameter for informing whether or not to begin emergency vaccination campaigns. The closely-related first 14 days spatial spread parameter, FFS, was developed subsequently [55]. Outbreak simulations indicate that there is a strong positive correlation between FFI/FFS and variables indicative of outbreak severity such as final size and outbreak duration [56,57]. Further studies have strengthened the evidence that simple metrics available early in outbreaks are able to predict final epidemic characteristics [56]. Such metrics can therefore form the basis of decision-making tools that present important information about the possible impacts of control interventions in a simplified format for policy-makers [57]. Other modelling analyses that seek to avoid unnecessary early control are based on real options, which attempt to understand and value potential benefits of waiting before introducing disease control [58]. Epidemiological applications include the timing of investment in antiviral drug stockpiles [59] and control of plant disease outbreaks [38,60]. The real options approach has illustrated the principle that when there is uncertainty in future pathogen/pest progression through the host population, combined with the sunk costs of control measures, there can be value in waiting instead of deploying control immediately [61–63]. How long the decision-maker should wait depends on a number of factors. These include the level of uncertainty in the future spread of the outbreak [60], whether or not proposed interventions are reversible [59] and policy-makers’ flexibility in choosing the timing of interventions [38]. However, previous studies using real options have not explicitly considered the fact that as the epidemic progresses and more data are collected parameter estimates become more precise. As we have demonstrated here, uncertainty in transmission parameter values is an important factor in determining the optimal timing for the introduction of control measures. Resolving other types of uncertainty might also be important in practice. So-called “Value of Information” analyses [64–67] have been proposed to estimate the costs associated with uncertainty. For example, the “Expected Value of Partial Perfect Information” was recently used in the context of foot-and-mouth disease to estimate the costs associated with uncertainty about a number of factors including vaccine efficacy and the time delay to immunity after vaccination [27]. Value of Information analyses can inform adaptive management strategies, in which disease controls are changed over time as more information becomes available and model uncertainty is resolved [68]. An adaptive management strategy has also been proposed to show how dynamically changing vaccination strategies can reduce the costs of epidemics [69]. However, the question of when to introduce control in the first place has not been considered in that type of study. Much closer to the current study is a dynamic methodology for deciding when to control an epidemic that was introduced in the context of influenza management by Ludkovski and Niemi [70]. In that approach, a set of outbreak statistics (observed states of the system and estimates of disease spread parameters) are chosen to characterise the current severity of the outbreak and the current level of predictability of future spread. Policy maps are then created using regression Monte Carlo, which specify values of these outbreak statistics for which control should be introduced immediately [70]. For efficient use, the policy maps must, however, be generated in advance of the outbreak. In contrast, the CSA does not require outbreak statistics to be chosen in advance, but instead uses all the information available to policy-makers during an emerging outbreak. The CSA also only simulates forwards from the current observed state of the system, thereby avoiding additional computational costs of exploring future behaviour that is no longer possible given the progress of the outbreak so far. We framed our analysis in terms of stochastic SIR dynamics, which we adopted here to represent only the necessary features of an outbreak for which there is uncertainty about transmission parameter values and future spread [24,71]. We used synthetic data generated using model simulations to illustrate the concept that the optimal time to introduce control might not be immediately, and also to test our results for a very large number of different outbreaks. In practice for any single particular outbreak, it would be necessary to use a detailed model more finely-tuned to the particular host-pathogen system in question. However, since our results and the CSA are largely based on forward simulation, in principle such alterations could be made without any significant methodological change. Indeed the elaborations to the CSA that we presented in S4–S9 Figs showed how additional detail can be included in the underlying methodology in simple cases. More exhaustive investigations exploring systematically how variations in model complexity–or alterations to the elaborations that we considered–impact on the performance of the CSA relative to the simpler methods for deciding when and how to control are left as future work. For example, an analysis analogous to S6 Fig in which the limit to the number of farms that can be controlled per day can take a range of values could be considered. The CSA could also be adapted for use with more complex epidemiological data and methods of parameter inference. Here we assumed that all epidemiological transitions made by all hosts were fully recorded to facilitate easy parameter estimation, but this is not necessary; the only requirement is that posterior estimates of epidemiological parameters can be generated. Other sources of uncertainty could also potentially be accounted for in the CSA, for example uncertainty in the number of currently infected host units arising either because of underreporting [72] or presymptomatic/asymptomatic infection [22], and uncertainty in which model out of a suite of plausible compartmental models most accurately represents disease spread [73]. However, we defer systematic testing of the performance of the CSA with these additional sources of uncertainty to future work. Other simplifications underlie our work. Control made susceptible farms immune to the disease for the remainder of the outbreak. Although simple, such behaviour is consistent with a range of pre-emptive controls, including vaccination, preventative culling/thinning and application of protectant fungicide or pesticide. However, other controls could instead have been adopted, singly or in combination. Complexities such as resource-driven limitations to the capacity for treatment [10] or delays between decision-making and control deployment [12] could also have been included. To demonstrate how the CSA can be extended to account for considerations like these, we built S6 Fig in which a limit to vaccination deployment was included in the analysis in a simple way, showing at least in principle that the CSA is robust to this additional complexity. Our metric for assessing the cost of outbreaks was also rather simple. Elaborations could include a discount rate [74], more complex notions of costs of infections and control [75], and multiple objectives [76,77]. As a proof of concept, we focused entirely on minimising the mean cost of an outbreak in this manuscript. This choice could potentially lead to outbreak features that might be undesirable for policy-makers. For example, in Fig 4D it can be seen that, despite the CSA having the minimum mean cost, on those occasions where costly outbreaks do occur the distribution of costs under the CSA is “top heavy” compared with the other strategies. Practical implementation of the CSA for disease outbreak management would require careful consideration of precisely which objective function to minimise. The mean cost of forward simulations may not be the preferred choice, since this quantity can be influenced by simulations with extremely high or low costs. For real outbreak response scenarios, policy-makers might therefore prefer to optimise quantities such as the median forecasted outbreak cost, or even to focus on an objective function taking a weighted average of a number of different metrics. Our underlying method could also handle other levels of risk aversion, for example by considering achieving optimal performance for different percentiles in the distribution of outbreak costs [12]. Since control and its cost only enter the CSA via forward simulation of the disease spread model, in principle interventions and cost functions of arbitrary complexity would be admissible. While additional features could be introduced into the underlying methodology that we presented, we note that efficient management is often beset with technical and operational challenges [4,78,79]. Identifying which factors to include in the model of spread and the cost function is an important problem, particularly for emerging outbreaks of novel pathogens during which there are numerous uncertainties. It must also be decided which interventions are likely to be feasible before the CSA can be used. Even if possible control strategies and factors that affect the impact of the outbreak can be identified, the optimal level of detail to incorporate in the modelling study is a fine balancing act. Communication between modellers and policy-makers is essential [79]. However, while extensions to the CSA are possible, our goal was to introduce the algorithm in a relatively simple setting. We therefore only considered a single management intervention, and assumed that control cannot be changed later in the outbreak. This type of restriction might well be in place in practice, since governments might only be able to set their policy once, or at most change it a limited number of times. Extending the CSA to account for multiple changes in policy would be interesting, although computationally challenging to trial on large numbers of simulated outbreaks. Even with a single time of control introduction, the CSA is computationally expensive to test on many simulated datasets due to the need to run forward simulations with control both at different times in the future and with different amounts of control deployed. However, we note that in practice decisions are required for only one single outbreak (i. e. practical use would require only an analogy of Fig 3, not of Fig 4), greatly reducing computation times. While the approach of waiting before deploying control when an outbreak is underway might pose an ethical dilemma for policy-makers, we have demonstrated the principle that delaying can minimise the expected cost. We have also introduced the CSA, a dynamic strategy for determining when to introduce management. As we have described, extensions could include considering multiple possible times at which interventions can be modified, additional complexity in epidemiological data and available controls, and applying the CSA with more complex models appropriate for outbreak data in a real outbreak response scenario. However, we have demonstrated that the underpinning question of whether or not to introduce control as quickly as possible should be carefully considered in future emerging outbreaks. The deterministic SIR model describing the spread of disease in a population of S+I+R = N host farms is dSdt=−βSI, dIdt=βSI−μI, dRdt=μI, in which βI is the per capita rate at which susceptible farms become infected, and μ is the rate at which farms are removed in the absence of control. The variables S, I and R represent the numbers of susceptible, infected and removed host farms in the system at time t. Our simulations use the analogous stochastic model (S4 Text) and are generated using the direct method version of Gillespie' s stochastic simulation algorithm [80]. For the simulated datasets underlying Figs 1 and 3, we choose parameters such that the farm-level basic reproduction number is R0 = βN/μ = 1. 5 (in particular N = 1,000, β = 0. 00015 per day, μ = 0. 1 per day). The variable R0 represents the average number of farms expected to be infected by a single infected farm if all other farms in the system are susceptible. To test the Control Smart Algorithm (CSA) more extensively, we later run simulations for R0 in the range 0–5 by varying β (Fig 4). We note that these values are within the range of estimated farm- or flock- level reproduction numbers for foot-and-mouth disease [81,82], avian influenza [83,84] and bovine tuberculosis [11], although detailed modelling of any particular disease is not the aim of our study. At a variable time, T, we consider introducing a control involving vaccination that completely removes a number of farms from the infection process for the duration of the outbreak. Although we consider vaccination specifically, a control focussed on treating healthy individuals is in keeping with other treatments of infectious diseases including removal of susceptible hosts and culling of hosts close to known infections [6,81]. The total cost of the outbreak is given by Costofepidemic=R (∞) +εQ, where R (∞) is the total number of farms lost to disease during the epidemic, Q is the total number of farms that are vaccinated and ε is the relative cost of control per farm compared to losing that farm to disease. This cost could represent a number of factors, including deploying the control itself [85,86], compensating farmers [87] and impacts on the rural economy and tourism [88]. We assume for simplicity in our main analyses that the infection rate between farms, β, is the only unknown outbreak parameter, and we assume that information on all epidemiological transitions made by all hosts is available. This allows us to generate easily a posterior distribution for the basic reproduction number, R0 (S5 Text). However, we note that in principle, any epidemiological data and/or method of parameter estimation could be integrated into the decision-making framework that we present. This includes extending the approach presented in the main text here to scenarios in which multiple parameters require estimation from disease spread data (S5 Fig) or in which there are informative priors concerning parameter values (S4–S9 Figs). The amount of control to deploy if interventions are to be introduced immediately can be chosen by running forward simulations of the stochastic model with different amounts of control according to the Control Amount Optimisation Algorithm, CAOA (S2 Fig; S1 Algorithm). In each forward simulation, to assess the expected effect of control given the range of transmission parameter values consistent with data up until the current time, parameter values are sampled from the current posterior distribution. By repeating this process for each possible amount of control (i. e. numbers of susceptible individuals to vaccinate, Q), the expected cost for each possible amount of control is estimated. The optimal amount of control to use is that corresponding to the minimum expected cost. The CSA (Fig 2; S2 Algorithm) generates an estimate of the expected cost of an ongoing outbreak if control is introduced immediately (i. e. at t = T), and the expected costs if control is instead introduced at each possible time in future (i. e. at t = T+τ, a delay of τ days before introducing control). If the expected outbreak cost is lower with control deployed now rather than at any time in future, then the decision to control immediately should be taken. If not, then a decision of waiting to learn more and reassessing at the following possible decision time is recommended. Here we present a more detailed description of how, at time t = T, an estimate of the cost of the outbreak with control at time t = T+τ is constructed. First, a posterior distribution for R0 is generated using disease spread data from the real ongoing outbreak up to current time t = T. This is the only time that real data are used in the CSA. Then, a value of R0 is sampled from this posterior and used in model simulations between t = T and t = T+τ. Control is then deployed in the simulation at time T+τ according to the CAOA (S1 Algorithm). However, the amount of control that is deployed in the simulation is calculated using the additional simulated information generated in the simulation between times T and T+τ. In this way the amount of control introduced in the simulation is the amount that would be introduced in reality, if the forward simulation was actually to occur. The simulation is then continued further from time T+τ to generate a single simulated outbreak cost. By repeating this for a large number of simulated outbreaks with different sampled values of R0, the estimated expected outbreak cost with control at time T+τ is generated. An interesting but subtle feature of an optimal control strategy is that it does not necessarily involve performing the intervention that is predicted to be optimal if it is the only one deployed until the end of the outbreak. Instead ideal control depends on the intervention options that will be available in future. For example, the optimal control intervention at the current time will be different if the next opportunity to intervene is near in future or a long time away. Similarly, if control if irreversible, then it might be better to deploy only small amounts of control initially, and intervene more intensely later when improved parameter estimates provide a greater weight of evidence that heavy handed control is likely to be beneficial. A key feature of the CSA is that the optimal future control decisions are programmed into the forward simulations, allowing the best decision accounting for all possible future changes in intervention policy to be taken, rather than simply recommending the control that is likely to be best if unchanged throughout the rest of the outbreak. In S6 Text we present tests showing that the CSA recovers the exact solution that can be derived using dynamic programming in the artificial case in which the values of all parameters are known (see also S10 and S11 Figs). In comparing the CSA with threshold-based strategies to determine when to introduce control (Fig 4), we consider the following strategies. For strategies 2,3 and 4, the amount of control to introduce is optimised in advance of the outbreak (S2 Text).

Infectious disease outbreaks in human, animal and plant populations can have devastating consequences. Where and how much control should be deployed are difficult questions to answer. However, mathematical modelling is increasingly used by policy-makers to underpin these–sometimes contentious–decisions. When to manage disease is often viewed as more straightforward. Surely interventions should be introduced quickly, in an attempt to contain the outbreak as soon as possible? In practice, there are a number of unavoidable uncertainties at the beginning of any outbreak, such as whether or not the initial cases will lead to a major epidemic, as well as uncertainty about the exact transmissibility of the pathogen. We show that waiting to resolve these uncertainties before starting interventions can lead to less costly outbreaks, even though waiting means there might then be a larger outbreak to control. We have developed a novel algorithm that can be used in real-time during emerging outbreaks to decide when and how to introduce control. While for human diseases it might be controversial for policy-makers to recommend delaying interventions, for pathogens of plants or animals such a strategy will often be optimal and ethically justifiable. Our work shows how this approach might be implemented in practice.

Abstract Introduction Results Discussion Materials and methods

medicine and health sciences pathology and laboratory medicine infectious disease epidemiology applied mathematics pathogens immunology simulation and modeling algorithms preventive medicine mathematics infectious disease control vaccination and immunization veterinary science research and analysis methods public and occupational health infectious diseases veterinary diseases epidemiology biology and life sciences physical sciences

2018

Control fast or control smart: When should invading pathogens be controlled?

How does education affect cortical organization? All literate adults possess a region specialized for letter strings, the visual word form area (VWFA), within the mosaic of ventral regions involved in processing other visual categories such as objects, places, faces, or body parts. Therefore, the acquisition of literacy may induce a reorientation of cortical maps towards letters at the expense of other categories such as faces. To test this cortical recycling hypothesis, we studied how the visual cortex of individual children changes during the first months of reading acquisition. Ten 6-year-old children were scanned longitudinally 6 or 7 times with functional magnetic resonance imaging (fMRI) before and throughout the first year of school. Subjects were exposed to a variety of pictures (words, numbers, tools, houses, faces, and bodies) while performing an unrelated target-detection task. Behavioral assessment indicated a sharp rise in grapheme–phoneme knowledge and reading speed in the first trimester of school. Concurrently, voxels specific to written words and digits emerged at the VWFA location. The responses to other categories remained largely stable, although right-hemispheric face-related activity increased in proportion to reading scores. Retrospective examination of the VWFA voxels prior to reading acquisition showed that reading encroaches on voxels that are initially weakly specialized for tools and close to but distinct from those responsive to faces. Remarkably, those voxels appear to keep their initial category selectivity while acquiring an additional and stronger responsivity to words. We propose a revised model of the neuronal recycling process in which new visual categories invade weakly specified cortex while leaving previously stabilized cortical responses unchanged. In both human and nonhuman primates, the ventral visual cortex comprises multiple specialized subregions that are involved in the visual recognition of image categories such as objects, faces, or places [1–5]. What is striking in humans, however, is that this mosaic of specialized regions is partially changed by the culture we live in. The acquisition of musical [6], mathematical [7,8], and reading abilities [9] leads to systematic changes in ventral visual organization. In particular, in all adults who have learned to read—regardless of the writing system—a small region of the left ventral visual cortex within the depth of the left occipitotemporal cortex systematically activates in response to written words [9,10]. This region has been termed the visual word form area (VWFA). Although there is still debate about the exact function of the VWFA, which may vary according to a posterior-to-anterior gradient [11,12], it is widely accepted that the responsivity of this region is an excellent marker of reading acquisition [13]. For instance, a whole-brain comparison of brain activity evoked by letter strings in literate and illiterate adults isolates a specific site at the location of the VWFA, the activation of which is proportional to reading speed [14]. Similarly, the VWFA site appears when comparing functional magnetic resonance imaging (fMRI) images of children who have or have not learned to read, either as a group [15,16] or within the same individual [17–19]. In both children and adults, those changes are accompanied by a massive enhancement and left lateralization of the N170 component of event-related potentials evoked by written words [e. g. 17,20,21–23]. In the present work, we aimed to provide novel longitudinal data on the impact of the acquisition of reading on the representations of other visual categories in the ventral visual cortex. In adults, the VWFA systematically lands at a fixed location relative to a reproducible mosaic of regions partially specialized for objects, faces, bodies, and places [10,14,24–26]. Little is known, however, about the development of this system in young children. In 9-year-old normal readers, the adult mesial-to-lateral gradient of preferred responses to houses, faces, and words is already present, with the expected left–right hemispheric asymmetries for words versus faces [15], but fMRI studies in younger children have underlined the protracted development of this mosaic of regions [27,28]. In particular, a selective response to faces is often difficult to isolate in the fusiform region in early childhood [27,28, but see 29] although a coarse mesial–lateral functional division at the level of the fusiform lobe can be suspected since infanthood [30]. The response to faces increases with age until late adolescence, while the activation for places and tools appears more stable along childhood [27,28,31,32]. A recent study indeed suggests that the face region is more plastic than the place region and continues to exhibit structural changes until adulthood [33]. In this context, it has been hypothesized that the acquisition of reading takes advantage of the preexisting organization and plasticity of this ventral visual cortex [34]. The theory of neuronal recycling proposes that cultural learning takes advantage of the prior organization of the cortex and repurposes some of its circuitry [35]. This could explain why expertise in reading primarily encroaches on the lateral sector of the ventral visual cortex, where late-childhood plasticity is maximal [33]. A combination of constraints, including preexisting connections to language areas [18,36], sensitivity to line junctions [37], and high-resolution representation of fovea shapes [26] would conspire to single out a specific cortical location as the most appropriate for the visual recognition of written words [38]. Reading acquisition would then displace the preexisting mosaic of visual categories in this region, leading to a reorganization that “makes space” for letter knowledge. In support of those ideas, the comparison of literate and illiterate adults [14,23] has revealed that the responses to written words overlap with those responding to objects, faces, and checkerboards and that as reading scores increase, face responses were slightly reduced in the left hemisphere and strongly increased in the right hemisphere. Similarly, in children, Monzalvo et al. [15] further showed that the right lateralization of the activation to faces increased with reading performance. Behavioral and event-related potentials have further supported the notion of a competition of words and faces for cortical space [23,39–41]. One problem, however, is that those studies relied on a comparison of distinct groups of subjects with variable ages and literacy scores. Such group comparisons are necessarily imprecise. Smoothing and intersubject averaging may lead to an apparent overlap between the cortical responses to different categories, even though those regions actually occupy well-delimited territories in individual subjects. Evidence on the development of reading within individual children is simply lacking. In the present study, our primary aim was therefore to obtain enough longitudinal data on a few individual children that they could be submitted to a single-subject analysis. To this aim, we scanned 10 children longitudinally at 6 different times, spread at approximately 2-month intervals before and throughout the first year of schooling (8 of them also came back for a seventh scan 1 year later). Furthermore, to better understand early cortical maps and their reorganization with reading, we presented the children with a broad array of age-appropriate pictures covering the 6 categories of letters, numbers, objects, faces, bodies, and places. These longitudinal single-subject data depicting the evolution of ventral visual responses to words and other visual categories should allow us to clarify the topographical changes, dynamics, and cortical competition underlying reading development, and we aimed to clarify the following questions: how quickly does the VWFA emerge during reading acquisition? Does it immediately land at its usual adult location, or does it move during development? Do we observe a transient invasion of broader cortical territories followed by selective shrinking? Do other categories remain stable, or are they shifted away from the site that becomes specialized for letters? Can the VWFA site be predicted by a prior pattern of specialization for other categories, such as faces? Or, on the contrary, do word-specific voxels fall upon a sector that is initially poorly specified? And what is the relation between the development of face and word responses: are these categories competing for the same resources in a plastic region? This study is part of the project" Etude multimodale en neuropsychologie et imagerie du développement cérébral et cognitif et de ses relations avec la variabilité génétique," approved on June 16,2011, by the ethical committee (CPP Kremlin Bicêtre, N° 11–008). We received ethical permission to scan 10 healthy children at approximately 2-month intervals, from the end of kindergarten through the first year of school (July, September, November, January, March, and June; the school year started 15 September). An extension allowed us to rescan 8 of them for a seventh time at the end of the next year of school (June). The 10 children (5 boys and 5 girls) were selected from an initial sample of 14 children who were scanned at session 1 (July, kindergarten). The purpose of that first session was 2-fold: (1) for the children and their parents to realize what a scanning session was and (2) for us to see which children had trouble staying quiet in the scanner. We then selected the first 10 children who seemed quiet enough and agreed to come back throughout the year. All children and both their parents signed a written consent at the first session. Then, at each session, they were asked whether they agreed to continue. At the first scan, children were aged 6 years 2 months on average (range 5: 7 to 6: 7). We ensured that they possessed little or no reading ability (number of words read in 1 minute = 0 to 7), thus probably selecting children in the lower half of the normal range. None had any known risk factor of reading impairment in their family history: their development was judged normal by their teacher and parents, and they exhibited normal-range performance in verbal and perceptual intelligence quotient (IQ) subtests (WISC IV) and Raven’s Colored Progressive Matrices. We assessed their reading level with 2 tests: “L’alouette, ” a classic standardized French reading test, which consists of reading as fast and accurately as possible a meaningless text of 265 words within 3 minutes [42], and “Lecture en une minute” (LUM), a standardized list of words that children are asked to read as fast as possible for 1 minute [43]. At the end of the first year of schooling, their reading age (“alouette”) was, on average, +1 month (range −9 to +15 months) relative to their civil age, indicating normal development, and this value was −2. 4 months (range −17 to +39) at the end of the second year (Fig 1). The number of correct words read in 1 minute (LUM) was 33. 4 at the end of the first year (range 16 to 54) and 59. 25 at the end of the second year (range 33 to 89), which is within or above the standardized norm at the end of the second year of school, i. e. , 36. 7 (±15. 8) words per minute. We recently received news from 9 of the 10 children. All 9, now in their first year of secondary school (sixth grade), have followed a normal school curriculum without reading difficulties. Just before each fMRI session, the reading level was assessed with a new list of words similar to the standardized LUM list. We also probed knowledge of the grapheme–phoneme code (BATELEM test) and other abilities affected by reading acquisition, including the following: rapid automatic naming of pictures (RAN), verbal short-term memory (forward and backward digit span; sentence span: correct repetition of sentences of increasing length), phonological awareness (EVALEC test, comprising deletion of the first syllable in 10 trisyllabic words, then of the first phoneme in 12 CVC words and in 12 CCV words [44]), vocabulary level (DEN48 test of picture naming [45]), number reading, and number dictation. Finally, to study the effect of reading acquisition on face processing, the face recognition test from NEPSY [46] was evaluated. After having memorized 16 children’s faces, the child had to point to the studied face among 3 faces, across 16 trials. This test was done immediately after the encoding phase and after a 30-minute delay. Except on the first, sixth, and seventh sessions (1 year apart) —for which we used standardized tests—in the intermediate sessions, we constructed equivalent stimulus materials in order to avoid test–retest effects as much as possible. For the fMRI paradigm, subjects were presented with stimuli belonging to the categories of houses, objects, faces, bodies, words, and numbers (see Fig 2 and S2 Fig). Two additional categories of high-frequency and low-frequency grids with variable orientations were also presented. Each category comprised 60 different exemplars, either using black and white pictures (for houses, objects, faces, and bodies) or 4-character strings (for letters and numbers). The words were frequent, regular words encountered by young readers, as specified in Manulex, a lexical database compiling the frequency of occurrence of words in 54 scholarly French reading books [47]. Faces were front views of male and female children’s faces. Bodies were standing male and female adult bodies; to avoid face responses, their head fell outside the picture frame (see an example in Fig 2). Objects were pictures of objects frequently encountered in a child’s daily life (scissors, spoons, shoes, etc.). Six subsets comprising 10 exemplars of each category were created, to be successively used in the 6 scanning sessions. The order of the subsets was different for each child. We used a miniblock design. Blocks of 6 images belonging to the same category were randomly selected and presented during 1 second each, thus forming a 6-second block. Blocks were separated by a variable interblock interval of 2. 4,3. 6, or 4. 8 s (mean of 3. 6 seconds). Thus, a new series of images was presented on average every 9. 6 seconds. The order of the categories was randomly chosen, with the constraint that each category was presented 3 times in a functional run (8 categories × 3 repetitions = 24 blocks of 6 images each). As in our previous work [14,15], we used an easy incidental target-detection task, the sole purpose of which was to maintain attention in all miniblocks, independently of reading or schooling level. Within each block, a target (the picture of the cartoon character Waldo) had a 33% chance to appear, replacing 1 of the 6 images (excluding the first 2 images of the block). Therefore, an average of 8 targets appeared during a run of 24 blocks. Children were instructed to press a button as soon as they detected Waldo. This task was used to keep the child’s attention focused toward the visual stimuli. The total run duration was 3’ 58”. In each fMRI session, 4 runs were acquired, except for the first and sixth sessions, in which only 3 runs were acquired due to the additional tests and sequences proposed to the children in these sessions. In the seventh session, 4 runs were acquired except for 2 children who asked to stop the acquisition after the third run. After training to remain still in a mock scanner (only for the first session), children were brought to the 3T MRI scanner (Siemens Trio). They were protected with noise-protection earphones, and a mirror system above their head allowed them to see the visual stimuli presented on a screen at the end of the tunnel. The images were viewed from a distance of 120 cm with an approximate view angle of 6 degrees. Stimulus presentation and behavioral responses collection were performed using PsychToolbox (http: //psychtoolbox. org), a free Matlab toolbox (MathWorks, Natick, MA, US). In each session, T1 images were first acquired (voxel size = 1 × 1 × 1 mm), then functional images were collected (100 EPI volumes with TR = 2. 4 seconds, TE = 30 millisecond matrix 64 × 64 × 40, voxel size = 3 × 3 × 3 mm in each run). At the first, sixth, and seventh sessions, this sequence was completed by diffusion tensor imaging (DTI) and a second fMRI sequence studying audiovisual representations, which are not reported in the current paper. To reduce head motion, the quality of the MRI images was checked after each sequence acquisition and functional run, and verbal feedback was given to the child. To correct for motion within each run, images were first realigned using the corresponding tool provided by SPM8 (http: //www. fil. ion. ucl. ac. uk/spm/), including both estimation and reslicing steps. The target image was the mean of all images, except if movement during acquisition corrupted the mean image. In that case, the first image was used. S1 Fig shows the average amount of detected movement, computed as the maximum absolute value of the three translation and three rotation parameters provided by SPM. Average movement amounted to a few millimeters in translation and a fraction of a degree in rotation, often due to a sudden movement (e. g. , cough), whereas the child was quiet most of the time. Each fMRI volume was then visually inspected one by one. Using the Matlab toolbox ArtRepair (http: //cibsr. stanford. edu/tools/human-brain-project/artrepair-software. html), images affected by excessive intravolume movement artifacts (stripes, severe shape or size distortion) were replaced by linear interpolation of previous and subsequent images or by nearest-neighbor interpolation when the damaged volume was the first or the last or when several consecutive images were affected [48]. The percentage of rejected images remained low, with the majority of sessions requiring no correction at all (44/68 fMRI runs) or less than 4 images (61/68 = 90%). The mean rejection percentage across the 68 sessions was 0. 77% (standard deviation = 1. 1%; range 0%–16%). Corrected images were then ready to undergo the rest of the preprocessing, i. e. , slice timing, coregistration to the anatomy acquired on the same session, and normalization. For normalization, the T1-weighted anatomical images were first normalized to the standard European adult MNI template. This step segmented the images automatically into different tissue classes (grey matter, white matter, and nonbrain, i. e. , cerebrospinal fluid and skull) using the “New Segmentation” option in SPM8. By averaging those segmented images across all 10 subjects and all 7 sessions, 3 tissue probability maps were obtained. The original T1 images were then normalized a second time, this time using as a target template those average images arising from our own cohort of children. The highly accurate alignment of the 6 or 7 anatomies obtained from the same child was visually verified using the CheckReg tool in SPM8. Finally, the normalization matrix was applied to all EPI images of the corresponding session, with a final resampled voxel size of 2 × 2 × 2 mm. All behavioral variables assessing reading—which were collected on each fMRI session during the first year of school—except RAN and vocabulary showed a significant increase with days-at-school: reading speed (r2 = 64%, p < 10−13), knowledge of grapheme−phoneme relations (r2 = 85%, p < 10−25), metaphonological performances (r2 = 42%, p < 10−7), backward digit span (r2 = 18%, p =. 0004), word span (r2 = 26%, p < 10−4), number and digit reading (respectively, r2 = 14%, p = 0. 003 and 22%, p = 0. 0001), and number and digit dictation (respectively, r2 = 15%, p = 0. 002 and 28%, p < 10−4). Time spent at school affects forward digit span only when the seventh session was included (r2 = 36%, p < 10−7). Fig 1 illustrates this relationship for 2 key variables: knowledge of grapheme−phoneme relations and reading speed (number of words read in 1 minute). Both were low and stable for the first 2 sessions before school, then sharply increased. At the end of first grade, most grapheme-knowledge relationships were mastered by all children, but reading speed remained highly variable. Two regression models were used to attempt to characterize this variability. First, a multiple regression with both days-at-school and age indicated, unsurprisingly, an effect of the former but not the latter (respectively, p < 10−5 and p = 0. 14, r2 = 65% for 6 sessions; and p < 0. 001 and p = 0. 24, r2 = 69% when 7 sessions were considered). Second, another multiple regression examined whether this effect was mediated by the evolution of 4 key cognitive variables: knowledge of grapheme−phoneme relations, metaphonological capacities, RAN, and effortful short-term memory (backward digit span). The results indicated that days-at-school was no longer significant and that knowledge of grapheme−phoneme relations and RAN mostly contributed to reading speed during the first year of school (respectively, p < 10−6 and p = 0. 03, r2 = 81%). Because reading speed thus appears to summarize many aspects of the evolution of reading ability both within and across subjects, this variable was selected as the main regressor to study the evolution of brain activity as in our previous work in adults and dyslexic children [14,15]. Face memory ability also increased with days-at-school during the same period of 2 years (r2 = 26%, p <. 10−4). The same regressions as above indicated that (1) days-at-school was a better predictor when pitted against age (respectively, p = 0. 04 and p = 0. 70, r2 = 24%); and (2) none of the above 4 cognitive variables were better predictors than days-at-school although backward digit span approached significance (p =. 07); and (3) the same conclusion was reached when reading speed was pitted against days-at-school (respectively, p = 0. 88 and p = 0. 02; r2 = 24%). Those results suggest that face recognition may be influenced by schooling yet independently of reading. During fMRI, only a minimal detection task (detection of a specific picture of a cartoon character Waldo) was required. There were no misses, and the mean reaction time (RT) to the target was 978 milliseconds (882 to 1,004 milliseconds across subjects and 933 to 1,001 milliseconds across sessions). The mean RT was stable across sessions (F[6,52] = 1. 7; p > 0. 1). We first examined category-specific activations when pooling across all 7 sessions (Fig 2 and S2 Fig for glass-brain views of the category-specific activations). When compared to the other categories, grids induced larger activation mainly in the calcarine scissures and areas of the dorsal occipitoparietal pathway, probably due to perceived movement induced by the constant change in line orientation (S2A Fig) across the successive images of the block. Because this category was so different from the others, we did not consider this condition further and only compared each category to the mean of the other “pictures” categories (i. e. , tools, words, numbers, faces, bodies, and houses). For these visual categories, a mosaic of specific responses to each category was observed in ventral extrastriate areas, similar to that reported in adults. From medial to lateral, the preference shifted from houses to bodies and objects (Fig 2 and S2 and S3 Figs). Tools activated 2 separated ventral foci as described by Hasson et al. [50] in adults. Faces mainly activated both fusiform gyri and amygdala, plus the right superior temporal sulcus. Interestingly, bodies induced a much larger response than faces in the inferior temporal and posterior middle temporal regions, while the converse effect was seen in the medial occipital region (S2 Fig). Numbers did not elicit any larger response relative to the other visual categories. Words, however, elicited an extended pattern of activation, with significant clusters observed not only in the VWFA but also in posterior temporal sulcus, parietal, and inferior frontal regions, all in the left hemisphere (Fig 2). We also pitted our two symbolic categories—letters and numbers—against each other (S2B Fig). Relative to numbers, words elicited stronger activations in the left frontal (240 vox, pcor < 0. 001, z = 5. 11 at [−52 10 24]), left parietal (115 vox, pcor < 0. 001, z = 4. 64 at [−32 −52 44]), and left fusiform regions (56 vox, pcor < 0. 001, z = 4. 06 at [−40 −56 −8] and z = 4. 00 at [−48 −55 −8]), confirming the presence of an early specialization for letters over numbers at the left VWFA site even at this early age. No significant cluster was observed for the reverse comparison (numbers > words). Second, we examined the changes in those activations across the 7 sessions, spread over the first year of reading acquisition, with an additional measurement at the end of the second year of schooling. Fig 3 shows an example of the evolution of the activation to words and to faces in an individual child (see S3 Fig for the other visual categories). In the group analysis, there was a linear increase with sessions of the activation to words relative to rest in several areas of the left hemisphere: in the anterior part of the fusiform region (VWFA), the occipital area, the posterior superior temporal sulcus, the precentral region, and both inferior parietal areas (see coordinates and Z scores in Table 1). A similar increase with sessions was seen for the activation to numbers relative to rest in the right inferior parietal region (77 vox, pcor < 0. 001, z = 4. 14 at [28 −60 36]) and for houses at [32 −78 32] (z = 4. 16,56 vox, pcor = 0. 002). No significant linear effect was observed for the other categories. No area showed a significantly larger increase to one category than to the others. However, when compared to grids, a specific increase was seen for words in the VWFA (74 vox, pcor < 0. 001, z = 4. 72 at [−50 −58 −14]). Session-by-session comparisons (Fig 2) indicated that the difference between words and other categories was absent from sessions 1 and 2 (before or at the onset of schooling) and first became significant in session 3 for prefrontal and temporal activations or session 4 for the VWFA, i. e. , about 2 or 4 months, respectively, after school onset. We next used regressions to evaluate whether behavioral measures of reading would predict the evolution of brain activity across sessions, independently of age (see Methods and Fig 4). A significant linear effect of reading speed (variable LUM = number of words read in 1 minute) was found on activation to words relative to rest in left occipitotemporal cortex, at around the VWFA site (75 vox, pcor < 0. 001, z = 4. 66 at [−42 −66 −16]), at a more posterior occipital area (86 vox, pcor < 0. 001, z = 4. 72 at [−38 −80 −10]), and in the right cerebellum (101 vox, pcor < 0. 001, z = 4. 26 at [20 −50 −30]). This correlation was stronger for words than for other categories in left occipital (52 vox, pcor = 0. 003, z = 4. 58 at [−42 −78 −12]) and occipitotemporal areas (110 vox, pcor < 0. 001, z = 5. 25 at [−44 −64 −8]), indicating that these activations were reading specific. In addition, interestingly, activation to categories other than words also correlated with reading speed. For faces relative to rest, an increase of activation with reading speed was found in the right-hemispheric fusiform gyrus, at or near the FFA site (48 vox, pcor = 0. 005, z = 6. 08 at [38 −52 −10]). Furthermore, this increase remained significant when contrasting faces versus other categories (78 vox, pcor < 0. 001, z = 6. 67 at [38 −52 −12]). This finding confirms the prior finding in adults (literate versus illiterate [14]) and in children (normal readers versus dyslexic [15]) that reading acquisition correlates with a right-hemispheric shift of face responses. For numbers relative to rest, an increase with reading speed was found in a left occipital area (46 vox, pcor = 0. 01, z = 5. 00 at [−36 −82 −6]), but no difference with other categories was seen. For other categories, there was no significant correlation. Also, in those regressions, no specific increase in activation was found with age or with the number of days at school, for any category relative to all others. The temporal profiles of activation to words presented in Fig 2 (bottom row) suggested that, in addition to monotonic increases, there might also be transient activations in the first months of reading acquisition that later reduced or vanished. To evaluate this possibility, we examined a quadratic contrast across the 7 sessions for word activations. The quadratic effect was significant in several left-hemispheric clusters: inferior frontal gyrus pars opercularis (pcor =. 043, z = 4. 53,33 vox at [−52 8 24]), anterior cingulate (36 vox, pcor =. 026, z = 4. 33 at [−6 12 48]), anterior occipitotemporal sulcus (42 vox, pcor =. 012, z = 4. 27 at [−48 −46 −8]), and anterior insula (34 vox, pcor =. 037, z = 4. 18 at [−30 18 8]). Left inferior parietal cortex was also present at a less stringent false discovery rate (FDR) –corrected threshold (26 vox, pFDRcor =. 025, pFWEcor =. 13, z = 4. 50 at [−26 −66 32]). Only the left parietal cluster remained significant when compared to the other visual categories or to the grids. Overall, the group analysis confirmed that ventral occipitotemporal organization in the present group of 6-year-old children was similar to previous studies of adults. Reading acquisition primarily impacted word and number activations in the left occipitotemporal region. Face- and body-selective responses were seen at the earliest age, and face selectivity increased with reading speed in the right hemisphere. To better understand how these responses were organized, we next examined the voxel-specific evolution of category specificity in each subject. We asked whether the VWFA could be identified in each individual subject using the contrast for activation to words relative to other visual categories, once reading was established (we used sessions 6 and 7 or—in the 2 subjects who could not return for session 7—session 6 only). At our standard threshold (p < 0. 001, cluster–FWE-corrected p < 0. 05), 8 out of 10 subjects showed a significant cluster at, or near, the classical VWFA location in the left occipitotemporal area (see S4 and S5 Figs for individual brain maps and S1 Table for coordinates). Concerning the last 2 children, 1 had a cluster at the correct location but not significant when FWE-corrected (z = 5. 67 at [−50 −62 −16], 40 vox, pcor =. 092, p =. 035 FDRcor), and the other, who was the second worst reader, showed only a significant occipital activation for the same contrast (83 vox, p cor <. 001, z = 5. 88 at [−16 −94 −10]) and a very small peak at the VWFA site (z = 3. 39,6 vox at [−46 −56 0]). To further examine individual responses to the different visual categories and how they were modified by reading acquisition, we restricted the next analyses to subject-specific masks encompassing the fusiform region in the left and right hemisphere (Fig 5). First, we asked whether there was a systematic expansion in the number of category-specific voxels, both for reading and for other categories, during the course of reading acquisition. We therefore computed, for each subject and each session, the number of voxels significantly more activated by a given category than by the others (p < 0. 001; hereafter called “specific voxels” for short). We entered these data into an ANOVA with factors of category (6 levels), sessions (1–6), and hemisphere (left and right). The results revealed a main effect of sessions (F[5,45] = 5. 6, p <. 001), category (F[5,45] = 9. 0, p <. 001), hemisphere (F[1,9] = 10. 1, p =. 011), and an interaction session × category × hemisphere (F[25,225] = 2. 1, p =. 002), which reflected a significant session × category interaction in the left but not the right fusiform region (Left: F[25,225] = 1. 7, p =. 024; Right: F[25,225] <1). Not surprisingly, only words showed a significant change in the number of selective voxels across sessions in the left hemisphere (F[5,45] = 5. 2, p <. 001), reflecting a sudden increase after session 2, i. e. , the start of school. No other visual categories showed a significant change of volume over the 6 sessions (Fig 5), in either hemisphere. The only other significant effects concerned hemispheric differences: there was a left-hemispheric asymmetry for words (F[1,9] = 16. 6, p =. 003) and tools (F[1,9] = 32, p <. 001), whereas bodies (F[1,9] = 48, p <. 001), houses (F[1,9] = 14, p =. 005), and faces (F[1,9] = 12. 6, p =. 006) induced larger volumes of activation in the right hemisphere. No effect of hemisphere was seen for numbers (F[1,9] < 1). We then asked whether the emergence of the VWFA modified the location of the responses to other categories. We therefore computed, for each image category and for each session, the distance (in mm) of its activation barycenter from the subject’s mean VWFA barycenter determined in sessions 6 and 7 (see Methods) and entered this distance into an ANOVA with category (4 levels, excluding words and numbers) and sessions (1–6). Only the main effect of category was significant (F[3,27] = 6. 61, p =. 002), reflecting the obvious differences in locations of the activations for each visual category. There was no linear effect of session (F[1,9] < 1), no session × category interaction (F[3,27] < 1), and no effect of session within each category indicating an absence of change in the position of activations relative to the VWFA. This was also the case when we considered the peaks of activation of houses, bodies, and tools relative to the left and right FFA barycenter, suggesting that the peaks of activation of the visual categories were not displaced by the development of the VWFA. Therefore, the development of the activation to words in the left fusiform region during the first year of school did not affect the volume or location of the activation to other categories. We next asked whether the voxels that become specific for written words in sessions 6 and 7 could already be distinguished by a particular response profile in previous sessions (1 to 5) and notably before reading acquisition. Within our ventral mask, we therefore distinguished RR and nRR voxels, using a voxelwise threshold of p < 0. 001 on the words > other visual categories in sessions 6 and 7. On average, we recovered 172 RR voxels (range 39–388 voxels), i. e. , 1,376 mm3 representing an average 3. 8% of the search volume (range 0. 87%–8. 57%). We could then go back in time and examine the properties that distinguished RR voxels from nRR voxels in the preceding sessions (Fig 6, left panel). First, were these voxels already specific for words before school, or did a selectivity for words emerge only after the onset of reading acquisition? A comparison of the mean activation evoked by words versus other categories—in RR versus nRR voxels—revealed no significant difference in session 1, approximately 2 to 3 months before the start of school (t[9] < 1). There was a marginal effect in session 2, which occurred around school onset (t[9] = 2. 23, p =. 053). A clear difference emerged in session 3 (t[9] = 2. 73, p =. 023), session 4 (t[9] = 2. 40, p =. 040), and session 5 (t[9] = 3. 46, p =. 007). Similar findings were found when analyzing only the RR voxels for a difference in responsivity to words versus other categories (S7 Fig, left panel). Those findings show that, even with a sensitive analysis targeted precisely at subject-specific voxels that ultimately become RR, we could not identify an early responsivity to words prior to schooling. In this respect, words behaved differently from other categories: when the same analysis was replicated with voxels specific for other categories of visual images in sessions 6 and 7 (294 voxels sensitive to tools, 167 to houses, 150 to faces, and 212 to bodies in the left hemisphere), these voxels were already specific in sessions 1 to 5 (Fig 6 and S6 Fig, all p <. 003). Even for numbers—although a reduced set of voxels showed a preference for numbers than for other categories in sessions 6 and 7 (mean = 22 voxels, range = 1–63) —a difference in activation to numbers versus other categories was already found in all previous sessions except session 2 (p = 0. 13, for all other sessions, p between 0. 025 and <. 001). Therefore, only for reading did we see the emergence of a novel cortical selectivity during the first year of schooling (which, in France, focuses almost entirely on reading acquisition). Second, we asked whether the ultimate preference of RR voxels for words could be anticipated by an early preference of those voxels for another category—thus testing, at the single-subject level, previous suggestions that reading might specifically encroach, for instance, on face-related circuits [14,15,51]. To this aim, we computed the average activity evoked by each of the 4 nonsymbolic image categories (tools, houses, faces, and bodies) relative to rest in the RR voxels and in each of the 5 initial sessions. We submitted these data to an ANOVA with categories and sessions as within-subject factors. A main effect of category was present (F[3,27] = 6. 91, p =. 001), indicating that RR voxels responded to other categories in the following order: tools (mean activation = 0. 51), bodies (0. 32), houses (0. 11), and faces (0. 09). Posthoc paired t tests analyzing each possible pair using Holm correction for multiple comparisons revealed that all pairwise comparisons were significant (ps ≤. 01) except faces versus houses, which had a similar weak level of activation in RR voxels (p >. 1), and bodies and tools (p =. 097), which both produced a larger response in these voxels. Those results indicate that RR voxels showed an initial response to tools and bodies. Importantly, there was no interaction with sessions (F[3,27] < 1), again indicating that those biases did not change during reading acquisition. Another important question, however, is whether such preferences are sufficiently unique to RR voxels that they would suffice to determine which voxels ultimately become specialized for reading. We operationalized this question by asking whether RR voxels differed from nRR voxels in their profile of responsivity to nonletter categories (Fig 6). Interestingly, RR voxels, relative to nRR, preferred numbers over other categories early on (Fig 6, left panel, light blue curve). This difference was not significant in sessions 1 (t = 2. 25) and 2 (t < 1) but became clear from sessions 3 to 7 (t[9] = 3. 10,3. 67,2. 70,2. 6, and 4. 62, respectively), suggesting that the development of letter responsivity was accompanied by an additional responsivity to numbers. With respect to other image categories, however, the responses of RR voxels did not differ from those of nRR voxels. Only a small preference for tools tended to be present in RR voxels more than in nRR voxels in session 1 and 5 (t[9] = 2. 72, p =. 024 and t[9] = 2. 27, p =. 049). Therefore, on average, RR voxels were not especially distinguished from nRR voxels in their initial commitment to a specific category, with the possible exception of a small bias for tools. A related question is whether RR voxels are simply less specialized overall. According to this hypothesis, reading would “land” in voxels that are not already strongly committed to a particular visual category and are therefore available for learning. To evaluate this possibility, for each voxel, we also calculated the F-test quantifying any difference between the 4 nonsymbolic image categories (tools, houses, faces, and bodies). We then examined whether this F-test differed for RR versus nRR voxels. Again, no significant difference was found in any session (ps >. 05), indicating that RR voxels were not particularly distinguished by a reduced initial commitment. We did find, however, a small but significant interaction with a linear contrast over sessions 1 through 7 (F[1,9] = 7. 26, p =. 025), indicating another type of difference between RR and nRR voxels: nRR voxels exhibited a significant linear increase of the F-test, indicating that their selectivity for nonsymbolic image categories increased (F[1,9] = 15. 1, p =. 004), while this was not true for RR voxels (F[1,9] = 2. 5, p =. 15). Those findings led us to ask, symmetrically, what were the initial preferences and temporal evolution of voxels that ultimately preferred nRR categories. When we selected voxels that ultimately preferred tools, faces, houses, bodies, or numbers over the other categories, we found that their preferences were temporally stable (unlike what was found for words): as shown in Fig 6, when going back in time, these voxels already exhibited a strong and significant selectivity for their preferred category in the first scanning session and a corresponding lack of responsivity to their nonpreferred categories. For instance, house-responsive voxels in sessions 6 and 7 showed a strong preference for houses in sessions 1 and 2 (ps <. 001), accompanied by a mild response to tools (ps >. 05) and a significantly smaller response to faces, bodies, words, and numbers (ps <. 035) compared to the responses found in non–house-responsive voxels. These preferences were therefore entrenched early on and remained stable or slightly increased over time (Fig 6), as confirmed by the above F-test. Most importantly, voxels selective for tools, houses, faces, or bodies systematically showed an early negative responsivity to words relative to other categories (Fig 6, green curve, ps <. 05). This finding implies that, if a voxel was strongly selective for a category other than words, it tended to be less responsive to written words than other voxels—and this antipreference remained stable over the time course of early reading acquisition. The above conclusions remained when we performed a number of variants of the above analysis. Namely, the stability of visual preferences in nRR voxels was observed when we selected category-selective voxels in sessions 1 and 2 and evaluated their responses in sessions 3 through 7 (S7 Fig), when we considered the entire set of voxels within anatomically defined regions: fusiform gyrus (FG) 1 and FG2 [52], and when we performed the same analyses in the right hemisphere. Averaging over an entire set of RR voxels, as we did in the above analyses, may mask the presence of fine-grained activity patterns that are specific to a given subject or a given category [53,54]. In a final analysis, we therefore used multivariate pattern analyses in order to quantify the stability and the evolution of subject-specific activation patterns within the entire ventral mask and, more specifically, within the VWFA. We first performed this analysis over the entire ventral mask and probed the reproducibility of the pattern of activation between one session and the next (i. e. , over a delay of approximately 2 months) for each category relative to rest (S8 Fig). We performed an ANOVA on the correlation coefficients within category (e. g. , correlation between the vectors of activity for faces in 2 successive sessions) versus across categories (e. g. , correlation between the vector for faces at session n and for houses at session n + 1) over the 6 sessions of the first school year. As shown in Fig 7A, the within-category correlation was systematically positive (ranging around approximately 0. 5–0. 6) and systematically higher than the between-category correlation, indicating a robust replicability of the category-specific activation patterns across successive scanning sessions (left hemisphere: F[1,9] = 58; right: F[1,9] = 49, ps <. 001). This result was observed for each category in both hemispheres (ps <. 007), with the sole exception of words in the right hemisphere (F[1. 9] = 3. 35, p =. 1). Interestingly, the activation pattern for bodies (p =. 004) and faces (p =. 02) was significantly more reproducible in the right hemisphere than in the left, whereas the reverse was true for words and tools (ps =. 008). Therefore, this multivariate analysis revealed that activity patterns could be reliably measured in our young children for all categories. We next examined how those patterns evolved over sessions (Fig 7B). For nonsymbolic images, this reliability was stable over time, whereas for words (and numbers), the reliability increased between sessions 3 and 4. This conclusion was confirmed by an interaction between category, correlation type, and a linear contrast for sessions (F[5,45] = 2. 6, p =. 035) in the left ventral visual cortex, due to an interaction between sessions × correlation type for words (F[1,9] = 6. 9, p =. 028) and numbers (F[1,9] = 6. 99, p =. 027) but not for the other categories, bodies, houses, faces, and tools (all F[1,9] < 1). Armed with this analysis, we could next ask whether the VWFA itself is a site that is initially devoid of category-specific patterns or, on the contrary, whether it is already traversed by reproducible activations for visual categories other than words; and if the latter is true, whether those activations change over time. We therefore performed the same analyses as above on voxels restricted to the VWFA (RR voxels defined on sessions 6 and 7). Unsurprisingly, a reproducible pattern of activation was found for words (F[1,9] = 15. 1, p =. 004) and numbers (F[1,9] = 11. 4, p =. 008) but also, remarkably, for nonsymbolic images (F[3,27] = 27. 7, p <. 001) mainly due to tools (F[1,9] = 31, p <. 001) and to a lesser degree bodies (F[1,9] = 10. 8, p =. 009; no significant effect for houses and faces, ps >. 1). Importantly, the pattern reliability for tools was present in sessions 1 and 2 (F[1,9] = 12, p =. 007) prior to reading and remained stable over time (interaction with sessions, F < 1). Therefore, RR-related voxels contain reliable multivariate activity patterns for visual categories other than words, and the emergence of reliable activation to words (Fig 7D) in the course of reading acquisition did not occur at the expense of other categories. Similar results were observed for the other category-specific ROIs (defined on sessions 6 and 7). Most importantly, the pattern of activation for words was reproducible in the tool ROIs (left and right, ps <. 001), house ROIs (left: p =. 011; right: p =. 005), left FFA (p <. 001), and left-body ROI (p =. 02) (Fig 7E and 7F). Therefore, activation patterns were distributed beyond the strict boundaries of their significant category-selective clusters [53]. As discussed below, those results are compatible with a superposition principle according to which visual categories are encoded by overlapping activity patterns over the same voxels. In 10 individual children, we visualized the emergence of reading circuits by reproducibly scanning them with a battery of visual stimuli during, and 1 year after, the first year of schooling. Prior to schooling, the VWFA could not be detected, although selectivity for faces, houses, bodies, or tools was clearly present. Within the first 2 to 4 months of schooling, the VWFA emerged at the group level and in 8 out of 10 individual children (in the remaining 2, activation was present and significant at the voxel level but not after correction for multiple comparisons at the whole-brain level). The VWFA immediately appeared at its adult location, with a subject-specific topography that was stable over time. An inverted-U curve indicated that—in most of the reading circuit, including anterior VWFA and posterior parietal cortex—the onset of schooling was associated with a peak of activation that then slightly receded over the following months. Most importantly for theories of reading acquisition, the data allowed us to retrospectively examine what the VWFA voxels were responding to prior to reading acquisition. Reading did not recruit voxels that were selective for faces but systematically encroached on slightly more lateral sectors of cortex, within the left occipitotemporal sulcus, in a region that was not strongly selective to any category but did respond more to tools than to other visual categories. The emergence of the VWFA occurred without radically altering the preferences or the topographical organization of ventral visual responses to faces, bodies, houses, or tools, although at the group level we detected a positive correlation between reading performance and the amount of right fusiform activation to faces. Those results suggest that the VWFA emerges quickly, at a fixed and constrained location within a well-organized mosaic of preferences, and by superimposing itself onto existing preferences while minimally altering them. These points, which constrain theories of brain development, will now be discussed in turn. The first question that our study aimed to answer is how quickly the VWFA emerges in children during the course of reading acquisition. In our subjects, the VWFA was undetectable prior to schooling: in our first 2 scans, we observed no specific response to letter strings relative to other visual categories. This is a negative result and should therefore be taken cautiously, but it is strengthened by the fact that we could easily detect reliable preferences for all nonsymbolic visual categories (faces, houses, bodies, and tools). The absence of the VWFA could be due to the fact that we selected children with very little prior knowledge of letters (Fig 1) in order to control their exposition to letters and reading when they entered school. Other studies show that preschoolers who possess some knowledge of letters already exhibit letter-specific steady-state responses that are detectable in a few minutes of electroencephalography (EEG) recording [22]. Our results suggest that, by 2 to 4 months after school onset, a change in the volume of selective activation to words is already detectable (Fig 5) and is associated with the establishment of a stable pattern of subject-specific activation for words relative to other categories (Fig 7). This finding is entirely coherent with another longitudinal study [17] in which left occipitotemporal responses emerged in fMRI and event-related potentials once preschoolers had been exposed to a few weeks of training with the GraphoGame, a software game that teaches grapheme−phoneme correspondences. Lochy et al. (2016) and Maurer et al. [21,55,56] likewise observed a rapid growth of left-lateralized event-related potentials evoked by letter strings (N170) in the course of reading acquisition, both in children and in adults acquiring a new script. Our results are compatible with those prior findings in as much as they show that VWFA responses emerge within the first few months of reading acquisition. Although reading acquisition was necessarily confounded with age in the present within-subject study, we still observed an effect of reading speed on the VWFA activation even when the effect of age was taken into account (Fig 4A). In other between-subject studies, the 2 variables have been clearly decorrelated, and the results indicate that age alone does not suffice to induce the observed changes. For an equal age of approximately 6 years, children who have already learned to read by themselves prior to formal schooling already exhibit a specialized VWFA, contrary to those who have not [57]. Even adults, in the absence of schooling, do not develop a specific ventral visual response to letter strings relative to other visual categories [14,23]. The remarkable speed with which word-specific responses emerged in 6-year-old children fits with other findings indicating a high degree of plasticity in children’s ventral visual cortex. Indeed, the fusiform cytoarchitectonic area FG2, where the FFA and VWFA are located, shows a prolonged development and does not fully mature until late childhood, in contrast with the nearby area FG1 [33]. Its plasticity supports not only reading acquisition but also other forms of visual expertise. For instance, the acquisition of music reading, starting at 3 to 4 years of age, has been shown to induce a large activation of lateral ventral temporal cortex to printed music in professional musicians and to induce a displacement of the nearby VWFA [6]. In adults, fusiform plasticity is sufficient to acquire a novel expertise for birds and cars [58], greebles [59,60], children’s faces in teachers [61], or Braille reading in nonblind adults [62]. Its plasticity might be reduced in adults compared to 6-year-olds. In exilliterate subjects who learned to read as adults, the VWFA is present but only moderately activated in proportion to reading speed, and written words induce a coarser pattern of activation extending into bilateral ventral visual regions [14]. In a recent single-case study [63], we used longitudinal fMRI in a single illiterate subject to follow the trajectory of adult literacy acquisition over the course of 2 years. Unlike the steep nonlinear increase shown here in children, this person’s VWFA emerged slowly and with a continuous increase over a period of several months, paralleling a slow increase in behavioral reading ability. Therefore, the fast changes induced by reading acquisition that we observed here betray the intense plasticity of children’s immature ventral occipitotemporal cortex. This conclusion is also confirmed by a recent animal model of symbol recognition that shows that, while both juvenile and adult monkeys can acquire a behavioral ability to recognize Arabic numerals and letters, only the juveniles develop a category-specific response to such symbols in inferotemporal cortex [64]. Most previous studies of reading acquisition have either been cross-sectional [14,23,65–67] or longitudinal with few data points spread over a long time period, typically 6 months to several years [19,21,68,69]. Here, we aimed to clarify the developmental trajectory of the VWFA and other areas of the reading circuit by obtaining detailed single-subject measurements spaced every 2 months around the onset of literacy. Theories of child development have contrasted several accounts of the emergence of functional brain specialization (reviewed in [70]): according to a maturational account, the adult reading circuit should progressively emerge at its normal location, with increasing size and intensity as increasingly larger populations of neurons become tuned to written words. An alternative account, the interactive specialization model, predicts that the reading circuit should be initially disorganized, diffuse, distributed over broad cortical territories, and that it should dynamically change in a stochastic manner until it settles into a minimal set of regions the interactions of which optimally fit the task. Finally, the skill-learning model postulates an initial phase of effortful processing, relying on nonspecific prefrontal and parietal areas, followed by a progressive reduction of this activity and its transfer to specialized posterior areas. Our data show features of both the maturational and the skill-learning models. On the one hand, the VWFA landed immediately at its adult-like location: just 2 months after school onset, it was already detectable, at its final location (e. g. , Fig 3), and with a fine-grained topographic pattern that showed little or no change over the next 1. 5 years (Fig 7D). Those findings argue against a stochastic search or dynamic change in interregional interactions and instead suggest the rapid maturation of a highly constrained brain circuit (the nature of those constraints is discussed below). On the other hand, several regions of the reading circuit showed an inverted-U activation as a function of time, peaking around the first year of schooling (Figs 4 and 5) and then decreasing steadily. In fact, activation at several parietal and inferior frontal sites eventually decreased down to an undetectable level (Fig 3). In the left occipitotemporal cortex, the volume of RR activation increased sharply and then was nearly halved (Fig 5C). This volume change arose because activation in the posterior part of the VWFA followed an inverted-U pattern (Fig 2), while activation at the peak of the VWFA proper steadily increased in parallel to behavior (Figs 2 and 6). In other words, there was a progressive concentration of activation in a specific ventral area and a progressive disengagement of parietoprefrontal areas, in agreement with the skill-learning model. This developmental pattern fits with previous evidence that reading progressively switches from a slow, serial, effortful mode to a fast, parallel, efficient mode. The posterior parietal cortex, in particular, participates in a top-down attention circuit that has been implicated in letter-by-letter reading, both in adult patients with pure alexia following lesion at or near the VWFA [71] and in normal adult readers confronted with unusual formats such as vertical or s t r e t c h e d words [72]. Its strong activation early on in literacy acquisition is therefore congruent with the classic observation that young readers show a strong effect of word length on reading time [73,74]. This effect progressively vanishes as reading becomes automatized and may therefore explain the inverted-U profile of parietal activity that we observed. The present data thus confirm the existence of a parietal circuit for effortful reading and underline its transient but important contribution during the early phase of reading acquisition. It fits with the suggestion that some children with developmental reading deficits suffer from an “attentional” subtype of dyslexia that prevents them from focusing on each letter in a sequential left-to-right order and avoiding illusions, conjunctions, and letter migrations from previous words [75–78]. In the future, it would be interesting to use the present fMRI task with subtypes of dyslexic children and examine whether the attentional subtype shows disorganized parietal activation while the more classical phonological subtype shows left temporal impairments, as shown in Peyrin et al. [79]. Another goal of our study was to examine whether a specific profile of functional brain activity predicts which voxels are going to become specialized for written words during reading acquisition. By identifying the VWFA approximately 1. 5 years after the onset of schooling and then examining its activity in prior fMRI sessions, we could ask this question within individual subjects at a single-voxel resolution. This is an important improvement over our previous between-subject study of literacy [14], in which intersubject averaging could have caused an artificial overlap between unrelated areas. The results indicated that, prior to schooling, the voxels that will ultimately become the VWFA are not very strongly specialized for faces, bodies, houses, or tools (Fig 6 and S6 Fig). They are clearly unresponsive to faces and show only a modest response to tools, clearly smaller than the peak response found at a slightly posterior site. Therefore, the VWFA lands at an initially relatively uncommitted cortical site, the main functional characteristic of which seems to be a low responsivity to pictures. This finding suggests that ventral visual cortex becomes progressively committed, first to visual categories such as faces and places [30] and later to cultural acquisitions such as letters and numbers and that the latter wave of cortical specialization is constrained to sites that were left partially or totally uncommitted by the first wave. Additionally, the present findings confirm that the VWFA emerges at a systematic location relative to other functional landmarks, i. e. , lateral to the fusiform face responses (as previously reported in [80]) and overlapping with the most anterior part of the lateral object-responsive cortex (Fig 2). Because the VWFA location is so reproducible, factors other than the mere lack of commitment to other categories are likely to play an important role in its spatial delineation [38]. Candidate factors that have been identified in other studies include (1) cytoarchitectony and extended plasticity [33], (2) preferred connectivity to a dedicated set of left-hemispheric targets that include classical spoken language areas [18,36,81], (3) high-resolution foveal bias [26], (4) preference for high spatial frequencies [82], (5) preference for line junctions that are characteristic of letter shapes [37], and (6) efficient detection of invariant shape features [38,83]. Animal studies in which juvenile monkeys were trained to recognize letters, cartoon faces, or tetris-like pentominos indicate that each of these categories landed at a dedicated site that was reproducible across monkeys, suggesting the existence of a biased proto-organization prior to learning [84]. Human adults are able to learn a new script based on human faces and activate the left, not the right, fusiform region for the facefont words, confirming the strong left bias for phonetic-based script, but they were less efficient than the group trained with Korean fonts, and the location of the activation was slightly displaced in an anteromedial position relative to the location of the VWFA [85]. A third goal of our study was to probe the organization of ventral visual cortex prior to reading and to examine how it is changed by literacy. As a selectivity for words emerges in occipitotemporal cortex, does it lead to a reorganization of the responses to other categories? By 5 to 6 years of age, our results indicate that the mosaic of specialization for visual categories that has been reported in adults [4,86] is already largely in place and changes very little in the course of almost 2 years. There was an initial controversy as to whether the cortical representation of faces—as opposed to that of objects and houses/landscapes—develops relatively late in childhood [27,28] or, on the contrary, is present early on [29,30]. Our results concur with the latter studies in showing that, by the age of 5, category selectivity is already deeply entrenched. In fact, a recent study showed a clear difference in responses to faces versus places in 3- to 8-month-old infants [30], and this was corroborated by a longitudinal fMRI study of juvenile monkeys [87]. In both studies, however, the selectivity was initially imperfect, for instance, showing no clear selectivity for faces over objects in human infants, and the longitudinal study in monkeys indicated that highly selective face patches emerged during the first year of life [87]. Once these regions are in place, do they change with reading acquisition? Our results uncover that, in fact, they are remarkably stable over time. Word selectivity appears against a background of largely unchanged responsivity in nearby voxels specialized for faces, bodies, tools, or places (Figs 2 and 6). Even within the same voxels that ultimately become the VWFA, multivariate responses to tools stay essentially unchanged during literacy acquisition (Fig 7). The results therefore strongly question the hypothesis that words and faces share the very same cortical circuitry [51,88]. There is, in fact, much evidence that face and word recognition engage nearby but systematically distinct cortical sectors in normal adult readers [14,80]. Neuropsychological dissociations have also been reported between both domains, both in adult subjects [89–92] and in developmental cases [93,94]. The present study is, we believe, the first to directly demonstrate that orthographic representations develop in ventral visual cortex without encroaching directly on preexisting face cortex but rather by invading nearby but more lateral and relatively uncommitted ventral visual cortex. The results are, however, compatible with a broader view of cortical competition and neuronal recycling [35]. According to this model, word and face recognition emerge at distinct cortical locations, possibly because of their distinct connections to distant areas [18,36], but they both rely on a similar hierarchical architecture common to all ventral visual cortex and therefore compete for expansion into the same overall region of cortex during development [14,35]. During the acquisition of literacy, the VWFA emerges at a site just lateral to the FFA, thus blocking the further growth of face responses in the left hemisphere and forcing them to preferentially develop in a right-lateralized manner [14,15]. Several experimental results fit with this view. First, Golarai et al. [28] observed that, across development, the peak FFA activation evoked by faces does not change with age in either hemisphere. What increases with age is the activation evoked by faces in peripheral voxels within concentric shells that center on the FFA peak—the FFA slowly expands. Second, in a previous study of adult literacy [14], we found that, in adults with variable degrees of schooling and literacy, greater reading scores were associated with a small decrease of face responses at the usual left-hemispheric site of the VWFA. It is crucial to remember that this finding was based on a group comparison and therefore does not contradict the present within-subject results. Finer-grained analysis, inspired by Golarai et al. [28], in fact revealed that the peak response to faces was unaffected by literacy: it was only in voxels that lay at a distance of 8 or 12 mm from the FFA peak that a difference between literates and illiterates was observed (Figure S6 in [14]), again suggesting that cortical competition occurs at the periphery, where literacy shifts cortical boundaries. Third and finally, our adult study also showed a highly significant increase in right-lateralized fusiform activation to faces as the literacy score increased, suggesting that literacy competed with the development of face responses in the left hemisphere [14]. This finding was also observed in 9-year-old normal readers compared to dyslexics [15] and replicated in the present study (Fig 4). It is corroborated by several behavioral and event-related potential studies (for a review, see [13,23,40,41]). Our study also probed the development of responses to written Arabic numerals. We expected to observe the emergence of number-related responses lateral to the VWFA, at a site known as the visual number form area (VNFA) and more responsive to digits than to letters in educated adults [7,8, 95]. The results, however, were weak. There was no evidence for the emergence of any region specialized for numbers. Instead, we found that ventral visual voxels responsive to numbers were also responsive to letters (Fig 6, S3 and S6 Figs). Reading speed correlated with an increased fMRI response to numbers but only at an occipital site posterior to the VWFA (Fig 4). The fact that we observed no clear separation between number-form and letter-form areas, as it was also observed in younger 4-year-old children in Cantlon et al. [29], nor any intraparietal activation specific to numbers is compatible with the hypothesis that, at a young age, digits are primarily processed as readable letter-like symbols but not as meaningful quantities. At least, such quantity processing does not seem to be automatically engaged when the task does not require any calculation but a mere intruder detection (“find Waldo”) as was the case here. Furthermore, we used 4-digit numbers (matched in length to the words) that probably challenged children’s quantity comprehension. Our results are fully compatible with behavioral studies of the number-size interference effect, which indicates an absence of automatic quantity processing in first graders [96]. Finally, beyond the overlap of numbers and letters, our results on visual categories clearly indicate that, in our children, category selectivity is not absolute. Even within subject-specific voxels that were defined by their selective response to one category relative to all others, there were highly detectable differences between the other nonpreferred categories, both in average activation (Fig 6 and S6 Fig) and in multivariate activation pattern (Fig 7 and S8 Fig). This result, which was first observed in adults [53], suggests that at the relatively modest resolution used here (3 mm isotropic), neural responses to different visual categories are spatially intermixed. It does not, however, exclude that entirely selective patches would emerge in higher-resolution scans, particularly for faces [97]. Fig 8 presents a schematic model of the emergence of ventral occipitotemporal organization in relation to literacy. This model is meant as a graphic summary of our findings and an illustration capable of accounting for the present and several earlier observations within a single framework [14,28–30,53,98–101]. The model starts with the observation that, prior to schooling, category selectivity for faces, houses, and tools is already present (as well as responses to bodies, which are omitted from the Figure for simplicity). Such macroscopic category-selective fMRI responses probably arise from the existence of columns or patches of neurons responsive to each category and/or its visual features and clustered in a reproducible sector of cortex. Indeed, the existence of such dedicated cortical patches was directly demonstrated in monkeys (e. g. , [97,99,101]). Developmental fMRI in humans [27–30] and monkeys [87] suggests that such category selectivity emerges within the first year of life and progressively expands in subsequent years. Therefore, we postulate that, prior to reading, the child’s ventral visual cortex comprises a mosaic of specialized sites as well as other more labile sites (respectively appearing in color and in gray in Fig 8). The key result of the present study is that, with education to literacy, responses to written words emerge at a fixed location within this mosaic (the VWFA), at initially weakly specialized sites, and without altering their (weak) preexisting responsivity to other images. This is captured in Fig 8 by assuming that words invade the patches that were left labile at the time of schooling and lie near those already activated by objects and faces. Within the same fMRI voxel, we may therefore find patches of neurons selective for words and for other images, particularly objects. This “superposition principle” may explain why, throughout the emergence of the VWFA, across 7 scans, we continuously remain able to decode a stable representation of objects within the same voxels (Fig 7B and 7D). For simplicity of illustration, the superposition principle is illustrated in Fig 8 at the level of adjacent patches, each specializing for a given category. However, superposition may also exist within the same populations of neurons. Neurophysiological studies have revealed neuronal vector codes whereby the same neurons may be engaged in the simultaneous coding of several features of the stimuli along orthogonal vector dimensions [102,103], compatible with the hypothesis that a given neural population may contain multiple superimposed or multiplexed codes in a high-dimensional space [86,104,105]. According to this view, word responsivity may emerge through the progressive differentiation of new principal axes within a preexisting neural population, without altering its pattern of responsivity to other categories. Future studies could use high-resolution fMRI (<1 mm) to attempt to separate the two models of cortical specialization (dedicated patches versus overlapping vectors). Fig 8 also illustrates why, in the course of the development, superposition may give way to cortical competition. In literate children, the progressive dedication of an increasing number of neuronal patches to written words progressively prevents the expansion of the nearby object and face patches, which occurs in illiterates in the absence of schooling. The model nicely explains how (1) the VWFA emerges at a cortical site that is unresponsive to faces in young children; and yet (2) at a later age, this site shows a greater response to faces in illiterates than in literate subjects [14] and in dyslexics who are not able to acquire fluent reading than in normal 9-year-old readers [15]. Our data indicate a remarkable degree of stability and reproducible order in the organization of children’s ventral visual cortex in at least 2 respects. First, prior to reading, the ventral mosaic of preferences for faces, houses, tools, and bodies is already in place and reproducible across individuals; and second, during reading acquisition, word responses quickly emerge at a reproducible location distinct from those already committed to other preexisting categories. Such reproducibility implies that the architecture of human ventral visual cortex must be strongly preconstrained, and the precise identification of those constraints, whether cytoarchitectural or connectional, is an important goal for further research. Given such constraints, it should perhaps not be surprising that a novel visual category as different as letter strings, which involves specific computational and connectivity requirements, lands at its own distinct location. We may also reverse the reasoning and wonder whether the shapes of the most successful cultural objects, such as letters and numbers, were selected across cultural evolution in order to be quickly learnable [34]. If so, one constraint might have been to avoid direct cortical overlap with any preestablished categories such as faces or tools (see [85] for a nice example in adults trained to a new script). Therefore, biological constraints may explain why the evolution of scripts systematically involves a progressive abstraction away from iconicity and towards abstract shapes (e. g. , from hieroglyphic to demotic in ancient Egypt). The success of education might also rely on the right timing to benefit from the highest neural plasticity. Our results might also explain why numerous academic curricula, even in ancient civilizations [106], propose to teach reading around 7 years.

Reading acquisition is a major landmark in child development. We examined how it changes the child’s brain. Ten young children were scanned repeatedly, once every 2 months, before, during, and after their first year of school. In the scanner, they watched images of faces, tools, bodies, houses, numbers, and letters while searching for a picture of “Waldo. ” As soon as they started to acquire reading skills, a specific region of the visual cortex of the left hemisphere—called the visual word form area (VWFA) —started to selectively respond to written words. In every child, it was then possible to go backward in time and ask what this region was doing prior to reading. We found that written words invaded a sector of visual cortex that was initially weakly specialized, slightly responsive to pictures of tools, and that lay next to a face-selective region. Reading acquisition did not displace those initial responses but blocked their development, such that face-selective responses became stronger in the right hemisphere. Those results provide direct evidence for how education recycles the human brain by repurposing some visual regions towards the shapes of letters.

Abstract Introduction Methods Results Discussion

children medicine and health sciences diagnostic radiology functional magnetic resonance imaging education sociology brain social sciences neuroscience cerebral hemispheres learning and memory left hemisphere magnetic resonance imaging face recognition perception age groups academic skills cognitive psychology right hemisphere brain mapping cognition literacy memory neuroimaging families research and analysis methods imaging techniques schools visual cortex people and places psychology radiology and imaging diagnostic medicine anatomy biology and life sciences population groupings cognitive science

2018

The emergence of the visual word form: Longitudinal evolution of category-specific ventral visual areas during reading acquisition

The type VI secretion system (T6SS) is a widespread protein secretion system found in many Gram-negative bacteria. T6SSs are highly regulated by various regulatory systems at multiple levels, including post-translational regulation via threonine (Thr) phosphorylation. The Ser/Thr protein kinase PpkA is responsible for this Thr phosphorylation regulation, and the forkhead-associated (FHA) domain-containing Fha-family protein is the sole T6SS phosphorylation substrate identified to date. Here we discovered that TssL, the T6SS inner-membrane core component, is phosphorylated and the phosphorylated TssL (p-TssL) activates type VI subassembly and secretion in a plant pathogenic bacterium, Agrobacterium tumefaciens. Combining genetic and biochemical approaches, we demonstrate that TssL is phosphorylated at Thr 14 in a PpkA-dependent manner. Further analysis revealed that the PpkA kinase activity is responsible for the Thr 14 phosphorylation, which is critical for the secretion of the T6SS hallmark protein Hcp and the putative toxin effector Atu4347. TssL phosphorylation is not required for the formation of the TssM-TssL inner-membrane complex but is critical for TssM conformational change and binding to Hcp and Atu4347. Importantly, Fha specifically interacts with phosphothreonine of TssL via its pThr-binding motif in vivo and in vitro and this interaction is crucial for TssL interaction with Hcp and Atu4347 and activation of type VI secretion. In contrast, pThr-binding ability of Fha is dispensable for TssM structural transition. In conclusion, we discover a novel Thr phosphorylation event, in which PpkA phosphorylates TssL to activate type VI secretion via its direct binding to Fha in A. tumefaciens. A model depicting an ordered TssL phosphorylation-induced T6SS assembly pathway is proposed. The type VI secretion system (T6SS) is the most recently described protein secretion system encoded as one or multiple copies in ∼25% of all sequenced Gram-negative bacteria [1]. T6SS is highly regulated and exhibits cytotoxicity to eukaryotic or bacterial hosts in a contact-dependent manner [2]–[6]. Growing evidence of the T6SS structure and imaging analyses documented that T6SS assembles into a contractile phage tail-like structure consisting of a TssB/TssC tubule structure [7]–[9], which is proposed to wrap around the Hcp tail tube and function to propel Hcp and effector secretion [10]. In support of T6SS functioning as a contractile phage tail-like structure, Hcp and VgrG are also detected on the cell surface and directly interact with each other and the TssB/TssC tubule [11]. Importantly, Hcp can interact with known or putative secreted effectors [11]–[13] and was recently identified to function as a chaperone and receptor of secreted substrates [12]. Time-lapse fluorescent microscopy further allowed for visualizing the dynamic T6SS activity that occurs between pairs of interacting cells, termed “T6SS dueling” [14], and killing activity at single-cell levels [15]–[17]. Bioinformatics and mutagenesis analyses revealed that T6SS consists of approximately 13–14 conserved components of type VI secretion (Tss, nomenclature proposed by Shalom et al.) [18] required for type VI secretion [10], [11], [13], [19]. Among them, TssM and TssL are the core inner-membrane proteins forming a stable inner-membrane complex connecting an outer-membrane complex [13], [20], [21]. TssM exhibits ATPase activity, and ATP binding-induced conformational change and subsequent ATP hydrolysis are important for the recruitment of Hcp to the TssM-TssL complex likely by directly interacting with the periplasmic domain of TssL [22]. The crystal structure of the N-terminal cytoplasmic domain of TssL has been reported [23], [24]. The cytoplasmic domain of TssL forms dimers, and this self-interaction is required for its function in enteroaggregative Escherichia coli Sci-1 T6SS [23]. Structural analysis of Francisella novicida cytoplasmic TssL revealed a surface-exposed groove that may represent a functional site for T6SS function [24] and is proposed to serve as a cytosolic hook to recruit the secreted substrates to the T6SS [25]. Interestingly, a subset of T6SS gene clusters also encode orthologs of Ser/Thr protein kinase (PpkA), phosphatase (PppA), and forkhead-associated (FHA) domain-containing proteins that are involved in regulating the Thr phosphorylation event [1], [26]–[28]. To date, our knowledge of the Thr phosphorylation regulatory mechanism is mostly due to several excellent studies in P. aeruginosa, in which the Hcp secretion island 1-encoded T6SS (H1-T6SS) is regulated positively by PpkA and negatively by the cognate phosphatase PppA via Thr phosphorylation on an FHA domain-containing protein, Fha1 [29]. A recent phosphoproteomic study also revealed that PpkA-dependent phosphorylation of Fha is required for T6SS activation in Serratia marcescens [28]. In addition, type VI secretion associated genes Q, R, S, T (TagQRST) function upstream of PpkA to promote kinase activity and subsequent type VI secretion in P. aeruginosa [30], [31]. TagF was identified as the repressor, whose absence caused the activation of a Thr phosphorylation-independent T6SS pathway [27]. Interestingly, Thr phosphorylation-dependent type VI activation could be stimulated by a cue during P. aeruginosa surface growth [27]. Also, PpkA, PppA, and TagT were required for the activation of the P. aeruginosa T6SS dueling activity when encountering the T6SS attack via contact with Vibrio cholerae cells [17]. Thus, the Thr phosphorylation-positive regulator TagQRST localized in the membranes [30], [31] may function to perceive and transduce signals to PpkA for activation of the local T6SS assembly for counterattack [17]. Although quite common in eukaryotes, this Ser/Thr phosphorylation regulatory mechanism is an emerging theme in prokaryotic signaling [32]. The FHA domain is a specific phosphothreonine (pThr) recognition unit [33]–[35] that was first identified as a conserved region via sequence analysis in a subset of forkhead-type transcription factors [36]. Numerous studies revealed that FHA domain-containing proteins play critical roles in diverse cellular processes in eukaryotes [35] and in a few prokaryotes, with a wealth of information obtained from Mycobacterium tuberculosis encoding 11 Ser/Thr protein kinases and 7 FHA domains in 6 proteins [37]–[39]. The importance of the FHA domain in activating type VI secretion and activity was also clearly demonstrated in the H1-T6SS of P. aeruginosa [29], [30]. Fha1 focally co-localized with ClpV1, an AAA+ ATPase specifically binding to the contracted TssB/TssC tubule for disassembly and cycling [7]–[9]. Importantly, Fha1 is required for ClpV1 focal recruitment, which suggests that Fha1 is a core scaffold protein of the H1-T6SS [29]. Interestingly, PpkA is crucial for both Fha1 and ClpV1 focal localization [29] and ClpV1 dynamics [17], but Fha1 phosphorylation is not required for ClpV1 recruitment but is critical for Hcp1 secretion [30]. Thus, PpkA may phosphorylate additional T6SS component (s), whereby its phosphorylation status may be critical for Fha and ClpV focal recruitment and subsequent type VI secretion. However, no additional T6SS protein has been identified as the PpkA target to date. In this study, we investigated the involvement of Thr phosphorylation in regulating type VI secretion in Agrobacterium tumefaciens, a soil phytopathogen capable of causing crown gall disease in wide range of plants. We discovered that TssL, a T6SS core component of the inner-membrane complex [20], is phosphorylated in a PpkA-dependent manner when A. tumefaciens is transcriptionally activated by the ExoR-ChvG/ChvI signaling cascade upon sensing the acidic signal [40]. We demonstrated that TssL is phosphorylated at Thr 14 and that this phosphorylation is crucial for stimulating secretion of the type VI hallmark protein Hcp and Atu4347, a putative toxin effector homologous to the anti-bacterial toxin secreted small protein (Ssp) in S. marcescens [6], [11]. Remarkably, TssL phosphorylation triggered TssM conformational change and promoted its binding to Hcp and Atu4347. The pThr binding ability of Fha is critical for specific binding to phosphorylated TssL (p-TssL) and this specific p-TssL-Fha interaction is required for efficient binding of Hcp and Atu4347 to p-TssL and activation of type VI secretion. We proposed that PpkA kinase initiates the complex assembly by phosphorylating TssL to increase the ATP binding ability of TssM ATPase for energy production and leads to the recruitment of Fha for activation of type VI secretion. The presence of genes encoding the Fha-family protein, PpkA kinase, and PppA phosphatase in a subset of T6SSs suggested the involvement of a Thr phosphorylation regulatory pathway in these bacteria [1], [26], [27]. However, this Thr phosphorylation regulation of T6SS has only been demonstrated in H1-T6SS of P. aeruginosa [29] and S. marcescens [28]. Bioinformatic analysis revealed the homology of Atu4330 to PpkA, Atu4331 with N-terminal domain and C-terminal domain to TagF and PppA respectively (thus renamed as TagF-PppA), and Atu4335 to Fha encoded in the T6SS gene cluster of A. tumefaciens strain C58 (Figure 1A) [11], [26], [29], [32], [41]. We previously identified that Hcp secretion was abolished in Δfha and its abundance was greatly reduced in ΔppkA, whereas Hcp secretion level was not significantly altered with the deletion of tagF-pppA (atu4331) [11]. These data suggested that the Thr phosphorylation pathway (TPP) may play a role to regulate type VI secretion in A. tumefaciens, but the molecular mechanisms underlying this potential Thr phosphorylation regulation remain unknown. Thus, we first performed complementation tests and overexpression analysis to determine the role of PpkA and TagF-PppA in regulating Hcp secretion. The reduced Hcp secretion in ΔppkA could be fully restored by trans complementation (Figure 1B). In contrast, Hcp secretion was abolished when TagF-PppA was overexpressed on plasmids in both C58 and ΔtagF-pppA strains (Figure 1B). The negative effect of overexpressed TagF-PppA in Hcp secretion was specific because overexpressed PpkA in C58 did not affect Hcp secretion levels. Therefore, PpkA positively and TagF-PppA negatively regulated Hcp secretion in A. tumefaciens. In P. aeruginosa, PpkA phosphorylates Fha1; both its phosphorylation status and FHA domain responsible for pThr binding ability are critical for Hcp1 secretion [29]. In A. tumefaciens, Fha contains a putative N-terminal FHA domain with amino acid residues (R30 and S46) corresponding to the conserved pThr binding residues [33], [35] (Figure 1C) critical for Hcp1 secretion in P. aeruginosa [29]. Thus, we generated various fha mutants, including the FHA-domain deletion mutant (fhaΔFHA) and alanine substitution mutants at Arg 30 residue (fhaR30A), Ser 46 residue (fhaS46A), and both residues (fhaR30AS46A) to determine the roles of the FHA domain and its putative pThr binding residues in Hcp secretion. Hcp secretion was completely abolished in FhaΔFHA and FhaR30AS46A strains, whereas FhaR30A and FhaS46A caused reduced Hcp secretion as compared to the wild-type C58 (Figure 1D). Because the protein abundance remained the same in all Fha point mutation variants as in the wild type (Figure 2A and Figure S1), we suggested that the pThr binding ability of Fha is indeed critical for Hcp secretion. Importantly, the evidence that the protein abundance of all analyzed T6SS components in these fha mutants remained at wild-type levels as well as in the absence of ppkA and tagF-pppA strongly suggests a PpkA- and Fha-dependent post-translational regulatory pathway for Hcp secretion from A. tumefaciens (Figure S1). To date, Fha-family protein is the only T6SS component identified to be phosphorylated by PpkA [28], [29]. In P. aeruginosa, Fha1 is phosphorylated by PpkA, and its Thr 362 phosphorylation site is required for Hcp secretion [29]. This PpkA-dependent Thr phosphorylation on Fha was also recently identified by a phosphoproteome screen and found to be critical for T6SS activation in S. marcescens [28], which suggests a common TPP in regulating type VI secretion. However, this Thr phosphorylation site is not universally conserved because no corresponding Thr could be identified in the A. tumefaciens Fha (Figure S2). Thus, PpkA may phosphorylate Fha at other sites or other T6SS component (s), in which its phosphorylation status is critical for interaction with Fha and activating type VI secretion in A. tumefaciens. Western blot analysis revealed that only TssL but not other analyzed T6SS components had slower migration in the wild-type C58 as compared with the ΔppkA mutant (Figure 2A and Figure S1; marked with an asterisk), which suggests that TssL may be phosphorylated by PpkA. To determine whether TssL is indeed phosphorylated, His-tagged TssL, which is functional in mediating the secretion of Hcp [20] and the putative T6SS toxin effector Atu4347 [6], [11] (Figure S3A), was expressed and purified by Ni-NTA resins for phosphorylation analysis. We first used Phos-tag SDS-PAGE for unambiguous separation of phosphorylated and unphosphorylated proteins on the gel. Western blot analysis revealed two TssL-His protein bands, with one faint lower band and one major upper band (Figure 2B). Importantly, with calf intestinal alkaline phosphatase (CIAP) treatment, the upper TssL-His band disappeared and the lower band intensity increased. Therefore, the upper protein band represented the phosphorylated TssL (p-TssL-His) and the lower band the unphosphorylated form (Figure 2B). Similar results were observed when total protein extracts were used for analysis (Figure S3B). Thus, TssL is indeed phosphorylated as the major form in A. tumefaciens when T6SS is activated in the acidic condition. In A. tumefaciens, TssL is a bitopic inner-membrane protein with an N-terminal VasF-like domin (35–296 aa) facing the cytoplasm and a C-terminal peptidoglycan binding domain (367–470 aa) exposed to the periplasm (Figure 3A) [20]. To identify the phosphorylated amino acid of TssL, the TssL-His expressed in A. tumefaciens was purified for phosphorylation site mapping by mass spectrometry (MS). MS analysis of purified TssL-His proteins identified phosphorylated Thr 14 (T14) located within its N-terminal cytoplasmic domain (residues 1–255) (Figure 3A, 3B) [20]. Because Thr 14 was the only phosphorylation site identified by MS/MS ion search with between 75% and 85% coverage of TssL-His proteins from 3 independent experiments (Figure S4), Thr 14 may be the sole phosphorylation site of TssL. This conclusion is also supported by the absence of a p-TssL-His protein band on Phos-tag gel when the TssL Thr 14 residue was substituted with alanine (A) (TssLT14A) (Figure 4). Therefore, TssL is a phosphoprotein phosphorylated at Thr 14. To investigate whether PpkA kinase activity is required for TssL phosphorylation, we next generated a ppkA mutant encoding PpkAD161AN166A with alanine substitutions in 2 conserved Asp 161 (D161) and Asn 166 (N166) residues located within the kinase magnesium binding loop (Figure S5) [42]. The corresponding Asp 129 of PpkA was previously demonstrated to be critical for Fha1 phosphorylation in P. aeruginosa [30]. In support of the role of PpkA and its kinase activity in phosphorylating TssL, we detected only an unphosphorylated TssL-His protein band in ΔppkAΔtssL and ppkAD161AN166AΔtssL expressing TssL-His, as shown by Phos-tag SDS-PAGE (Figure 4). Moreover, no phosphorylated residues could be identified by MS analysis of TssL-His proteins purified from ΔppkAΔtssL and ppkAD161AN166AΔtssL (data not shown). Not surprisingly, purified TssL-His remained with the wild-type phosphorylation pattern in fhaR30AS46A by Phos-tag gel (Figure 4) and MS analysis (data not shown), and cellular TssL expressed in various fha mutants migrated to the same position as that for C58 on regular SDS-PAGE (Figure 2A and Figure S1). In conclusion, we provide evidence that PpkA kinase activity is required for TssL phosphorylation. The requirement of PpkA and its conserved kinase catalytic motif for TssL phosphorylation strongly suggests the importance of TssL phosphorylation in regulating type VI secretion in A. tumefaciens. Thus, we determined whether the disruption of PpkA kinase catalytic sites affects type VI secretion activity, which was monitored by the secretion of Hcp [20] and the putative T6SS toxin effector Atu4347 [6], [11]. PpkAD161AN166A, which is incapable of phosphorylating TssL (Figure 4), also caused the largely reduced Hcp and Atu4347 secretion levels similar to that in ΔppkA (Figure 5A). Although PpkAD161AN166A protein levels are slightly lower than in wild-type C58, the complete loss of phosphorylated TssL in this kinase mutant strongly suggests that PpkA kinase activity plays a crucial role in regulating type VI secretion. Next, we investigated whether TssL phosphorylation status is critical in regulating type VI secretion. In addition to generating the phosphorylation-inactive mutant TssLT14A (Figure 4), we generated 2 additional tssL mutants with aspartic acid (D) or glutamic acid (E) substitutions (TssLT14D and TssLT14E). We observed complete loss of Hcp secretion from ΔtssL and fhaR30AS46A and low levels of secreted Hcp proteins from ΔppkA and TssL phosphorylation site mutants (TssL T14A, TssLT14D, and TssLT14E) (Figure 5B). Consistent with reduced Hcp secretion in ΔppkA or TssL phosphorylation site mutants, Atu4347 secretion was highly attenuated and barely detected in the same mutants. Therefore, we suggested that TssL phosphorylation status is critical in regulating type VI secretion. The complete loss of both Hcp and Atu4347 secretion in fhaR30AS46A suggests an essential role of the Fha pThr binding site in type VI secretion. Because FHA domains are well known to bind to pThr residues specifically [33]–[35], we hypothesized that Fha specifically binds Thr 14-phosphorylated TssL to activate type VI secretion. To test this hypothesis, we first performed isothermal titration calorimetry (ITC) analysis with purified Fha proteins and synthetic phosphorylated- and unphosphorylated TssL N-terminal peptides. Because of the instability of full-length Fha protein purified from E. coli (data not shown), we purified truncated Fha proteins with different lengths for the ITC experiments (Figure S6A). Fha7-267WT containing the intact FHA domain was capable of binding to the 8 mer phosphorylated TssL peptide (DLPpTVVEI, p-TssL8 mer), although with relatively low affinity (Kd 1. 58±0. 35 mM) (Figure 6A). The binding affinity was ∼3-fold higher (Kd 0. 54±0. 07 mM) with a longer phosphorylated TssL peptide (DNPSSWQDLPpTVVEITEESR, p-TssL20 mer) (Figure 6B). These data suggest that the TssL sequence or structure adjacent to the phosphorylation peptide motif contributes to formation of the pTssL-Fha complex. We could not detect the binding affinity with the unphosphorylated TssL peptide (DLPTVVEI, TssL8 mer) or the 6 mer phosphorylated CHK2 peptide (VSpTQEL, p-CHK26 mer) as controls [43] (Figure 6C, 6E). Notably, the truncated Fha proteins with a mutated FHA domain (Fha7-267R30AS46A) completely lost the ability to interact with the 8 mer p-TssL peptide (Figure 6D). We showed specific binding to the p-TssL peptide but not unphosphorylated TssL peptide with a longer version of the truncated Fha protein (Fha7-309WT) (Figure S6B). These data strongly suggest that Fha specifically binds to Thr 14-phosphorylation TssL peptide via its pThr binding motif of the FHA domain. The increased binding affinity for the FHA domain with a longer phosphorylated TssL peptide also suggests that stronger interaction may occur in vivo, where Fha interacts with full-length TssL. To our knowledge, this is the first demonstration of specific binding of an FHA domain-containing protein to the T6SS phosphorylated target. Next, we investigated the p-TssL-Fha complex formation in vivo by pulldown assays using functional Strep-tagged TssL (Figure S3A). To avoid the non-specific protein–protein interactions occurring when proteins are released into solution after cell breakage and to detect weak and dynamic interactions, we used the cleavable and membrane permeable cross-linker dimethyl 3,3′-dithiobispropionimidate (DTBP) to cross-link interacting proteins before cell lysis [44] for interaction studies in A. tumefaciens. The known TssL-interacting inner-membrane protein TssM [20], [22] co-precipitated with TssL-Strep in all analyzed samples (Figure 7A), which indicates that TssL interacts with TssM independent of its phosphorylation status or association with Fha. In contrast, Fha was pulled down by wild-type TssL-Strep but not TssL-Strep with a T14A substitution or by TssL-Strep in the FhaR30AS46A strain. All interaction proteins could not be co-precipitated by Strep-Tactin resin from the negative control, ΔtssL expressing wild-type TssL without Strep-tag. As well, the two non-T6SS proteins, soluble protein ActC [20], [45] and outer membrane protein AopB [46], were not pulled down with TssL-Strep (Figure 7A). Our previous study showed that Hcp is recruited into the TssM-TssL complex via directly interacting with the TssL [22]. Thus, we determined whether TssL phosphorylation status and its specific binding to Fha are critical for its interaction with Hcp. Abundant Hcp was pulled down by wild-type TssL-Strep but only small amounts of Hcp bound to TssLT14A-Strep or wild-type TssL-Strep in the FhaR30AS46A strain (Figure 7A). Because TssL remained highly phosphorylated in the FhaR30AS46A strain (Figure 4), these data suggest that Fha binding to p-TssL is critical for efficient Hcp-TssL interactions. Interestingly, Atu4347 was also co-purified with TssL, in which its association with TssL was stimulated by phosphorylation of TssL and pThr binding ability of Fha (Figure 7A). Because TssM undergoes an ATP binding-induced conformational change and the subsequent ATP hydrolysis energizes Hcp recruitment into the TssM-TssL complex and activates Hcp secretion [22], we next tested whether TssL phosphorylation and/or its interaction with Fha is required for TssM structure transition. TssM undergoes a structural transition on ATP binding, as seen by the higher protease resistance of wild-type TssM than the ATP binding-deficient walker A mutant [22]. Thus, we performed a spheroplast protease susceptibility assay to determine whether the TssM becomes more susceptible to protease digestion in mutants expressing unphosphorylated TssL or fhaR30AS46A. The relative protein abundance of TssL and TssM was quantified and normalized with an internal control, the cytoplasmic protein GroEL resistant to protease digestion [22], [47]. Protein abundance of both TssL and TssM was slightly lower in all three TssL phosphorylation site mutants (TssLT14A, TssLT14D, and TssLT14E) as compared with the wild type and fhaR30AS46A strains before protease treatments (Figure 7B). However, no further increased degradation could be detected for 3 TssL variants with phosphorylation-site mutations relative to that of wild-type TssL. In contrast, TssM protein abundance was further dose-dependently decreased in the TssL phosphorylation-site mutants after protease treatment (Figure 7B). Differently, TssM remained with the same degree of protease resistance in fhaR30AS46A regardless of protease concentration used. These data suggest that TssL phosphorylation but not p-TssL-Fha complex formation is responsible for TssM conformational change. Our discovery of TssL as a new PpkA substrate in A. tumefaciens, together with previous findings of PpkA-mediated Fha phosphorylation from P. aeruginosa [29], [30] and S. marcescens [28], have revealed both the conservation and uniqueness of Thr phosphorylation regulatory pathways in different bacteria. These studies demonstrate that both PpkA kinase activity and phosphorylation of target proteins are critical for type VI secretion. In addition, the pThr binding motif of Fha proteins from both P. aeruginosa and A. tumefaciens are crucial for Hcp secretion, which indicates the importance of this Thr phosphorylation event and that Fha interaction with phosphoprotein is a conserved and key step for subsequent type VI assembly and secretion. However, although the Fha Thr phosphorylation site is conserved in several Fha-family proteins across different bacterial species [28], we could not detect the corresponding Thr in the A. tumefaciens Fha protein sequence (Figure S2). Our attempt to detect Fha phosphorylation by Phos-tag SDS-PAGE was also unsuccessful (Figure S3C), which suggests that Fha may not be the phosphorylated target of PpkA in A. tumefaciens. The N-terminal cytoplasmic domain harboring the Thr 14 (T14) phosphorylation site of TssL is highly conserved in all analyzed Agrobacterium and Rhizobium species that also encode Thr phosphorylation components (Figure S7) but not in other TssL orthologs encoded by distantly related bacterial species (Figure S8). However, Thr residues are widely distributed at the N-terminal domain of all analyzed TssL-family proteins across different bacterial species (Figure S8). In considering the sequence diversity of the FHA domain recognition site (pT+3) [34], [35], TssL may be a conserved phosphorylation target of PpkA not only in Agrobacterium and Rhizobium species but also in other bacteria outside of Rhizobiaceae, which suggests a broader function of TssL phosphorylation in regulating type VI activity. Low but detectable Hcp secretion from the absence of PpkA or PpkA with mutation in its catalytic site suggested that PpkA kinase activity is not absolutely required to activate type VI secretion in A. tumefaciens. This finding is consistent with the low Hcp1 secretion when ppkA was deleted in the P. aeruginosa wild-type strain, in which a Thr phosphorylation-independent pathway could be derepressed by the deletion of tagF [27], [31]. The presence of a TagF-PppA fusion protein encoded by the T6SS gene cluster suggested that T6SS may be negatively regulated by TagF in a Thr phosphorylation-independent manner in A. tumefaciens. In P. aeruginosa, Thr phosphorylation is stimulated by a cue during the surface growth [27], which implies a signal derived from the cell-to-cell contact. Indeed, the Thr phosphorylation regulatory components PpkA, PppA, and TagT are required for the activation of the P. aeruginosa T6SS dueling activity when encountering the T6SS attack by contacted V. cholerae cells [17]. The authors suggested that Thr phosphorylation-positive regulation by TagQRST localized in the membranes [30], [31] may function to perceive and transduce signals to PpkA for activation of the local T6SS assembly for counterattack [17]. In A. tumefaciens, Thr phosphorylation is not absolutely required but plays an important role to enhance type VI secretion activity. Less than 5% of Fha1 is phosphorylated in P. aeruginosa grown in liquid culture and up to ∼20% of p-Fha1 is stimulated by growth on solid-phase medium [27]; however, most if not all of TssL is phosphorylated when A. tumefaciens is grown in acidic liquid culture. Moreover, the absence of ppkA and loss of TssL Thr phosphorylation caused a similar degree of reduced Hcp and Atu4347 secretion when A. tumefaciens was grown on agar (Figure S9) and liquid culture (Figure 5B). Therefore, surface-derived cell contact may not be a signal to stimulate the Thr phosphorylation event in A. tumefaciens. This notion is also supported by the lack of TagQRST orthologs in A. tumefaciens. From the presence of both Thr phosphorylation-independent and -dependent type VI activation in A. tumefaciens, we propose that the initial acid-induced type VI gene expression may somehow trigger a signal to activate TPP. The signal may come from the assembly of the T6SS subassembly from the same cell or by sensing the secretion activity from other cells. However, the evidence that TssL remains highly phosphorylated even when type VI secretion is abolished with the loss of Fha or its functional FHA domain suggests that the signal is unlikely to be Hcp or effectors secreted from neighboring cells. Alternatively, the signal may be derived from the acid-induced but T6SS-independent components [48]–[50]. Future work to identify the genetic requirements stimulating Thr phosphorylation will shed light on the molecular mechanisms and biological significance of this posttranslational regulation in A. tumefaciens. Only the authentic phosphate group at Thr 14 but not the phospho-mimic residues exhibiting the full function of TssL in activating type VI secretion is consistent with the notion that the FHA-domain phosphopeptide binding site is highly pThr-specific [51], [52]. Interestingly, Hcp secretion was slightly higher from ΔppkA than from all 3 TssL phosphorylation site mutants, with almost no detection of Hcp secretion in the T14A mutant (Figures 5B and S9), which suggests that these Thr 14 (T14) substitutions may partially affect other TssL function (s) besides its phosphorylation status. The observation that all 3 TssL variants are less stable and subsequently affect the stability of TssM (Figure 7B) suggests that either Thr 14 or its phosphorylation status affects TssL stability. Because TssM is highly unstable in the absence of TssL [20], [22] but remains at similar levels in ΔppkA as in the wild type grown in liquid culture or on agar (Figure 5B and S9), the reduced stability of TssL and TssM may be due to the substitution of T14 instead of TssL phosphorylation status. However, the slightly reduced TssL and TssM protein levels in the TssL phosphorylation site mutants cannot account for the high attenuation of Hcp and Atu4347 secretion because overexpression of TssLT14A was unable to restore Hcp or Atu4347 secretion in ΔtssL as did the wild type TssL (Figure S10). Furthermore, while the 3 TssL variants remain the same protease resistance level like wild-type TssL, TssM become more susceptible to protease in these TssL phosphorylation-site mutants. These data suggested that TssL phosphorylation status indeed affect TssM conformational change. Thus, we conclude that PpkA kinase activity and phosphorylation of TssL at Thr 14 are critical for TssM conformational change and type VI secretion. Why Hcp and Atu4347 secretion is completely abolished in the fhaR30AS46A mutant but the loss of TssL phosphorylation in ΔppkA retains low but detectable Hcp secretion activity is unclear (Figure 5B and S9). Because no additional Ser/Thr kinase gene could be identified in the A. tumefaciens C58 genome by annotation and BLAST analysis (data not shown) and no phosphorylation of TssL could be detected in ΔppkA (Figure 4), PpkA may be the sole Thr kinase in A. tumefaciens. Therefore, the complete loss of type VI secretion activity in the fhaR30AS46A mutant is unlikely due to the loss of its ability in binding to another pThr site other than TssL. The formation of the TssM-TssL inner-membrane complex independent of TssL phosphorylation is consistent with the detection of TssM-TssL interaction when expressed in heterologous E. coli [20], [22]. Although we could not detect a conformation change for TssL by protease susceptibility assay, TssL may undergo a structural transition on phosphorylation by PpkA, thereby leading to the accessibility or increased affinity of interacting TssM for ATP binding and trigger the structural transition of TssM. The detection of the ATP binding-induced TssM conformational change in the FhaR30AS46A strain suggests that Fha binds to the p-TssL after TssM binds with ATP. Furthermore, Hcp and Atu4347 can only bind efficiently with TssL when TssL is phosphorylated and forms the p-TssL-Fha complex (Figure 7A). Together with previous findings that TssM and its ATPase activity are required for recruiting Hcp into TssM-TssL complex via direct interaction with TssL and activating Hcp secretion [22], we suggested that Fha likely interacts with p-TssL residing in p-TssL-TssM inner membrane complex instead of p-TssL alone. To investigate whether TssM is required to trigger TssL phosphorylation by PpkA and subsequently causes the reduced Hcp and Atu4347 binding to TssL, we determined the phosphorylation status of TssL in the tssM deletion mutant by regular and Phos-tag SDS-PAGE. To our surprise, TssL remains phosphorylated at wild-type levels in the absence of TssM in A. tumefaciens (Figure S11). These data indicated that TssM and its ATPase activity are not required for TssL phosphorylation but critical for recruiting Hcp and Atu4347 into TssM-p-TssL-Fha ternary complex for type VI assembly and secretion. Taken together, we proposed a hypothetical model for the TssL phosphorylation-induced T6SS assembly pathway (Figure 8). The formation of the TssM-TssL inner-membrane complex likely occurs (step I) before phosphorylation of TssL (step II). TssL phosphorylation triggers TssM conformation switch (step III), which occurs before cytoplasmic Fha is recruited into the TssM-p-TssL complex via interacting with N-terminal Thr 14 phosphopeptide motif of TssL (step IV). The formation of this TssM-p-TssL-Fha complex combined with TssM ATP hydrolysis activates the efficient recruitment of Hcp and Atu4347 into this T6SS inner-membrane subassembly for type VI tube assembly and effector secretion across outer membrane (step V). The direct evidence for formation of this inner-membrane subassembly remains lacking and other alternative or multiple pathways may exist. Future work to investigate whether and how TssM ATP binding/hydrolysis influence Fha recruitment to activate type VI complex formation and secretion shall provide further mechanistic insights in type VI assembly and secretion. Recent phosphoproteomic analysis in S. marcescens identified Fha but not TssL as PpkA-phosphorylated substrate [28]. Thus, the phosphorylation events for Fha and TssL have not been concurrently identified in the same organism. However, we do not exclude that both proteins may be phosphorylated and coordinated to regulate type VI assembly and effector secretion. In P. aeruginosa, Fha1 focally co-localizes with ClpV1 and is required for ClpV1 focal recruitment [29] and dynamics [17], which suggests that Fha1 is a T6SS core scaffold protein. The requirement of tssM (icmF) for ClpV1 foci recruitment in P. aeruginosa [53] indeed supports our proposed model for the recruitment of Fha into TssM-p-TssL inner-membrane complex. Our findings now extend the knowledge that phosphorylation of TssL may function to initiate the inner-membrane subassembly formation, for which Fha may come into play at a later stage for recruiting ClpV1 for TssB-TssC tubule disassembly. Future work to determine the conservation or uniqueness of Fha and TssL phosphorylation in different bacteria and to investigate their hierarchy and roles for type VI complex assembly will elucidate the mechanistic role of Thr phosphorylation in T6SS. Only a subset of T6SS-containing bacterial species encodes the Thr phosphorylation components [1], [26], [27], so Thr phosphorylation-dependent regulation may be required for only certain bacteria that encounter complex environmental signals. The evolution of the conservation and divergence of Thr phosphorylation components, phosphorylated targets, and regulatory mechanisms in different bacterial species is an interesting question for future study. Strains, plasmids, and primer sequences used in this study are in Tables S1 and S2. A. tumefaciens and Escherichia coli strains were grown at 28°C in 523 [54] and at 37°C in LB [55], respectively. The plasmids were maintained by the addition of 50 µg/ml gentamycin (Gm) for A. tumefaciens and 100 µg/ml ampicillin (Ap), and 50 µg/ml Gm for E. coli. A. tumefaciens cells grown in liquid AB-MES medium (pH 5. 5) [56] at 25°C for 6 hr were concentrated by trichloroacetic acid precipitation (TCA) for Hcp and Atu4347 secretion assays as described [20], [41]. All in-frame deletion and amino acid substitution (s) mutants were generated in A. tumefaciens C58 via double crossover using the suicide plasmid pJQ200KS [57] as described [20], [41]. The detailed procedures for the construction of plasmids and mutant strains are described in Supporting Information S1. Dephosphorylation analysis by calf intestinal alkaline phosphatase (CIAP) was performed according to the user manual (New England Biolabs, Beverly, MA, USA) with minor modifications. Equal amounts of Ni-NTA resins with purified TssL-His or total protein extracts isolated from various A. tumefaciens strains were resuspended in 1× CIAP buffer containing 100 mM NaCl, 50 mM Tris-HCl (pH 7. 9), 10 mM MgCl2,1 mM DTT, and 1× protease inhibitor cocktail (EDTA-free) with CIAP at 1 unit per µg of protein. The protein samples treated with or without CIAP were incubated at 37°C for 90 min. An equal volume of 2× SDS loading buffer was added and incubated at 96°C for 20 min and analyzed by Phos-tag SDS-PAGE. The Phos-tag SDS-PAGE analysis was performed according to the user manual for Phos-tag Acrylamide AAL-107 (Wako Pure Chemical Industries, Osaka, Japan) with minor modifications. Protein samples were separated on 7% polyacrylamide gels containing 0. 35 M Bis-Tris-HCl (pH 6. 8), 35 µM Phos-tag Acrylamide AAL-107, and 100 µM ZnCl2, with electrophoresis conducted at 40 mA/gel under a maximum voltage of 90V in a cold room. After electrophoresis, Phos-tag gels were washed with transfer buffer (25 mM Tris, 192 mM glycine, 20% methanol) containing 1 mM EDTA for 15 min with gentle shaking followed by a second wash in transfer buffer without EDTA for 15 min. The gels were washed with transfer buffer containing 1% SDS for 15 min before transfer to PVDF membranes with a submarine blotting apparatus. Phosphorylation site mapping was determined by in-gel trypsin or trypsin/chymotrypsin digestion of purified TssL-His followed by nanoLC/nanospray/tandem mass spectrometry (LC-ESI/MS/MS) analysis. TssL-His proteins expressed in A. tumefaciens were purified by use of Ni-NTA resins and separated by 12% SDS-PAGE followed by Coomassie blue staining. TssL-His protein bands were cut out for in-gel trypsin or trypsin/chymotrypsin digestion as described [58], [59]. The extracted tryptic peptides were subjected to the LC separation followed by a linear quadrupole ion trap-Fourier transform (LTQ-FT) ion cyclotron resonance mass spectrometer (Thermo Fisher Scientific) equipped with a nanoelectrospray ion source (New Objective, Woburn, MA, USA) for protein identification and phosphorylation site mapping. To avoid the non-specific protein–protein interactions occurring when proteins are released into solution after cell breakage, we used the cleavable and membrane permeable cross-linker dimethyl 3,3′-dithiobispropionimidate (DTBP) to cross-link interacting proteins before cell lysis [44] for the TssL-Strep pulldown assays from A. tumefaciens. The pulldown assay with Strep-Tag was performed according to the user manual for StrepTactin Sepharose (GE Healthcare). In total, 500 ml A. tumefaciens cell cultures were centrifuged and washed 3 times with 12 ml phosphate buffer (20 mM sodium phosphate, pH 7. 6; 20 mM sodium chloride) and resuspended in the same buffer adjusted to OD600 about 4. Membrane-permeable cross-linker dimethyl 3,3′-dithiobispropionimidate (DTBP) was added at a final concentration of 5 mM, and the mixture was incubated at room temperature for 45 min. The reaction was stopped by adding Tris-HCl (pH 7. 6) to a final concentration of 20 mM for 15 min. The cells were collected by centrifugation and washed twice with 12 ml of 50 mM Tris-HCl (pH 7. 6) before pulldown assay. The harvested cells were resuspended in binding buffer (100 mM Tris-HCl, pH 8. 0; 150 mM NaCl; 1 mM EDTA) containing 0. 05% Triton X-100 and underwent sonication on ice. The cell lysate was centrifuged twice at 10,000× g for 15 min at 4°C. The soluble fraction was passed through Strep-Tactin resins (GE Healthcare) and washed 6 times with binding buffer (100 mM Tris-HCl, pH 8. 0; 150 mM NaCl; 1 mM EDTA) containing 0. 05% Triton X-100. The bound proteins were eluted by use of elution buffer (100 mM Tris-HCl, pH 8. 0; 150 mM NaCl; 1 mM EDTA; 2. 5 mM desthiobiotin). The fractions were examined by western blot analysis. ITC analysis was used to determine the binding affinity and stoichiometry between Fha and TssL T-14 peptide with use of the MicroCal iTC200 system (GE Healthcare) and analyzed with ORIGIN software. Truncated His-tagged Fha7–267 and Fha7–309 proteins were overexpressed and purified from E. coli BL21 (DE3) as described [43]. The proteins and various synthetic peptides were dissolved in buffer containing 50 mM Tris-HCl (pH 8. 0), 150 mM NaCl. Synthetic phosphorylated TssL (DLPpTVVEI, p-TssL8 mer; DNPSSWQDLPpTVVEITEESR, p-TssL20 mer), unphosphorylated TssL (DLPTVVEI, TssL8 mer), and phosphorylated CHK2 (VSpTQEL, p-CHK26 mer) [43] peptides, were used to titrate the various Fha7–267 or Fha7–309 proteins at 25°C with a total 25 times of injections with 200 µM protein in the sample cell and 6 mM peptide in the injection syringe. The titration heat was calculated to eliminate the effect of heat generated from titrating the ligand into buffer. Thermal data were fitted to the One Set of Sites binding model with the N value fixed at 1 to yield the value of the equilibrium dissociation constant (Kd). Spheroplast preparation and protease susceptibility assay were performed as described [22] with modifications. A. tumefaciens cells were grown in liquid AB-MES medium (pH 5. 5) [56] at 25°C for 6 hr. Cells were harvested by centrifugation at 10,000× g for 15 min at 4°C and washed once with 50 mM Tris-HCl (pH 7. 5). To prepare spheroplasts, cell pellets were resuspended gently in buffer containing 50 mM Tris-HCl (pH 7. 5), 20% sucrose, 2 mM EDTA, 0. 2 mM DTT, and 1 mg/ml lysozyme and incubated on ice for 1 hr. After lysozyme treatment, spheroplasts were treated with Streptomyces griseus protease (Sigma) at a final concentration of 12. 5 or 25 µg/ml for 10 min on ice in the presence of 10 mM MgSO4. The reaction was stopped by adding an equal amount of 2× SDS loading buffer and incubated at 96°C for 20 min before SDS-PAGE. Western blot analysis was performed as described [56] with primary polyclonal antibodies against proteins (PpkA, PppA, TssK, Fha, TssE, TssC41, TssB, TssA, ClpV, Atu4347, VgrGs, RpoA, AopB) [11], TssL [20], TssM [20], Hcp [41], ActC [45], GroEL [47], polyclonal antibodies against His (Sigma), or monoclonal antibodies against Strep (IBA-Life Sciences, Goettingen, Germany), followed by a secondary antibody horseradish peroxidase (HRP) -conjugated goat anti-rabbit IgG (chemichem) and detected by use of the Western Lightning System (Perkin Elmer, Boston, MA). Chemiluminescent bands were visualized on X-ray film (Kodak, Rochester, NY) or were quantified with the UVP BioSpectrum 600 imaging system (Level Biotechnology, Inc.).

The bacterial type VI secretion system (T6SS) resembles a contractile phage tail structure and functions to deliver effectors to eukaryotic or prokaryotic target cells for the survival of many pathogenic bacteria. T6SS is highly regulated by various regulatory systems at multiple levels in response to environmental cues. Post-translational regulation via threonine (Thr) phosphorylation is an emerging theme in regulating prokaryotic signaling, including T6SS; the knowledge is mainly contributed by studies of Hcp secretion island 1-encoded T6SS (H1-T6SS) of Pseudomonas aeruginosa. Here, we discover a new phosphorylated target, a T6SS core-component TssL, and demonstrate that this Thr phosphorylation event post-translationally regulates type VI secretion in a plant pathogenic bacterium, Agrobacterium tumefaciens. We provide the first demonstration that the specific binding of Fha, a forkhead-associated domain-containing protein, to the phosphorylated target is required to stimulate type VI secretion. Genetic and biochemical data strongly suggest an ordered TssL-phosphorylation–dependent assembly and secretion pathway.

Abstract Introduction Results Discussion Materials and Methods

biochemistry plant biochemistry transmembrane proteins protein interactions plant microbiology proteins regulatory proteins biology microbiology bacterial pathogens

2014

Fha Interaction with Phosphothreonine of TssL Activates Type VI Secretion in Agrobacterium tumefaciens

Understanding how a pathogen colonizes and adapts to a new host environment is a primary aim in studying emerging infectious diseases. Adaptive mutations arise among the thousands of variants generated during RNA virus infection, and identifying these variants will shed light onto how changes in tropism and species jumps can occur. Here, we adapted Coxsackie virus B3 to a highly permissive and less permissive environment. Using deep sequencing and bioinformatics, we identified a multi-step adaptive process to adaptation involving residues in the receptor footprints that correlated with receptor availability and with increase in virus fitness in an environment-specific manner. We show that adaptation occurs by selection of a dominant mutation followed by group selection of minority variants that together, confer the fitness increase observed in the population, rather than selection of a single dominant genotype. The extreme mutation rates of RNA viruses and the highly diverse populations they generate in few replication cycles are considered the basis for their rapid adaptation to new environments [1,2]. Such adaptive steps result in the emergence of new variants capable of escaping immune responses, resisting antiviral approaches, altering tissue tropism or crossing species barriers. In the past, experimental evolution of viruses in different host environments has proven to be a useful tool in quantifying fitness increases and the dynamics of adaptation. By classic sequencing techniques, some of the key genetic determinants responsible have been identified [3,4], but until the advent of deep sequencing, analysis of the mutational composition of RNA virus populations was hampered by lack of depth of sequence coverage. The potential to describe the whole virus mutant spectrum and detect variants that otherwise would be overlooked by conventional sequencing is fundamental to studying virus evolution and understanding emergence [5]. Recent work shows that deep sequencing can identify the emergence of escape mutants in experimental and clinical samples [6,7], and can be used to characterize the entire mutant spectrum of a virus population [8]. One of the goals in the field of emerging infectious diseases is to determine whether adaptation to novel hosts (species, tissues or cell types) can be identified for a recently introduced pathogen that is confronted with a less than optimal host environment [9–11]. Viruses are well-suited for studying adaptation and evolution for several reasons: i) high mutation rates ii) short generation time and iii) large population sizes. We used Coxsackie virus B3 (CVB3) as a model, since the genetics of this virus and the interactions between the cell receptors and viral capsid proteins (VP1, VP2 and VP3) are well characterized. CVB3 enters the cell through a primary receptor, the Coxsackie and Adenovirus Receptor (CAR) [12], while certain strains may use as co-receptor the Decay Accelerating Factor (DAF) [13,14], also known as CD55. To study expansion of host tropism, we passaged virus in two cellular environments, a highly permissive one and a less permissive one. By deep sequencing longitudinal samples of experimentally evolved populations, we identify the emergence of host environment-specific mutations undergoing positive selection. We show that Coxsackie virus adapts differently to two cell types according to receptor and co-receptor availability in a multi-step adaptation sequence that involves group selection of minority variants. Importantly, we reveal the significant contribution of several minority variants to the overall fitness of the entire population. Our results underscore the importance of characterizing RNA virus quasispecies during adaptation and virus evolution. To monitor the evolution of CVB3 towards novel and less permissive host environments, we selected human lung A549 cells, which gave similar final virus yields as the highly permissive HeLa cells, but after two days rather than one day of infection. CVB3 was thus serially passaged 40 times in six biological replicate series in both cell types. Virus yields were constant throughout the passage series suggesting that no significant genetic drift or accumulation of detrimental mutations through population bottlenecking had occurred (Fig 1A and 1B). The time required to reach peak titers was reduced in A549 cells over the first ten passages from 48 hours to 24 hours, suggesting that fitness increases and adaptation occurred in this novel environment. We measured the relative fitness of the passaged viruses by competing the passage 1,20 and 40 populations from each replicate with a genetically marked neutral CVB3 virus in a quantitative fitness assay [15]. Increases in fitness were observed in both cell types, with the most significant increase found by passage 40 in A549 cells (Fig 1C and 1D), suggesting that adaptation had occurred in this less permissive cell type. Whole-genome deep sequencing of these populations revealed significant increases in overall genetic variation, throughout the genome, between the 1st, 20th and 40th passages in both HeLa (Fig 1E) and A549 (Fig 1F) cells (p<0. 0001). The total number of minority variants within the P1 structural coding region also significantly increased over the passage series (Fig 1G), yet there were no significant differences between the numbers observed in Hela and A549 cell passage. The vast majority (>98%) of these variants were low frequency (<1. 0% of the total population), suggesting that moderate genetic drift, rather than positive selection, was responsible for most of this variance. The increase in diversity in both cell types is likely the result of general population expansion in sequence space, since all passage series were started from homogenous in vitro transcribed RNA derived from an infectious clone. The data also confirmed that severe bottlenecking did not occur during the passage series where expansion of diversity should have been stalled or lost. Although the HeLa and A549 cell passage series presented similar mean genetic diversity at passage 40 (Fig 1E–1G, P = 0. 314), we mined the deep sequence data for all minority variants above 1% frequency that might explain adaptation in each condition (Table 1). In both cases, several mutations recurring in multiple replicates mapped to residues known to be part of the CAR receptor and DAF co-receptor footprints [16]; however, the distribution of these mutations was different for the HeLa- and A549-adapted viruses. In HeLa cells, the most abundant variants involved mutations strictly in the CAR footprint (VP1-259,2–7% frequency), in the CAR/DAF shared footprint (VP2-138,20–80%), and in the DAF footprint (VP3-234,2–73% frequency; VP3-63,3–14% frequency). In contrast, viruses passaged in A549 cells presented no variants strictly involved in the CAR footprint. Instead, in addition to CAR/DAF residue VP2-138 (2–81% total), a larger cluster of exclusively DAF-footprint variants were identified (VP3-63,3–35% frequency; VP3-234,3–82% frequency and VP1-271,2–13% frequency). Strikingly, the most abundant mutation in every replicate passaged in A549 cells, and not observed in HeLa cell passage, was VP3-76 (61–95% total). This residue is not known to participate in either footprint. Considering these CAR/DAF-specific differences, we hypothesized that expression levels of these molecules must vary between these cells. There were no significant differences in CAR expression in either cell by both flow cytometry (Fig 2A) or Western blot (Fig 2C). On the other hand, while DAF was very highly expressed in HeLa cells (Fig 2B and 2D), expression was one order of magnitude lower in A549 cells by flow cytometry (Fig 2B) and barely detectable by Western blot (Fig 2D). Normalized quantification of Western blot signals revealed that DAF expression was 4-fold higher in HeLa than in A549 cells (Fig 2E). To characterize the localization of CAR and DAF in these cell types, confocal microscopy was performed. In HeLa cells, CAR (Fig 2F) and DAF (S1 Fig) were expressed throughout the surface of the cell. Interestingly, in A549 cells CAR predominantly localized at cell-to-cell contacts (Fig 2G), while DAF was diffused throughout the surface of the cell (S1B Fig), albeit at considerably lower levels than in HeLa cells (see also, S1 and S2 Movies). It has been shown that adaptation is often a multi-step process, especially in asexual populations like viruses. Usually, until the mutation with the largest beneficial effect becomes fixed, secondary beneficial mutations with less effect are unable to compete, thereby rendering the adaptation process sequential [17–19]. Given the different frequencies of minority variants observed at passage 40, we examined the patterns of emergence and sequential adaptation to novel environments by deep sequencing every second passage in the A549 cell series (Fig 3 and S1 Dataset). A number of step-wise trends were observed across the replicates. In all six replicates, the VP3-E76G mutation was the first mutation to emerge (already above 0. 1% at passage 1) and peak by passage 11 to become the most abundant mutation in the population (between 65 and 95%). Furthermore, the increase in frequency of E76G over the first 11 passages (Fig 3) correlated with the increases in fitness over the same period (Fig 4A), underscoring a considerable contribution of this single mutation to population fitness. Indeed, the E76G variant was generated by reverse genetics and found to confer significantly enhanced fitness in A549 cells (Fig 4B). Although the VP3-76 residue was not previously identified as a contact between virus and receptor, this residue maps to the icosahedral three fold region of the capsid (Fig 4C). In the three dimensional virus-receptor structure, there is an interaction between the C-terminal 6-His tags of three symmetry related DAF molecules [20]. This interference by the His-tag may have restricted the normal interactions of DAF SCR3/4 with the virus surface, masking a potential role for residue 76. In a biolayer interferometry assay using BLItz technology [21] (Fig 4D), E76G bound DAF with the same affinity as wildtype virus, but no increased binding could be detected. Despite the clear fitness advantage conferred by the E76G mutation alone (approximately 10-fold increase with respect to wildtype), it could not account for the more significant fitness increases observed in the passaged populations, particularly at late passage stages (nearly 100-fold increases). Following the fixation of this first mutation, the frequency of VP3-E76G plateaued until a second increase after passage 27, in each of the six replicates, coinciding with the emergence of different combinations of mutations, all mapping to the DAF footprint (Fig 3). Closer examination of the patterns of emerging mutations revealed potential synergistic and antagonistic epistasis among variants. Despite the advantage afforded by extreme depth in coverage, the Illumina technology used here set a limit of 69 nucleotides for read length. In principle, this limitation forfeits the possibility of linking more distant mutations and identifying haplotypes. However, the longitudinal deep sequence data revealed that several mutations increase with parallel kinetics and frequency, suggesting that: i) each mutation appeared in individual genomes and the variants were selected as a group and/or ii) the mutations are accumulating in the same genome, resulting in the selection of single haplotypes. To distinguish between the above cases, phylogenetic trees describing the mix of haplotypes at each passage were inferred from the longitudinal data using maximum likelihood estimation in a Bayesian model. The best-fit, predicted haplotypes were then generated by reverse genetics and their relative fitness values were measured. As expected, this analysis illustrated the quick rise of the E76G genotype that continues to coexist with, although generally dominating over, the original WT genotype (Fig 5A–5F), with a significant increase in relative fitness (Fig 5G). For residue VP3-234, mutations were predicted to arise on either the WT background (Fig 5C and 5F) or the E76G genotype (Fig 5A and 5E). Mutations in residue VP2-138 were also predicted to arise on either WT (Fig 5B) or E76G (Fig 5C, 5D and 5F) genotypes at approximately the same time in the passage series as residue 234. However, their frequencies on the WT background tended to remain lower than on the E76G background where they seemed to be better tolerated. To confirm these modelled predictions, we generated each mutation on both backgrounds and measured the relative fitness. Indeed, the D138G mutation alone on the WT background conferred nearly a ten-fold drop in fitness and the Q234K mutation conferred up to 100-fold drop in fitness (Fig 5G), while the fitness costs of these mutations on the E76G background resulted in neutral fitness relative to WT. The data suggest that epistasis between these mutations and the E76G mutation rescues the fitness of these double variants and permits their positive selection in the viral population. Interestingly, the D138G and Q234K variants seemed to entirely exclude one another from the population (Fig 5A, 5B, 5D and 5E), and the MLE analysis suggested that if both are present in the population, they must occur on separate genomes. We thus generated the E76G-D138G-Q234K haplotype and confirmed that the triple mutant bears a significant fitness cost, rendering this haplotype less fit than the original WT genotype (Fig 5G). Finally, shortly after the appearance of residue 138 and 234 variants, a third mutation appeared during the passage series at residue 63 (Fig 5A, 5B and 5E). Once again, computational modelling predicted that it too exists on the E76G background, and on separate haplotypes than the position 138 and 234 mutations. Furthermore, data from two replicates (Fig 5A and 5E) suggested that the E76G-N63Y double variant bears higher fitness than E76G-Q234K, as its frequency increased towards the end of the passage series as the other' s decreased. Indeed, the residue 63 variants presented higher relative fitness than both the 138 and 234 variants on the E76G background (Fig 5G). Although we could not determine whether residue 76 plays a role in receptor interaction directly, and the role of accompanying mutations are inferred from previously published studies, to rule out that these mutations impact virus fitness on activities downstream of receptor binding and entry, we transfected cells with in vitro transcribed RNAs corresponding to some of these variants and assayed virus production at 8 hours (before a new round of infection can occur). The data revealed that the fitness advantages (E76G, E76G-N63Y) and disadvantages (WT-Q234K, E76G-Q234K) observed during infection of cells (Fig 5G) do not appear to exist when the binding and entry step is bypassed (Fig 5H). Although we initially expected selection of fitness-increasing adaptive mutations to occur by step-wise accumulation of mutations on a single genotype, our computational analysis followed by fitness measurements of double and triple mutation-bearing genotypes suggested otherwise. Importantly, none of these single, double and triple mutations conferred the same fitness increases as those observed in the passage 40 virus population used to identify mutant composition (Fig 5G). Since previous work suggested that overall population fitness results from cooperative interactions among key variants in the mutant swarm, we examined whether a reconstituted composition of the most predominant variants within the passage 40 population could manifest comparable fitness. We thus generated artificial quasispecies presenting mixtures of position 63,138 and 234 mutations on the E76G background and tested their relative fitness. Interestingly, artificial quasispecies presenting 63: 138,138: 234 or 63: 234 combinations at 1: 1 ratios all presented fitness increases as high as, or higher than, the individual values for each variant (Fig 5G). On the other hand, none of these combinations resulted in the highest fitness values that were observed for the passage 40 populations of each replicate used to identify these individual mutations (Fig 5G, p40 a-f). To address whether the minority variant composition may dictate observed population fitness, we reconstituted an artificial quasispecies based on the average frequency of each variant in passage 40 populations. In contrast to the individual variants or the 50: 50 combinations described above, a mixture of 50: 30: 10: 10 of the four most predominant genotypes E76G, E76G-N63Y, E76G-D138G, and E76G-Q234K, reproducibly reached the same fitness values as passage 40 samples themselves (Fig 5G), demonstrating that the global fitness of a virus population is determined by cooperative contribution of minority variants, rather than the dominant genotype alone. Taken together, the sequence data and in vitro characterization of CAR/DAF expression provide a model for the adaptive dynamics of CVB3 to the two host environments presented here (Fig 6). In HeLa cells, where both CAR and DAF are highly and ubiquitously expressed on the surface, adaptive mutations mapped to both footprints. It is not clear whether the CAR-specific mutations (e. g. VP1-K259M) observed in HeLa cells increased interactions with CAR, or conversely, decreased interactions to facilitate the appearance of other mutations related to the DAF footprint. Because the particular strain of CVB3 Nancy used to initiate the passage series already contains some DAF-specific binding residues not found on other Nancy clones, we cannot speculate with confidence which way evolution would go in HeLa cells. In retrospect, inclusion of a CAR-exclusive binding strain of Coxsackie virus in the passage series would have helped discern the direction of evolution in HeLa cells. In A549 cells, on the other hand, the principal receptor CAR is mainly located in tight cell-cell contacts and likely inaccessible to the virus during initial stages of infection; while DAF is present throughout the surface, but at relatively low levels compared to HeLa cells. Thus, the focus of selection is entirely on residues involved in DAF binding (VP3-63, VP3-234) or shared in the CAR-DAF footprints (VP2-138). The CAR-specific mutations observed at residue VP1-259 in HeLa cells, were not observed in A549 cells. The dominant mutation, VP3-E76G, occurred in all six A549-passaged replicates and in none of the HeLa passages—a residue that was not identified as a determinant by conventional structural studies of receptor usage [20]. In the original structural studies, the interaction of residue VP3-76 could not be determined due to steric hindrance of the 6 His-tag between the DAF short consensus repeat domains, SCR3/4, and the virus surface. When CAR is not accessible, DAF is thought to facilitate translocation of the virus to tight junctions containing CAR [22,23]. The interaction with CAR mediates the transition to the A-particle, a required entry intermediate of expanded structure relative to native virus [24]. Nevertheless, it is possible that E76G improves the fitness of the virus in aspects not related to receptor binding, even if we could not identify these mechanisms when comparing replication cycles. Our work reveals the power of deep sequencing to monitor the population dynamics of virus adaptation in new environments. By Sanger sequencing, only the E76G mutation would have been identified in each replicate. The remainder of mutants found to be positively selected during adaptation to host environment were all minority variants that could have otherwise been missed in multiple replicates. An issue inherent to deep sequencing technology is the estimated error of the chemistry (0. 1% for Illumina sequencing), which has been regarded as a caveat to properly describe RNA virus quasispecies or mutant spectra. This problem was recently resolved by elegant molecular biology techniques to remove background error [8,25]. By applying more stringent bioinformatic treatment in our study, between 1700 and 2500 individual, statistically significant point mutations were identified in the structural protein region of the six replicates of A549-adapted populations; yet only mutations in positions related to the DAF footprint underwent selection and amplification over time (S1 Dataset). We thus show that robust bioinformatic treatment coupled with longitudinal data can circumvent error-related issues by identifying changes in variant frequency that would indicate positive or negative selection. This is particularly relevant to in vivo or clinical samples where the quantity of RNA genomes may be too low for more direct sequencing approaches and would require PCR amplification. Similar, relatively simple experimental studies could be designed to understand the population dynamics and evolution of pathogens in new environments during host adaptation and host switching, by identifying the selection over time of one or more mutations as single or multiple genotypes. However, one must keep in mind that although in vitro studies, such as ours, using immortalized cell lines may facilitate studying the dynamics of adaptation; but the specific mutations that are identified may not be indicative of the panel of mutations that would arise in vivo. In this work, adaptation to a novel host environment was a multistep process involving the emergence of an initial mutation (E76G), followed by selection of a number of minority variants on the E76G background. Initially, we expected mutations to accumulate in a step-wise manner in a single genotype; however, fitness assays revealed that genotypes harboring double and triple combinations of these mutations presented fitness decreases relative to the wildtype and/or E76G backgrounds. Instead, computational inference of haplotypes suggested that such variants existed as a heterogeneous population of distinct genotypes. An intriguing observation was that no single variant (nor combination of two variants) conferred the high fitness values observed in the passage 40 populations. Strikingly, only the combination of the four most predominant genotypes, as an artificial quasispecies with the same frequencies observed in the passage 40 populations, was able to confer the same fitness values as the p40 replicate samples. This phenomenon is reminiscent of previous work that suggested that minority variants within an RNA virus quasispecies may contribute significantly to phenotype [26]. In that study, a high-fidelity poliovirus with restricted quasispeces composition was unable to infect the central nervous system (CNS); while the same virus stock that was chemically mutagenized to present wildtype numbers of minority variants restored the ability to disseminate to the CNS. Unfortunately, deep sequencing technology was not yet available and the authors could not uncover the identity of these presumed minority variants by Sanger technology to confirm their hypothesis. Here, we succeed in identifying the variants involved. Importantly, we provide evidence for group selection within the virus population and show that only the group contribution of these positively selected minority variants confers the fitness phenotype observed in the original samples. Our results thus illustrate the significant role of minority genomes in the fitness and phenotype of a virus population, providing further evidence for the quasispecies behavior of RNA viruses under certain conditions. It is important to note, however, that the group selection observed in this study resulted from relatively large population size in passages that reached high MOI by the time cell monolayers were lysed for subsequent passage. It is possible that in conditions where MOI is low, or when population bottlenecks occur (particularly in vivo), that the emergence of such minority variants would be delayed or impeded. HeLa and A549 cells (American Type Culture Collection) were maintained in DMEM medium with 10% new-born calf serum. Coxsackie virus B3 (Nancy strain) was recovered from a pCB3-Nancy infectious cDNA plasmid [27] that was linearized with Sal I and in vitro transcribed using T7 RNA polymerase. It should be noted that unlike other Nancy strains, this infectious clone already has VP2-138D and VP3-234Q residues, known to facilitate binding to DAF. 4 μg of transcript were electroporated into 4 x 106 Vero cells that were washed twice in PBS and resuspended in PBS at 107 cells/ml. For A549 cells, 15 μg of transcript were electroporated into 10 x 106 cells. Electroporation conditions were as follows: 0. 4mm cuvette, 950 μF, 250V, maximum resistance, exponential decay in a Biorad GenePulser XCell electroporator. Cells were recovered in DMEM-10% NCS. For each passage (40 passages total), virus was titrated by TCID50 and 400 μl of medium containing 2 x 105 TCID50 was used to infect 2 x 106 Hela or A549 cells in 6 well plates using a multiplicity of infection (MOI) of 0. 1. Cells were incubated with virus for 45 minutes with frequent rocking, the supernatant was removed and monolayers were washed twice with 2ml PBS, then replenished with 2ml of complete medium. For each passage, virus was harvested at total cytopathic effect (CPE) by one freeze-thaw cycle, representing 2–3 viral generations. Six biologically independent stocks and passage series were generated. Ten-fold serial dilutions of virus were prepared in 96-well flat-bottom plates in DMEM. Dilutions were performed in octuplate and 100 μl of dilution were transferred to 104 Vero cells plated in 100 μl of DMEM-10% NCS. After 5 days living cell monolayers were colored by crystal violet. TCID50 values were determined by the Reed and Muensch method. No significant differences were observed between TCID50 values when using Vero, HeLa or A549 cells as the cellular substrate. 5x108 virion from passaged samples were RNA extracted and RT-PCR amplified by RT (Superscript III) and PCR (Phusion) using primers sets that covered the whole genome, in 3–4 kb fragments. For consensus sequencing, the resulting PCR products were purified, sequenced and analyzed using Lasergene software (DNAStar Inc). For deep sequencing, PCR fragments were purified via the Nucleospin Gel and PCR Clean-up kit (Macherey-Nagel) and total DNA was quantified by Nano-drop. PCR products were then fragmented (Fragmentase), linked to Illumina multiplex adapters, clusterized and sequenced with Illumina cBot and GAIIX technology. Sequences were demultiplexed by CASAVA with no mismatches permitted. Clipping was performed using the fastq-mcf tool, removing common adapter contaminants and trimming low quality bases (Phred<30). Clipped reads were aligned to the Coxsackie virus B3 Nancy sequence as reference with a maximum 2 mismatches per read, and no gaps, using BWA v0. 5. 9. Alignments were processed using SAMTools to obtain a pileup of the called bases at each position. An in-house pipeline, termed ViVAN (Viral Variant ANalysis) [28] was used to identify statistically significant variants above the background noise due to sequencing error, in every sufficiently covered site (>100x). Briefly, for each position throughout the viral genome, base identity and their quality scores were gathered. Each variant was determined to be true using a generalized likelihood-ratio test (used to determine the total number of minority variants) and its allele rate was modified according to its covering read qualities based on a maximum likelihood estimation. Additionally, a confidence interval was calculated for each allele rate. In order to correct for multiple testing, Benjamini-Hochberg false-discovery rate of 5% was set. The total allele rates passing these criteria, across the whole genome, were used to calculate the mean variation rates (diversity) at different passages. The variation rate at position i is defined as the proportion (F) of significant non-reference alleles (k) and is denoted Vi: Vi=∑j=1kFij The region-wide variation rate is the averaged variation rate across all covered positions in the genome (denoted n): V=∑i=1nVin Hela and A549 cells were plated onto coverslips, fixed with 2% paraformaldehyde for 20 minutes at room temperature, and then washed with PBS. Before staining, non-specific staining was blocked by incubating the cells with 5% FCS and 0. 05% saponin in PBS during 10 minutes. Staining antibodies were diluted also in this buffer, and it was used for washes between antibodies. Cells were incubated with either CAR (Santacruz) or DAF (Abcam) primary antibody, washed, stained with a secondary antibody coupled to the appropriate fluorophore and washed. Cells were analyzed using a Zeiss LSM-700 confocal microscope. 3D reconstruction of the images was performed using the Imaris software. One million cells of HeLa and A549 cells were lysed using a buffer containing 1% Triton-X and 1% sodium deoxycholate with protease inhibitors (Sigma). A fraction of the lysate was run for 1 hour on a 4–15% gradient gel (Biorad) on denaturalizing conditions. After the run, we performed the transfer to a nitrocellulose membrane. We washed the membrane with PBS-T (PBS 1X and 0. 1% Tween-20) and blocked for 1 hour in PBS-T plus 5% milk. After the blocking we washed again with PBS-T and left overnight with each one of the antibodies, anti-CAR and anti-DAF (Santacruz Biotechnology). We washed the membrane and we added the secondary fluorescent antibodies (DyLight 680 and DyLight800 conjugated, Thermo Scientific) for 1 hour. We washed the membrane one last time and measure the fluorescence in the Odyssey system (Li-Cor). In order to prepare cells for cell cytometry analysis, cells were washed twice with PBS 1x and trysinized, washed again twice in PBS. Cells were stained for 30 minutes with either CAR-PE or DAF-FITC antibodies (Millipore and Abcam) on ice. Unbound antibody was discarded and cells were washed again with PBS 1x. Cells were resuspended in 1% Parafolmaldehyde (PFA, Electron Microscopy Sciences) and kept in the dark for 15 minutes. After the incubation PFA was discarded and cells were resuspended in 200 μl of PBS 1x. Cells were kept at 4C until analysed. For each cell type used (HeLa and A549) specific instrument settings were set according to the size and complexity of the cell type, as well as antibodies fluorescence. Samples were analysed using the MACSquant flow cytometer (Miltenyi Biotec) using 96 well plates and obtaining 10,000 events per sample. Mock samples were also used in each plate to setup the baseline. Results were analysed using Flowjo software v10. Relative fitness values were obtained by competing each virus population with a marked reference virus that contains four adjacent silent mutations in the polymerase region introduced by direct mutagenesis. Co-infections were performed in triplicate at an MOI of 0. 01 using a 1: 1 mixture of each variant with the reference virus for 24 hours. The proportion of each virus was determined by real time RT-PCR on extracted RNA from the infection supernanant, using a mixture of Taqman probes labelled with two different fluorescent reporter dyes. MGB_CVB3_WT detects WT virus (including the fidelity variants) with the sequence CGCATCGTACCCATGG and labelled at the 5’ end with a 6FAM dye (6-carboxyfluorescein) and MGB_CVB3_Ref containing the four silent mutations: CGCTAGCTACCCATGG was labelled with a 5’ VIC dye. Each 25 μl-reaction contained 5ul of RNA, 900nM of each primer (forward primer, 5’-GATCGCATATGGTGATGATGTGA-3’; reverse primer, 5’-AGCTTCAGCGAGTAAAGATGCA-3’) and 150 nM of each probe. The relative fitness was determined by the method described by Carrasco et al. [15]. Briefly, the formula W = [R (t) / (R (0) ) ] ^ (1/t), represents the fitness, W, of each mutant genotype relative to the common competitor reference sequence, where R (0) and R (t) represent the ratio of mutant to reference virus densities in the inoculation mixture and t days post-inoculation, respectively. The fitness of the normal wildtype to reference virus was 1. 019, indicating no significant differences in fitness due to the silent mutations engineered in the reference virus. CVB3 parental strain, and CVB3-E76G were propagated in HeLa cells and purified as described previously [14]. The membranes of infected cells were broken by three freeze-thaw cycles. Virus was concentrated by pelleting through sucrose and purified by tartrate step gradient ultracentrifugation. The virus bands were collected and exchanged into PBS and virus concentration and quality was estimated by measuring the absorbance at 260,280, and 310 nm. Binding assays were performed by biolayer interferometry (BLI) measured by the BLItz from Fortebio [21]. Briefly, BLI measures binding by sending white light down a glass fiber-based biosensor, which is reflected back up to the instrument from two interfaces: 1) the interface between the glass fiber and the biosensor, and 2) the interface between the surface chemistry and solution. Since the two reflections come from the same white light source in the instrument, they both contain the same wavelengths. When molecules bind to the surface of the biosensor, the path length of the reflection (the one reflecting from the interface between surface and solution) increases while that of the other reflection remains the same. 0. 4% BSA and 0. 08% Tween was added to virus and DAF to prevent non-specific binding to the sensor during the assay. Purified, His-tagged DAF was diluted to 0. 1 mg/ml and attached to a Ni-NTA sensor. The DAF loaded sensor was then dipped into virus diluted to 0. 5 mg/ml. Let X be a phylogenetic tree, taken to include relative population sizes of the branches, and let Xt = (stpt) T be the state of the tree at time t = 1,2, …, T. Here st denotes the structure of the tree, i. e. the set of branches existing at time t and pt is a vector with the relative population sizes of all existing branches. Furthermore, let Y be the set of measurements, where Yt ∊ ℤ+64xN is the number of reads supporting each of the 64 codons at each of the N different positions in the genome. By Bayes' theorem, P (X|Y) ∝P (Y|X) P (X) The output probability P (Y | X) follows a multinomial distribution for each position. The dynamics of the phylogenetic tree are naturally assumed to be Markovian, i. e. P (X) =P (X1) ∏t=2TP (Xt|Xt−1), where the transition probability can be expressed using the structural and population dynamics parts separately, P (Xt|Xt−1) =P (pt|st, st−1, pt−1) P (st|st−1, pt−1) =P (pt|st, pt−1) P (st|st−1, pt−1). A simple random walk model is used for the population dynamics P (pt | st, pt-1), the population change from t−1 to t is taken to be lognormally distributed with standard deviation σh where h is the length of the time step. The structural part, P (st | st−1, pt−1), depends on the mutation frequency of the virus and the population size of the haplotype in which the mutation occurs. The posterior probability is thus, assuming that s1 and p1 are known, P (X|Y) ∝P (Y|X) ∏t=2TP (pt|st, pt−1) P (st|st−1, pt−1) To make a maximum likelihood estimation (MLE) of X from the posterior tractable, some approximations are made. The attention is limited to a small set of variants, greatly reducing the number of possible tree structures. The most prevalent variants in the data should be included as they dominate the dynamic behaviour, but minority variants of particular interest can also be added. Every possible tree structure matching the selected set of variants is generated. The time of appearance for each variant is set to where it is first seen in the measurements, i. e. the first time the frequency is above a small threshold. An MLE of P (pt | st, Yt) is then computed for each tree and time point. The rationale is that there is very little freedom for the population sizes to deviate from a point which can explain the output data. By construction of the tree structure, the number of non-reference haplotypes equals the number of variants at each time point in every tree. Hence, there is at most one convex combination of the haplotypes that match the event probabilities of the multinomial distribution for the output data. Due to the high number of reads in the deep sequencing data, moving away from the optimal point will cause a quick drop in the posterior probability. Dependencies between variants that are close enough to be covered by a single read are included in the model (amino acid residues 63 and 76). The MLE of X can be found by evaluating the posterior for each generated tree and picking the most likely. All generated trees contain the same number of mutations. Since the value of the posterior probably only needs to be known up to a multiplicative constant, the effect of the overall mutation frequency of the virus on P (st | st−1, pt−1) cancels out. Hence, P (st | st−1, pt−1) is simply proportional to the population size of the haplotypes in which the mutations occur.

When RNA viruses replicate, they do so with a high rate of error; hence, their populations are not composed of a single genotype, but of a swarm of different, yet related, genomes. This mutant spectrum has been described as the viral quasispecies, and its composition has important consequences for evolution, adaptation and emergence. In this study, we analysed adaptation in fine detail thanks to the use of the deep sequencing, and we determined the adaptative pathway of a model RNA virus, Coxsackievirus B3, to a new environment, A549 cells. Our results demonstrate that adaptation occurred in response to a differential expression of the virus receptors in the new cellular environment, compared to the former. Our experiments and mathematical analyses established that the corresponding increase in fitness resulted from the selection and contribution of a group of genotypes, including low frequency variants, and not to the effect of a single, dominant genome. Our work underscores the importance of considering group effects when studying RNA virus biology and evolution.

Abstract Introduction Results Discussion Materials and Methods

2015

Group Selection and Contribution of Minority Variants during Virus Adaptation Determines Virus Fitness and Phenotype

Eukaryotic striatin forms striatin-interacting phosphatase and kinase (STRIPAK) complexes that control many cellular processes including development, cellular transport, signal transduction, stem cell differentiation and cardiac functions. However, detailed knowledge of complex assembly and its roles in stress responses are currently poorly understood. Here, we discovered six striatin (StrA) interacting proteins (Sips), which form a heptameric complex in the filamentous fungus Aspergillus nidulans. The complex consists of the striatin scaffold StrA, the Mob3-type kinase coactivator SipA, the SIKE-like protein SipB, the STRIP1/2 homolog SipC, the SLMAP-related protein SipD and the catalytic and regulatory phosphatase 2A subunits SipE (PpgA), and SipF, respectively. Single and double deletions of the complex components result in loss of multicellular light-dependent fungal development, secondary metabolite production (e. g. mycotoxin Sterigmatocystin) and reduced stress responses. sipA (Mob3) deletion is epistatic to strA deletion by supressing all the defects caused by the lack of striatin. The STRIPAK complex, which is established during vegetative growth and maintained during the early hours of light and dark development, is mainly formed on the nuclear envelope in the presence of the scaffold StrA. The loss of the scaffold revealed three STRIPAK subcomplexes: (I) SipA only interacts with StrA, (II) SipB-SipD is found as a heterodimer, (III) SipC, SipE and SipF exist as a heterotrimeric complex. The STRIPAK complex is required for proper expression of the heterotrimeric VeA-VelB-LaeA complex which coordinates fungal development and secondary metabolism. Furthermore, the STRIPAK complex modulates two important MAPK pathways by promoting phosphorylation of MpkB and restricting nuclear shuttling of MpkC in the absence of stress conditions. SipB in A. nidulans is similar to human suppressor of IKK-ε (SIKE) protein which supresses antiviral responses in mammals, while velvet family proteins show strong similarity to mammalian proinflammatory NF-KB proteins. The presence of these proteins in A. nidulans further strengthens the hypothesis that mammals and fungi use similar proteins for their immune response and secondary metabolite production, respectively. Signaling pathways that regulate morphological and physiological processes in response to stimuli are often highly conserved throughout eukaryotes, signifying their importance. Striatin is one of the regulatory proteins proposed to act as a signalling hub for the control of many cellular processes including development, cellular transport and signal transduction [1,2]. It forms a scaffolding platform to build the striatin-interacting phosphatase and kinase (STRIPAK) complex which is a large multimeric protein complex highly conserved in eukaryotes [3]. The STRIPAK complex influences mammalian cell size, morphology and migration [1]. It also plays a role in the polarisation of the golgi apparatus and is implicated in the process of mitosis through tethering vesicles of the golgi to the nuclear membrane and centrosomes [4]. The mammalian STRIPAK complex consists of a multitude of core members which include (i) Striatins (ii) Striatin-interacting proteins (STRIP1/STRIP2), (iii) monopolar spindle one-binder (Mob3/phocein) protein, [5], (iv) cerebral cavernous malformation 3 protein, CCM3 (v) and the phosphatase 2A subunits PP2AA and PP2Ac that have structural and catalytic functions, respectively. Further associated proteins include cortactin-binding proteins (CTTNBP), suppressor of IKKε (SIKE) and sarcolemmal membrane associated protein (SLMAP) and multiple germinal centre kinases (GCKIII), such as STK24, STK25 and MST4 that belong to the STE20 kinase family [3,6]. The GCKs were discovered to be involved in the control of the cell cycle, polarity and migration [7,8] and their functionality is reliant on CCM3 which is involved in stabilising the kinases [9]. The STRIPAK complex in the fruit fly Drosophila melanogaster acts as a negative regulator of the Hippo signaling pathway [10]. In Saccharomyces cerevisiae, the homologous complex is termed the Far complex which is implicated in cell cycle arrest and acts as an antagonist towards target of rapamycin complex 2 (TORC2) signaling [11,12]. In the fission yeast Schizosaccharomyces pombe, the STRIPAK complex is known as the SIP complex (Septation initiation network [SIN] Inhibitory PP2A complex). This complex is necessary for coordinating mitosis and cytokinesis in S. pombe [13]. In closely related fungi Neurospora crassa and Sordaria macrospora, the STRIPAK complex controls cell-cell recognition, cell fusion and fruit body formation [6,14,15]. In S. macrospora, STRIPAK complex was initially discovered to be composed of PRO22 (STRIP1/2), PRO11 (STRN), PP2AA and PP2Ac1 by using Tandem Affinity Purification (TAP) coupled to mass spectrometry (MS) [14]. The SLMAP homolog PRO45 was also shown to be part of the complex [16]. The N. crassa STRIPAK complex, which is required for aerial mycelium formation, conidiospore formation and cell-cell signaling, is composed of the core HAM2 (STRIP), HAM3 (STRN), catalytic subunit PPG1 (PP2Ac) components and the accessory proteins HAM4 (SLMAP) and MOB3 (Mob3) [15]. In both fungi, the STRIPAK complex is made of STRIP, STRN, PP2AA, PP2Ac1, SLMAP and several GCKs [16]. It has very recently been shown that a SIKE homolog of S. macrospora, STRIPAK complex interactor (SCI1), a small coiled-coil protein, interacts with striatin (PRO11) and is required for hyphal fusion and fruit body development [17]. In both fungi, several subunits of the STRIPAK complex are associated with the nuclear envelope [15–17]. Furthermore, N. crassa STRIPAK is required for nuclear accumulation of mitogen activated protein kinase (MAPK), MAK-1, which is important for the cell-wall stress pathways as well as for cell-cell communication [15]. Fungi produce secondary metabolites (SM) ranging from beneficial antimicrobials, insecticides and antitumor agents to deleterious mycotoxins. SM genes, which are usually clustered, are expressed in response to environmental signals such as light, starvation and stress conditions [18,19]. A model fungus frequently used to study SM production is Aspergillus nidulans. This fungus reproduces by forming asexual (conidia) or sexual (ascospores) spores [20,21]. Germination of either type of spores leads to long hyphal filaments that in response to different light regimes develop into either asexual or sexual organs. In the light, the hyphae mainly form aerial conidiophores that produce clonal conidia. In the dark, the fungus switches to sexual reproduction and formation of ascospores that are formed inside of multicellular higher-ordered spherical fruit bodies (cleistothecia). Defects in the different developmental programs frequently impair SM production [20]. A. nidulans produces many metabolites, including the carcinogenic mycotoxin Sterigmatocystin (ST), antibiotic Penicillin (PN) and antitumour agent Terrequinone (TQ) [22]. Development and SM production are controlled by regulatory complexes, including the light-operated heterotrimeric VeA-VelB-LaeA complex. The velvet complex is formed in the nucleus where it controls gene expression required for sexual development and SM production [23]. The velvet complex is further controlled by the MAPK MpkB (yeast Fus3 ortholog), which phosphorylates VeA and is involved in regulation of cell-cell fusion, sexual development and SM production. All upstream kinases of MpkB, like MAP3K SteC (yeast Ste11p), MAP2K MkkB (yeast Ste7p) and the adaptor protein SteD (yeast Ste50p) also participate in sexual development and SM production [24,25]. In A. nidulans, striatin (StrA) is localized to the nuclear envelope and loss of StrA results in pleitropic effects, including reduced hyphal growth, conidiation and loss of ascospores [26]. However, the molecular composition of the AnSTRIPAK complex, where it localizes, how it controls growth and development and whether it is involved in SM production are currently unknown. Therefore, we have attempted to understand mechanistic functions of the AnSTRIPAK complex. We report the composition of the AnSTRIPAK complex, its subcellular assembly and its roles in fungal development and SM production. Furthermore, we show the presence of SIKE-like protein as a component of the AnSTRIPAK in fungi in addition to S. macrospora coiled-coil domain protein SCI1 [17] and present the functional interplay between the AnSTRIPAK complex and two MAPK pathways, involved in development and stress responses. To reveal and understand the molecular interaction network of StrA (AN8071), a fully functional StrA-TAP fusion expressed from its native locus under the control of the endogenous promoter was used for tandem affinity purification (TAP) and liquid chromatography-mass spectrometry (LC-MS/MS) (Fig 1A, S1–S6 Tables). Six proteins associated with StrA, termed striatin interacting proteins (Sips), were identified from vegetative cultures grown for 24 hours: AN6190/SipA, AN1010/SipB, AN6611/SipC, AN4632/SipD, AN0164/SipE (PpgA) and AN4085/SipF. For consistency, SipE and PpgA refer to the same protein within the text. StrA was expressed at higher levels during early stages of asexual and sexual development in comparison to vegetative growth (S1 Fig). Therefore, TAP-MS/MS was performed from light and dark induced cultures (6,24,48h). StrA interacted with the same set of proteins SipA-F during early sexual and asexual development (6h). However, it recruited only SipA at late developmental time points (24,48h) (Fig 1A). With the exception of SipB, all other interacting partners of StrA are conserved components of the STRIPAK complex in fungi (S7 Table). SipA is an ortholog of human and fungal Mob3 (Phocein) with 480 amino acids (Fig 1B and 1C). Mob3p is a kinase co-activator and a part of the STRIPAK complex in humans, fruit flies and filamentous fungi [15]. Interestingly, The C-terminus of SipB (444 aa) contains a suppressor of IKKε (SIKE) -like domain (S2 Fig), an ortholog of which has only recently been shown to exist in S. macrospora [17]. SIKE-like proteins contain a coiled-coil domain which is conserved in fungi. Alignment of two human SIKE isoforms with the c-terminus of A. nidulans SipB indicates 46% similarity between the proteins (S2 Fig). SipC (1063 aa) is the largest of all with two putative domains, an N1221family domain and a domain of unknown function (DUF3402). It is an ortholog of human STRIP1/2, NcHAM-2 and ScFar11. SipD (747 aa) is an ortholog of human SLMAP, NcHAM-4, SmPRO45 and ScFar10. Both proteins HAM-4 and Far10p control similar developmental events as HAM-2 and Far11p. SipE (PpgA) (320 aa), the smallest of all, is an ortholog of the catalytic subunit of human phosphatase (PP2Ac), NcPP2Ac and ScPppg1. SipF (616 aa) is the homolog of the human phosphatase regulatory subunit (PP2AA) with fungal homologs NcPP2AA, ScTpd3, SpPaa1, respectively. Since SipA interacted with StrA during all developmental stages, expressions of the functional Sip-GFP fusions (S3 Fig) were monitored during vegetative, asexual and sexual stages (Fig 1D). The catalytic subunit SipE (PpgA), but not the regulatory subunit of phosphatase (SipF) was used for expression studies since sipF deletion was lethal. A 79kDa SipA-GFP and a 76kDa SipB-GFP fusion were present during almost all developmental time points. A 145kDa SipC-GFP fusion was poorly expressed during developmental stages. A 107kDa SipD-GFP fusion was constantly expressed at all stages except for late sexual development (24 and 48h). Interestingly, the 63kDa SipE-GFP fusion was only present during vegetative growth and degraded and disappeared at both asexual and sexual stages. In order to define the core STRIPAK complex more precisely, reciprocal TAP-MS/MS was performed. TAP of SipA recruited StrA along with SipB to SipF (Fig 1E). Similarly, TAP of SipB, SipC, SipD or SipE also recruited all members of the complex (S8–S12 Tables). These interactome data clearly underline that a heptameric STRIPAK complex made of a striatin StrA, a Mob3 kinase ortholog SipA, a SIKE-like protein SipB, a STRIP1/2 ortholog SipC, an SLMAP ortholog SipD, and the phosphatase subunits SipE (PpgA) and SipF exists in A. nidulans. To understand the roles of the AnSTRIPAK complex in fungal development, individual sip deletions, sip; sip double deletions and combinations of sip deletions with strAΔ were created (S4 Fig) and subjected to developmental tests (Fig 2). Loss of sipB, sipC, sipD and sipE all resulted in similar phenotypes to that seen in the strAΔ strain, characterized by slower growth rate, reduced conidiation and lack of fruit bodies. sipE (ppgA), encoding one of the two phosphatase 2A catalytic subunits of A. nidulans was shown to influence growth together with more than twenty other phosphatases [27]. All attempts to delete sipF encoding PP2A-A regulatory subunit failed, suggesting that the gene is essential for viability of A. nidulans as it is in N. crassa [15]. The defects were complemented by introducing the corresponding genes into the deletion strains (Fig 3). Furthermore, sipΔ/sipΔ double deletions and sipΔ/strAΔ double deletions were similar to single deletion strains. Surprisingly, sipAΔ showed an opposite phenotype with significantly increased radial growth and two-fold more conidiation than the WT and production of normally shaped fruit bodies, which were devoid of ascospores (Fig 2). Moreover, strAΔ/sipAΔ phenocopied sipAΔ, suggesting an epistatic effect of sipA over the strA. In contrast, sipA was not epistatic to sipB to sipE since sipA double deletion combinations with other sip genes behaved similar to single sip deletions, showing that sipA is only epistatic to StrA yet not the other members of the AnSTRIPAK complex. These results show that almost all members of the AnSTRIPAK complex are equally important for growth and light-dependent development except for SipA. However, SipF is likely essential for the viability of A. nidulans. The STRIPAK complex mutants showed developmental defects, suggesting an essential role of the complex in signal transduction of developmental processes. In order to see whether the AnSTRIPAK complex is also involved in stress responses, mutants were subjected to various stress conditions (Fig 3). The radial growth of all single and double deletions except for sipAΔ were significantly reduced in oxidative stress (H2O2, Menadione) and cell wall stress (Congo Red) media in comparison to WT (Fig 3A–3C). sipAΔ was more resistant to oxidative and cell wall stressors than the WT. However, under stress conditions sipAΔ did not show epistatic effects over the strAΔ since the sipA/strA double mutant was as sensitive as the strA single deletion and the double deletions of strA and sip genes to both stress conditions. Strains were also monitored to see how they cope with DNA damage, amino acid starvation (3-AT), caffeine and osmotic stress. Similar to oxidative and cell wall stress, sipAΔ displayed more robust vegetative growth than the WT under all tested conditions. strA double deletions with sip genes, interestingly including the strA/sipA double mutant, were extremely sensitive to all three types of DNA damage conferred by HU, MMS and EMS (S5 Fig). Amino acid starvation mainly influenced strA/sip double deletions (S6 Fig). All of the deletion strains excluding the sipA deletion were slightly sensitive to osmotic stress induced by NaCl. These data clearly indicate that the lack of AnSTRIPAK complex results in drastic problems in combating various types of stressors. Furthermore, the epistatic effect of SipA over the StrA deletion is abolished in the presence of stress conditions, suggesting an interplay between SipA and StrA in regulating stress responses. All mutants of the AnSTRIPAK complex, except for sipAΔ, exhibited drastic changes in asexual and sexual development, both of which are regulated by a cascade of transcription factors. We consequently determined the effects of our deletions on expression of these transcription factors by qRT-PCR. Expression of major transcription factors that control conidiophore (abaA, brlA) and sexual (nsdD, steA) development were significantly reduced in the STRIPAK mutants except for sipAΔ (Fig 4A and 4B) which caused 3-fold higher brlA, abaA and 2-fold nsdD, steA expression. This increase was consistent with the increased asexual development of the sipA mutant. AnSTRIPAK complex mutants were sensitive to oxidative stress. Therefore, expression of two ROS scavenging enzyme encoding genes, catalase catC and superoxide dismutase sodB were monitored in the mutants. Expression of both genes was significantly decreased in all AnSTRIPAK mutants except for sipAΔ (Fig 4C), where they were slightly upregulated, which was consistent with the higher resistance of this strain to stressors when compared with WT. These expression data demonstrate that the AnSTRIPAK complex is required for the balanced expression of developmental regulators and ROS scavenging enzyme genes. A defect in fungal development particularly in fruit body formation (sexual development) is often associated with changes in SM production. All members of AnSTRIPAK complex are involved in growth and fruit body formation, excluding the SipA protein which is specifically required for ascospore formation in fruit bodies in A. nidulans. Therefore, levels of the fungal mycotoxin sterigmatocystin (ST) were measured in mutants by HPLC (Fig 4D). Production of ST was significantly reduced in mutants in comparison to the WT. strA and sipA mutants produced less ST than WT but more than sipB, sipC, sipD and sipE (ppgA) mutants. ST production of the strA/sipA double mutant was slightly less than the respective single mutants. However, double deletion of strA with sipB, sipC, or sipE resulted in extremely reduced ST (less than 10% of WT). In order to examine if the drastic alterations in SM production in AnSTRIPAK mutants were due to changes in transcript levels of the SM regulators, the expression profiles of the velvet complex along with ST gene cluster as well as two additional SM clusters were determined. Expression of the velvet complex (veA/velB/laeA) was generally diminished in the absence of AnSTRIPAK complex (Fig 4E). The most drastic decrease was observed in velB and veA expression. The reduced expression of the velvet complex is translated into expression of the ST gene cluster since expression of the transcription factor aflR and the two structural genes stcQ and stcE sharply dropped in AnSTRIPAK mutants except for sipAΔ (Fig 4F). Antibiotic penicillin (PN) and antitumour terrequinone (TQ) genes were tested in addition. Surprisingly, expression of acvA, ipnA and aatA required for PN production was reduced in AnSTRIPAK mutants. ipnA expression in sipBΔ and aatA expression in sipDΔ did not change (Fig 4G). Similarly, tdiA and tdiB genes of the TQ cluster were significantly down-regulated in AnSTRIPAK mutants except for the sipA mutant, which was higher (Fig 5E). These metabolite and expression data reveal that the AnSTRIPAK complex is important for production of ST and expression of the velvet complex. Furthermore, full expression of at least three different gene clusters ST, PN and TQ require an intact AnSTRIPAK complex, except for sipA. Striatin (StrA) is localized to the nuclear periphery and endomembrane systems in A. nidulans [26] (S1 Fig). However, it is not known if the entire STRIPAK complex is also associated with the nuclear envelope in A. nidulans. We have found that SipA to SipF interact with StrA constituting the heptameric AnSTRIPAK complex. A functional StrA-mRFP fusion was co-expressed with SipA, SipB, SipC, SipD and SipE-GFP fusions (Fig 5). The StrA-mRFP fusion, which clearly decorated the nuclear periphery where the nuclear envelope is found, was not primarily found inside the nucleus. Furthermore, StrA was also present on long string-like extensions, presumably representing endomembrane systems such as endoplasmic reticulum. The SipA-GFP fusion was also found to be accumulated around the nucleus and in string-like extensions but was present in the nucleus at trace levels and overlapped with StrA-mRFP signals (Fig 5A). Similarly, GFP fusions of SipB and SipC were found to be co-localised with StrA around the nucleus (Fig 5C). In co-localizations with histone H2A, SipD-GFP and SipE-GFP showed clear perinuclear localization, very similar to those of SipA-GFP, SipB-GFP and SipC-GFP (Fig 6). Colocalizations of SipD and SipE (PpgA) -GFP with StrA-mRFP showed a somewhat different pattern. The staining was less punctate and the localization around the nucleus was visible but weaker than that of SipA- to SipC-GFP (Fig 5D and 5E). We speculate that the mRFP tag on StrA weakens the binding of SipD and SipE to StrA. Like StrA, SipA to SipE were all co-localized with StrA around the nuclear envelope and partially in endomembrane systems. Since sip double deletions with strA led to more sensitive phenotypes suggesting the key role of StrA for the molecular function of the AnSTRIPAK complex, SipA to SipE-GFP fusions were expressed in a strain devoid of StrA (Fig 6). The absence of StrA did not influence expression of the Sip proteins except for SipD, which showed a higher molecular weight as well as a thicker lower molecular band (Fig 6D). Interestingly, lack of StrA led to loss of nuclear envelope localization of SipA, which became more diffuse in the cytoplasm. SipB-GFP also lost its nuclear envelope localization in the absence of StrA and dispersed in the cytoplasm and, interestingly, was present in the nucleoplasm except for the nucleolus (Fig 6B). SipC also dispersed from the nuclear envelope and was relatively uniformly distributed in the cytoplasm but it was at least partially excluded from nuclei (Fig 6C). SipD remained punctate in the absence of StrA, but it was no longer concentrated at the nuclear envelope, and many punctae were seen in the nucleoplasm (Fig 6D). SipE (PpgA) showed a similar localization pattern to SipB in the absence of StrA, diffuse in the cytoplasm and nucleoplasm but excluded from the nucleolus (Fig 6E). In summary, our expression data and confocal imaging data reveal that all of the members of the AnSTRIPAK complex localize to the nuclear envelope and endomembrane system. They require the molecular scaffold StrA for normal localization and all but SipD disperse in the absence of StrA. SipD does not disperse in the absence of StrA but its localization pattern is altered. Given the fact that StrA is required for appropriate cellular localization of the AnSTRIPAK complex, it became intriguing to ask whether the mislocalizations reflect the interdependent interactions of the complex proteins. As assayed by TAP purification followed by MS, SipA to SipE all complexed with each other with high peptide numbers in the presence of StrA (Fig 6F, S13–S17 Tables). However, surprisingly, in the absence of StrA, TAP purification of SipA did not pull down any other members of the complex. SipB and SipD only reciprocally copurified with each other. SipC and SipE (PpgA) reciprocally pulled down the regulatory subunit of phosphatase SipF as well as a second regulatory subunit B (PabA; presumably reflecting a distinct, STRIPAK-independent PP2A complex). Surprisingly, in the absence of SipA, StrA was able to establish a form of the AnSTRIPAK complex lacking only SipA (S18 Table). These TAP data comparing the physical interaction dynamics of AnSTRIPAK complex in the presence and absence of StrA clearly display that (I) SipA only interacts with StrA, therefore it is recruited to the AnSTRIPAK complex via StrA and StrA does not need SipA to establish the AnSTRIPAK complex, (II) SipB-SipD form heterodimers and then presumably associate with StrA, (III) SipC-SipE-SipF form a heterotrimeric subcomplex, which is then attached to StrA to form the fully functional AnSTRIPAK complex. It was shown that activation of the N. crassa cell-wall stress pathway MAPK MAK-1 as well as its nuclear accumulation was reduced in core components of STRIPAK complex mutants [15]. The same study also showed that MAK-2, which regulate hyphal fusions in N. crassa, phosphorylates MOB-3 component of STRIPAK. Striatin mutants in A. nidulans show developmental and SM defects. Furthermore, the mutants also show sensitivity to cell-wall and oxidative stressors. Therefore, we wondered how common MAPK pathways are influenced by the STRIPAK complex and whether the influence of STRIPAK on MAPK pathways was similar to that of N. crassa. In addition to MpkB which controls sexual development and SM production, there are three mitogen activated protein kinases (MAPK): MpkA is mainly responsible for cell-wall regulation, MpkC and SakA (yeast Hog1 ortholog) play roles in stress responses, particularly oxidative and osmotic stress responses [24,28]. To determine if StrA influences the localizations of these kinases, the three MAPKs, MpkA (cell-wall stress), MpkB (sexual development and SM production) and MpkC (oxidative and osmotic stress) were expressed as GFP fusions in the presence and absence of StrA (Fig 7). All kinases were expressed similarly in the presence or absence of StrA (Fig 7A). Surprisingly, activation phosphorylation (P-44/42) of MpkB, which is necessary for fruit body formation, was almost totally lost in the absence of StrA. MpkB localization, which was not influenced by lack of StrA, was mainly nucleo-cytoplasmic (Fig 7B). Furthermore, MpkB (yeast Fus3p) interacted with MAP2K MkkB (yeast Ste7p), recently characterized scaffold protein HamE and transcription factor SteA (yeast Ste12p) in the absence of StrA (S19 and S20 Tables). In contrast to N. crassa MAK-1, MpkA exhibited nuclear localization both in WT and strAΔ strains, which was not influenced by cell-wall damaging antifungal drug caspofungin (Fig 7C). Interestingly, MpkC, which was mainly found in the cytoplasm under non-stress conditions in the WT, imported into the nucleus in the absence of StrA (Fig 7D). Caspofungin treatment had no influence on the localization of MpkC. Both oxidative and osmotic stress conditions led to nuclear accumulation of MpkC in WT. However, lack of StrA resulted in loss of MpkC nuclear enrichment under oxidative stress whereas osmotic stress had no effect on MpkC localization in the strA deletion. These results imply that StrA acts differently in A. nidulans and mainly prevents MpkC from entering into the nucleus under non-stress conditions to respond to stress and is required for nuclear accumulation of MpkC under oxidative stress conditions. Furthermore, StrA controls phosphorylation of MpkB, which is required for coordination of development and SM production. The STRIPAK complex is highly conserved in eukaryotes and is involved in many cellular functions [1,6]. In this study, we have revealed the molecular nature and functions of the AnSTRIPAK complex, which consist of at least seven proteins StrA/STRN, SipA/Phocein, SipB/SIKE1-like, SipC/STRIP, SipD/SLMAP, SipE (PpgA) /PP2Ac and SipF/PP2AA. Detailed phenotypic, genetic, biochemical, live cell imaging and chemical approaches identified StrA as the core scaffolding protein which assembles at least six other members at the nuclear envelope to control intracellular signaling events (Fig 7F). StrA is found in a heterodimer state with SipA/Phocein and further recruits the heterodimer SipB/SipD (SIKE/SLMAP) and the heterotrimer SipC/SipE/SipF (STRIP/PP2Ac/PP2AA) to establish a heptameric complex. This complex controls gene expression for the sexual reproductive cycle, including formation of multicellular fruit bodies and furthermore, it is also a key regulator for production of SMs. Although the molecular composition of the STRIPAK complex is conserved, the described functions of the complex show diverse roles in fungi, flies and mammals. In the baker´s yeast S. cerevisiae, the FAR (STRIPAK) complex acts as an antagonist of target of rapamycin (TOR) pathway and counteracts recovery from pheromone arrest [11,12] whereas in the fission yeast S. pombe, the STRIPAK complex acts as a negative regulator of septation initiation [13]. In closely related filamentous fungi S. macrospora and N. crassa, STRIPAK regulates hyphal fusion and fruit body formation [14,15]. In two Fusarium species, which are plant pathogenic fungi, striatin is required for pathogenicity on host organisms [29,30]. In the fruit fly, the STRIPAK complex controls epithelial cell movement and tissue size by modulating two different signaling pathways [31,32], whereas in the nematode C. elegans, members of the complex control polarity establishment during embryogenesis [33]. In humans, STRIPAK complex governs embryonic stem cell differentiation, proper cardiac function, dendritic spine morphology and cancer [34,35]. Lack of fully assembled AnSTRIPAK complex results in loss of proper light response as a result of defective asexual and sexual development. The AnSTRIPAK complex controls light-dependent fungal development. In this fungus, light controls asexual reproduction through the various light receptors. The velvet complex physically and functionally interacts with the red and blue light receptors [36]. Improper expression of the velvet complex in STRIPAK mutants might influence interaction dynamics of the complex with light receptors, therefore, disrupting the light-dependent development. Particularly, two major asexual transcription factors abaA and brlA are induced by AnSTRIPAK complex, which drives asexual responses. The role of AnSTRIPAK complex in sexual development might be somewhat complicated. It controls formation of fruit bodies by properly dosing expression of sexual transcription factors such as nsdD and steA. It is known from N. crassa and S. macrospora that the STRIPAK complex is involved in cellular fusion, which finally leads to fruit body formation in these fungi. In A. nidulans, fruit body formation is also controlled by formation of cell-cell fusions. Loss of fruit bodies also indicate that there are defects in cell-cell fusions in the absence of AnSTRIPAK. In N. crassa, the NRC-1-MEK-2-MAK-2 kinase cascade are the central components of self signalling machinery [37]. The MOB-3 component of the NcSTRIPAK complex interacts with MAK-2. However, nuclear localization of MAK-2 is not influenced by STRIPAK but MAK-1 localization is altered. In A. nidulans, the sexual pathway is controlled by a pheromone response (SteC-SteD-MkkB-MpkB) module which migrates from the plasma membrane to the nuclear envelope to deliver MpkB into the nucleus. Phenotypes of MAPK module mutants are very similar to STRIPAK mutants [24]. Interestingly, in time course purifications at sexual stages, SteD, which is the adaptor domain of the pheromone response pathway, was repeatedly co-purified with the StrA-TAP fusion (S1–S6 Tables). Both MkkB and MpkB localizations did not change and MpkB interacted with both MkkB and SteA in the absence of StrA. However, interestingly, MpkB was not activated by phosphorylation in the absence of StrA. Since MpkB interacts with SteA-VeA and phosphorylates VeA, which is necessary for sexual development and SM production, the AnSTRIPAK complex presumably activates fruit body formation and SM production by phosphoactivation of the MAPK module. The sipA mutant and its strA combinations showed an opposite phenotype to STRIPAK component deletions, indicating an epistatic function of SipA over the StrA but not over the SipB-SipC-SipD-SipE complex. The sipA deletant grows as well as WT or even better. It produces 2. 5-fold more asexual spores and is more resistant to different stress conditions. SM production was slightly influenced in the absence of sipA and expression of all other gene clusters was upregulated in sipAΔ, which strongly suggests an inhibitor role for SipA in these processes. How does SipA perform this function within the AnSTRIPAK complex? This is obvious from interaction dynamics, because SipA interacts only with StrA during all developmental stages. Furthermore, in the absence of SipA, StrA is able to form a hexameric complex with SipB, SipC, SipD, SipE and SipF. This partial AnSTRIPAK (-SipA) complex without SipA is presumably more active than an intact complex, and it promotes excessive growth and asexual sporulation by unknown mechanisms. However, it does not sufficiently fulfill the meiotic functions of the WT complex, because the sipA deletant cannot produce ascospores. MOB-3 (SipA homolog) in N. crassa interacts with the NRC-1-MEK-2-MAK-2 kinase self-signaling cascade [38]. However, although SipA does not interact with the components of the MAPK pathway in the absence of StrA in TAP studies, it might transiently interact with the MAPK pathway to elicit its effect on development. The partial complex functions (AnSTRIPAK (-SipA) ) will require more understanding at experimental level. In N. crassa and S. macrospora, the involvement of the STRIPAK complex in SM has not been reported. Production of the mycotoxin ST is positively controlled by the AnSTRIPAK complex, which requires proper expression of the velvet complex. Accordingly, the expression of the ST gene cluster is also drastically reduced by loss of the AnSTRIPAK complex. Expression of PN and TQ clusters are similarly diminished in the STRIPAK mutants. A. nidulans produces many more metabolites than these three molecules. Although only three gene clusters were examined here, the effects of the loss of STRIPAK might be more extensive. How does the complex control SM production? Transcriptional downregulation of the velvet complex might be the primary reason why SM production is drastically affected in STRIPAK mutants. Another scenario might be that AnSTRIPAK is important for vital functions of the SteC-SteD-MkkB-MpkB module. Because this module uses the nuclear envelope to interact with the nucleus and deliver the active MAPK MpkB into the nucleus. As discussed previously, reduction in signal fidelity of the pheromone response pathway in the absence of StrA presumably results in reduced SM production. In other eukaryotes, STRIPAK complex acts as a negative regulator of kinases, because the GCKIII kinase family member Mst3 and Mst4 are hyperphosphorylated in the mutants of PP2A subunit in human cells or okadaic acid treated cells, respectively [39]. The scenario in A. nidulans and N. crassa is different, however. In A. nidulans MpkB loses its phosphorylation in the absence of striatin and MpkB is a MAPK, not a GCK type kinase. In contrast, in N. crassa, MAK-1 loses its activity under resting and stress conditions [15]. Two GCKs SmKIN3 and SmKIN24 were found to be functionally and physically interacting with S. macrospora striatin [40,41]. However, functional control of these kinases by striatin remains to be shown. The SIKE-like domain containing SipB is part of the AnSTRIPAK complex and it forms a heterodimer with SipD (SLMAP), which is similar to mammals where SIKE and SLMAP form a heterodimer. However, it is not surprising to see a mammalian anti-inflammatory protein conserved in fungi, including A. nidulans since it was also shown that velvet family proteins of A. nidulans contain a DNA binding domain structurally similar to the proinflammatory Rel family of NF-KB proteins although they show only 13% similarity at the amino acid level [42]. Presence of NF-KB like velvet family proteins as well as SIKE-like proteins in fungi strengthens the hypothesis that fungal and mammalian defense systems have some degree of analogy since SMs of fungi often act as a defense mechanism against competitors or predators [43]. A SIKE-like protein as part of the AnSTRIPAK complex participates in production of SMs such as the mycotoxin ST to help the fungus to compete with other organisms in soil. The AnSTRIPAK complex mutants are sensitive to oxidative and cell-wall stress, however they are not extremely influenced by osmotic stress. At least two genes catC and sodB are down-regulated in the absence of the AnSTRIPAK complex. In N. crassa, transport of MAK-1 (cell wall regulator kinase) into the nucleus is facilitated by NcSTRIPAK complex in a MAK-2-dependent manner. However, the scenario is somewhat different in A. nidulans. A. nidulans stress responses are mediated by three MAPK proteins, MpkA, SakA and MpkC. MpkA regulates cell-wall integrity [28]. SakA and MpkC MAPKs control oxidative stress responses interdependently in A. nidulans [44,45]. Striatin has no influence on MpkA, which is a homologue of N. crassa MAK-1. MpkC nuclear accumulation is restricted by StrA in the absence of stress conditions. It is known that MpkC orthologs are activated in the cytoplasm in response to stress and enter into the nucleus. The function of the AnSTRIPAK complex is presumably to keep MpkC in the cytoplasm in the absence of stress, which allows subsequent activation under stress conditions. In conclusion, this study revealed the composition and assembly hierarchy of the AnSTRIPAK complex. It was surprising that STRIPAK also regulates light and stress responses and the production of SMs. Since A. nidulans is representative of more than three hundred biotechnologically and medically relevant members of the genus Aspergillus, the information gained through this study is applicable to other important fungi to understand their biology and to use their full potential in SM and enzyme production. Paradoxically, deletion of sipA resulted in an increase in transcription of a number of SM genes and this deletion may be a useful tool in eliciting expression of silent gene clusters, a goal of genome mining for useful SMs. A functional link between STRIPAK and MAPK pathways has been established and further information on this connection will also reveal how fungi sense various signals and how they control signal influx on the nuclear envelope and how they convert these signals into appropriate responses to control their growth, development and pathogenicity. Fungal strains used in this study are listed in S21 Table. A. nidulans AGB551, which has a WT veA allele, was used for all deletion and epitope taggings. Stellar (Clontech) and MACH-1 (Invitrogen) competent Escherichia coli cells were used for recombinant DNA preparations. The WT and transformed A. nidulans strains were grown in Glucose Minimal Medium (GMM), supplemented with appropriate amounts of vitamins. For vegetative stage experiments, fungal spores were grown submerged in liquid GMM with 180 rpm rotation for 20 or 48h. For induction of development, vegetative mycelia were then filtered through miracloth and placed on solid GMM with 2% Agar. To induce the cultures asexually, cultures were grown vegetatively for 20h and shifted to the plates and were further incubated in the presence of light for 6,12,24h. For sexual induction, plates were covered by aluminium foil and incubated in the dark for 6,12,24,48h. E. coli were grown in LB broth agar or liquid LB with Ampicillin (100 μg/ml) at 37°C overnight. Transformation of E. coli and A. nidulans were performed as explained in detail [46,47]. The plasmids and oligonucleotide sequences used and created in this study are listed in S22 and S23 Tables, respectively. In order to create the strA deletion construct, the 5’ UTR region of strA was amplified from wild-type genomic DNA (AGB551) using primers OZG1025/1110 and 3’ UTR with OZG1112/1028 for ptrA. 5’ UTR and 3 UTR were amplified with (OZG1025/1111, OZG1113/1028) for AfpyroA. The two fragments were fused to the ptrA and AfpyroA markers and amplified by a fusion PCR, using oligos OZG1025/1028 (4–4. 2 kbp). Both deletions cassettes were cloned in SmaI site of pUC19 leading to the plasmids pOB525 (strAΔ: : ptrA) and pOB526 (strAΔ: : AfpyroA). Both plasmids were digested with PmeI (MssI) and linear deletion fragments were transformed into AGB551 generating strA deletions STRA-DEL1 and STRA-DEL2, respectively. pOB525 was transformed into ANNE21, ANNE22, ANNE23, ANNE24 and ANNE25, resulting in ANNE31, ANNE32, ANNE33, ANNE34 and ANE35 strains. For complementation of the strA deletion, the strA genomic locus (6. 6 kbp), containing 2 kbp promoter and terminator regions, were amplified from gDNA (NE94/NE95) and cloned into the PmeI site of pOSB114, which yielded pNE33. Then, pNE33 was introduced into strA deletion strains (STRA-DEL1), generating the ANNE46 strain. For the creation of strA: : sgfp and strA: : ctap fusions, the promoter and ORF (4. 4 kbp) (OZG1025/1026), and terminator sequences (OZG1025/1028) of strA were amplified. These two fragments were fused to sgfp: : natR, ctap: : natR cassettes in SmaI site of pUC19 by using in-fusion HD cloning kit leading to plasmids pOB480 (strA: : sgfp) and pOB481 (strA: : ctap). In order to create the strA: : mrfp fusion, the promoter as well as ORF of strA (OZG1025/1114) and terminator sequences (OZG1115/1028) were amplified and fused to mrfp: : AfpyrG by cloning in SmaI site of pUC19 leading to pOB527. These cassettes were released by PmeI digestion and introduced into AGB551 strain and ANNE26, ANNE27, ANNE28, ANNE29 and ANNE30. All deletions and epitope tags were confirmed by the Southern blots (S3 and S4 Figs). For the creation of the sipA deletion plasmid construct, the 5’ UTR region of sipA was amplified from AGB551 (WT) genomic DNA using primers (NE1/3 and NE1/7) resulting in approximately 1. 5 kbp fragments. The 3’ UTR region was amplified with (NE4/N5). The two fragments were fused to the AfpyrG and AfpyroA markers and inserted into SmaI site of pUC19 leading to the plasmids pNE1 (sipAΔ: : AfpyrG) and pNE2 (sipAΔ: : AfpyroA) which were then transformed into MACH-1 competent Escherichia coli cells as described previously. Plasmid constructs were isolated using Qiagen Mini-Prep Plasmid Purification (Cat no: 27104) kit. To amplify the deletion cassette, NE2/6 primers were used and 4. 2kbp linear fragment was extracted from the gel. These linear deletion fragments were used to transform AGB551, generating the sipA deletion strains, termed (ANNE1. 1 and ANNE1. 2), and StrA-DEL1 for double deletion combinations with strA (ANNE1. 3). The sipA complementation plasmid was generated by amplifying sipA genomic locus including, 1. 9 kbp promoter and 1. 9 kbp terminator regions using primers (NE84/85) and cloned into the PmeI site of pOB114 that yielded pNE28, which was introduced into sipA deletion strain (ANNE1. 1), generating the ANNE41 complementation strain. All deletion strains were verified by Southern hybridization (S4 Fig). For the creation of sipA: : sgfp and sipA: : ctap fusions, the promoter (NE1/61) and ORF (5. 5kbp), along with the terminator sequences (NE5/62) of sipA were amplified from gDNA. These two fragments were fused to sgfp: : AfpyrG, sgfp: : AfpyroA and ctap: : AfpyrG cassettes in SmaI site of pUC19 by in-fusion HD cloning kit creating pNE13 (sipA: : sgfp: : AfpyrG), pNE23 (sipA: : sgfp: : AfpyroA), pNE14 (sipA: : ctap: : AfpyrG) plasmids respectively. The nest oligos (NE2/6) amplified these cassettes from pNE13, pNE23, and pNE14, which were introduced into the wild-type (ABG551) yielding strains ANNE16, ANNE26, ANNE21. These plasmids were also introduced into strA deletion strain STRA-DEL1 as explained in S21 Table. The construction of sipB deletion strains (ANNE2. 1 and NANE2. 2) was performed as explained above, using the primers NE8/10 and NE8/14 for amplifying the 5’ UTR and NE11/12 for amplifying the 3’ UTR. These fragments were fused to AfpyrG and AfpyroA markers and inserted into SmaI site of pUC19 leading to pNE3 (sipBΔ: : AfpyrG) and pNE4 (sipBΔ: : AfpyroA). The nest oligos NE9/13 were used to amplify deletion cassettes (4. 2 kbp), which were ultimately transformed into ABG551 to generate sipA deletion (ANNE2. 1 and ANNE2. 2) and double deletion combinations with strA (ANNE2. 3), sipA (ANNE6), sipC (ANNE10), sipD (ANNE11), and sipE (ANNE12). The sipB genomic locus (5. 4kbp), containing the 2kb promoter region and 2 kbp terminator regions, was amplified from genomic DNA (NE86/87) and used for the complementation of sipB deletion strains. This fragment was cloned into the PmeI site of pOSB114 that yielded pNE29 that was introduced into sipB deletion strains to generate ANNE42 strain. The sipB: : sgfp and sipB: : ctap cassettes were generated by using AfpyrG or pyroA markers in the same way that was explained previously, using primers NE8/63 and NE12/64 and pNE15 /pNE16. The resulting strains were named ANNE17 (sipB: : sgfp: : AfpyrG), ANNE27 (sipB: : sgfp: : AfpyroA) and ANNE22 (sipB: : ctap: : AfpyrG). In order to create ANNE3. 1 and ANNE3. 2, the 5’ UTR region of sipC was amplified from WT gDNA using primers NE15/17 or NE15/21. The 3’ UTR regions were amplified with NE18/19. The two fragments were fused to the AfpyrG and pyroA markers and inserted into SmaI site of pUC19 leading to pNE5 (sipCΔ: : AfpyrG) and pNE6 (sipCΔ: : AfpyroA). Linear deletion cassettes (4. 2kbp), which were amplified by PCR, using nest oligos (NE16/20), were transformed to AGB551 generating sipC single deletion (ANNE3. 1 and ANNE3. 2) and double mutant combinations of sipC with strA (ANNE3. 3), sipA (ANNE37), sipD (ANNE13) and sipE (ANNE14). For construction of complementation plasmid, pNE30, a 7kbp sipC genomic fragment (NE88/89) was cloned into the PmeI site of pOSB114. The pNE30 was then transformed into sipC deletion ANNE3. 1 Yielding ANNE43 as described previously. To create pNE17, pNE18 and pNE25, the sipC promoter and ORF was amplified with NE15/65 and the terminator region with NE19/66, using AGB551 genomic DNA as a template. These fragments were fused to sgfp: : AfpyroA, sgfp: : AfpyrG or ctap: : AfpyrG cassettes and cloned in pUC19 as described for the sipA and sipB. The primers NE16/20 amplified 10 kbp linear fragments from (pNE17, pNE18 and pNE25). These fragments were transformed into the wild type to generate ANNE18, NANE23, NANE28, respectively. Gene replacement events were verified by the Southern hybridization. For the creation of the sipD deletion construct, the 5’ UTR (NE41/42 or NE59/42) and the 3’ UTR (NE43/44) were amplified and fused to the AfpyrG and AfpyroA markers and inserted into SmaI site of pUC19 leading to the plasmids pNE9 (sipDΔ: : AfpyrG) and pNE10 (sipDΔ: : AfpyroA), respectively. Deletion cassettes were amplified with NE45/46 and used to transform AGB551, generating the sipD single deletion ANNE4. 1 and ANNE4. 2 and double deletion combinations with strA/sipD (ANNE4. 3), sipA/sipD (ANNE8), and sipE/sipD (ANNE15). For complementation of the sipD deletion, sipD genomic DNA including 1. 97 kbp of the promoter region and 2 kbp of the terminator, was amplified (NE90/91) and cloned into pOSB114 at the PmeI site, generating pNE31, which was then transformed into ANNE4. 1. This yielded the ANNE44 strain. To generate the sipD: : sgfp and sipD: : ctap strains, the primers NE41/67 were used to amplify the sipD promoter and ORF for sgfp and ctap cassettes, while the primers NE44/68 were used to replicate the sipD terminator. These fragments were fused to sgfp: : AfpyrG, sgfp: : AfpyroA and ctap: : AfpyrG with NE45/NE46. Created endogenous fragments (pNE19, pNE26, pNE20) were transformed into AGB551, yielding ANNE19, ANNE29 and ANNE24 respectively. To generate the sipE (ppgA) deletion construct, sipE 5’ UTR (NE50/51 and NE50/60) and the 3’ UTR (NE52/53) were amplified. The two fragments were fused to the AfpyrG and AfpyroA markers and inserted into SmaI site of pUC19 yielding pNE11 (sipEΔ: : AfpyrG) and pNE12 (sipEΔ: : AfpyroA). The liner deletion fragments, which were amplified with nest oligos NE54/55, then used to transform AGB551, generating sipE deletion strains (ANNE5. 1 and ANNE5. 2, respectively) and double deletions sipE/strA (ANNE5. 3), sipE/sipA (ANNE9). For complementation of the sipE deletion, the sipE genomic locus (6. 8 kbp), containing 2 kbp of the promoter region and 2 kbp of the terminator regions, was amplified (NE92/93) cloned into the pOSB114 (pNE32). Then pNE32 was introduced into sipE deletion strain (ANNE5. 1), generating the ANNE45 strain. To generate the sipE: : sgfp and sipE: : ctap strains, primers NE50/69 were used to amplify the sipE promoter and ORF for sgfp and ctap, while the primers NE71/70 were used to amplify the sipE terminator. These fragments were fused to sgfp: : AfpyrG, sgfp: : AfpyroA and ctap: : AfpyrG cassettes with NE54/55. Endogenous fragments (pNE21, pNE27, pNE22) were transformed into AGB551, yielding ANNE20, ANNE30 and ANNE25, respectively. The pOB340 plasmid was digested with PmeI at 37°C overnight, then transformed into ANNE16, ANNE17, ANNE18, ANNE19 and ANNE20, resulting in generation of ANNE47, ANNE48, ANNE49, ANNE50 and ANNE51. The same strains were transformed with pME3857, which led to strains ANNE57, ANNE58, ANNE59, ANNE60 and ANNE61. These strains were introduced with pOB526 (strAΔ: : AfpyroA) in order to delete strA gene generating the strains ANNE62, ANNE63, ANNE64, ANNE65 and ANNE66. In order to create MpkB/strAΔ combination, pOB526 strA deletion cassette was introduced into AGB655 strain, which yielded ANNE73. Generation of MpkA and MpkC fusions were performed as follows: mpkA and mpkC promoter and ORFs were amplified (BK546/547 for mpkA, BK552/553 for mpkC) from gDNA. 3 UTRs for both genes were amplified (BK548/549 for mpkA, BK554/555 for mpkC) and fused to sgfp: : AfpyrG cassette in SmaI site of pUC19, which yielded pBK125 (mpkA: : sgfp: : AfpyrG) and pBK126 (mpkC: : sgfp: AfpyrG), respectively. Both plasmids were transformed into AGB551, which led to ANNE67 (MpkA: : sGFP) and ANNE68 (MpkC: : sGFP). Then, mRFP: : H2A plasmid pME3857 was introduced to ANNE67 and 68, which resulted in ANBK116. 1 and 117. 2, respectively. Finally, strA gene was deleted with pOB526 in ANBK116. 1 and 117. 2 yielding ANBK119 (MpkA: : sGFP/strAΔ) and 120 (MpkA: : sGFP/strAΔ), respectively. Southern hybridization experiments were performed as given in detail [48]. Fungal genomic DNA was prepared from plates using the ZR Fungal/Bacterial DNA MiniPrep TM (Zymo Research) kit. 700 ng of isolated genomic DNA was used for restriction enzyme digestion. The southern hybridization was performed with non-radioactive probes by using DIG labeling (Roche) as described in the user protocol. The phenotypes and quantification of sporulation for WT and deletion strains were examined as follows: Fungal spores were counted using a haemocytometer. 5 x103 spores (5μl) were used to point inoculate solid GMM, containing appropriate supplements. Plates were incubated in the light (for asexual development) and in the dark (for sexual development) for 4–5 days at 37°C. Colonies were observed using the Olympus szx16 microscope with Olympus sc30 camera. Digital pictures were taken and processed with the Cell Sens Standard software (Olympus). Quantifications were performed from three independent biological replicates. Fungal mycelia were obtained from liquid cultures and broken using liquid nitrogen. Protein extracts were prepared by re-suspending the smooth mycelia in protein extraction buffer (B300 buffer) which contains 300 mM NaCl, 100 mM Tris-Cl pH: 7. 5,10% Glycerol, 1 mM EDTA, 0. 1% NP-40 and added with 1mM DTT, 1x Protease inhibitor mix (Roche), 1. 5mM Benzamidine, 1x Phosphatase inhibitor mix (Roche) and 1mM PMSF. Protein concentrations were calculated by performing Bradford assays. 100 μg of total protein extract was run on various percentages SDS gels as required (10% or 12%) and transferred to protean membrane with 0. 45μm pore size (GE Healthcare). Antibody conditions and dilutions: For TAP, the primary rabbit α-calmodulin binding peptide (CBP) antibody (05–932, Millipore) as a 1: 1000 dilution in TBS with 5% milk. The secondary antibody (goat α-rabbit, G-21234, Life technologies) as a 1: 1000 dilution in TBS and 5% milk. For GFP, the primary α-GFP (SC-9996, Santa Cruz) as a 1: 500 dilution in TBST with 5% milk. The secondary goat α-mouse (170–6516, Biorad) as a 1: 1000 dilution in TBST and 5% milk. For detection of SkpA, custom made rabbit α-SkpA 1: 2000 dilutions in TBST and 5% milk (Genscript). TAP experiments, GFP pull-downs and preparation of the protein crude extracts and analysis of the proteins were performed as explained in detail [23,49]. TAP experiments were performed from two independent biological replicates. TAP of the WT strain eluates were used as non-specific control. Proteins identified from StrA, SipA, SipB, SipC, SipD and SipE TAP experiments were filtered from non-specific control purifications. The final lists of the proteins were presented in excel format in Supplementary tables (S1–S6 and S8–S18 Tables). GFP pull-downs of MpkB interaction partners in the presence or absence of StrA originates from either one or two biological replicates (S19 and S20 Tables). The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium (http: //proteomecentral. proteomexchange. org/cgi/GetDataset) via the PRIDE partner repository [50] with the dataset identifier PXD011927. Green and monomeric red fluorescent protein (GFP and mRFP) expressing strains were grown in 500 μl liquid GMM media, with appropriate supplements in sterile Lab-Tek Chambered Coverglasses with covers (8 wells per coverglass) (Thermo Scientific) for 17–20 hours at 30°C. Localizations of the proteins were captured and recorded as published previously [51]. The WT and all deletion strains were grown as follows: 5x103 spores were inoculated on GMM supplemented with vitamins and 1% Oatmeal at 37°C for 5 days. Metabolites were extracted by using chloroform and drying them in speed vac. Samples were suspended in 200μl Methanol. 1mg/ml sterigmatocystin (Sigma) was used as the standard (2. 5 μl of standard added in 47. 5 μl of 100% Methanol). RP-HPLC analysis was carried out on a Shimadzu RP-HPLC with a photodiode array detector (PDA). 20 μl of standard or sample was injected onto a Luna omega 5μm polar C18 (LC column 150 x 4. 6 mm) and separated across a water: acetonitrile gradient with 0. 1% (v/v) TriFluoroacetic Acid (TFA). Gradient conditions of 5–100% acetonitrile over 30 min with a flow rate of 1 ml/min were used with PDA detection at 254 nm. 100 mg of mycelia was used for RNA isolation by using RNeasy (Qiagen). 1 μg RNA was used for each cDNA synthesis using the Transcriptor First Strand cDNA Synthesis Kit (Roche). Different primers of target genes were used for qPCR reaction as described in the user protocol, by using LightCycler 480 Sybr Green I Master (Roche). A house-keeping gene, benA was used as a standard. Relative Expression Analysis was performed by Light Cycler 480 Software. All experiments were performed on three independent occasions and numerical data (S1 Dataset) are expressed as the mean ± SD and standard error. The corresponding means were compared for significant differences via the student t-test and One-way ANOVA methods, by using the software Graphpad Prism Version 6.

The multisubunit STRIPAK complex has been studied from yeast to human and plays a range of roles from cell-cycle arrest, fruit body formation to neuronal functions. Molecular assembly of the STRIPAK complex and its roles in stress responses are not well-documented. Fungi, with an estimated 1. 5 million members are friends and foes of mankind, acting as pathogens, natural product and enzyme producers. In filamentous fungus Aspergillus nidulans, we found a heptameric STRIPAK core complex made from three subcomplexes, which sits on the nuclear envelope and coordinates signal influx for light-dependent fungal development, secondary metabolism and stress responses. STRIPAK complex controls activities of two major Mitogen Activated Protein Kinase (MAPK) signaling pathways through either promoting their phosphorylation or limiting their nuclear localization under resting conditions. These findings establish a basis for how fungi govern signal influx by using multimeric scaffold protein complexes on the nuclear envelope to control different downstream pathways.

Abstract Introduction Results Discussion Materials and methods

cell physiology cell fusion cellular stress responses aspergillus cell processes fungal molds aspergillus nidulans developmental biology fungi model organisms experimental organism systems morphogenesis mapk signaling cascades research and analysis methods animal studies gene expression signal transduction eukaryota cell biology neurospora crassa sexual differentiation genetics biology and life sciences yeast and fungal models cell signaling neurospora organisms signaling cascades

2019

Assembly of a heptameric STRIPAK complex is required for coordination of light-dependent multicellular fungal development with secondary metabolism in Aspergillus nidulans

Gram-negative bacterial pathogens of plants and animals employ type III secreted effectors to suppress innate immunity. Most characterized effectors work through modification of host proteins or transcriptional regulators, although a few are known to modify small molecule targets. The Xanthomonas type III secreted avirulence factor AvrRxo1 is a structural homolog of the zeta toxin family of sugar-nucleotide kinases that suppresses bacterial growth. AvrRxo1 was recently reported to phosphorylate the central metabolite and signaling molecule NAD in vitro, suggesting that the effector might enhance bacterial virulence on plants through manipulation of primary metabolic pathways. In this study, we determine that AvrRxo1 phosphorylates NAD in planta, and that its kinase catalytic sites are necessary for its toxic and resistance-triggering phenotypes. A global metabolomics approach was used to independently identify 3’-NADP as the sole detectable product of AvrRxo1 expression in yeast and bacteria, and NAD kinase activity was confirmed in vitro. 3’-NADP accumulated upon transient expression of AvrRxo1 in Nicotiana benthamiana and in rice leaves infected with avrRxo1-expressing strains of X. oryzae. Mutation of the catalytic aspartic acid residue D193 abolished AvrRxo1 kinase activity and several phenotypes of AvrRxo1, including toxicity in yeast, bacteria, and plants, suppression of the flg22-triggered ROS burst, and ability to trigger an R gene-mediated hypersensitive response. A mutation in the Walker A ATP-binding motif abolished the toxicity of AvrRxo1, but did not abolish the 3’-NADP production, virulence enhancement, ROS suppression, or HR-triggering phenotypes of AvrRxo1. These results demonstrate that a type III effector targets the central metabolite and redox carrier NAD in planta, and that this catalytic activity is required for toxicity and suppression of the ROS burst. A central theme in gram-negative bacterial pathogenesis is the injection of Type III-secreted effectors (T3E) into host cells. Plant pathogen effectors may suppress two tiers of innate immunity: PAMP-Triggered Immunity (PTI), triggered by perception of pathogen-associated molecular patterns (PAMPs) by pattern recognition receptors, or Effector-Triggered Immunity (ETI), triggered by perception of specific T3E by cognate nucleotide binding-leucine rich repeat resistance (R) genes [1]. With the exception of the DNA-binding Transcriptional Activator-Like effectors of Xanthomonas, the known molecular functions of T3E were once thought to be limited to the mimicry or covalent modification of host proteins involved in signaling [2]. Recently, the Ralstonia solanacearum effectors RipTPS and RipAY were reported to synthesize the sugar trehalose-6-phosphate and degrade the small peptide glutathione, respectively, demonstrating that small molecules might also be strategic targets of T3E in plants [3,4]. AvrRxo1 (also called XopAJ) is a T3E produced by several species of Xanthomonas, Acidovorax, and Burkholderia plant pathogens of monocot and dicot plants. Putative homologs lacking the T3 secretion signal are also found in a variety of environmental bacteria with no known pathogenic role [5]. AvrRxo1 has been implicated in several different T3E functions; it triggers a type III secretion-dependent hypersensitive resistance response (HR) in maize or transgenic rice plants expressing the resistance protein Rxo1 [6,7], enhances virulence of Xanthomonas oryzae on rice [8], and suppresses nonhost resistance to X. oryzae on tobacco [9]. Homologs of AvrRxo1 are always encoded upstream of a gene encoding a small protein binding partner, Arc1. The solved structure of the X. oryzae homolog of AvrRxo1 revealed a T4 polynucleotide kinase (T4pnk) domain [8], typically involved in phosphorylation of polynucleotides and nucleoside small molecules [10]. Among T4pnk domain proteins, AvrRxo1 shares the strongest structural homology with a zeta (ζ) toxin, part of a toxin-antitoxin (TA) system from Streptococcus pyogenes [8,11]. TA systems are ubiquitous gene modules comprised of a bacterial growth-suppressing toxin and a neutralizing antitoxin, which function in metabolic stress management and DNA maintenance [12]. Consistent with this structural homology, AvrRxo1 is bacteriostatic to Escherichia coli in the absence of Arc1, demonstrating that AvrRxo1 functions as TA system toxin in addition to a T3E [5]. AvrRxo1 also triggers growth suppression and watersoaked tissue collapse when expressed in yeast and plant cells, respectively [8,13]. The bacteriostatic function of AvrRxo1 is dependent on two key functional residues conserved among T4pnk domain proteins. A catalytic aspartic acid (D193) is predicted to coordinate the hydroxyl group on the target carbon of the substrate, activating it for nucleophilic attack, and a Walker A threonine (T167) is predicted to coordinate the ATP [8,14]. AvrRxo1’s toxic effects in prokaryotes and eukaryotes suggests that it has a universally essential target. While many TA toxins work by disrupting aspects of polynucleotide replication or translation, zeta toxin phosphorylates a product of central metabolism, the peptidoglycan precursor uridine diphosphate-N acetylglucosamine (UNAG), to generate a nonfunctional and inhibitory analog [15]. Despite its structural similarity to zeta toxin, AvrRxo1 differs at most predicted substrate-coordinating residues and does not phosphorylate UNAG [8]. Recently, Schuebel and colleagues used a rational bottom-up approach to demonstrate that AvrRxo1 acts as a nucleotide kinase that phosphorylates NAD and its derivative NAAD at the 3’ hydroxyl position in vitro, and also generates the product 3’-NADP in targeted assays in E. coli [16]. However, the presence and broader implications of this kinase activity in eukaryotic and plant host systems were not investigated, and the relevance of NAD phosphorylation to toxicity or immune suppression remains unclear. Here, we used a top-down global metabolomics approach to identify molecular targets of AvrRxo1. Accumulation of 3’-NADP was the most substantial detected metabolic consequence of AvrRxo1 expression in yeast and bacteria. Targeted assays revealed 3’-NADP accumulation in plant tissue upon transient expression of avrRxo1 in leaves of N. benthamiana, and during infection of rice with an avrRxo1-containing strain of X. oryzae. The catalytic aspartic acid D193 was essential for NAD phosphorylation and for the AvrRxo1 phenotypes of toxicity, Rxo1-triggered immunity, and ROS suppression. A mutation in the Walker A threonine strongly reduced 3’-NADP accumulation and abolished AvrRxo1 toxic phenotypes, but did not eliminate the ability of AvrRxo1 to trigger Rxo1 or suppress the flg22-mediated ROS burst in plants. These results demonstrate that T3E can directly modify a central primary metabolite in eukaryotic cells, representing a novel strategy for immune triggering and suppression by secreted effectors. Structural analysis of AvrRxo1 identified several conserved residues predicted to be required for catalytic activity. Mutation of two of these, the catalytic aspartate D193 and the Walker A threonine T167, abolished growth-suppressive activity of AvrRxo1 in E. coli [8]. Here, we asked whether the predicted sites D193 and T167 are necessary for other reported phenotypes of AvrRxo1, including yeast growth suppression, watersoaking upon transient expression in plants, and triggering the Rxo1-dependent HR. We previously found that growth suppression by AvrRxo1 was more pronounced in E. coli liquid cultures induced from a lower initial density than from a higher density [5], suggesting that toxicity is reduced above a threshold cell density, as was observed in a zeta toxin [15]. For a sensitive comparison of bacteriostasis by site-directed AvrRxo1 mutants in E. coli, we assayed for AvrRxo1 and catalytic site mutant growth suppression at multiple cell densities on solid media. 6xHis-AvrRxo1, but not T167N and D193T mutant derivatives, abolished E. coli growth when expression was induced below the density of 2X107 CFU/mL, but not at 4X107 CFU/mL or above (Fig 1A). AvrRxo1 expression similarly inhibited growth of Saccharomyces cerevisiae in a manner dependent on the two catalytic residues; here growth suppression was obvious at all densities tested (Fig 1B). Previous work demonstrated that transient expression of YFP-tagged AvrRxo1 induced a toxic-like watersoaking and cell collapse phenotype in several plant species, and that this was abolished by the T167N mutation [8]. We found that transient expression of HA-tagged AvrRxo1 triggered toxic cell collapse by 50 hours post-infiltration (hpi) in leaves of N. benthamiana, and both the T167 and D193 residues were necessary for this phenotype (Fig 1C). Expression of the catalytic mutants was confirmed using Western blot analysis (S1 Fig); although expression of HA-AvrRxo1 could not be detected, possibly due to cell death associated with expression. We next investigated whether residues D193 and T167 were required for triggering Rxo1-mediated resistance in rice. We and others have been unable to obtain a knockout mutant of avrRxo1 from the model X. oryzae pv. oryzicola strain BLS256 in repeated attempts over many years [7], perhaps because of its function as part of a TA system and addiction module. Therefore, avrRxo1: arc1 gene module and mutant derivatives were introduced into the weakly pathogenic X. oryzae strain X11-5A on the low-copy cosmid vector pHM1. X11-5A has an arc1 antitoxin gene in its chromosome, allowing avrRxo1: arc1 to be introduced without detrimental growth effects. At five days post infiltration (dpi) into leaves of Kitaake Rxo1 transgenic rice, vector control X11-5A (pHM1) caused watersoaking, while X11-5A (pavrRxo1/ pHM1-AvrRxo1) inoculation sites were characterized by brown color development and absence of watersoaking, characteristic of the Rxo1 resistance response (Fig 1D). X11-5A (pHM1-AvrRxo1-D193T) caused a watersoaking phenotype similar to the control, indicating that the catalytic aspartate is necessary for triggering Rxo1. However, the X11-5A (pHM1-AvrRxo1-T167N) also exhibited a browing phenotype and did not show signs of watersoaking (Fig 1D). This finding, in addition to our previous observation that the T167N mutant retains virulence enhancement of X. oryzae in rice [8], suggests that the T167N mutation uncouples AvrRxo1 toxicity from immune triggering or suppressing functions. A western blot confirmed expression of AvrRxo1 and catalytic site mutants during rice infection (S2 Fig). Unlike during expression in tobacco, AvrRxo1-T167N accumulated to lower levels than AvrRxo1-D193T or AvrRxo1, which may affect the phenotype of AvrRxo1-T167N in rice. We used untargeted metabolomics to uncover small molecule targets of ectopically expressed AvrRxo1 in bacteria and yeast. Having demonstrated that a substitution in the predicted substrate-binding residue D193T abolishes diverse known phenotypes of AvrRxo1, we selected AvrRxo1-D193T as the inactivated AvrRxo1 control for the study along with the control vectors used in Fig 1A and 1B. AvrRxo1 expression in E. coli cultures was induced at a density well above the threshold for discernible growth repression as observed in Fig 1A, to minimize off-target effects of AvrRxo1 toxicity on the metabolome. Triplicate cultures were subjected to metabolic profiling using zwitterionic-hydrophilic interaction liquid chromatography (ZIC-HILIC) coupled to time-of-flight mass spectrometry (TOF-MS). Analysis revealed that AvrRxo1 expression affected bacterial and yeast central metabolism in distinct but overlapping ways. In both organisms, the most dramatic AvrRxo1-dependent change was the strong accumulation of a single peak annotated as NADP (Fig 2A and 2B. S1 File). Other effects, some of which were specific to E. coli, included modest (1. 5 to 2-fold) increases in relative abundance of predicted metabolites involved in the TCA cycle (citric acid, succinic acid, malic acid, and pantothenic acid), in some amino acids (aspartic acid, lysine, leucine/isoleucine, phenylalanine, tyrosine, and tryptophan), in glutathione, and in pyrimidine nucleotides. Heatmaps summarizing relative changes in abundance of 79 annotated metabolites are found in Supplemental S3 and S4 Figs, and boxplots of individual metabolites are found in Supplemental s1 File. In both organisms, some metabolite changes were observed in both AvrRxo1 and AvrRxo1-D193T (S3 Fig and S4 Fig), suggesting that AvrRxo1 can exert some metabolic changes in a manner independent of the catalytic aspartic acid. For example, the NADP precursor NAD decreased in relative abundance as a result of expression of both AvrRxo1 and AvrRxo1-D193T in E. coli, but increased slightly in abundance during the expression of both treatments in yeast (pg. 12 of S1 File). We also profiled tissues of Kitaake rice 48 hours after inoculation with X. oryzae strain X11-5A carrying avrRxo1: arc1 and the empty vector pHM1, but no significant differences were observed (S1 File). Given that a large accumulation of NADP observed upon AvrRxo1 expression in yeast and bacteria would be unlikely due to metabolic turnover, spiking experiments were performed to determine whether the annotation was accurate. Lysates from AvrRxo1-expressing E. coli cells were amended with commercial NADP and subjected to TOF-MS/MS analysis. NADP formed a distinct peak from the AvrRxo1-dependent metabolite, with a retention time roughly 10s longer (S5A and S5B Fig). However, the mass spectra of NADP and the AvrRxo1-dependent metabolite were identical (S5C and S5D Fig) indicating that the peak represented a close analog of NADP. Interpretation of the MS/MS fragmentation pattern suggested that the most probable NADP analog that would generate a nearly identical spectrum would be 3’-NADP. Spiking of AvrRxo1-expressing E. coli lysates with commercially available 3’-NADP standard or simultaneously with both NADP isomers confirmed that the AvrRxo1-dependent compound has an identical retention time to 3’-NADP (Fig 2C). In addition, the AvrRxo1-dependent molecule was a spectral match to the 3’-NADP standard, confirming its identity as 3’-NADP (S6 Fig). Thus, AvrRxo1 expression in yeast and E. coli results in a strong accumulation of 3’-NADP, and not in other polar metabolites detectable through the ZIC-HILIC method. To determine whether AvrRxo1 mediates 3’-NADP accumulation through direct phosphorylation of NAD, 6xHis-AvrRxo1 and 6xHis-AvrRxo1-D193T were purified and tested for NAD kinase activity in an in vitro assay (S7 Fig). 3’-NADP accumulated in the presence of 6xHis-AvrRxo1 (S7C Fig) but not in the presence of 6xHis-AvrRxo1-D193T (S7D Fig), confirming that AvrRxo1 is an NAD kinase dependent on the D193 catalytic site. AvrRxo1 is a T3E that has been associated with virulence enhancement and suppression of nonhost resistance when expressed from X. oryzae, and with phytotoxic cell collapse when transiently expressed in planta [8,9]. To determine whether AvrRxo1 phosphorylates NAD in plant cells, HA-AvrRxo1 and HA-AvrRxo1-D193T were transiently expressed in leaves of N. benthamiana. Leaf discs were collected at 36 h after Agroinfiltration, prior to the initial onset of watersoaking symptoms in AvrRxo1-expressing leaves. Both NADP and 3’-NADP were detected by LC-MS/MS in the AvrRxo1-expressing leaves (Fig 3A), but only NADP was detected in leaves infiltrated with the AvrRxo1-D193T construct (Fig 3B). We next tested whether X. oryzae pv. oryzicola (Xoc), the pathogen that produces AvrRxo1, can generate 3’-NADP during the infection of rice by wild-type X. oryzae. Because of our aforementioned difficulty in inactivating avrRxo1, we were unable to generate a deletion mutant in BLS256 mutant to conclusively determine that avrRxo1 is the sole source of 3’-NADP during infection. However, some leaf streak-causing Xoc isolates from Africa naturally lack avrRxo1. Profiling of diverse Xoc strains has shown that avrRxo1 gene presence or absence is tightly linked to the phenotypes of Rxo1 activation and suppression of tobacco nonhost resistance [17]. The Xoc strain MAI10 was used as an avrRxo1-negative leaf streak pathogen in this study. MAI10 is a close genetic relative of BLS256 that produces similar symptoms and encodes a nearly identical complement of non-TAL effector genes other than avrRxo1, although it differs in TAL effector composition [17–19]. Leaves of rice variety Kitaake were infiltrated with strain MAI10, BLS256 or sterile water, and samples were collected and analyzed at 12,24 and 48 hpi. 3’-NADP was detected in samples inoculated with BLS256, but not with MAI10 or a water control (Fig 4A–4C). To confirm that avrRxo1 is the key genetic factor conferring the difference in 3’-NADP accumulation between BLS256 and MAI10, we introduced pHM1-AvrRxo1 and pHM1 into MAI10 via triparental mating. A very low but distinct 3’-NADP signal was detected in three out of four replicates of the inoculations with MAI10 (pHM1-AvrRxo1), but not in inoculations with MAI10 alone or MAI10 (pHM1) (S8 Fig). These results demonstrate that AvrRxo1 NAD kinase activity occurs during Xoc infection of rice. 3’-NADP was detectable within 12 hpi of infection with BLS256, with relative abundance increasing through 48 hpi (Fig 4D and 4E, S9 Fig). Unlike the heterologous expression conditions in which 3’-NADP accumulates to a very high relative abundance, relative peak areas of 3’-NADP were far smaller than those of native NADP in rice tissues within 48h. This is consistent with the lack of observable increase in the NADP/3’-NADP peak identified in our initial metabolomic profile in infected rice. Although this experiment does not conclusively prove that the source of the 3’-NADP was the rice and not and intercellular X. oryzae, it is well-established that AvrRxo1 is introduced into and active in rice cells during infection [20], and that AvrRxo1 activity is strongly inhibited in bacteria, in vitro, and in planta in the presence of the antitoxin Arc1 [8,16]. We have not been able to introduce an expressed avrRxo1 gene into X. oryzae strains in the absence of the arc1 gene, thus it is presumed that free AvrRxo1 is toxic to bacteria and thus is likely inactivated in BLS256 cells. Although NADP and 3’-NADP are both synthesized from NAD, there was no significant difference in total (2’) -NADP levels associated with the presence of avrRxo1 during X. oryzae pathogenesis on rice (Fig 4A–4C). Together, these results demonstrate that AvrRxo1 catalyzes 3’-NADP production in plants and that 3’-NADP is produced early in bacterial leaf streak infection. In previous work, we demonstrated that AvrRxo1 enhances early proliferation of pathogenic Xanthomonas in rice leaves. The T167N substitution in the Walker motif abolished growth suppression in E. coli and cell collapse in plants, but still retained AvrRxo1 virulence enhancement on rice [8] and recognition by Rxo1 (Fig 1D). Because the Walker A threonine has been reported essential for the activity of other Pnk domain kinases [21], we initially hypothesized that the Walker site mutant may be catalytically inactive but retain selected phenotypes due to retention of an intact substrate-binding site, perhaps due to sequestration of the hypothetical substrate in the cell [8]. Alternatively, AvrRxo1 could be recognized structurally by Rxo1. In light of AvrRxo1 3’ NAD kinase activity, we decided to revisit the activity of the T167N mutant. Specifically, we wanted to determine whether the T167N substitution uncouples NAD kinase activity from immune suppression or recognition in the plant. We first determined whether AvrRxo1-T167N could produce 3’-NADP in yeast. Induced culture lysates of yeast strains depicted in Fig 1B were subjected to metabolic profiling using HILIC-LC-MS/MS. AvrRxo1-T167N generated detectable 3’-NADP in yeast, although the level was low compared to AvrRxo1 (Fig 5A–5C), accumulating roughly in proportion to NADP. Thus, yeast expression of AvrRxo1-T167N retains a level of kinase activity that is insufficient to cause visible toxicity. In addition to enhancing virulence, AvrRxo1 is a suppressor of flg22-mediated innate immunity in Arabidopsis [9,22]. We determined the impact of transient AvrRxo1 expression on generation of the PAMP-triggered burst of reactive oxygen species (ROS), a critical early event in plant defense signaling. The initial apoplastic oxidative burst is largely generated by plasma membrane NADPH oxidases from NADPH or NADH in the cytoplasm [23], where AvrRxo1 is also localized [22]. NAD and NADP are also thought to have signaling roles in generation of the oxidative burst [24,25], and therefore we hypothesized that AvrRxo1 activity might suppress ROS formation. In a luminol-HRP based chemiluminescence assay performed on leaves prior to the onset of cell collapse, HA-AvrRxo1 transient expression in N. benthamiana strongly suppressed flg22-induced ROS accumulation in comparison with HA-AvrRxo1-D193T (Fig 5D). AvrRxo1-T167N also strongly suppressed ROS accumulation (Fig 5D), while no signal was detected in mock-induced leaves (Fig 5E). While this is not a quantitative comparison of HA-AvrRxo1 and HA-AvrRxo1-T167N ROS suppression efficacy- HA-AvrRxo1 was not detected by Western Blot, preventing comparison of expression levels with HA-AvrRxo1-T167N (S1 Fig) - these results demonstrate that AvrRxo1 suppresses the PAMP-triggered ROS burst even in a mutant that generates subtoxic levels of AvrRxo1 activity. Finally, we asked whether AvrRxo1 and the catalytic site mutants were able to suppress ROS generation in other eukaryotes, or whether ROS suppression was specific to the flg22-induced burst in plants. Menadione is a redox cycling agent that stimulates ROS production from mitochondrial NAD (P) H in yeast [26]. In a fluorescence assay, AvrRxo1 expression reduced levels of menadione-induced ROS compared with AvrRxo1-D193T. AvrRxo1-T167N did not suppress yeast ROS levels (Fig 5F). These findings demonstrate that AvrRxo1 is a suppressor of multiple mechanisms of ROS generation in diverse eukaryotes, and determined that although the mutant AvrRxo1-T167N can suppress the flg22-mediated ROS burst in plants, it does not suppress mitochondrial ROS accumulation in yeast. AvrRxo1 is part of an ancient family of bacterial toxins co-opted by the type III secretion system for secretion into plants. This study demonstrates that AvrRxo1 phosphorylates NAD in eukaryotic cells, and that a catalytic site necessary for this kinase function is required for the toxic and immune-triggering functions of AvrRxo1. NAD is a coenzyme and redox carrier universally essential for metabolic function, and host metabolism is a common target for pathogens. In human pathogenic bacteria, glycohydrolase domain toxins metabolically poison target host or microbial cells by degrading NAD, gaining access via cytolysin-mediated translocation or through the type VI secretion system [27,28]. Pathogens of plants and animals are known to affect diverse host primary metabolites through metabolic crosstalk or manipulation of host transcription and signaling [29,30]. A few type III secreted effectors directly target host small molecules, including sugars and glutathione [3,4], challenging the former paradigm of effectors as protein modifiers. This work demonstrates that a type III secreted effector can phosphorylate a universally essential cofactor in the host. The direct modification of a universal metabolic player and redox carrier such as NAD represents a novel pathogen strategy for manipulation of host function. AvrRxo1 exhibits toxicity upon ectopic (i. e. , high-level) expression in bacteria, yeast, and plants, and these phenotypes are abolished upon mutation of the D193 or T167 catalytic sites. The drastic accumulation of 3’-NADP in yeast and bacteria, and the lack of detection of any other novel AvrRxo1-dependent metabolites in the global metabolomic profile, suggests that NAD is a primary small molecule target of AvrRxo1 function. One putative avenue for AvrRxo1 toxicity is enzyme inhibition by its product 3’-NADP. The AvrRxo1 structural homolog PezT, which phosphorylates UDP-GlcNac to generate a non-native analog, exerts toxicity both through UDP-GlcNac depletion and through inhibition of the UNAG-utilizing enzyme MurA by the phosphorylated product [31]. Unlike UNAG, NAD and NADP are required for hundreds of essential reactions, and therefore determining the in planta enzyme inhibition effects of 3’-NADP may not be as straightforward. Accumulation of 3’-NADP upon AvrRxo1 expression in bacterial and yeast cells suggests that 3’-NADP is not an enzyme substrate, nor is it turned over rapidly in these organisms. Early biochemical studies found little indication of 3’-NADP utilization or competitive inhibition of numerous central NADP-specific enzymes from various organisms, including the dehydrogenases of the pentose phosphate pathway, isocitrate dehydrogenase, glutathione reductase, and NADP phosphatases and transhydrogenases [32–35]. AvrRxo1 might also confer toxicity through depletion or sequestration of its substrate, NAD, and a resulting alteration in NAD (H) homeostasis and redox environment. We did not observe changes in relative NAD abundance that were dependent on catalytically active AvrRxo1 in bacteria, yeast, or infected rice. However, due to tightly maintained homeostasis of NAD levels, experimental manipulation of NAD synthesis or turnover often does not elicit measurable changes in NAD abundance, but often rather affects the rate of NAD metabolism [36,37]. Therefore, the lack of association of NAD levels with AvrRxo1 activity does not rule out NAD depletion as a mechanism of immune suppression. Profiling of diverse organisms chemically or genetically manipulated to alter NAD production have frequently reported effects on abundance of metabolites pertinent to NAD-related functions, notably TCA cycle intermediates, nucleotides, and glutathione species [36–40]. These trends are consistent with the profile of metabolites modestly affected by AvrRxo1 expression in yeast and bacteria (S1 File, S3 Fig, S4 Fig). Enhancement of NAD consumption or inhibited NAD production is usually associated with a decreased abundance of these metabolites, and vice-versa [36–39], so it is interesting that the NAD-modifying enzyme AvrRxo1 caused them to increase in our study. AvrRxo1 metabolic effects were studied here in conditions under which no toxic effects were observable (i. e. , early timepoints or high cell densities), and may only reflect early events in AvrRxo1 metabolic modification. NAD (H) quantification and assessment of metabolic flux and redox changes during the process of AvrRxo1 intoxication, as well as during immune suppression, will be needed to determine how metabolic perturbation by AvrRxo1 affects growth suppression or cell collapse. Expressed from a low-copy vector in X. oryzae on rice, AvrRxo1 shows no signs of toxicity or of major metabolic remodeling (S1 File), but it does exhibit several classic phenotypes of effectors in the plant infection process, including suppression of PTI, triggering of ETI, and suppression of a nonhost HR [8,9]. The Walker A site mutant AvrRxo1-T167N, which has no growth suppressing or cell collapse phenotype, retains an ability to trigger HR, suppress ROS, and enhance pathogen proliferation (summarized in Table 1). In N. benthamiana, this mutation uncouples the toxic effects of AvrRxo1 from plant immune functions, suggesting that the latter may be more sensitive to subtoxic levels of AvrRxo1 activity. As with toxicity, plant immune functions could be modulated through NAD utilization and/or through generation of 3’-NADP. AvrRxo1 is localized to the plant cytoplasm [9,22], which is thought to contain a small fraction of the tightly compartmentalized NAD in the eukaryotic cell [41,42]. Cytoplasmic NAD is an critical early signal in plant defense against diverse pathogens that is required for the PAMP-induced ROS burst and activation of stomatal immunity [43], salicyclic acid signalling, and callose deposition [25]. Manipulation of NAD metabolism by AvrRxo1 could disrupt a delicate redox balance key for NAD function in early defense signaling [24]. The observation that AvrRxo1-T167N inhibited the cytoplasmic NAD (P) H-derived ROS burst in plants, but not the mitochondria-localized ROS burst in yeast, would be consistent with the hypothesis that the smaller cytoplasmic pool of pyridine nucleotides is especially sensitive to subtoxic levels of AvrRxo1 activity. NAD is also required for production of the NADPH and cyclic ADP-ribose that function in the oxidative burst and calcium signaling, and for posttranslational defense signaling by sirtuins and poly-ADPribosyltransferases [24]. 3’-NADP could also be a critical inhibitor of the flg22-mediated ROS burst. 3’-NADP did not inhibit many central respiratory enzymes in older studies [32], but was later found to effectively inhibit maize NADP-malic enzyme in a noncompetitive fashion [44]. NADP-malic enzyme is a cytosolic electron donor that donates the electrons necessary for the defense-related oxidative burst in plants; the Magnaporthe oryzae effector AVR-Pii targets the rice malic enzyme Os-NADP-ME2 to inhibit the ROS burst [45]. Another logical ROS-blocking mechanism of 3’-NADP could be the inhibition of synthesis of NAADP from NADP by ADP-ribosyl cyclases. NAADP is a second messenger involved in calcium release, but its synthesis is not well characterized in plants. AvrRxo1 was also recently found to phosphorylate the NAD precursor, NAAD, in vitro at a similar efficiency to NAD, generating the product 3’ NAADP [16]. NAAD is a very low-abundance metabolite and we did not detect evidence of 3’ NAADP accumulation through global metabolome profiling, but this does not rule out NAAD as a potentially relevant target of AvrRxo1-induced phenotypes. Perhaps the most intriguing question arising from this work is that of how AvrRxo1 triggers activation of resistance mediated by the canonical NB-LRR R protein Rxo1. The lack of a phenotype in the D193T mutant suggests that AvrRxo1 kinase activity is required for Rxo1 activation. However, bacterial avirulence effectors are generally recognized through their activity on a protein target. The finding that AvrRxo1 modifies a small molecule prompts many questions on the nature of resistance activation- might the recognized signal be 3’-NADP, a change in the redox environment, or a change in NAD-dependent post-translational modification of proteins? By establishing NAD as an effector phosphorylation target during infection, this work demonstrates that direct “cofactor engineering” may be an effective strategy for pathogen manipulation of the host- and possibly, an effective signal for host recognition of the pathogen. The bacterial strains used in this study included E. coli DH5α for plasmid maintenance, E. coli BL21 (DE3) for protein expression, Agrobacterium tumefaciens strain GV2260 for transient expression in N. benthamiana, Xanthomonas oryzae pv. oryzicola strains BLS256 and MAI10 for infection of rice plants, and Xanthomonas oryzae strain X11-5A for detection of the Rxo1- mediated HR. Saccharomyces cerevisiae strain Y2H gold (TakaraBioUSA, Mountain View, CA) was used for AvrRxo1 expression. E. coli strains were routinely cultured in Luria Bertani (LB) medium at 37° C unless stated otherwise and A. tumefaciens strain GV2260 was grown in YEB medium at 28° C with appropriate antibiotics. X. oryzae pv. oryzicola strains BLS256 and MAI10 and Xanthomonas oryzae strain X11-5A were grown on peptone sucrose agar (PSA) at 28° C. Saccharomyces cerevisiae was grown in SD (Synthetic defined) media lacking tryptophan and containing either glucose (2%) or galactose (2%) or galactose/raffinose (2%/1%). Plasmids used in this study are listed in Table 2. For protein expression studies in E. coli, previously published pDEST527-derived clones were used for expression of 6xHis-AvrRxo1 and the catalytic site mutants 6xHis-AvrRxo1-D193T and 6xHis-AvrRxo1-T167N [8]. pDESTcv, used as a second control strain for pDEST studies, replaces the toxic Gateway cassette with the first 300 nt of the 5’ end of avrRxo1; this construct was previously shown to lack any growth suppression phenotype in bacteria [5]. For bacterial assays in rice, pAvrRxo1 (also referred to as pHM1-AvrRxo1) [7], pHM1-AvrRxo1-T167N [8] and pHM1-AvrRxo1-D193T (this study) were used with pHM1 (R. Innes, Indiana University) as the empty vector control. For construction of pHM1-AvrRxo1-D193T, the plasmid pAvrRxo1 [7] was used as a template for amplification of SacI-PstI flanked avrRxo1: arc1 region using the primers avrRxo1: arc1for, caccgagctcgacgcatttttatagcttcgttcg, and avrRxo1: arc1rev, gatcctgcagacagaggactcggattgaaccagt. The fragment was cloned into pENTR-D-Topo (Invitrogen). A single-site substitution was introduced using primers pHM1AvrRxo1D193Tfor, gggaaaggttgcgtgaatccaactgtattcaagagttcgcttgcg, and pHM1AvrRxo1-D193Trev, cgcaagcgaactcttgaatacagttggattcacgcaacctttccc, in a mutagenic PCR designed according to the manufacturer’s instructions for QuikChange II Site-Directed Mutagenesis Kit (Agilent), to generate pENTR-D-Topo-pAvrRxo1-D193T. Mutagenesis was confirmed by sequencing. The SacI-PstI fragment from vector pENTR-D-Topo-pAvrRxo1-D193T was cloned into pHM1. Inserts of all pHM1 and Gateway constructs were sequenced to confirm that avrRxo1-containing inserts were identical except for the single site mutations. pHM1 constructs were introduced into X. oryzae strain X11-5A by electroporation as previously described [7], and into MAI10 by triparental mating as previously described [46]. For Agrobacterium-mediated transient assays, entry vector pENTR-D-Topo clones carrying wild type avrRxo1, avrRxo1-D193T and avrRxo1-T167N [8] were used for recombination of the desired inserts into the Gateway compatible destination vector pEARLEYGATE 201 [47] using LR clonase according to manufacturer’s instructions (Invitrogen, Carlsbad, CA). The plasmids were transformed into A. tumefaciens via electroporation. For inducible expression in yeast, the Gateway destination vector pESC-Trp-DEST was constructed by cloning the SalI fragment of destination cassette C. 1 (Gateway Vector Conversion System, ThermoFischer Scientific) into the MCS2 of pESC-TRP (Agilent). Wild type avrRxo1, avrRxo1-T167N, and avrRxo1-D193T were recombined into pESC-Trp-DEST vector from the corresponding pENTR-D-Topo clones [8] via LR clonase reactions; the incorporated stop codon prevents expression of the C-terminal Myc tag. The plasmids were transformed into yeast using the Frozen-EZ Yeast Transformation Kit (Zymo Research, Irvine, CA) according to manufacturer’s instructions. Freshly streaked strains of E. coli transformed with IPTG inducible plasmid pDEST527 carrying wild type 6xHis-avrRxo1,6xHis-avrRxo1-T167N, 6xHis-avrRxo1-D193T and the pDESTcv plasmid were grown for 4 h. Exponentially growing cells were diluted to OD600 0. 2 and then 2-fold serial dilutions were spotted onto LB agar plates with and without the inducer 1 mM IPTG. Colonies were photographed after 24 h incubation at 37° C. Freshly streaked strains of S. cerevisiae transformed with pESC-TRP-DEST vectors carrying wild type avrRxo1, avrRxo1-D193T or avrRxo1-T167N were grown overnight in media containing 2% glucose. Exponentially growing cultures were serially diluted and spotted on SD agar plates containing galactose/raffinose (2%/1%). Overnight cultures of E. coli transformed with pDEST527-derived plasmids were suspended to OD600 0. 2 and expression was induced by addition of 1 mM IPTG. At 6 hpi, cells were harvested and washed thrice with sterile ice cold HPLC grade water. Cell pellets were frozen in liquid nitrogen and stored at -80° C prior to analysis. For protein expression in yeast, cultures of Saccharomyces cerevisiae transformed with pESC-TRP-DEST vectors carrying wild type avrRxo1, avrRxo1-D193T or pESC-TRP alone were grown in SD media containing 2% glucose for 3 days. Cultures were washed, resuspended in SD media containing 2% galactose and adjusted to OD600 0. 5. At 6 hpi in galactose-containing media, cells were harvested and washed thrice in sterile ice cold HPLC grade water. Cell pellets were snap-frozen and stored at -80° C until further use. Overnight cultures of E. coli transformed with pDEST527-AvrRxo1 and pDEST527-AvrRxo1-D193T were diluted 1: 100 with LB medium and grown at 30° C. When the culture density reached OD600 0. 4, expression was induced by addition of 1 mM IPTG. After 5 h, cells were harvested, resuspended in TBS (Tris buffered Saline, pH 7. 2), lysed by sonication and incubated in bacterial protein extraction reagent (ThermoFisher Scientific, Rockford, IL). for 30 min before centrifugation of pellet. Proteins were purified using HisPur Colbalt Resin (ThermoFisher Scientific, Rockford, IL) according to manufacturer’s instructions and eluted in TBS with 0. 3 M imidazole. Samples were desalted using Zeba Spin Desalting columns, 7K MWCO (ThermoFisher Scientific, Rockford, IL) according to manufacturer’s instructions, and eluted in water. Purified proteins were stored at -80° C until further use. The in vitro NAD kinase assay was performed according to Kornberg et al. [48], with modifications. Briefly, a reaction mixture comprising of purified 6xHis-AvrRxo1 or 6xHis-AvrRxo1-D193T (3 μM), 5 mM NAD+, 5 mM ATP, 10 mM MgCl2,100 mM Tris-HCl (pH 7. 5) was incubated for 1. 5 h or overnight at 29° C. The samples were snap-frozen and stored at -80° C until further use. A. tumefaciens strain GV2260 was used for transient expression in N. benthamiana. A. tumefaciens strains carrying different constructs were re-suspended in Agro-infiltration buffer (1M MgCl2,1 M MES, and 200 μM acetosyringone) to OD600 0. 8. Leaves of 4-5-week old N. benthamiana were infiltrated using a needleless syringe on the abaxial surface and the plants were incubated in 16 h daylight. For LC-MS based metabolic profiling, leaf discs were collected 36 hpi, macerated in liquid nitrogen and stored at -80° C until further use. For recording of watersoaking, photographs were taken 50 hpi. Leaves of 6-week-old Oryza sativa ssp. Japonica cv. Kitaake were inoculated with cell suspensions of X. oryzae pv. oryzicola strains (OD600 0. 2) prepared in water from a 48- to 72-h-old PSA plate cultures. The inoculations were done by infiltration of the suspensions on the abaxial leaf surface using a needleless syringe. For LC-MS based metabolic profiling, leaf discs were collected 12,24,48 hpi, macerated in liquid nitrogen and stored at -80μ C until further use. Four-week-old transgenic rice plants (cv. Kitaake) expressing Rxo1 [6] were grown in a growth chamber, and fully-expanded leaves were inoculated with 108 cfu/mL X. oryzae strain X11-5A carrying pHM1, pHM1-AvrRxo1, pHM1-AvrRxo1-T167N, or pHM1-AvrRxo1-D193T by leaf infiltration using needleless syringe as described by [6]. Inoculation sites were photographed at 5 days post infiltration. For samples from E. coli and S. cerevisiae expressing avrRxo1 or avrRxo1-D193T, 500 μL of MTBE solution (2: 3 75% methanol: methyl tert-butyl ether v: v) was added to cell pellets, followed by 90 min of sonication at room temperature. 125 μL of H2O was added to induce phase separation, followed by 5 min of vortexing. After centrifugation at 2095 × g for 15 min at 4° C, the bottom aqueous phase was removed and analyzed by LC-MS. For in vitro kinase assay reaction mixtures and infiltrated N. benthamiana leaf disc samples, 500 μL MTBE solution was added followed by 30 min of sonication, 30 min vortexing, and again 30 min of sonication at room temperature. 300 μL of H2O was added to induce phase separation followed by 15 min vortexing. After centrifugation at 2095 × g for 15 min at 4° C, the bottom, aqueous phase was removed, dried down under a gentle stream of nitrogen, resuspended in 150 μL of 40% methanol and then analyzed by LC-MS. For inoculated rice leaf samples, 1000 μL MTBE solution was added to each sample followed by 90 min of sonication at room temperature. 250 μL of H2O was added to induce phase separation followed by 5 min of vortexing. After centrifugation at 2095 × g for 15 min at 4° C, the bottom aqueous phase was removed and analyzed by LC-MS. Liquid chromatography coupled to mass spectrometry (LC-MS) was performed on a Waters Acquity UPLC system coupled to a Waters Xevo G2 time-of-flight (TOF) mass spectrometer (MS) for untargeted metabolomics analysis. For high resolution ms/ms experiments, a Waters Q-TOF MS was used to selectively fragment the parent ion of interest. A ZIC-pHILIC stationary phase (Merck Millipore, 150 x 2. 1 mm, 5 μM) was used for the separation of polar metabolites. Mobile phase A was H2O with 10 mM NH4HCO3 adjusted to pH 9. 6 with NH4OH and mobile phase B was acetonitrile. Flow rate was 270 μL/min and the column was held at 50° C. Injection volume was 2 μL. The gradient was as follows: 0 min 90% B, 1. 5 min 90% B, 8. 5 min 62% B, 11 min 40% B. The column was washed at 150 μL/min with 5% B for 1 min, then equilibrated at starting conditions with 7. 3 column volumes of solvent. For experiments performed on the TOF-MS, source and desolvation temperatures were at 150° C and 350° C, respectively. Desolvation gas flow was 850 L/hr. The MS was operated in full scan mode with an m/z range of 50–1200. Capillary voltage in negative ion mode was 2. 2 kV. For experiments performed on the QQQ-MS, source and desolvation temperatures were at 150° C and 500° C respectively. Desolvation, cone, collision, and nebuliser gas flows were 1000 L/hr, 150 L/hr, 0. 2 mL/min, and 7 Bar, respectively. The MS was operated in selected reaction monitoring mode. 2’- and 3’-NADP have a precursor ion of 744. 08 and product ions of 604 (collision energy = 20) and 136 (CE = 35). Capillary voltage in positive ion mode was 3. 1 kV. Detection of the flg22-based oxidative burst was performed using a luminol-HRP-based chemiluminescence assay. 24 h after infiltration of N. benthamiana leaves with A. tumefaciens, leaf discs (5 mm diameter) were collected and incubated overnight in H2O. Per measurement, two leaf discs were removed from H2O and transferred into cuvette containing 300 μL assay solution (20 mM luminol, 1 μg/ml horseradish peroxidase and 100 nM flg22 (EZBiolab, Carmel, IN, USA) or H2O (mock) ). Luminescence was measured over a period of 30–40 min using an FB12 single tube Luminometer (Berthold Detection Systems GmbH, Pforzheim, Germany). The oxidant-sensitive probe H2DCFDA (ThermoFisher Scientific) was used to determine the intracellular levels of ROS generated in yeast cells after treatment with menadione. Yeast cells were grown overnight in media containing glucose. They were then washed and diluted in SD media containing galactose/ raffinose for induction of expression. Exponentially growing cells were harvested at OD600 0. 8, washed and resuspended in 10 mm potassium buffer (6. 15 mM K2HPO4,3. 85 mM KH2PO4, pH 7. 0) followed by incubation for 30 min in the same buffer with 10 μM of H2DCFDA. After 60 min exposure to 8 mM menadione, cells were washed and resuspended in distilled water and disrupted using glass beads. Fluorescence of the cell lysate was measured at λEX = 490 nm and λEM = 519 nm using Synergy HI microplate reader (Biotek). The values were further normalized by protein concentration. A DNA fragment containing AvrRxo1 (65-421aa) was amplified from the genomic DNA of Xanthomonas oryzae pv. oryzicola strain BLS256, and cloned into the protein expression vector pGEX4T-1 (GE Healthcare Bio-Sciences, Pittsburgh, PA). To facilitate the removal of GST tag, a TEV protease cleavage site was inserted between the GST and targeted protein. The GST-AvrRxo1 (65-421aa) fusion protein was expressed in E. coli and purified as previously described [8]. The GST tag was removed by cleavage with AcTEV (ThermoFisher Scientific, Waltham, MA) and separated on PAGE. The protein band corresponding to AvrRxo1 (65-421aa) was used to generate anti-rabbit polyclonal antibodies (Cocalico Biologicals, Inc. , Reamstown, PA). To confirm expression of AvrRxo1 and derivatives in X11-5A, leaves of six-week-old O. sativa cv. Kitaake plants were inoculated with cell suspensions of X. oryzae pv. oryzicola strains as described above. Inoculated 3-cm leaf sections were collected at 48h, homogenized in liquid nitrogen, and then ground in 100 uL crude extraction buffer (50 mM Tris-HCl, pH 7. 5,2% SDS, 10 mM DTT, and 1 mM PMSF). Lysates were mixed 4: 1 with urea-amended SDS loading dye, for a final concentration of 1 M urea. Roughly 50 μg of total protein sample was separated on 10% SDS/PAGE gel and transferred to a PVDF membrane or stained with GelCode Blue Stain Reagent (ThermoFisher Scientific, Rockford, IL). Immunodetection was performed in a 1: 3000 dilution anti-AvrRxo1 polyclonal antibody, followed by a 1: 5000 dilution of goat anti-rabbit IgG-HRP conjugate. Commercial antibodies were obtained from ThermoFisher Scientific (Rockford, IL). The blot was incubated with ECL Plus Western blotting detection reagents (GE Healthcare, Amersham) following the manufacturer' s instructions, and blot was developed for two minutes on autoradiography film. To confirm expression of AvrRxo1 in transient expression experiments, N. benthamiana leaf disks were collected at 42 hours after infiltration with Agrobacterium strains for transient expression of HA-AvrRxo1, HA-AvrRxo1 (D193T) and HA-AvrRxo1 (T167N); this is after initial appearance of watersoaking in the HA-AvrRxo1 treatment but prior to total leaf tissue collapse. Lysis and separation was performed using the protocol above. Immunodetection was performed in a 1: 1000 dilution mouse anti-HA primary antibody followed by a 1: 3000 dilution of goat anti-mouse IgG-HRP conjugate. Blot developing was performed as described above except that the blot was visualized under a Chemi-Doc-It 2 chemiluminescence imager (UVP, Upland, CA).

Infectious bacteria have many strategies to weaken the defenses of their hosts. One common strategy is to inject proteins called effectors into the host cells. Many effectors inactivate proteins that transmit defense signals, disabling the alarm systems that alert the plant to defend against the pathogen. In this paper, we show that pathogens can also target very important small molecules in eukaryotes. We found that the plant pathogen effector AvrRxo1 adds a modification to NAD, a molecule required for hundreds of respiratory and signaling reactions, when expressed inside yeast or plant cells. This modification produces an unusual molecule, 3’-NADP, that is not able to be used efficiently by the cell. This finding demonstrates a new way that effectors can target essential central metabolites to manipulate the host.

Abstract Introduction Results Discussion Materials and methods

plant anatomy medicine and health sciences yeast infections pathology and laboratory medicine immunology immune suppression toxicology toxicity fungi plant science rice model organisms metabolites signs and symptoms experimental organism systems plant pathology fungal diseases plants saccharomyces research and analysis methods infectious diseases grasses plant bacterial pathogens leaves yeast biochemistry plant and algal models diagnostic medicine plant pathogens biology and life sciences yeast and fungal models saccharomyces cerevisiae metabolism organisms

2017

The effector AvrRxo1 phosphorylates NAD in planta

Identifying environmentally-specific genetic effects is a key challenge in understanding the structure of complex traits. Model organisms play a crucial role in the identification of such gene-by-environment interactions, as a result of the unique ability to observe genetically similar individuals across multiple distinct environments. Many model organism studies examine the same traits but under varying environmental conditions. For example, knock-out or diet-controlled studies are often used to examine cholesterol in mice. These studies, when examined in aggregate, provide an opportunity to identify genomic loci exhibiting environmentally-dependent effects. However, the straightforward application of traditional methodologies to aggregate separate studies suffers from several problems. First, environmental conditions are often variable and do not fit the standard univariate model for interactions. Additionally, applying a multivariate model results in increased degrees of freedom and low statistical power. In this paper, we jointly analyze multiple studies with varying environmental conditions using a meta-analytic approach based on a random effects model to identify loci involved in gene-by-environment interactions. Our approach is motivated by the observation that methods for discovering gene-by-environment interactions are closely related to random effects models for meta-analysis. We show that interactions can be interpreted as heterogeneity and can be detected without utilizing the traditional uni- or multi-variate approaches for discovery of gene-by-environment interactions. We apply our new method to combine 17 mouse studies containing in aggregate 4,965 distinct animals. We identify 26 significant loci involved in High-density lipoprotein (HDL) cholesterol, many of which are consistent with previous findings. Several of these loci show significant evidence of involvement in gene-by-environment interactions. An additional advantage of our meta-analysis approach is that our combined study has significantly higher power and improved resolution compared to any single study thus explaining the large number of loci discovered in the combined study. Identifying environmentally specific genetic effects is a key challenge in understanding the structure of complex traits. In humans, gene-by-environment (GxE) interactions have been widely discussed [1]–[12] yet only a few have been replicated. One reason for this discrepancy is the inability to accurately control for environmental conditions in humans as well as the inability to observe the same individuals in multiple distinct environments. Model organisms do not share such difficulties and for this reason can play a crucial role in the identification of gene-by-environment interactions. For example, in many mouse genetic studies the same traits are examined under different environmental conditions. Specifically, knock-out or diet-controlled mice are often utilized in the study of cholesterol levels. The availability of these studies presents a unique opportunity to identify genomic loci involved in gene-by-environment interactions as well as those loci involved in the trait independent of the environment. In order to utilize genetic studies in model organisms to identify gene-by-environment interactions, one needs to directly compare the effects of genetic variations in studies conducted under different conditions. This practice is complicated for a number of reasons, when combining more than two studies. First, environmental conditions are often variable across studies and do not fit to the standard univariate model for interactions. For example, in one study, cholesterol may be examined under different diet conditions (eg. low fat and high fat) and then in another study cholesterol is examined using gene knockouts. In this case, it is not straightforward to analyze these studies in aggregate using a single variable to represent the environmental condition. Applying a multivariate model, one in which the environment is represented using multiple environmental variables, results in increased degrees of freedom and low statistical power. Second, model organisms such as the mouse exhibit a large degree of population structure. Population structure is well-known for causing false positives and spurious associations [13], [14] in association analysis and can be expected to complicate the ability to combine separate studies. In this paper, we propose a random-effects based meta-analytic approach to combine multiple studies conducted under varying environmental conditions and show that this approach can be used to identify both genomic loci involved in gene-by-environment interactions as well as those loci involved in the trait independent of the environment. By making the connection between gene-by-environment interactions and random effects model meta-analysis, we show that interactions can be interpreted as heterogeneity and detected without requiring uni- or multi-variate models. We also define an approach for correcting population structure in the random effects model meta-analysis, extending the methods developed for fixed effects model meta-analysis [15]. We show that this method enables the analyses of large scale meta-analyses with dozens of heterogeneous studies and leads to dramatic increases in power. We demonstrate that insights regarding gene-by-environment interactions are obtained by examining the differences in effect sizes among studies facilitated by the recently developed m-value statistic [16], which allows us to distinguish between studies having an effect and studies not having an effect at a given locus. We applied our approach, which we refer to as Meta-GxE, to combine 17 mouse High-density lipoprotein (HDL) studies containing 4,965 distinct animals. To our knowledge, this is the largest mouse genome-wide association study conducted to date. The environmental factors of the 17 studies vary greatly and include various diet conditions, knock-outs, different ages and mutant animals. By applying our method, we have identified 26 significant loci. Consistent with the experience of meta-analysis in human studies, our combined study finds many loci which were not discovered in any of the individual studies. Among the 26,24 loci have been previously implicated in having an effect on HDL cholesterol or closely related lipid levels in the blood, while 2 loci are novel findings. In addition, our study provides insights into genetic effects on several disease loci and their relationship between environment and sex. For example, we identified 3 loci (Chr10: 21399819, Chr19: 3319089, ChrX: 151384614), where female mice show a more significant effect on HDL phenotypes than male mice. We also identified 7 loci (Chr1: 171199523, Chr8: 46903188, Chr8: 64150094, Chr8: 84073148, Chr10: 90146088, Chr11: 69906552, Chr15: 21194226) where male mice show a more significant effect on HDL than female mice. In addition, many of the loci show strong gene-by-environment interactions. Using additional information describing the studies and our predictions of which studies do and do not contain an effect, we gain insights into the interaction. For example, locus on chromosome 8 (Chr8: 84073148) shows a strong sex by mutation-driven LDL level interaction, which affects HDL cholesterol levels. Part of the reason for our success in identifying a large number of loci is that our study combined multiple mouse genetic studies many of which use very different mapping strategies. Over the past few years, many new strategies have been proposed beyond the traditional F2 cross [17] which include the hybrid mouse diversity panel (HMDP) [18], [19], heterogeneous outbred stocks [20], commercially available outbred mice [21], and the collaborative cross [22]. In our current study, we are combining several HMDP studies with several F2 cross studies and benefit from the statistical power and resolution advantages of this combination [15]. The methodology presented here can serve as a roadmap for both performing and planning large scale meta-analysis combining the advantages of many different mapping strategies. Meta-GxE is publicly available at http: //genetics. cs. ucla. edu/metagxe/. The Meta-GxE strategy uses a meta-analytic approach to identify gene-by-environment interactions by combining studies that collect the same phenotype under different conditions. Our method consists of four steps. First, we apply a random effects model meta-analysis (RE) to identify loci associated with a trait considering all of the studies together. The RE method explicitly models the fact that loci may have different effects in different studies due to gene-by-environment interactions. Second, we apply a heterogeneity test to identify loci with significant gene-by-environment interactions. Third, we compute the m-value of each study to identify in which studies a given variant has an effect and in which it does not. Forth, we visualize the result through a forest plot and PM-plot to understand the underlying nature of gene-by-environment interactions. We illustrate our methodology by examining a well-known region on mouse chromosome 1 harboring the Apoa2 gene, which is known to be strongly associated with HDL cholesterol [23]. Figure 1 shows the results of applying our method to this locus. We first compute the effect size and its standard deviation for each of the 17 studies. These results are shown as a forest plot in Figure 1 (a). Second we compute the P-value for each individual study also shown in Figure 1 (a). If we were to follow traditional methodology and evaluate each study separately, we would declare an effect present in a study if the P-value exceeds a predefined genome-wide significance threshold (P). In this case, we would only identify the locus as associated in a single study, HMDP-chow (M) (P =). On the other hand, in our approach, we combine all studies to compute a single P-value for each locus taking into account heterogeneity between studies. This approach leads to increased power over the simple approach considering each study separately. The combined meta P-value for the Apoa2 locus is very significant (), which is consistent with the fact that the largest individual study only has 749 animals compared to 4,965 in our combined study. In order to evaluate how significantly different the effect sizes of the locus are between studies, we apply a heterogeneity test. The statistical test is based on Cochran' s Q test [24], [25], which is a non-parametric test for testing if studies have the same effect or not. In this locus, the effect sizes are clearly different and not surprisingly the P-value of the heterogeneity test is significant (). This provides strong statistical evidence of a gene-by-environment interaction at the locus. Below we more formally describe how heterogeneity in effect size at a given locus can be interpreted as gene-by-environment interaction. If a variant is significant in the meta-analytic testing procedure, then this implies that the variant has an effect on the phenotype in one or more studies. Examining in which subset of the studies an effect is present and comparing to the environmental conditions of the studies can provide clues to the nature of gene-by-environment interactions at the locus. However, the presence of the effect may not be reflected in the study-specific P-value due to a lack of statistical power. Therefore, it is difficult to distinguish only by a P-value if an effect is absent in a particular study due to a gene-by-environment interaction at the locus or a lack of power. In order to identify which studies have effects, we utilize a statistic called the m-value [16], which estimates the posterior probability of an effect being present in a study given the observations from all other studies. We visualize the results through a PM-plot, in which P-values are simultaneously visualized with the m-values at each tested locus. These plots allow us to identify in which studies a given variant has an effect and in which it does not. M-values for a given variant have the following interpretation: a study with a small m-value () is predicted not to be affected by the variant, while a study with a large m-value () is predicted to be affected by the variant. The PM-plot for the Apoa2 locus is shown in Figure 1 (b). If we only look at the separate study P-values (y-axis), we can conclude that this locus only has an effect in HMDP-chow (M). However, if we look at m-value (x-axis), then we find 8 studies (HMDPxB-ath (M), HMDPxB-ath (F), HMDP-chow (M), HMDP-fat (M), HMDP-fat (F), BxD-db-5 (M), BxH-apoe (M), BxH-apoe (F) ), where we predict that the variation has an effect, while in 3 studies (BxD-db-12 (F), BxD-db-5 (F), BxH-wt (M) ) we predict there is no effect. The predictions for the remaining 6 studies are ambiguous. From Figure 1, we observe that differences in effect sizes among the studies are remarkably consistent when considering the environmental factors of each study as described in Table 1. For example, when comparing study 1–4, the effect size of the locus decreases in both the male and female HMDPxB studies in the chow diet (chow study) relative to the fat diet (ath study). Thus we can see that when the mice have Leiden/CETP transgene, which cause high total cholesterol level and high LDL cholesterol level, effect size of this locus on HDL cholesterol level in blood is affected by the fat level of diet. Similarly, when comparing study 12–15, the knockout of the Apoe gene affects the effect sizes for both male and female BxH crosses. However, in the BxD cross (study 8–11), where each animal is homozygous for a mutation causing a deficiency of the leptin receptor, the effect of the locus is very strong in the young male animals, while as animals get older and become fatter, the effect becomes weaker. However in the case of female mice, the effect of the locus is nearly absent at both 5 and 12 weeks of age. Thus we can see that sex plays an important role in affecting HDL when the leptin receptor activity is deficient. We note that there are many genes in this locus and the genetic mechanism of interactions may involve genes other than Apoa2. Despite this caveat, the results of Meta-GxE at this locus provides insights into the nature of GxE and can provide a starting point for further investigation. We note that an alternate explanation for differences in effect sizes between studies is the presence of gene-by-gene interactions and differences in the genetic backgrounds of the studies. While this is a possible explanation for differences in effect sizes between the different crosses and the HMDP studies, in Figure 1, we see many differences in effect sizes among studies with the same genetic background. Thus gene-by-gene interactions can only partially explain the differences in observed effect sizes. Gene-by-environment interactions, random effects meta-analysis and heterogeneity testing are closely related. Suppose we have studies each conducted under different environmental conditions. We define the following linear model, where is the observed phenotype for study, is the phenotype mean for study, is the genetic effect on the phenotype for study, is the genotype, and is the residual error. (1) Since each environment is different, the effect size is partially determined by environmentally-specific factors and partially determined by factors common to all studies. Given that we can decompose the effect into environment-independent and environment-dependent factors. Then we define the following linear model, where is the environment-independent genetic effect and is the environment-dependent genetic effect for study. (2) In order to test for the presence of an effect shared across environments, we test the null hypothesis and to test for the presence of a gene-by-environment interaction, we test the hypothesis that. In the random effects meta-analysis, we assume that the effect size is sampled from a normal distribution with mean and variance, denoted. Under this assumption, we test the null hypothesis and, in order to obtain a study-wide P-value. Additionally, we perform a heterogeneity test to test the null hypothesis versus the alternative hypothesis. We posit that by conducting hypotheses tests in the meta-analysis framework, we are simultaneously testing for the presence of environmentally-independent and environmentally-specific effects and that by applying heterogeneity testing we are testing for only environmentally-specific effects. Consider that in the meta-analysis framework is analogous to and the variation () around is analogous to variation among s. In the random effects meta analysis testing framework we are testing if and. This is equivalent to testing both environmentally-independent () and environmentally-dependent () effects simultaneously. In heterogeneity testing, we test the null hypothesis versus the alternative hypothesis. When the environmentally-dependent effect () is 0 it means that and thus. When, we expect that will vary around, so that we do not expect that. Since the variation () of around is analogous to the variable, heterogeneity testing in the meta-analysis framework is approximately equivalent to testing for environmentally-specific effects. The presence of heterogeneity in the effect size at causal genetic loci due to gene-by-environment interactions is naturally expected in mouse genetic studies when combining studies with varying environmental conditions. One extreme example comes from a knock-out experiment. If the knocked-out gene is causal for a particular trait, then we can expect that the gene would have no effect on a knock-out mouse, while the gene would have an effect on the wild type mouse. This is a binary form of heterogeneity. In a less extreme form of heterogeneity, the effect of a given gene may be affected by an environmental factor which varies in different mice – ranging from small effects to large effects. To see the relationship between significance of the association and gene-by-environment interactions, we compute and compare this P-value for each SNP from the 17 studies using the random effects meta-analysis to a measure of heterogeneity. Heterogeneity can be assessed by statistic, which describes the percentage of variation across studies that is due to heterogeneity rather than chance [26]. Figure 2 compares statistic with the meta-analysis P-value for each SNP. In this figure, we see that is uniformly distributed for the non-significant SNPs (blue dots), while it is right skewed for significant SNPs (red dots), indicating that more significant SNPs have a greater potential for exhibiting heterogeneity in effect. Since heterogeneity in this case can be interpreted as representing gene-by-environment interactions, as heterogeneity is induced by differences in the environment, we see that the presence of a GxE interaction confers higher power to detect an association. The power to identify both gene-by-environment and main effects in a meta-analysis of mouse studies depends on both the main effect size and the amount of heterogeneity. We performed simulations using the genotypes of the 17 mouse studies analyzed in this paper. We simulated a range of main effect (mean effect) sizes and a range of gene-by-environment effects. We are simulating the realistic scenario in which we do not know exactly the set of covariates which are responsible for the gene-by-environment effects. We simulated gene-by-environment effects by drawing the effect in each study from a distribution with a mean given by the main effect size and a variance controlling the magnitude of gene-by-environment interactions. If this variance is small, then all of the studies have close to the same effect size and there are few gene-by-environment effects. If the variance is high, then there are strong gene-by-environment effects. Figure 3 shows the results of our simulations. 1000 simulated phenotypes were generated for each mean and variance pair. Statistical power is estimated by computing the proportion of the datasets in which a simulated effect is detected. We observe that the power is high for a wide range of main effect sizes and gene-by-environment effect sizes which is explained by the large sample size of the study. We also observe that even for small main effects, if there are strong gene-by-environment effects, we can still identify the locus. This is because in this case a subset of the studies will have strong effect sizes due to gene-by-environment effects. Our approach is not the only way to analyze a meta-analysis study. We compare the power to two other meta-analytic approaches. The first is the traditional meta-analysis strategy which uses a fixed effects model (FE) in which all of the effect sizes across studies are assumed to be the same. We utilize an extension of the fixed effects model which corrects for population structure [15]. A second alternate strategy is to simply apply the heterogeneity test (HE), which in our framework is only applied to loci first identified using random effects meta-analysis. The HE test follows the intuition that loci with high heterogeneity will harbor gene-by-environment interactions. For the purposes of the comparison we refer to Meta-GxE as the random effects (RE) model. The level of gene-by-environment interactions can be simulated by changing both the environment-dependent and environment-independent effect simultaneously, when simulating the phenotype. Figure 4 (a) – (c) shows the power of the three approaches (RE, FE, HE) respectively when we vary the mean and variance of the effect size distribution we sampled from. In this simulation study, mean effect represents shared effect and variance of the effect size represents interaction effect. As expected, RE has high power in cases where the shared effect or the interaction effect is large. FE has high power when the shared effect is large and the HE test has high power when the interaction effect is large. Figure 4 (d) shows the heatmap which is colored with the color of highest powered approach. FE is most powerful at the top-left region, HE is most powerful at the bottom-right region, while RE is most powerful for a majority of the simulations. In the Text S1, we show through simulations that our methodology outperforms the alternative fixed effects and heterogeneity testing approaches when the effect is present in a subset of the studies, which is another possible interaction model we can assume. We also show in the Text S1 that our approach is more powerful than the traditional uni- or multi-variate gene-by-environment association approach which assumes knowledge of the covariates involved in gene-by-environment interactions. For the traditional uni- or multi-variate approach, required knowledge includes kinds of variable (e. g. sex, age, gene knockouts) and encoding of the variables (e. g. binary values, continuous values). In the Text S1, we also show the our proposed approach controls the false positive rate. ? We applied Meta-GxE to 17 mouse genetic studies conducted under various environmental conditions where each study measured HDL cholesterol. Table 1 summarizes each study. More details are provided in the Materials and Methods section and in Text S1. We analyzed all 17 studies together and we also analyzed the 9 male and 8 female studies separately. Some significant associations are shared and some associations are specific to males and females. The Manhattan plots in Figure 5 show the meta-GxE result when applied to the 17 studies, 9 male only studies and 8 female only studies. Table 2 summarizes 26 significant peaks (P) showing the P-values obtained by applying meta-GxE to the male only studies (9 studies), the female only studies (8 studies) and the male+female studies (17 studies). For each significant locus, we computed m-values, interpreted as the posterior probability of having an effect on the phenotype and report the number of studies with an effect (E), the number of studies with ambiguous effect size (A) and the number of studies without an effect (N). We also report the number of individual studies where the locus was significant (P). As seen in the table, many of the loci were not significant in any of the individual studies and would not have been discovered without combining the studies. We note that we use a more stringent genome wide threshold of P than was used in the original studies. The Genes in Region and Gene Refs columns contain the gene names near the locus previously known to affect HDL cholesterol level or closely related lipid level in the blood and associated literature citations. Among the 26 loci that we identified by applying Meta-GxE, 24 loci are near the genes (mostly genes are located within 1MB of the peak) known to affect HDL or closely related lipid level in the blood, while 3 loci are novel. For example, we identified 3 loci (Chr10: 21399819, Chr19: 3319089, ChrX: 151384614) female mice show a more significant effect on HDL phenotypes than male mice. We also identified 7 loci (Chr1: 171199523, Chr8: 46903188, Chr8: 64150094, Chr8: 84073148, Chr10: 90146088, Chr11: 69906552, Chr15: 21194226) where male mice show a more significant effect on HDL than female mice. Interestingly, we observed that in 3 loci (Chr10: 21399819, Chr19: 3319089, ChrX: 151384614), female mice are more highly affected, while in 7 loci (Chr1: 171199523, Chr8: 46903188, Chr8: 64150094, Chr8: 84073148, Chr10: 90146088, Chr11: 69906552, Chr15: 21194226) male mice are more highly affected. Among 26 loci, many show a significant heterogeneity in effect sizes between the 17 studies, which we interpret as gene-by-environment interactions. One interesting example showing strong gene-by-environment interaction is a locus in Chr8: 84073148. This locus is located near the gene, which is known to affect the abnormal lipid levels in blood [27]. Figure 6 shows the forest plot and PM-plot for this locus. If we look at the forest plot of the locus in Figure 6, we can easily see that there are two groups: 12 studies with an effect (red dots) and 5 studies with an ambiguous prediction of the existence of an effect (green dots). Interestingly, the log odds ratios of effect size for the 12 studies with an effect is about the same (around 0. 2). The common characteristic in 4 of the 5 studies (HMDPxB-chow (F), HMDPxB-ath (F), BXH-apoe (F), CXB-ldlr (F) ) is that they are female mice with high LDL levels in the blood. In addition, in all 4 cases, these high LDL levels are caused by mutant genes. Mice in HMDPxB-chow and HMDPxB-ath studies have transgenes for both Apoe Leiden and for human Cholesterol Ester Transfer Protein (CETP), while mice in the BXH-apoe and CXB-ldlr studies carried knockouts of the genes for Apoe and LDL receptor, respectively. This is a strong evidence that there is an interaction between sex×mutation-driven LDL levels through this locus (Chr8: 84073148) when affecting HDL levels in mice. Figures S5, S6, S7, S8, S9, S10, S11, S12, S13, S14, S15, S16, S17, S18, S19, S20, S21, S22, S23, S24, S25, S26, S27, S28, S29, S30 show the forest plots and PM-plots for each locus, which show information such as effect sizes, the direction of the effect, which study has an effect and which study does not have an effect for each of 17 studies at the given locus. In this paper, we present a new meta-analysis approach for discovering gene-by-environment interactions that can be applied to a large number of heterogeneous studies each conducted in different environments and with animals from different genetic backgrounds. We show the practical utility of the proposed method by applying it to 17 mouse HDL studies containing 4,965 mice, and we successfully identify many known loci involved in HDL. Consistent with the results of meta-analysis in human studies, our combined study finds many loci which were not discovered in any of the individual studies. A point of emphasis is that in our study design, in each of the combined studies, all of the individuals in the study are subject to only a single environment. This is distinct from other approaches for discovery of gene-by-environment interactions using meta-analysis such as those described in [28]. In these approaches, in each of the combined studies, the individuals in the study are subject to multiple environments and information on each individual' s environment is collected. Gene-by-environment statistics are then computed in each study and then combined in the meta-analysis. In our study design, we compute main effect sizes for each SNP and then look for variants where the effect sizes are different suggesting the presence of a gene-by-environment interaction. In our meta-analysis approach, we assume that we do not have any prior knowledge of the effect size in any particular study. However one might incorporate prior knowledge of the specific environmental effects. In some cases, one might know that some of the studies have similar effect sizes as compared to others. Or the prior knowledge might suggest that one specific study needs to be eliminated in the meta analysis. If we utilize such prior knowledge, we may be able to achieve even higher statistical power. In this paper we have addressed how to perform meta-analysis when the studies have different genetic structures, building off the results of our previous study [15]. While in this paper we combine 7 HMDP studies with 10 genetic crosses, the approach in principle can be used to combine any variety of study types. Recently, several strategies for mouse genome-wide association mapping have been proposed [29] [17]. These include HMDP [18], collaborative cross [30] and outbredstock [21] [17]. The approach presented here can be utilized to combine these different kinds of studies and is a roadmap for integrating the results of different strategies for mouse GWAS. Although we have focused on explaining heterogeneity by gene-by-environment interaction, it is possible that the differences in effect sizes can be caused by gene-by-gene interactions on different genetic backgrounds, where the interacting variants differ in frequency in the different studies. While gene-by-gene interactions certainly contribute to locus heterogeneity, we predict that, in combining studies with similar genetic structures, locus heterogeneity more likely arises from gene-by-environment interactions. In any case, determining whether or not these heterogeneous loci are environment-driven or interaction-driven is an important and interesting direction for future study. In the model organism studies for which we can control the environment, the standard study design for testing gene-by-environment interactions is to combine multiple cohorts whose environments are known. The environmental value that we vary is typically a quantitative measure that we can model with a single random variable. Thus, the standard univariate linear model can be appliedwhere is vector of phenotype measurements from individuals, is the phenotype mean, is the main environmental effect mean, is environmental status vector, is the genetic effect, is genotype vector, is GxE interactions effect, denotes the dot-product between two vectors, and is the residual error, which follows normal distribution. In this model, vector is a vector of indicators which describes the environmental status of each individual. study. For example, Suppose the environmental condition of one study is wildtype and that of another is gene knockout. In this case, the environmental condition of wildtype is described as 0 and that of knockout is described as 1. In order to test if there are interactions, we test the null hypothesis versus the alternative hypothesis. Another possible testing strategy is to test the interactions effect together with the genetic effect, that is, the null hypothesis versus the alternative hypothesis. This strategy is powerful in detecting loci exhibiting both the genetic effects and the interactions effects. For more complicated scenarios where the different environments can not be modeled with a single variable, a straightforward extension of the standard univariate interactions model is the multivariate model. Suppose that there are k different possible environments and the information on the environments of each individual are captured by a matrix D which has k columns where each column corresponds to one environment. Then, the standard multivariate interactions model will be (3) is the column of the D matrix, is the environment specific mean, denotes the phenotype measurements, denotes the genotypes, denotes the fixed genetic effect, denotes GxE interactions effect of environmental variable and, and denotes the residual error. Then the testing will be between the null hypothesis versus the alternative hypothesis. The test statistic will bewhere is the z-score corresponding to. follow under the null. Similarly to the univariate model, if we want to test the interactions effect together with genetic effect, we add the z-score corresponding to into the statistic, in which case the statistic will follow. Before we describe the relationship between gene-by-environment interactions and meta-analysis, we first describe the standard fixed effects and random effects meta-analysis in details. Here we explain more about the relationship between gene-by-environment interactions and meta-analysis based on the explanation in Results section. If we do not consider the interactions, it has been already known that the fixed effects model meta-analysis is approximately equivalent to the linear model of combined cohorts [35]. That is, the fixed effects model equation (5) gives approximately equivalent results to the combined linear model (8) where is the combined genotype vector from all cohorts, is a matrix that includes indicator columns which identify which individual is in each cohort, is the column of matrix A, and is the cohort specific mean. The two methods are approximately equivalent because they both test the fixed mean effect (in equation (8) and in equation (5) ). The subtle difference between the two models is that in equation (8), we assume the error follows a single normal distribution (e. g.), whereas in equation (5), the variance of the distributions may differ between studies (e. g. for each). In other words, under the constant error variance assumption (), the two models become equivalent and in equation (8) equals in equation (5), Similarly, by considering interactions, we extend this argument to show the relationship between gene-by-environment interactions and meta-analysis. We consider the relationship between equation (3) and equation (4). For simplicity of the notation, we consider the case where the matrix D is defined in such a way that each individual is only in one environment such that the D matrix is equivalent to the matrix A described above. If we assume the constant error variance assumption, we establish the following relationship, where the left hand side is the coefficient of the genotype of study from the meta-analysis equation (4) and the right hand side is the same coefficient of (the study' s part within the combined genotype matrix) from the equation (3). Suppose that there are no interactions (null hypothesis of interaction testing). Then, for each study. Thus, the effect size of meta-analysis is equivalent to, the genetic effects that are invariant across studies. Therefore, (null hypothesis of heterogeneity testing). On the other hand, suppose that (null hypothesis of heterogeneity testing). Naturally, for all studies (null hypothesis of interaction testing). This shows that the null hypothesis of the interactions test in the model (3) and the null hypothesis of the heterogeneity test in meta-analysis are equivalent. As a result, we can utilize meta-analytic heterogeneity testing to detect interactions. Using reasoning, it is straightforward to show that we can utilize the random effects model meta-analysis method to detect the mean effect and the interaction effect at the same time, which can be powerful for identifying loci bearing both kinds of effects. Model organism such as the mouse are well-known to exhibit population structure or cryptic relatedness [36], [37], where genetic similarities between individuals both inhibit the ability to find true associations and cause the appearance of a large number of false or spurious associations. Mixed effects models are often used in order to correct this problem [38]–[42]. Methods employing a mixed effects correction account for the genetic similarity between individuals with the introduction of a random variable into the traditional linear model. (9) In the model in equation (9), the random variable represents the vector of genetic contributions to the phenotype for individuals in population. This random variable is assumed to follow a normal distribution with, where is the kinship coefficient matrix for population. With this assumption, the total variance of is given by. A z-score statistic is derived for the test by noting the distribution of the estimate of. In order to avoid complicated notation, we introduce a more basic matrix form of the model in equation (9), shown in equation (10). (10) In equation (10), is a matrix with the first column being a vector of 1 s representing the global mean and the second vector is the vector and is a coefficient vector containing the mean and genotype effect. We note that this form also easily extends to models with multiple covariates. The maximum likelihood estimate for in population is given by which follows a normal distribution with a mean equal to the true and variance. The estimates of the effect size and standard error of the () are then given in equation (11) and equation (12), where is a vector used to select the appropriate entry in the vector. (11) (12) When we test gene-by-environment interactions with meta analysis approaches, one important step is correcting for population structure. This can be achieved by correcting for population structure within each study first as described above. For example, consider the random effects model meta-analysis method that we primarily focus on. We employ population structure control, using (11) and (12). Then the likelihood ratio test statistic will be (13) where and. After identifying loci exhibiting interaction effects, we employ the meta-analysis interpretation framework that we recently developed. The m-value [16] is the posterior probability that the effect exists in each study. Suppose we have number of studies we want to combine. Let be the vector of estimated effect sizes and be the vector of estimated variance of effect sizes. We assume that the effect size follows the normal distribution. (14) (15) We assume that the prior for the effect size is (16) A possible choice for in GWASs is 0. 2 for small effect and 0. 4 for large effect [43]. We also denote be a random variable whose value is 1 if a study has an effect and 0 otherwise. We also denote as a vector of for studies. Since has binary values, can be possible configurations. Let be a vector containing all the possible these configurations. We define m-value as the probability, which is the probability of study having an effect given the estimated effect sizes. We can compute this probability using the Bayes' theorem in the following way. (17) where is a subset of whose elements' value is 1. Now we need to compute and. can be computed as (18) where denotes the number of 1' s in c and B denotes the beta function and we set and as 1 [16]. The probability given configuration, , can be computed as (19) (20) (21) where where is the indices of 0 in and is the indices of 1 in, denotes the probability density function of the normal distribution with mean and variance. is the inverse variance or precision and is a scaling factor. (22) All summations appeared for computing, and are with respect to. The m-values have the following interpretations: small m-values (0. 1) represent a study that is predicted to not have an effect, large m-values (0. 9) represent a study that is predicted to have an effect, otherwise it is ambiguous to make a prediction. It was previously reported that m-values can accurately distinguish studies having an effect from the studies not having an effect [16]. For interpreting and understanding the result of the meta-analysis, it is informative to look at the P-value and m-value at the same time. We propose to apply the PM-plot framework [16], which plots the P-values and m-values of each study together in two dimensions. Figure 1 (b) shows one example of a PM-plot. In this example, studies with an m-value less than are interpreted as studies not having an effect while studies with an m-value greater than are interpreted as studies having an effect. For studies with an m-value between and, we cannot make a decision. One reason that studies are ambiguous () is that they are underpowered due to small sample size. If the sample size increases, the study can be drawn to either the left or the right side.

Identifying gene-by-environment interactions is important for understand the architecture of a complex trait. Discovering gene-by-environment interaction requires the observation of the same phenotype in individuals under different environments. Model organism studies are often conducted under different environments. These studies provide an unprecedented opportunity for researchers to identify the gene-by-environment interactions. A difference in the effect size of a genetic variant between two studies conducted in different environments may suggest the presence of a gene-by-environment interaction. In this paper, we propose to employ a random-effect-based meta-analysis approach to identify gene-by-environment interaction, which assumes different or heterogeneous effect sizes between studies. Our approach is motivated by the observation that methods for discovering gene-by-environment interactions are closely related to random effects models for meta-analysis. We show that interactions can be interpreted as heterogeneity and can be detected without utilizing the traditional approaches for discovery of gene-by-environment interactions, which treats the gene-by-environment interactions as covariates in the analysis. We provide a intuitive way to visualize the results of the meta-analysis at a locus which allows us to obtain the biological insights of gene-by-environment interactions. We demonstrate our method by searching for gene-by-environment interactions by combining 17 mouse genetic studies totaling 4,965 distinct animals.

Abstract Introduction Results Discussion Materials and Methods

genetics genome-wide association studies biology animal genetics

2014

Meta-Analysis Identifies Gene-by-Environment Interactions as Demonstrated in a Study of 4,965 Mice

Cryptosporidium are parasitic protozoa that infect humans, domestic animals, and wildlife globally. In the United States, cryptosporidiosis occurs in an estimated 750,000 persons annually, and is primarily caused by either of the Cryptosporidium parvum genotypes 1 and 2, exposure to which occurs through ingestion of food or water contaminated with oocytes shed from infected hosts. Although most cryptosporidiosis cases are caused by genotype 1 and are of human origin, the zoonotic sources of genotype 2, such as livestock, are increasingly recognized as important for understanding human disease patterns. Social inequality could mediate patterns of human exposure and infection by placing individuals in environments where food or water contamination and livestock contact is high or through reducing the availability of educational and sanitary resources required to avoid exposure. We here analyzed data from the National Health and Nutritional Examination Survey (NHANES) between 1999 and 2000, and related seropositivity to Cryptosporidium parvum to correlates of social inequality at the household and individual scale. After accounting for the complex sampling design of NHANES and confounding by individual demographics and household conditions, we found impaired household food adequacy was associated with greater odds of Cryptosporidium seropositivity. Additionally, we identified individuals of non-white race and ethnicity and those born outside the United States as having significantly greater risk than white, domestic-born counterparts. Furthermore, we provide suggestive evidence for direct effects of family wealth on Cryptosporidium seropositivity, in that persons from low-income households and from families close to the poverty threshold had elevated odds of seropositivity relative to those in high-income families and in households far above the poverty line. These results refute assertions that cryptosporidiosis in the United States is independent of social marginalization and poverty, and carry implications for targeted public health interventions for Cryptosporidium infection in resource-poor groups. Future longitudinal and multilevel studies are necessary to elucidate the complex interactions between ecological factors, social inequality, and Cryptosporidium dynamics. Cryptosporidium are parasitic protozoa that infect humans, domestic animals, and wildlife globally [1]. In humans, cryptosporidiosis is a major cause of global diarrheal illness, and in the United States an estimated 750000 cases occur annually [2]. Although direct mortality from Cryptosporidium infection is rare and often limited to immunocompromised individuals, cryptosporidiosis can cause significant morbidity that in turn can result in high healthcare expenses and losses to productivity [3–5]. Human cryptosporidiosis is primarily caused by either of the Cryptosporidium parvum genotypes 1 and 2, also known as Cryptosporidium hominis and Cryptosporidium parvum [6,7]. Human exposure to Cryptosporidium parvum occurs primarily through ingestion of food or water contaminated with oocytes shed from infected hosts [8,9]. While genotype 1 is limited to human transmission cycles [10], transmission of genotype 2 is based in livestock reservoir hosts such as cattle [11]. Although the majority of cryptosporidiosis cases are caused by genotype 1, the zoonotic capacity of genotype 2 is increasingly recognized as important for understanding human disease patterns [12–14]. For example, cryptosporidiosis cases in the United Kingdom are highest in agricultural areas that utilize cattle manure [15], and water-borne outbreaks in Ireland and the United States have been traced back to cattle [16,17]. Furthermore, exposure directly from livestock has also been observed, particularly among persons working closely with cattle such as children and farmworkers [18,19]. Despite potential linkages between livestock reservoir sources and cryptosporidiosis, we know less about how social inequality may mediate patterns of human exposure and infection. This is unfortunate, as Cryptosporidium is now included in the WHO Neglected Disease Initiative [20], and epidemiological evidence suggests low socioeconomic conditions may amplify human risk. A cross-sectional study in Venezuela found individuals residing in poor urban sectors and in thatched roof–style housing had higher Cryptosporidium prevalence [21]. In a similar setting in Brazil, household water and nearby domestic animals tested positive for Cryptosporidium oocysts [22]. Such work suggests living in physically impaired environments can increase exposure to contaminated water or animals harboring infection [23]. An impaired social environment could also influence patterns of human exposure, as individuals within these environments may lack resources necessary for proper sanitation or educational avoidance of transmission routes. For example, women in Kenyan agricultural communities had greater exposure to cattle and contaminated food and water owing to power hierarchies within households [24]. A major limitation of past work on social determinants of cryptosporidiosis is a focus on either the individual or household level at small spatial scales, restricting understanding interactions between scales and broader inference [25]. To simultaneously address individual- and household-scale socioeconomic drivers of cryptosporidiosis risk, we utilized the National Health and Nutrition Examination Survey (NHANES) from the United States to ask how poor physical and social conditions affect the odds of seropositivity to Cryptosporidium parvum. Several reviews of neglected infections in the United States have noted that cryptosporidiosis is without significant links to poverty or social marginalization [26,27], and socioeconomic factors remain absent in syntheses of risk factors for this disease in the United States [28]. Yet a prior analysis of NHANES found that Hispanics, African Americans, and women all have greater odds of Cryptosporidium seropositivity [29]. Incorporation of household-scale socioeconomics may therefore improve our understanding of how an impaired physical or social environment contextualizes these individual-level relationships. Furthermore, reorienting our focus on cryptosporidiosis towards socioeconomic determinants could offer tangible opportunities for public health interventions and environmental management. Our analysis used cross-sectional data from NHANES, a series of large nationally representative surveys conducted by the National Center for Health Statistics (NCHS) based on a stratified, multistage, probability cluster design. Data are collected through household interviews, standardized physical examinations, and collection of biological samples at mobile examination centers. A nationally representative sample is selected annually, but data are released in two-year cycles to protect confidentiality and increase statistical reliability. All data were obtained from NHANES between 1999 and 2000, the only two years for which Cryptosporidium serological testing occurred. To ensure adequate sample size, NHANES 1999–2000 oversampled low-income persons, adolescents 12–19 years of age, persons ≥ 60 years of age, non-Hispanic blacks, and Mexican Americans. Data were weighted to represent the total civilian non-institutionalized U. S. household population and to account for oversampling and nonresponse to the household interview and physical examination. The weights were further ratio-adjusted by age, sex, and race and ethnicity to the U. S. population control estimates from the Current Population Survey adjusted for undercounts. Of the 5663 study participants aged 6 to 49 and who underwent physical examination, 4359 individuals had serum samples available for evaluation and had data available for relevant socioeconomic and demographic covariates for this study. Participants with hemophilia or recipients of chemotherapy within four weeks were excluded. NHANES is reviewed and approved annually by the NCHS institutional review board, and informed written consent was obtained from all participants or their parents or legal guardians. All individual records were anonymized through unique respondent sequence numbers within NHANES. Infection with Cryptosporidium parvum is accompanied by the production of parasite-specific antibody (Ig) of all major classes [30,31]. NHANES used an experimental enzyme-linked immunosorbent assay (ELISA) to measure IgG antibodies to two surface antigens to Cryptosporidium parvum, 17kDA and 27kDA [32,33]. In experimental human studies, IgG reactivity to these antigens peaks within 4 to 6 weeks [34]. The IgG response to 17kDA declines to near background levels by 4 to 6 months, whereas the same antibody response to 27kDA can remain elevated for at least 6 to 12 months [35,36]. Evidence from animal and human studies suggests that this antibody response requires inoculation with Cryptosporidium parvum oocysts and that seropositivity to both antigens develops after either asymptomatic or symptomatic infection [34,37]. Hence the IgG response to both antigen groups likely reflects recent or current Cryptosporidium parvum infection and not merely exposure [36,38]. To determine seropositivity to Cryptosporidium parvum, serum samples were tested for reactivity to both antigen groups through the ELISA methods detailed by NHANES and [32,33]. Briefly, sample absorbance was measured using a Molecular Devices UVmax kinetic microplate reader, and IgG levels were assigned a unit value based on the eight-point positive control standard curve with a four-parameter curve fit. The 1: 50 dilution of the positive control was assigned a value of 6400 units; unknown samples with absorbance values above the standard curve were diluted further and reassayed. Cutoff values to determine seropositivity are not reported within NHANES; however, prior studies have used cutoffs for seropositivity to the 17kDA and 27kDA antigens as a sample absorbance greater than 86 units, exceeding the mean plus three standard deviations of the negative control, or exceeding 10% of the positive control [38–40]. NHANES reports seropositivity separately to the 17kDA and 27kDA antigen groups as binary outcomes. Because of our interest in the social determinants of Cryptosporidium seropositivity, which likely remain constant through the duration of both IgG antibody responses, we here report seropositivity as a positive IgG response to both 17kDA and 27kDA antigen groups. Following experimental studies, a seropositive response to both of these antigens represents a likely recent or current infection with either genotype 1 or genotype 2 of Cryptosporidium parvum [36,38]; however, a seropositive result does not distinguish between the distinct human or zoonotic sources of oocysts nor whether or not an individual is currently infected. We examined three household-level indicators of social inequality available through NHANES: food adequacy, annual income, and the poverty income ratio (PIR). Food adequacy was defined as households reporting “enough and the kinds of food wanted, ” “enough but not always the kinds of food wanted”, and “sometimes/often not enough to eat. ” This variable was recoded as “enough, ” “some, ” and “not enough” to eat and could serve as a proxy for socioeconomic status, in which resource-poor households are unable to access adequate quantities and qualities of food [41,42]. To consider direct effects of financial resources on Cryptosporidium seropositivity, annual income was defined as total combined family income and was divided evenly into categories of less than $25,000, between $25,000 and $45,000, and greater than $45,000. The PIR was calculated within NHANES to provide a relative measure of income relative to poverty thresholds. Annual family income was divided by the poverty guidelines specific to family size and the appropriate year and state. A PIR less than one indicates a household below the poverty threshold, whereas a ratio of one or greater indicates income above that poverty threshold. We reclassified the PIR into even categories of households below the poverty threshold (PIR < 1), households one to three times above the poverty threshold (PIR 1–3), and households with income more than three times above the poverty threshold (PIR 3+). In addition to these primary household socioeconomic variables, we also considered confounding by the source and treatment status of household water, as Cryptosporidium parvum is predominantly transmitted through water-borne pathways and untreated water may also be symptomatic of low socioeconomic status. The source of household water was defined as a private or public water company, a private or public well, or another source. Water treatment was defined as whether or not the following treatment devices were used within a household: pitcher water filter, ceramic or charcoal filter, water softener, aerator, or reverse osmosis. Our analysis also considered the size of a household, defined as the number of rooms per home, to account for larger households potentially stemming from greater wealth. We also considered confounding by demographic covariates of individual race and ethnicity, age, gender, country of birth, and education [29]. Age was defined in one-year intervals, and race and ethnicity were defined by self-report and categorized as non-Hispanic white, non-Hispanic black, Mexican American, and other. Country of birth was categorized as the United States, Mexico, or other. Education was defined by the highest grade of school completed, which we reclassified into a binary variable describing whether or not individuals completed high school. Lastly, associations between socioeconomic status and Cryptosporidium parvum could be confounded by individual health status, as immunocompromised persons are more susceptible to infection [1,43]. Lymphocytes play key roles in the immune defense against Cryptosporidium, and in particular T helper lymphocytes (CD4+ cells) are required for parasite clearance [31,44]. We therefore considered the number of thousand lymphocytes per microliter of blood in our analyses, derived from the total count of leukocytes times the differential count of percent lymphocytes. Although NHANES has directly quantified CD4+ counts, these data were only available for human immunodeficiency virus–positive individuals, which represent a very small subset of the sample tested for Cryptosporidium seropositivity (n = 34 records, < 1% of dataset). All statistical procedures were conducted to account for the design of multistage stratified, cluster-sampled, unequally weighted survey samples such as NHANES using the package survey in R [45,46]. All seroprevalence estimates were weighted to represent the total U. S. population and to account for over-sampling and nonresponse to interviews and physical examinations [47]. We first performed univariate analyses of primary exposures and potential confounders by fitting survey-weighted generalized linear models with the binomial outcome as the serological response to the 17kDA and 27kDA antigens. These logistic regressions were fit through pseudolikelihood and used inverse-probability weighting and design-based standard errors [48,49]. Standard error estimates were calculated using the Taylor series linearization method to account for the complex sampling design [50], and we used a Wald test to test if all coefficients associated with each covariate differed from zero to examine overall variable significance [51,52]. To adjust for the NHANES survey design, the degrees of freedom for Wald tests were calculated as the number of primary sampling units (n = 27) minus the number of strata (n = 13). This method is recommended for retaining power when considering individual-level covariates within a survey-adjusted analysis [49,53]. Due to this limited degrees of freedom imposed by the sampling design (df = 14), we only included covariates with p < 0. 20 from a Satterthwaite-adjusted F statistic as potential confounders to avoid overfitting final multivariable models (shown in Table 1). We next constructed three survey-weighted multiple logistic regression models. These models incorporated household socioeconomic variables separately owing to strong associations between each (S1 Table). For each household socioeconomic variable (food adequacy, annual income, PIR), we included confounder variables within each model to then test associations between impaired environments and Cryptosporidium seropositivity. We again used Wald tests as the omnibus test of covariate significance within each model, and those variables with p ≤ 0. 05 from a Satterthwaite-adjusted F statistic were considered significant. We also tested for differences between groups within each exposure variable after adjusting for the potentially inflated false-discovery rate associated with multiple comparisons using the Benjamini and Hochberg correction and multcomp package in R [54,55]. Crude and adjusted odds ratios and 95% confidence intervals were reported for all variables in the final survey-weighted models. Odds ratios and confidence intervals from the survey-weighted logistic regression models were also calculated using the adjusted degrees of freedom described above. Owing to the cross-sectional nature of NHANES, neither cause–effect relationships nor distinction between recent or current Cryptosporidium infection can be established, and hence odds ratios should be interpreted accordingly. Of the 4359 persons tested for seropositivity to the Cryptosporidium parvum 17kDA and 27kDA antigen groups in our NHANES sample, 925 persons were IgG positive. This corresponds to a weighted seroprevalence for individuals aged 6 and 49 years of 21. 2% (95% CI = 18. 5–23. 9%). Seroprevalence estimates specific to household food adequacy, family income, PIR, water source and treatment, household size, age, gender, race and ethnicity, country of birth, education, and immunocompetence (lymphocyte count) are shown in Table 1. Seropositivity was influenced by all household socioeconomic variables in our univariate analyses (all F2,14 > 2. 7, all p ≤ 0. 10), with living in low-income households, close to the poverty line, and with food inadequacy associated with increased odds of seropositivity. Demographics were also strong correlates of seropositivity, with odds increasing with age (F1,14 = 188. 96, p < 0. 001); with seroprevalence higher among non-Hispanic blacks, Mexican Americans, and other racial and ethnic groups than non-Hispanic whites (F3,14 = 81. 29, p < 0. 001); and with seroprevalence higher among persons born outside the United States compared to those born within the country (F2,14 = 52. 53, p < 0. 001). Odds of seropositivity differed little between men and women (F1,14 = 2. 2, p = 0. 162) and were not influenced by individual lymphocyte count (F1,14 = 0. 86, p = 0. 371). Lastly, we found significant associations between serostatus and home water treatment (F1,14 = 4. 91, p = 0. 045) but not with the source of household water (F2,14 = 1. 42, p = 0. 28); household size was also a weak negative predictor of seroprevalence (F1,14 = 2. 58, p = 0. 132). For multivariable models of household socioeconomic conditions and Cryptosporidium seropositivity, we therefore included individual age, reported race and ethnicity, country of birth, education, gender, household size, and household water treatment status as confounders. While univariate tests found no association between individual immunocompetence and Cryptosporidium seropositivity, we also included the total lymphocyte count in our multivariable models to adjust for immunosuppressed individuals being more susceptible to the parasite [31,44]. Additionally, although the effect of household socioeconomic status on Cryptosporidium seropositivity could depend on individual age and warrant age-stratified analyses, we found no support for an interaction between age and all three socioeconomic conditions (S2 Table, S1 Fig) and thus retained age as a separate fixed effect in all models. Our survey-weighted models showed that household socioeconomic conditions were significant correlates of Cryptosporidium seropositivity after adjusting for individual age, race and ethnicity, country of birth, education, gender, immunocompetence, and household size and water treatment status (Table 2). We first found that household food inadequacy was associated with elevated odds of Cryptosporidium seropositivity (F2,14 = 4. 06, p = 0. 04; Fig 1A). After adjustment, the odds of Cryptosporidium seropositivity were 1. 4 times higher for persons in households with some food inadequacy (OR = 1. 4, p = 0. 04) compared to those in food-adequate households. This trend appeared to be as strong for persons living in households with high food inadequacy, but this elevated odds of seropositivity compared to persons living in food-adequate households was not significant (OR = 1. 31, p = 0. 65). Likewise, we observed a negative trend between household annual income and seropositivity, although the overall association was not significant (F2,14 = 3. 22, p = 0. 07; Fig 1B). Compared to individuals in homes with an annual income of <$25,000 per annum, those in households with an income between $25,000 and $45,000 appeared to have lower risk, although this effect was not significant (OR = 0. 78, p = 0. 21). Yet relative to the lowest income bracket, individuals in households earning greater than $45,000 had 39% lower odds of Cryptosporidium seropositivity (OR = 0. 61, p = 0. 03). After adjustment for confounders, the PIR was the strongest household socioeconomic correlate of Cryptosporidium seropositivity and showed a non-linear relationship with seroprevalence (F2,14 = 8. 46, p < 0. 01). Relative to individuals in families living below the poverty threshold, those in households with an income one-to-three times above the poverty line had slightly elevated odds of seropositivity (OR = 1. 26, p = 0. 13), yet individuals in families with an income more than three times above than the poverty line had a suggested 25% reduction in risk (OR = 0. 75, p = 0. 14); however, neither of these differences was statistically significant. When comparing individuals in families with an annual income one-to-three times above the poverty threshold to those in families earning greater than three times the poverty line, however, this increase in income was protective against Cryptosporidium seropositivity (OR = 0. 60, p < 0. 001; Fig 1C). Our analyses also found individual demographics to be strong correlates of Cryptosporidium seropositivity. In our most conservative survey-weighted logistic model containing annual family income, we found that after adjustment for other covariates, risk of seropositivity increased by 6% per year increments in age (OR = 1. 06, p < 0. 001) and was at least 1. 76 times higher for non-Hispanic blacks and Hispanics compared to non-Hispanic whites (F3,14 = 11. 89, p < 0. 001; Table 2). The odds of Cryptosporidium seropositivity were also at least 2. 27 times higher for persons born outside the United States (F2,14 = 16. 25, p < 0. 001; Fig 1D). After adjusting for other covariates in this conservative model, we found no significant associations between Cryptosporidium seropositivity and household size, water treatment status, individual education, gender, or immunocompetence (Table 2). Our analysis of NHANES 1999–2000 sought to examine if and how social inequality, at both household and individual scales, influenced the odds of Cryptosporidium parvum seropositivity. In low- and middle-income countries, Cryptosporidium infection associates explicitly with poverty and social marginalization [56,57]. However, regions of high socioeconomic status report greater risk in some high-income countries [15]. Such findings have perhaps prompted the claim that cryptosporidiosis is without significant links to social inequality in the United States and resulted in socioeconomic status being absent in discussions of risk factors for this infection [26–28]. Our results suggest that, contrary to these assertions, cryptosporidiosis risk in the United States is highest for individuals living in households with poor food adequacy; in families where income is low and close to the poverty threshold; and for older, non-white, and foreign-born persons. First, we found food inadequacy was a strong household predictor of serological status, with persons in homes reporting some food inadequacy (enough to eat but not always of the foods desired) showing elevated odds of Cryptosporidium seropositivity. This result was robust to adjustment for household size, water treatment, age, race and ethnicity, education, gender, country of origin, and immunocompetence, yet after adjustment there was also no greater risk of seropositivity in homes with poor food adequacy compared to food-adequate households. As the direction of the food inadequacy effect suggested dose response, however, our finding of no difference in risk could be driven by the small sample size of food-inadequate households. More broadly, the significant association between food adequacy and cryptosporidiosis could suggest several pathways through which social inequality influences Cryptosporidium infection. Food inadequacy could first function as a proxy for household socioeconomic status, as food scarcity is typically driven by a broader lack of financial resources [41,42]. An effect of household food inadequacy through this poverty pathway could suggest reduced access to educational or sanitation resources that allow individuals to avoid Cryptosporidium transmission pathways such as contaminated water or contact with livestock reservoir hosts. For example, low levels of education and access to the media were both associated with poor hand washing practices and hence greater parasite exposure risk in Kenya [58]. Our multivariable models containing either annual family income or the household PIR both are suggestive of this poverty pathway, as a high annual income and living far above the poverty threshold were both protective against Cryptosporidium seropositivity. Yet a significant overall association between household financial resources and seroprevalence was only observed for the PIR, and for both variables we did not find a significant dose-response relationship after adjustment. Specifically, our analysis showed no difference in the odds of Cryptosporidium seropositivity between individuals in households living below the poverty line and those in households at any degree above the poverty threshold. Instead, we only found a difference in the odds of seropositivity between individuals in households close to the poverty threshold (1–3 times) and those in households with income greater than three times above the poverty line. These mixed findings for a direct effect of poverty on cryptosporidiosis could therefore suggest more immediate links between household food inadequacy and risk. Food inadequacy could first increase susceptibility to parasite infection through reductions to host nutritional status and therefore immunocompetence [59,60]. Food inadequacy could also push households towards accessing food products from avenues where food safety and sanitation may not be regulated, such as at open street markets [61]. This latter exposure-based pathway between household food inadequacy and cryptosporidiosis risk seems particularly plausible, as small vegetable markets have been identified as a source of Cryptosporidium in low-income regions [62]. Additionally, as the lymphocyte-based measure of immunocompetence had little effect on predicting seropositivity in univariate analyses and was incorporated into all multivariable models, the observed association in our multivariable model between household food inadequacy and cryptosporidiosis could be driven more by food-borne exposure rather than health status. Because household water quality was also included in our multivariable model in the form of water treatment, this potential exposure-based effect of household food inadequacy on Cryptosporidium seropositivity could also be driven more by consumption of contaminated produce rather than contaminated water. Along with this household socioeconomic correlate of Cryptosporidium seropositivity, our analysis also identified older individuals, those of non-white ethnicities and races, and individuals born outside the United States as at greater risk of cryptosporidiosis. Increased odds of seropositivity with age are to be expected, given that the risk of ever being infected by Cryptosporidium would accumulate over time. For the other individual risk factors identified, another analysis of NHANES similarly found non-white Hispanics and blacks to have higher seroprevalence to the Cryptosporidium antigen groups used here [29]. Our finding of higher seropositivity in immigrants across the United States is also consistent with smaller-scale findings that children living along the Texas–Mexico border have higher seroprevalence than non-border children and that immigrants from Mexico have higher cryptosporidiosis prevalence than American-born counterparts in Los Angeles [63,64]. These results again highlight the potential links between social marginalization and Cryptosporidium seropositivity, as non-white, immigrant populations are more prone to experience unemployment, live in economically poor neighborhoods, and have reduced access to resources [65,66]. Such groups may therefore be more likely to lack access to sanitation or educational resources for avoiding parasite exposure and be prone to live in physically impaired or remote environments where access to clean food is limited or where contact with domestic animals is frequent. For example, many low-income immigrants in the United States find their employment in agriculture [67,68], which likely amplifies exposure to Cryptosporidium oocysts through contact with soil and water contained with livestock excrement [69]. The recreational use of open natural water sources in such regions may also elevate the odds of parasite exposure. These results in turn suggest that public health interventions for cryptosporidiosis in the United States could focus on improving awareness of Cryptosporidium exposure routes in such marginalized and resource-poor groups. Further research could also monitor the potential for water and food contamination in regions where high-risk groups reside and test if structural aspects of the physical environment amplify Cryptosporidium seropositivity [21]. Future research on the social epidemiology of cryptosporidiosis in the United States could also utilize multilevel analyses to tease apart the relative contribution of individual and household socioeconomic determinants of seropositivity while accounting for potential neighborhood effects [70,71]. Specifically, we did not find clear evidence of an overall dose–response relationship between individual seropositivity to Cryptosporidium and household financial resources (annual income or PIR), despite suggestive trends. Although this could be due to a small sample size for select groups of the study population, the structuring of NHANES could also have limited identifying a strong income effect. Associated geographic data (e. g. , zip codes, census block) for NHANES are supplied as restricted access, and thus we only included household and individual correlates in these analyses. Yet a stronger effect of income could manifest spatially at the neighborhood scale, where low income level could cluster food-inadequate households and the demographic groups found to be at risk in our multivariable models [72,73]. An additional needed area of work on the social epidemiology of cryptosporidiosis in the United States is the pursuit of longitudinal rather than cross-sectional approaches. Owing to the cross-sectional design of NHANES within the two-year study period tested for Cryptosporidium seropositivity, we were unable to distinguish between IgG-positive individuals with Cryptosporidium infection and those that had only been recently infected and recovered. Longitudinal sampling across a socioeconomic gradient could help tease apart in which cases seropositivity is due to current or recent infections and how this varies by poverty and social marginalization, particularly if investigators measure IgM antibodies alongside IgG antibodies or changes in titers over time [74]. This could be particularly useful to account the positive association between age and seropositivity observed in this study and other analyses of NHANES [29], as IgG titers can increase with age owing to repeated Cryptosporidium infections. Together such studies would allow researchers to assess new Cryptosporidium infections in relation to acute exposures and relate direct infection to impaired social and physical conditions. Likewise, our analysis of NHANES was limited by serological testing with the 17kDA and 27kDA antigen groups, which can identify recent or current infection with Cryptosporidium parvum but cannot distinguish between genotype 1 and genotype 2 of the parasite (C. hominis and C. parvum, respectively; [6]. Rather, genotyping methods on human fecal samples could elucidate whether observed cryptosporidiosis or Cryptosporidium seropositivity is due to infection with the human-origin genotype 1 or zoonotic genotype 2 [7,75]. Differentiation of human versus animal sources of infection in combination with analyses of socioeconomic risk factors could further improve our understanding of how impaired physical and social conditions interact with Cryptosporidium transmission. For example, a location-based study in the UK found that Cryptosporidium hominis cases were more frequent in urban areas of high socioeconomic status, whereas Cryptosporidium parvum cases (zoonotic genotype 2) were more common in rural areas where more oocysts were detected in agricultural soil, presumably from cattle [15]. Within the United States, genotyping methods could be particularly useful to elucidate how the individual- and household-scale correlates of poverty and social marginalization identified in our analyses interact with the abundance of livestock reservoir hosts and hence human infection with zoonotic Cryptosporidium parvum. Specifically, the density of livestock reservoir hosts such as cattle could be related to regional socioeconomic status [76,77], in turn driving greater exposure of marginalized groups to Cryptosporidium parvum genotype 2 in the United States. One analysis of sporadic cryptosporidiosis cases in Scotland accordingly found Cryptosporidium prevalence to be highest in rural regions with high livestock density [78], suggesting a neighborhood poverty influence on seropositivity. Hence ecological and multilevel analyses could test for an interactive influence of livestock density and social marginalization variables identified here (e. g. , food adequacy, immigration status, race and ethnicity) on cryptosporidiosis while accounting for neighborhood income. Our analyses here demonstrate clear associations between social marginalization, poverty, and cryptosporidiosis in the United States, thereby carrying important implications for targeted public health interventions for this infection in resource-poor groups. Alongside direct effects of Cryptosporidium infection on mortality in immunocompromised individuals, morbidity from cryptosporidiosis can range from subtle to severe effects quality of life that can impose serious restrictions on economic wellbeing [3,4]. As cryptosporidiosis is estimated to occur in 750000 persons across the United States annually, these effects can scale up to over $100 million per year in healthcare costs and losses to productivity [2,5]. Therefore, understanding interactions between socioeconomic and environmental conditions in combination with longitudinal and genotyping approaches will be key to guiding prevention and intervention strategies to cryptosporidiosis within the United States. Analyses in this spirit will more broadly help address the complex relationships between ecological factors, social inequality, and infectious disease risk [79,80].

We examined if and how social inequality in the United States influences seropositivity to Cryptosporidium parvum. By using nationwide data on parasite seropositivity, demographics, and household metrics of socioeconomic status provided through the National Health and Nutritional Examination Survey, we quantified how measures of social inequality affect the odds of parasite infection. After adjusting for the complex sampling design of NHANES and potential confounding by individual demographics and household conditions, we found household food inadequacy was associated with greater parasite seropositivity. Additionally, we found that individuals of non-white races and ethnicities and those born outside of the United States have significantly greater odds of seropositivity than white, domestic-born counterparts. Furthermore, our study suggests evidence for direct effects of family wealth on cryptosporidiosis risk, in that persons in low-income households have elevated odds of parasite seropositivity relative to those in high-income families. These results refute the claim that cryptosporidiosis in the United States in independent of poverty and social marginalization and carry implications for targeted public health interventions for this parasitic infection in resource-poor groups.

Abstract Introduction Methods Results Discussion

2015

Household Socioeconomic and Demographic Correlates of Cryptosporidium Seropositivity in the United States

The World Health Organization (WHO) in collaboration with partners is developing a toolkit of resources to guide lymphatic filariasis (LF) morbidity management and disability prevention (MMDP) implementation and evaluation. Direct health facility inspection is the preferred method for documenting the readiness of a country programme to provide quality lymphoedema management services, one of the three MMDP criteria used to demonstrate the elimination of LF as a public health problem. As component of tool development, a Delphi consultation was implemented to gain consensus on six proposed domains and fourteen proposed tracer indicators to measure national programme readiness to provide quality health facility-based lymphoedema management services. A seven-point Likert-type scale was used to rank the importance of proposed domains and tracer indicators. Consensus for inclusion of the indicator was defined a priori as 70% or more of respondents ranking the proposed indicator in the top three tiers (5–7). Purposive sampling was used to select 43 representative experts including country representatives, programme implementers, and technical experts. A 55. 8% response rate (n = 24) was achieved for the survey. Analysis of the responses demonstrated that consensus for inclusion had been reached for all proposed domains including trained staff (mean = 6. 9, standard deviation (SD) = 0. 34), case management and education materials (mean = 6. 1, SD = 0. 65), water infrastructure (mean = 6. 3, SD = 0. 81), medicines and commodities (mean = 6. 3, SD = 0. 69), patient tracking system (mean = 6. 3, SD = 0. 85), and staff knowledge (mean = 6. 5, SD = 0. 66). The Delphi consultation provided an efficient and structured method for gaining consensus among lymphatic filariasis experts around key lymphoedema management quality indicators. The results from this analysis were used to refine the indicators included within the direct inspection protocol tool to ensure its ability to assess health facility readiness to provide quality lymphoedema management services. Lymphatic filariasis (LF) is a parasitic infection caused by filarial nematodes that are transmitted by mosquitoes. Chronic infection with LF can lead to clinical manifestations such as lymphoedema and hydrocele that have significant impacts on the mobility and quality of life of affected individuals. Further, individuals with lymphoedema are prone to painful and debilitating secondary bacterial infections, known as acute attacks, that are associated with diminished quality of life and progression of disease [1–4]. Approximately 947 million people are at risk for LF in more than 73 countries worldwide [5]. In an effort to reduce suffering, LF has been targeted for elimination as a public health problem by 2020 following the World Health Assembly Resolution 50. 29 [6]. The Global Programme to Eliminate LF (GPELF) has established a two-pillar strategy for elimination: (1) interruption of transmission through mass drug administration (MDA) and (2) alleviating the suffering of individuals affected by the chronic manifestations of LF infection through the provision of morbidity management and disability prevention (MMDP) services. To meet the criteria established by the Word Health Organization (WHO) for the MMDP pillar for elimination, national LF elimination programmes are asked to provide data on the number of patients with lymphoedema (or elephantiasis) and hydrocele, the number of health facilities designated to provide care, and the readiness and quality of the care provided [7]. Quality of care assessments can be used to understand what resources are needed to improve services as well as advocate for other sectors or departments within the Ministry of Health to supplement these services. While the implementation and evaluation activities for MDA have been clearly defined, there is a need for clearer guidance on the provision and assessment of MMDP services for national LF elimination programmes. In order to meet this need, WHO is developing a toolkit to guide LF MMDP implementation and evaluation. One component of the toolkit is a direct inspection protocol, a tool designed to measure readiness to provide quality health facility-based lymphoedema management services in accordance with WHO recommendations. Here, we summarize an expert consultation following the Delphi methodology to reach consensus on domains and indicators that should be used to evaluate health-facility readiness to provide quality lymphoedema management services [10]. The results of the Delphi consultation informed the refinement of the direct inspection protocol tool and will assist national LF elimination programmes demonstrating that they have achieved the requirements for validation of elimination of LF as a public health problem. The response rate for the online survey component of the Delphi consultation was 55. 8% (n = 24). The individuals who participated in the survey represented a range of professions (Table 2). A third of participants (n = 8,33. 3%) had more than 25 years of experience in their respective field. Participants’ responses to the domains and tracer indicators are presented in Tables 3 and 4. In the first round, there was consensus that all six domains of readiness and quality of MMDP services in health care facilities were important. None of the respondents ranked the domains in the bottom three categories. The strongest agreement was observed for trained staff and staff knowledge with 87. 5% and 62. 5% of respondents respectively indicating they felt that these domains were extremely important. Furthermore, consensus was reached for all thirteen evaluated tracer indicators, though a wider range of ranking was observed. The strongest consensus was observed for tracer indicators related to medicines and commodities—primarily the availability of medicines—as well as the tracer indicators for case management and education materials. Based on the sensitivity analysis using stricter criteria for consensus, all domains met consensus criteria under stricter conditions with between 79. 2% and 100% of respondents ranking the domains in the top two categories. All of the tracer indicators met the stricter consensus criteria except for staff training in the last two years (66. 7%) and at least one patient with lymphoedema recorded in the reporting system in the last 12 months (50. 0%). Common themes from the qualitative feedback included: the need for a more robust definition of training and refresher training, the importance of clinic staff being able to identify more than one sign, symptom, and management strategy, and a need for a more clearly stated definition for a patient tracking system. The World Health Assembly resolution to eliminate LF was built on a desire to mitigate the harm caused by LF, both by preventing future infection as well as by alleviating the suffering experienced by individuals who present with clinical manifestation as a result of infection [6]. Since the clinical sequelae of LF develop many years after infection and are chronic, national LF programmes must work closely within the health care system to ensure that MMDP services are well integrated, available, and sustainable. While the components of a minimum package of care for lymphoedema and hydrocele patients has been clearly defined [9], there is a need for standardization in the evaluating and reporting of the availability and quality of MMDP services in the provision of the minimum package of care at healthcare facilities. This Delphi consultation allowed input from multiple stakeholders and improved the practicality and acceptability of a standard survey for direct inspection of health facilities to assess readiness and quality of MMDP. Based on previous literature, framing questions using a Likert-type scale for Delphi consultations facilitates straightforward statistical analysis to assess for consensus across respondents [13,14,16]. A strength of the Delphi methodology is that it allows stakeholders from a variety of perspectives to offer their expert opinion on the key elements that need to be included in an evaluation of quality services. Using a Delphi consultation, we were able to gather input for indicator development from a range of stakeholders. Our hope is that this approach will lead to broader stakeholder support and acceptability. Based on the diversity of participants, we feel that the consensus achieved reflects the priorities of global partners working towards elimination of LF as a public health problem. However, due to limitations in accessibility we were unable to include the perspectives of two important stakeholders: health facility level staff and affected patients. Steps were taken to include feedback from staff and patients during the pilot testing of the direct inspectional protocol as discussed later. Though consensus was reached for the domains and indicators, through open-ended feedback experts proposed more stringent criteria to strengthen the indicators to measure the readiness of health facilities to provide quality lymphoedema management care. Citing the critical need for appropriate identification of lymphoedema in patients, experts suggested that a greater emphasis should be placed on the evaluation of staff knowledge. To address this, questions assessing staff knowledge were modified to require two correct responses instead of one for each tracer indicator. No significant changes were made to the components of the remaining domains and indicators. The fourteen tracer indicators, refined as part of the Delphi consultation, are intended to comprise the questionnaire component of a health facility inspection tool, allowing LF programmes to evaluate the readiness of health facilities to provide quality lymphoedema management services as a component of MMDP programmes. The inspection comprises a facility walkthrough and interview with key health personnel at randomly selected health facilities providing lymphoedema management services. The surveyor evaluates if the facility meets the criteria for each indicator, through direct observation where relevant (e. g. the presence of medicines and commodities). The results of the questionnaire generate a health facility score, by which the programme can evaluate highly performing and poorly performing health facilities. Programmes can also evaluate indicator scores across facilities to evaluate systematic strengths and weakness, in order to implement informed process-improvement steps to strengthen the quality of lymphoedema management services. In addition, the standardization offered through these tracer indicators provides programmes with the ability to compare lymphoedema management services across settings. We recognize that while the primary focus of this Delphi consultation was assessing lymphoedema management services, hydrocele care is also important in LF endemic countries. We feel confident that we could replicate similar procedures to develop indicators to assess health facility readiness to provide quality hydrocele care. Due to the unique components of care required for hydrocele patients, expert consultation with urologists with expertise in hydrocele management will be conducted to determine the domains and tracer indicators for evaluating readiness and quality of health-facility based hydrocele care. A standardized protocol, the WHO Surgical Assessment Tool (SAT), provides information on general surgical capacity including the availability of hydrocelectomy and is under revision. Efforts are under consideration to include a module specifically evaluating the readiness and quality of hydrocele care. By demonstrating that the global community is in agreement about the components of lymphoedema management that healthcare facilities must be prepared to provide their patients, there is evidence to support the inclusion of the direct inspection protocol tool in the MMDP toolkit for countries implementing the MMDP component of the GPELF strategy. In an effort to assess the measurability of these indicators in healthcare facilities, the quality domains and tracer indicators identified were included in a pilot direct inspection tool to assess the readiness of healthcare facilities to provide quality lymphoedema management services in Mali, Vietnam, and Haiti [17]. The pilot demonstrated that these indicators were feasible to implement and yielded useful information about the quality of services; however minor changes were incorporated to the survey based on the results of the pilot. The direct inspection protocol is intended to supplement SARA assessments, provide more detailed information on lymphoedema management services, and to ensure programme managers have the information needed to plan for services to meet the needs of individuals with LF.

Prior to this assessment, there was a need for clearly defined, measurable indicators for global lymphoedema management programmes to use to evaluate their lymphoedema morbidity management and disability prevention services. The results presented in this report outline a framework for indicators to assess healthcare facility-based readiness to provide quality lymphoedema management services. We describe our use of the Delphi methodology to obtain consensus on programme evaluation metrics as a model for elimination or control metrics and targets. The quality indicators developed from this Delphi consultation will be used to improve the direct inspection protocol tool that can be used to assess health facility readiness to provide quality lymphoedema management services LF endemic countries.

Abstract Introduction Results Discussion

medicine and health sciences hydrocele pathology and laboratory medicine education sociology tropical diseases social sciences parasitic diseases health care health services administration and management filariasis signs and symptoms global health neglected tropical diseases patients lymphatic filariasis medical education public and occupational health helminth infections diagnostic medicine medical humanities health care facilities

2018

A Delphi consultation to assess indicators of readiness to provide quality health facility-based lymphoedema management services

Hepatitis C virus (HCV) naturally infects only humans and chimpanzees. The determinants responsible for this narrow species tropism are not well defined. Virus cell entry involves human scavenger receptor class B type I (SR-BI), CD81, claudin-1 and occludin. Among these, at least CD81 and occludin are utilized in a highly species-specific fashion, thus contributing to the narrow host range of HCV. We adapted HCV to mouse CD81 and identified three envelope glycoprotein mutations which together enhance infection of cells with mouse or other rodent receptors approximately 100-fold. These mutations enhanced interaction with human CD81 and increased exposure of the binding site for CD81 on the surface of virus particles. These changes were accompanied by augmented susceptibility of adapted HCV to neutralization by E2-specific antibodies indicative of major conformational changes of virus-resident E1/E2-complexes. Neutralization with CD81, SR-BI- and claudin-1-specific antibodies and knock down of occludin expression by siRNAs indicate that the adapted virus remains dependent on these host factors but apparently utilizes CD81, SR-BI and occludin with increased efficiency. Importantly, adapted E1/E2 complexes mediate HCV cell entry into mouse cells in the absence of human entry factors. These results further our knowledge of HCV receptor interactions and indicate that three glycoprotein mutations are sufficient to overcome the species-specific restriction of HCV cell entry into mouse cells. Moreover, these findings should contribute to the development of an immunocompetent small animal model fully permissive to HCV. HCV is an enveloped virus with a positive sense single stranded RNA genome, belonging to the family of Flaviviridae [1]. Based on sequence comparison patient isolates are classified into seven genotypes which differ from each other by ca. 31–33% at the nucleotide level [2], [3]. Chronic hepatitis C is associated with severe liver disease including hepatitis, liver cirrhosis and hepatocellular carcinoma and it is the most frequent indication for liver transplantation [4]. Presently, neither therapeutic nor preventive vaccines are available and the only current therapy, a combination of pegylated interferon-α and ribavirin, is limited by resistance, toxicity and high costs [5]. Development of HCV-specific antivirals and vaccines has been impeded by lack of convenient animal models. Besides humans the only host for HCV is the chimpanzee and an immunocompetent small animal model is still lacking. The restricted tropism of HCV likely reflects specific host factor requirements for entry, RNA-replication, assembly and release of virions. Although mouse cells have been shown to sustain HCV RNA replication [6], the efficiency is low and release of progeny virus particles was not observed [7]. Moreover, cell entry of HCV into rodent cells is possible, but requires overexpression of essential human entry factors [8]. Therefore, propagation of HCV in mouse cells is likely restricted at multiple steps of the viral replication cycle. HCV cell entry is a complex process involving a number of host factors. Initial adsorption of viral particles consisting of core protein, envelope proteins 1 and 2 (E1, E2) and associated with lipoproteins [9]–[11], may be facilitated by attachment factors like glycosaminoglycans [12], the LDL receptor [13], [14] and lectins [15]–[17]. Beyond these, four essential HCV entry factors have been identified: the scavenger receptor class B type 1 (SR-BI) the tetraspanin CD81 and the tight junction proteins claudin-1 (CLDN1) and occludin (OCLN) [8], [18]–[22]. Usage of these four factors mediates clathrin-dependent endocytosis [23] of virus particles and ultimately permits viral membrane fusion in a low pH-triggered fashion [24]. Notably, ectopic expression of human SR-BI, CD81, CLDN1 and OCLN rendered all cell lines tested permissive for HCV suggesting that additional host factors necessary for host cell entry are broadly expressed and conserved among different mammalian species [8]. Meanwhile, accumulating evidence indicates that a species-specific interplay of HCV with these four entry factors restricts HCV entry into non-human cells and thus contributes to HCV' s narrow species tropism [25]–[27]. Particularly, inefficient usage of non-human CD81 and OCLN by HCV seems to limit viral entry into non-human cells [8]. However, more recently evidence was provided that mouse SR-BI and at least for genotype 2a also mouse CLDN1 supported HCV infection with slightly lower efficiency compared to their human homologs [26], [27], implying that incompatibility of these factors may also limit HCV entry into mouse cells. The interaction between the viral glycoprotein E2 and CD81 has been analyzed using truncated, soluble E2 and retroviral pseudoparticles bearing the HCV glycoproteins E1 and E2 (HCVpp). These studies highlight that defined residues within the large extracellular loop (LEL) of CD81, a protein domain that is highly variable between CD81 isoforms from different species, are crucial for interaction with HCV E2 [28], [29]. On the other hand, the viral binding site for CD81 seems to be a conformational epitope within E2: Based on reverse genetics and antibody competition experiments with E2-specific antibodies, multiple discontinuous epitopes have been implicated in CD81 binding [30]–[34]. To characterize viral determinants for CD81 usage and to assess if improved utilization of mouse CD81 is sufficient to permit entry of HCV into mouse cells, we adapted HCV to mouse CD81, and analyzed the consequences for virus receptor interaction and tropism. For adaptation of HCV to mouse CD81, we first created Huh7-Lunet cells [35] that are highly permissive for HCV RNA replication and virus production but that are essentially not infectable due to limiting expression of endogenous human CD81. This cell clone designated Lunet N [36] was created by fluorescence activated cell sorting for cells expressing very little CD81. The resulting cell population was subcloned and individual clones were analyzed for CD81 expression and permissiveness for HCV RNA replication. Ultimately, a subclone with minimal residual CD81 expression supporting high level HCV RNA replication was selected. In the absence of ectopically expressed CD81, these cells designated Lunet N cells are essentially refractory to HCV infection (Figure S1). Susceptibility of these cells to HCV infection was restored by ectopic expression of human or mouse CD81. However, in agreement with the results of Flint et al. [25], mouse CD81 supported infection approximately 100-fold less efficiently than human CD81 (Figure S1). Next, we transfected Lunet cells expressing mouse CD81 (Lunet N mCD81) with the HCV genotype 2a chimeric virus Jc1 that grows to high virus titers [37]. After five cell passages, naive Lunet N mCD81 cells were added to the culture and nine additional cell passages later, virus had spread to nearly all cells of the co-culture (data not shown). At this time point, four consecutive passages of cell-free culture fluid to naïve Lunet N mCD81 cells were conducted to select for virus variants that efficiently use mouse CD81 (Figure 1A). At individual passage steps of the procedure, we sampled the infectivity of the virus population present on cells expressing either human, mouse or no CD81 to monitor the progress of adaptation. In the course of this experiment, viruses were selected that remained dependent on CD81 but which were able to use mouse and human CD81 with comparable efficiency (Figure S2). HCV RNA derived from the final supernatant passage was isolated and sequenced. In the region of the genome encoding the viral structural proteins Core, E1, E2 as well as p7, NS2 and NS3, we identified four mutations that were conserved at least among 4 out of the 6 sequenced clones (Figure 1B). Three of these mutations were located in the viral glycoproteins; one in E1 (L216F) and two in E2 (V388G and M405T). Both mutations located in E2 reside within the hypervariable region 1 (HVR1), a stretch of 27 amino acids at the N-terminus of E2 which has been implicated to interact with SR-BI [38]. The fourth mutation was detected in the first transmembrane domain of the p7 protein (N767D). To analyze the influence of these amino acid changes on RNA-replication, virus production and infection of cells expressing human or mouse CD81, we transferred them individually or in combinations into the backbone of Jc1. These constructs were then transfected into naïve Huh-7. 5 cells, and intracellular as well as extracellular core protein levels were determined to assess replication and virus release (Figure 1C–1D). Finally, cell-free culture fluids were harvested, normalized for equal quantity of core and then used to infect Lunet N cells expressing either human or mouse CD81 (Figure 1E). None of the mutations individually or in combination markedly affected RNA-replication and virus release, as is evident from comparable levels of intracellular core 48 h post transfection of these constructs and from a similar efficiency of core protein release at this time point (Figure 1C–1D). The mutation in p7 did not affect virus entry into cells with mouse or human CD81 and thus CD81 usage. In contrast, all three mutations in the glycoproteins moderately increased the efficiency of the virus to infect cells carrying mouse CD81 compared to the parental Jc1 chimera (Figure 1E) thus suggesting that these changes enhance the ability of the virus to use mouse CD81. Interestingly, L216F and V388G at the same time reduced efficiency of infection via the human CD81 counterpart. In contrast, the M405T mutation stimulated infection through mouse CD81 without affecting entry via human CD81. Strikingly, the combination of all three glycoprotein-resident mutations as well as all four mutations together, increased infection mediated by mouse CD81 to the levels sustained by human CD81 (Figure 1E). Notably, infectivity of Jc1 carrying a combination of the L216F, V388G and M405T mutations (designated Jc1/mCD81) on Lunet N mCD81 cells was efficiently neutralized by antibodies specific to mouse CD81 but not by antibodies directed against human CD81, thus indicating that the virus indeed uses the mouse entry factor for infection (data not shown). To address if these mutations specifically permit infection via mouse CD81 or if they would also increase HCV infection through CD81 from other rodent species, we cloned rat and hamster CD81 and stably transduced Lunet N cells with these orthologs. These cells were challenged with equal quantities of Jc1 and Jc1/mCD81 particles. Similar to the mouse ortholog, infection through rat and hamster CD81 was approximately 100-fold more efficient with Jc1/mCD81 compared to parental Jc1 (Figure 2). Thus, these three mutations within HCV E1 and E2 enhance entry efficiency of HCV through mouse, rat and hamster CD81 to a comparable degree suggesting that the adaptation was not specific to mouse CD81 only. It is worth mentioning here that both parental Jc1 and Jc1/mCD81 infected Lunet N cells, albeit with very low efficiency (Figure 2). Since this low level infectivity was susceptible to neutralization by CD81-specific antibodies it is likely due to very low residual levels of human CD81 in these cells rather than due to a CD81-independent route of virus entry (data not shown). It is conceivable that improved usage of mouse CD81 by HCV may be linked with augmentation of the affinity between the adapted viral glycoproteins and CD81. Therefore, we assessed whether these mutations altered the interaction between the viral E1/E2 glycoprotein complexes and CD81. To this end, we utilized a recombinant fusion protein between GST and the human CD81 large extracellular loop (LEL) which encompasses the viral binding site and has previously been used to probe this interaction [28], [29]. To distinguish possible changes of virus receptor usage of mouse CD81 adapted glycoproteins from altered receptor usage caused by cell culture adaptation of HCV after long term passage, we created an additional virus mutant in the context of the Jc1 chimera which harbours a single G451R mutation within the E2 protein. This mutation was originally identified in the JFH1 isolate upon long term passage [39] and the G451 position is conserved between the JFH1 isolate and the J6CF strain of HCV as present in the Jc1 chimera. More recently, the G451R mutation was shown to increase the interaction of JFH1 E2 with CD81 and to reduce virus dependence on SR-BI [40]. Unlike Jc1/mCD81, Jc1/G451R still displays a strong preference for infection via human rather than mouse CD81 (Figure S3). Interestingly, when we performed infections in the presence of increasing doses of the GST hCD81 LEL, the fusion protein only poorly neutralized infection by parental Jc1 (Figure 3A). These data indicate that soluble human CD81 does not efficiently compete with cell surface resident CD81 during infection of parental Jc1. In contrast preincubation of Jc1 variants with 10 µg/ml hCD81 LEL reduced infection by these viruses more than 10-fold, suggesting that both adapted variants are more susceptible to inhibition by soluble human CD81 protein (Figure 3A). Similarly, both viruses with adapted glycoproteins were also precipitated more efficiently by hCD81LEL as is evident from 2-3-fold higher amounts of HCV RNA associated with hCD81LEL coated beads incubated with Jc1/mCD81 or Jc1/G451R compared to beads incubated with an equal amount of wildtype Jc1 particles (Figure 3B). Thus, both the cell culture adaptive mutation G451R and the triple mutation identified by us increased interaction with soluble human CD81 possibly by enhancing the exposure of the viral CD81 binding site. To assess if increased binding of Jc1/mCD81 to human CD81 correlates with altered receptor usage during virus entry, we used two complementary approaches: First, we analyzed susceptibility of Jc1/mCD81 infection to neutralization with anti-human CD81 antibodies (Figure 4A). Second, we quantified the threshold level of human CD81 needed for productive infection of Huh7-Lunet cells (Figure 4B and S4). Interestingly, selectively Jc1/mCD81 was more resistant to neutralization by anti-human CD81 antibodies compared to the parental Jc1 virus and compared to Jc1/G451R (Figure 4A). These results imply that Jc1/mCD81 competes more efficiently with CD81-antibodies for available CD81 binding sites on the cell surface. To determine the minimal CD81 threshold surface expression needed for infection by these viruses, parental Huh7-Lunet cells displaying low to moderate CD81 surface expression were mixed with Lunet N hCD81 cells expressing high levels of CD81 to create a mixed cell population containing cells with highly divergent CD81 abundance at the cell surface. Subsequently these cells were challenged with Venus-GFP-expressing HCV reporter viruses normalized for equal quantity of core protein [41]. Using this approach the relation between CD81 cell surface level and infection efficiency can be quantified using a dual colour FACS analysis [41]. Jc1/mCD81 infected up to 71% of cells with high CD81 expression (ca. 1,000 MFI) compared to Jc1 and Jc1/G451R infecting only 38. 9% and 34% of cells with this CD81 abundance (Figure 4B and S4). However, this increased efficiency of Jc1/mCD81 was not accompanied by a reduction of the minimal threshold density of CD81 required for entry since the turning point of the curves which describe the relation between CD81 receptor density and % infected cells was not significantly different between the three viruses compared (Jc1 MFI 55. 70; Jc1/mCD81 MFI 57. 74; Jc1/G451R MFI 54. 90; p-value >0. 5). Therefore, we conclude that all three viruses share a similar minimal human CD81 receptor density to permit infection. However, selectively Jc1/mCD81 has attained a higher efficiency of infection in cells displaying CD81 levels in excess of the minimal CD81 level. Together these results argue that Jc1/mCD81 uses human CD81 with elevated efficiency, suggesting either that the avidity (the combined strength of multiple interactions) of the virus for human CD81 has increased and/or that the threshold avidity needed for proper CD81 usage was lowered. Next, we analyzed if the adaptation of HCV to mouse CD81 has influenced the utilization of human SR-BI, CLDN1 or OCLN. To this end we employed SR-BI- and CLDN1-specific and control antibodies for receptor competition assays comparing parental Jc1 with Jc1/mCD81 and Jc1/G451R (Figure 5A–5C). Since OCLN-specific neutralizing antibodies are not available, we assessed the dependence of these viruses on OCLN surface expression using RNA interference (Figure 5D). Our data indicate that all viruses compared by us need CLDN1 for cell entry, since receptor-specific antibodies inhibited infection by each virus in a dose-dependent fashion (Figure 5A). Dependence of all three viruses on CLDN1 was also confirmed by inoculation of HuH6 cells which express only minimal amounts of CLDN1, rendering them refractory to Jc1 infection [26] and also to both adapted viruses (data not shown). Notably, virus neutralization through SR-BI-specific antibodies was clearly less efficient for Jc1/G451R and Jc1/mCD81 compared to parental Jc1 particles reducing infectivity to approximately 80% only compared to 5%, respectively (Figure 5B). These data confirm the findings of Grove et al. , who observed a similar phenotype for JFH1/G451R [40] and indicate that both virus mutants are less dependent on SR-BI for cell entry. Finally, knock down of OCLN reduced efficiency of cell entry for all three viruses, although the reduction was less pronounced for Jc1/mCD81 (p<0. 001) (Figure 5D). In conclusion these results suggest that adaptation to mouse CD81 has not only lowered dependence on human CD81 but also on SR-BI and OCLN expression. Since structural information of the HCV E1/E2 complex is not available, we performed a series of virus neutralization experiments using well defined E2-specific monoclonal antibodies with a conserved linear epitope or conformational binding sites in order to document structural changes of the E1/E2 complex that may be responsible for the phenotypic changes of adapted HCV. More specifically, we used AP33 a mouse monoclonal antibody recognizing a linear epitope within E2 [42], and three human monoclonal antibodies interacting with three distinct conformational domains on the E1/E2 complex designated A, B and C [43], [44]. Antibodies interacting with domain A like CBH4D are non-neutralizing, whereas antibodies to domains B or C (e. g. CBH5 or CBH23, respectively) inhibit E2 binding to CD81 and neutralize infection of HCVpp and HCVcc from genotypes 1 and 2 [45], [46]. As expected, even high doses of the non-neutralizing antibody CBH4D and the R04 antibody which recognizes a protein of the cytomegalovirus, did not interfere with infection of all three viruses tested (Figure 6). AP33, CBH5 and CBH23 did not markedly neutralize Jc1. In contrast, these antibodies neutralized Jc1/mCD81 and Jc1/G451R infection in a dose-dependent fashion, indicating that both adapted viruses are much more prone to neutralization by E2-specific antibodies (Figure 6). Notably, domain C antibody (CBH23) neutralized Jc1/mCD81 and Jc1/G451R to a different degree with Jc1/mCD81 being more susceptible to the antibody (IC90 of Jc1/mCD81 and Jc1/G451R: 12. 9 µg/ml and 157. 3 µg/ml, respectively, p<0. 001). In contrast, both viruses were similarly neutralized with the AP33 antibody (IC90 of Jc1/mCD81 and Jc1/G451R: 0. 27 and 0. 07 µg/ml, respectively, p = 0. 28) and with the domain B antibody (CBH5; IC90 of Jc1/mCD81 and Jc1/G451R: 0. 18 and 0. 07 µg/ml, respectively, p = 0. 18). The differential susceptibility to neutralizing antibodies was not due to overtly different buoyant density of these viruses (Figure S5) and is indicative of distinct structures of the respective viral glycoprotein complexes. These in turn are likely responsible for the different phenotypes regarding host entry factor usage and tropism. After cell surface binding and virus internalization into clathrin-coated pits, HCV infection is mediated by conformational changes of the viral glycoprotein complex E1/E2 which are triggered by low pH [24]. Notably, the responsiveness of HCV to low pH-triggered fusion seems to change over time: First, unlike for alphaviruses which also enter host cells in a low pH-dependent fashion and which are inactivated by low pH-treatment in solution, HCV resists such treatment and retains infectivity [24]. Second, exposing HCV particles directly after binding to permissive cells at 4°C, a temperature which allows binding of the viral particles, but no further steps of the entry process, to a low pH wash does not efficiently initiate infection. In contrast, chasing these particles for 1 h at 37°C increases their sensitivity to low pH which may indicate that HCV requires certain additional stimuli like for instance receptor interactions to prepare the virus for low pH-triggered fusion and infection [24]. Given these findings, we hypothesized that adaptation to mouse CD81 may have selected for altered E1/E2 complexes that are more responsive to conformational changes triggered by receptor interactions and/or low pH, thus permitting entry through use of an entry factor of non-human origin with poor binding affinity for HCV. To test this hypothesis we compared the responsiveness of Jc1, Jc1/mCD81 and Jc1/G451R to low pH-treatment directly after cell surface binding and 1 h later following the protocol described by Tscherne et al. [24]. Importantly, infection via acidified endosomes is blocked throughout these experiments by treatment of target cells with concanamycin A (ConA). Since this drug inhibits endosomal V-ATPases, acidification of endosomes is inhibited thus ablating infection through the natural endosomal route [24]. In agreement with the observation by Tscherne et al. , treatment with pH 5 directly after virus binding at 4°C stimulated Jc1 infection with very low efficiency, reaching only ca. 4% of the infection attained when acidification of endosomes was not blocked by addition of ConA and viruses were treated with pH 7 only (Figure 7A). Infectivity of both Jc1/mCD81 and Jc1/G451R was comparable to Jc1 when washed with pH 7 ranging between 1 and 3% of control infections in the absence of ConA. However, both viruses strongly responded to the pH 5 wash already directly after virus binding reaching up to 30% and 20% of the untreated control. After incubation of cell-bound HCV for 1 h at 37°C also Jc1 particles could be stimulated to infect ConA-treated cells (Figure 7B). Therefore, unlike wildtype HCV both adapted viruses can be triggered to infect ConA-treated cells already directly after cell surface binding suggesting that the requirements for induction of infection by low pH have been modified. Finally, we analyzed if the mutations selected by adaptation of Jc1 to mouse CD81 permit HCV infection of mouse cells in the absence of human factors. For these experiments we chose NIH3T3 cells which become susceptible to HCVpp by ectopic expression of human SR-BI, CD81, CLDN1 and OCLN [8]. While these cells express endogenous mouse CD81 and mouse SR-BI, we were unable to detect mouse CLDN1 and OCLN (Figure S6). Given these circumstances, we created NIH3T3-derived cell lines which ectopically express all four human or mouse HCV entry factors, designated NIH3T3-4xH or NIH3T3-4xM. Since mouse cells support very low HCV RNA replication, we used murine leukemia virus-based HCV pseudoparticles (HCVpp) transducing a Venus-GFP reporter gene to quantify HCV cell entry into these cells. As expected, HCVpp with wildtype or mouse CD81 adapted glycoproteins were unable to infect control NIH3T3 cells but efficiently entered Lunet N hCD81 cells (Figures 8A and S7A). Ectopic expression of human HCV entry factors in NIH3T3 cells rendered these cells permissive to both wildtype HCV and mouse CD81 adapted HCVpp, reaching infectivities as high as ca. 20% and 60% of the infectivity in Lunet N hCD81 cells, respectively. In contrast, NIH3T3 cells expressing mouse entry factors (NIH3T3 4xM) were only efficiently infected with HCVpp carrying mouse CD81 adapted glycoproteins. Importantly, HCVpp coated with mouse CD81 adapted glycoproteins infected NIH3T3 cells with human or mouse entry factors with comparable efficiency and infectivity of HCVpp was efficiently neutralized by anti-human or anti-mouse CD81 antibodies in the respective cell lines (Figures 8B and S7B). In agreement with our data described above, HCVpp coated with wildtype J6 glycoproteins were more prone to neutralization by anti-CD81 antibodies compared to J6/mCD81-coated HCVpp. In conclusion these results indicate that three adaptive mutations within HCV E1/E2 proteins are sufficient to permit efficient HCV entry into mouse cells in the absence of human factors. We were unable to detect productive entry of HCVpp with wildtype or mouse CD81-adapted glycoproteins into primary mouse hepatocytes. We cannot exclude that dominant restriction factors expressed in these cells may limit cell entry even of mouse adapted HCVpp. However, we observed very low expression of mouse CLDN1 and undetectable quantities of mouse OCLN in these cells (Fig. S8) which very likely precludes efficient HCV cell entry. In this study we took advantage of a human liver cell line expressing mouse in place of human CD81 to select for a population of HCV that utilizes this entry factor with high efficiency for infection of host cells. Employing reverse genetics we showed that the ability to utilize mouse CD81 was conferred by a combination of three mutations within the viral surface proteins E1 and E2. In principle, change of receptor usage mediated by these mutations may reflect an adaptation of the viral CD81 binding site to accommodate the mouse entry factor and to enhance binding affinity. However, neither these residues nor the domains in which they are located in have previously been implicated in the virus-CD81 interaction. Two of the identified mutations (V388G and M405T) are part of the so called hypervariable region 1 (HVR1). This highly variable domain at the N-terminus of E2 has been implicated in the interaction with SR-BI [47], [48] and is dispensable for binding of soluble E2 to CD81 [31]. In fact, deletion of HVR1 enhances binding of soluble E2 to soluble or cell surface-bound CD81 [31]. In addition viruses lacking HVR1 are only moderately impaired in entry and remain susceptible to neutralization by CD81-specific antibodies [48], thus strongly suggesting that the HVR1 is dispensable for the interaction with CD81. Consequently, it is unlikely that the HVR1 is part of the viral CD81 binding site and that the adaptive mutation within this domain directly enhance the binding to mouse CD81. Alternatively, adaptive changes may act indirectly by eliminating obstructions that otherwise sterically prevent access of mouse CD81 to the viral CD81 binding site. In the absence of structural information of the viral E1/E2 complex we probed the influence of these mutations on the interaction with CD81 using soluble human CD81 and CD81-specific antibodies. To phenotypically distinguish adaptation to mouse CD81 from cell culture adaptation we compared the mouse CD81-adapted virus to a recently described cell culture-adaptive mutation within E2 (G451R) that is known to modulate SR-BI and CD81 receptor usage and virus neutralization [49]. Interestingly, both mutant viruses interacted more efficiently with soluble human CD81-LEL and they were both more susceptible to neutralization by this protein (Figure 3). Previously, Flint et al. were unable to detect binding of HCV E2 protein to soluble mouse CD81 [25]. Our attempts to detect changes of HCV binding to soluble mouse CD81 did not yield conclusive results precluding a definite answer if the affinity to mouse CD81 was changed by the adaptation (data not shown). Therefore, it remains to be determined if parental or adapted HCV binds mouse CD81. Nevertheless, these results indicate that both virus mutants have a more accessible binding site for human CD81 which likely facilitates interaction and neutralization. Such a more open conformation may also assist interaction with mouse CD81. Notably, also the mouse CD81 adapted virus remains dependent on CD81 since it does not infect cells devoid of CD81. Therefore, either the adapted HCV indeed does interact with mouse CD81 or it utilizes an essential function of mouse CD81 as entry factor even in the absence of direct binding. As CD81 is a member of the protein family of tetraspanins, which assemble multi-protein complexes within membranes, such a function could for instance be the coordination of additional HCV entry factors into a functional receptor complex. In line with this notion, a direct interaction between CD81 and CLDN1 has been reported [50], [51]. Interestingly, the mouse CD81 adapted virus was more resistant to neutralization by CD81-specific antibodies and infected cells carrying a given abundance of human CD81 at their surface with increased efficiency compared to wildtype and the Jc1/G451R variant HCV (Figure 4). These findings suggest that only the mouse CD81-adaptive changes have enhanced human CD81 receptor usage for infection. Notably, this change however was not accompanied by a reduction of the minimal threshold CD81 density needed for viral entry (Figure 4B). It is worth mentioning here that Grove et al. noted decreased neutralization of JFH1 carrying the G451R mutation by CD81-specific antibodies [49]. Although we do not know the reason for this discrepancy it is possible that the phenotype of this mutation at least with regard to CD81 usage may differ dependent on the strain of glycoproteins (J6 and JFH1, respectively). Receptor competition experiments with antibodies against SR-BI and CLDN1 revealed a similar behaviour of the mutant viruses, both exhibiting a decreased SR-BI dependence, while CLDN1-dependence was indistinguishable between both mutant viruses and wildtype HCV (Figure 5). Finally, when reducing the number of available OCLN molecules by RNA interference, both wild type and the G451R adapted virus were clearly more inhibited than Jc1/mCD81. These data indicate that both cell culture adaptation as well as adaptation to mouse CD81 caused reduced SR-BI dependence, however only the latter procedure at the same time also reduced dependence on human CD81 and OCLN. A third possible mechanism that may permit utilization of mouse CD81 for entry is through facilitating essential conformational changes of the glycoprotein complex. This could for instance be mediated by an opening of the overall structure of the viral surface proteins. Our neutralization experiments with E2-specific antibodies revealed a much enhanced susceptibility to neutralization for both mutant viruses. Interestingly, susceptibility to domain C-specific antibodies was more pronounced for the mouse CD81 adapted virus than for Jc1 with the G451R mutation, indicating distinct conformations of the virus particle-resident E2 proteins at this domain. These findings are both suggestive of gross conformational changes and of increased exposure of key neutralizing epitopes. Since these changes were not accompanied by an altered distribution of these viruses in density gradients, we believe that they are not due to differential interaction with lipoproteins. As additional indicator for altered surface protein conformation and as a measure for viral responsiveness to triggers of cell membrane fusion, we compared the ability of wildtype and mutant HCV to infect host cells stimulated by a low pH wash. In agreement with previous findings, wildtype HCV was refractory to a low pH wash immediately after binding to the cell surface at 4°C and only infected cells after an incubation of 1 h at 37°C [24]. In contrast, both mutant viruses could be induced to infect cells by low pH directly after binding. This suggests that the requirements for induction of pH-dependent fusion have changed for both viruses. Since the structure of the HCV E1/E2 complex is unknown it is difficult to finally conclude which of the above mentioned mechanisms that are certainly not mutually exclusive, is responsible for the efficient utilization of mouse CD81. Our data support a model where this combination of mutations leads to an opening of the glycoprotein complex. This “unlocked” structure increases exposure of the CD81 binding site and likely in turn interaction with soluble CD81. Moreover, it also facilitates conformational changes which may permit the virus to utilize a weak CD81 binding (e. g. due to interaction with CD81 from non-human species which bind with low affinity) for its entry process. In addition, such a conformational change may contribute to decreased dependence on SR-BI and OCLN and a lower threshold for pH-triggered cell entry. This interpretation may also explain why these mutations selected on mouse CD81 increase the usage of rat and hamster CD81 to a comparable degree (approximately 100-fold). Given that these forms of CD81 differ from each other by a number of amino acids within the viral binding site [25], this would have been unlikely, if the viral binding site had specifically adjusted to mouse CD81. Finally, we show that HCV J6-derived glycoproteins adapted to mouse CD81 permit infection of mouse cells in the absence of human factors. Thus, three adaptive mutations within E1 and E2 are sufficient to overcome the barrier for cell entry into mouse cells. In our view, this finding lends further support to the notion that these mutations may facilitate receptor-dependent conformational changes of the glycoproteins thus permitting the use of mouse CD81 and ultimately infection of cells via solely mouse factors. Previous reports have established that besides CD81 at least OCLN poses a strong species barrier to HCV infection [8], [25]. In addition, evidence was reported that sub-optimal usage of mouse CLDN1 and SR-BI may also limit efficient infection of mouse cells [26], [27]. Therefore, poor entry of HCV into mouse cells seems to be determined by multiple entry factors, making it difficult to envision that mere adaptation of the viral CD81 binding site would be sufficient to allow usage of mouse SR-BI, OCLN, and CLDN1. Notably, while the G451R mutation only reduced SR-BI-dependence, the mouse CD81 adaptive mutations at the same time also reduced CD81- and OCLN-dependence which may be an important prerequisite for the virus to utilize the foreign entry factors which possibly interact with lower affinity. Likely due to very low RNA replication of HCV in mouse cells we were unable to observe productive infection of mouse cells by Jc1/mCD81 (data not shown). In turn these results highlight that additional host factors limit propagation of HCV in mouse cells downstream of virus cell entry. Nevertheless, our findings are proof of concept that a limited number of mutations can be sufficient to adapt the virus for usage of foreign host factors or possibly even host factor complexes and therefore encourage future efforts to adapt HCV for growth in non-human cells. Combined with genetic approaches and xenotransplantation [52] such efforts may ultimately lead to the development of immune competent small animal models for research of HCV pathogenesis, immune control as well as vaccine and drug development. The plasmids pFK Jc1, pFK-Luc-Jc1 and pFK-Venus-Jc1 have been described earlier [37], [41], [48]. The constructs pFK Jc1/L216F, pFK Jc1/V388G, pFK Jc1/M405T, pFK Jc1/N767D, pFK Jc1/V388G/M405T, pFK Jc1/L216F/V388G/M405T (equivalent to pFK Jc1/mCD81) and pFK Jc1/L216F/V388G/M405T/N767D were created via PCR-based site-directed mutagenesis. PCR inserts were sequenced to exclude off site mutations. The constructs pcDNA3ΔcE1E2J6/mCD81 (containing L216F, V388G and M405T) and pcDNA3ΔcE1E2J6/G451R are derivatives of pcDNA3ΔcE1E2J6 that has been described recently [26]. pRV Venus was created by insertion of the gene encoding Venus-GFP [53] into the retroviral vector pczCFG5IZ [54]. Lentiviral plasmids pWPI humanCD81 BLR, pWPI ratCD81 BLR, pWPI hamsterCD81 BLR, pWPI mouseCD81 BLR, pWPI humanCD81 Gun, pWPI mouseCD81 Gun, pWPI humanOCLN Gun, pWPI mouseOCLN Gun, pWPI humanSR-BI Gun, pWPI mouseSR-BI Gun, pWPI human CLDN1 BLR and pWPI mouse CLDN1 BLR encode the human, rat, hamster or mouse orthologs of the four HCV receptors in the context of the self inactivating lentiviral vector pWPI [55]. In this context the transgene is expressed by an internal human elongation factor 1 alpha (EF1-α) promoter. The transcribed unit also contains an IRES element from EMCV that allows internal initiation of translation of a blasticidin S deaminase of Aspergillus terreus or a GFP-ubiquitin neomycinphosphotransferase fusion protein (Gun) as selectable markers. Exact cloning strategies and primer sequences can be obtained on request. Huh-7. 5, Huh7-Lunet, HuH6,293T and NIH3T3 cells were cultured in Dulbecco' s modified Eagle medium (DMEM; Invitrogen, Karlsruhe, Germany) supplemented with 2 mM L-glutamine, non-essential amino acids, 100 U of penicillin per ml, 100 µg of streptomycin per ml, and 10% fetal calf serum (DMEM complete) at 37°C and 5% CO2. Lunet N cells were generated by FACS sorting of CD81 low expressing cells within the Lunet cell population and subsequent subcloning by limiting dilution. Three clones were analyzed further with regard to CD81 expression and permissiveness for HCV RNA replication (clones #3, #4, and #7). Of these subclones, number #4 and #7 were described recently [36] and subclone #3 was used throughout this study and was designated as Lunet N. Stable cell lines were generated via lentiviral gene transfer as described recently [56] using the three plasmids pCMVΔR. 74 [57], a pWPI derivative (either coding for a resistence against blasticidine (blasticidine S deaminase; BLR) of Aspergillus terreus or a GFP-ubiquitin-neomycin fusion protein (Gun) and the respective gene of interest) and pcz VSV-G [58] in a ratio of 3∶3∶1. Selection was carried out in the presence of either 5 µg/ml Blasticidin or 0,75 mg/ml G418. HCVcc particles and firefly luciferase HCV reporter viruses were generated as reported previously [48]. In brief, plasmid DNA was linearized and transcribed into RNA, which was then electroporated into Huh-7. 5 cells. Virus-containing culture fluids of transfected cells were harvested 48 h and 72 h after transfection. Luciferase reporter virus infection assays were carried out and analyzed as described [48] using Lunet N hCD81 cells as target cells. Wildtype HCV particles were titrated by using a limiting dilution assay [59]. The 50% tissue culture infectious dose (TCID50) was calculated based on the methods described by Spearman and Kärber [60], [61]. Murine leukemia virus (MLV) -based pseudotypes carrying vesicular stomatitis virus glycoproteins (VSV-G) or the HCV J6-derived E1 and E2 proteins as well as their mouse CD81 adapted derivatives (J6/mCD81) were generated by cotransfection of 293T cells. Briefly, 1. 2×106 293T cells were seeded into 6-cm-diameter plates 1 day before transfection with 2. 6 µg envelope protein expression construct pczVSV-G, pcDNA3ΔcE1E2-J6, pcDNA3ΔcE1E2-J6/mCD81 or an empty-vector control, 2. 6 µg MLV Gag-Pol expression construct pHIT60, and 2. 6 µg firefly luciferase transducing vector by using Lipofectamine 2000 (Invitrogen). The medium was replaced 6 h after transfection, and the supernatants containing the pseudoparticles were harvested 48 h later. The supernatants were cleared of cells by passage through a 0. 45-µm-pore-size filter, concentrated ca. 10-fold by ultrafiltration with Amicon Ultra centrifugal filter units (molecular weight cut off 100 kDa, Millipore, Schwalbach, Germany) and used to infect Lunet N hCD81 and NIH3T3 cells and its derivatives expressing HCV entry factors of human and mouse origin. For infection, target cells were seeded at a cell density of 2×104 per well of a 48-well plate 24 h prior to inoculation. Cells were inoculated for 4 h with retroviral pseudoparticles and cells were cultured an additional 48 h prior to harvesting and fixation in 1% paraformaldehyde (PFA) w/v. Venus GFP-expression as marker for productively infected cells was quantified using a FACS-Calibur flow cytometer (Becton Dickinson, Heidelberg, Germany) and the Flow Jo software package. Raw FACS scatter plots of one representative experiment of five independent repetitions are depicted in supplementary Figure S7. A representative experiment carried out in duplicates from 5 independent experiments is depicted in Figure 7. The background Venus-GFP signal from non-enveloped pseudoparticles (Env-pp) was first subtracted from the VSV-Gpp and HCVpp signals. The HCVpp signal was then normalized to VSV-Gpp infectivity [ (HCVpp-NE) / (VSVG-NE) ]. Finally, these values were expressed relative to HCVpp infectivity determined on our human reference cell line Lunet N hCD81. As a result, the data reflect the relative ability of HCVpp to infect mouse NIH3T3 cells expressing different human or mouse entry factors compared to highly permissive human host cells (Lunet N hCD81). CD81 LEL GST fusion proteins were prepared as described [25]. In brief the expression constructs were transformed in E. coli rosetta gami. A 1 l culture of the bacteria was grown to an OD592 of 0. 8 at 37°C and induced to express the GST fusion proteins with 1 mM IPTG (Sigma-Aldrich, Steinheim, Germany) at RT over night. Bacteria were first lysed with 1 mg/ml chicken egg lysozyme (Sigma) and 500 U benzonase in PBS and in a second step with 0. 5% nonidet P-40 (NP40) in PBS (lysis buffer) the supernatant was incubated with 2 ml of a 1∶1 slurry glutathione agarose (Sigma) for 2 h at RTunder gentle agitation washed three times with 25 ml lysis buffer and afterwards the bound proteins were eluted with 100 mM Glutathione (Sigma-Aldrich, Steinheim, Germany) in PBS and dialysed in PBS over night. Concentrations were determined by Bradford. To determine the interaction between the recombinant CD81 protein and virus particles, 500 µl virus-containing culture fluid harvested 48 h post transfection of Huh-7. 5 cells and normalized for the amount of Core protein were preincubated with 50 µl glutathione agarose beads (1∶1 slurry). These preadsorbed virus preparations were incubated with 50 µl glutathione agarose beads that had been coupled with 10 µg recombinant GST-CD81-LEL proteins and the mixture was incubated on a rotator at 4°C over night. Subsequently, beads were washed 3 times with PBS and then resuspended in 200 µl of PBS. The amount of bead-associated HCV core protein was determined by a commercial core-specific ELISA (Wako Chemicals, Neuss, Germany). HCV RNA associated with beads suspended in 100 µl PBS was extracted using the Nucleo Spin RNAII kit (Macherey-Nagel, Düren, Germany) as recommended by the manufacturer. Two microliters of the RNA sample were used for quantitative RT-PCR analysis using a LightCycler 480 (Roche, Mannheim, Germany). HCV-specific RT-PCRs were conducted in duplicate employing a one-step RT-PCR LightCycler 480 RNA Master Hydrolysis Probes kit (Roche, Mannheim, Germany) and the following JFH1-specific probe (TIB Molbiol, Berlin, Germany) and primers (MWG-Biotech, Martinsried, Germany): A-195 [5′-6FAM (6-carboxyfluorescein) -AAA GGA CCC AGT CTT CCC GGC AAT T-TAMRA (tetrachloro-6-carboxyfluorescein) -3′], S-146 (5′-TCT GCG GAA CCG GTG AGT A-3′), and A-219 (5′-GGG CAT AGA GTG GGT TTA TCC A-3′). Luc-Jc1 viruses were incubated with different concentrations of the indicated antibodies for 30 minutes at RT prior to inoculation of Lunet N hCD81 cells which had been seeded at a density of 0. 7×104 cell/96-well the day before. Each neutralization assay was conducted in triplicates with a total volume of 100 µl/well. 4 h after inoculation with the virus/antibody mixture another 100 µl of medium was added. Anti-E2 neutralization assays were conducted using the human E2-specific mAbs CBH4D, CBH23, CBH5, and the R04 antibody which is directed against a cytomegalovirus protein as control [46], [62], and the E2-specific mouse monoclonal antibody AP33 [63]. For receptor neutralization assays the mouse monoclonal anti CD81 antibody 5A6 (Santa Cruz, Santa Cruz, CA) and 1. 3. 3. 22 (Ancell, Bayport, MN) or anti-CD13 (Beckton Dickenson, Heidelberg, Germany) were utilized. IgGs from polyclonal rat sera specific to SR-B1 and CLDN1, raised by genetic immunization of Wistar rats with a plasmid expressing the respective full length cDNA, were produced and purified as described [64], and employed at the indicated concentrations starting with 50 µg/ml. SiRNA pools from Dharmacon (Lafayette, CO) specific to human OCLN were transfected via reverse transfection according to the manufacturer' s instructions using RNAiMax (Invitrogen) into Lunet N hCD81 cells. Briefly, 150 pmol siRNA were mixed with 200 µl OptiMEM in a 12-well, and 1. 5 µl of RNAiMax were added. A total of 6×104 Lunet N hCD81 cells suspended in 1 ml of plain medium with 10% FCS were seeded on top. Knock down efficiency was analyzed 48 h after transfection via Western Blot directly at the time of inoculation with HCV luciferase reporter viruses. Infection efficiency was determined 48 h later using luciferase assays as described above. The assay was conducted as described earlier [24], [65]. Lunet N hCD81 cells were seeded at a density of 2. 5×105 cells/6-well. 24 h later, cells were pretreated with Concanamycin A (ConA; 5 nM) for 1 h at 37°C. Subsequently, cells were shifted to 4°C and were then infected with F-Luc Jc1 variants in the presence of ConA for 2 h at 4°C. Cells were washed twice with cold PBS either directly followed by an incubation in citric acid buffer (McIllivaine' s buffer system pH 7 or 5) for 5 min at 37°C. Alternatively, after virus binding at 4°C, cells were first shifted to 37°C for 1 h under continued presence of ConA to 37°C, and only then incubated with citric acid buffer as described above. In both protocols, cells were maintained for another 3 h in medium containing 5 nM ConA and then the medium was replaced. Cells were lysed after 48 h and the luciferase signal was measured. Luciferase values are expressed relative to control infections which were conducted in the absence of ConA and with an incubation step with citric acid buffer of pH 7. HCV particles were harvested 48 h after electroportation of Huh-7. 5 with the indicated viral RNA. 1 ml of virus containing supernatant was mixed with 2 ml of a 60% (w/v) iodixanol stock solution (Optiprep Axis Shield, Oslo, Norway) and layered under a 0–30% iodixanol gradient. Gradients were centrifuged at 154,000×g for 18 h and afterwards 1 ml fractions were collected from bottom to top. The density of the different fractions was determined via refractometry, the infectivity was determined in limiting dilution assay and the amount of particles was assessed by the determination of RNA copies as described above. 5×105 cells were stained with antibodies specific for hCD81 (5A6,1∶200, Ancell, Bayport, MN) for mCD81 (EAT2-PE, 1∶20, Santa Cruz, Santa Cruz, CA) or with a polyclonal rabbit serum directed against SR-BI (Novus Biologicals, Littleton, CO 1∶50) in PBS containing 2% FCS for 30 minutes on ice and washed with PBS. In case of noncoupled primary antibodies bound antibodies were detected with secondary antibodies coupled to phycoerythrin or allophycocyanin (1∶200, eBioscience, San Diego, CA) for 30 minutes on ice. After a next washing step the cells were analyzed with a FACS Calibur (Becton Dickinson, Heidelberg, Germany) and the results were analyzed using the Flow Jo Software. Cells were washed with PBS and lysed in RIPA buffer (0. 3 M NaCl, 20 mM TrisHCl pH 8,1% Sodium dexoxycholat; 0. 1% SDS and 1% Triton) for 30 minutes on ice. The total protein content was determined by Bradford assay. Equal protein amounts for each sample were mixed with 2× denaturing protein sample buffer (200 mM Tris-HCl [pH 8. 8], 5 mM EDTA, 0. 1% bromophenol blue, 10% sucrose, 3. 3% sodium dodecyl sulfate [SDS], 2% 2-mercaptoethanol [2-ME]) for 5 minutes at 98°C, loaded onto a 12. 5% SDS-gel and resolved by electrophoresis. Subsequently, proteins were transferred with a semiydry blotter to a polyvinylidene difluoride membrane. The membrane was blocked with 5% milk in PBS containing 0. 5% Tween (PBS-T) for 1 h at RT. CLDN1, OCLN and actin were detected with specific monoclonal antibodies (Zymed, San Francisco, CA and Sigma-Aldrich, Steinheim, Germany) and a secondary antibody coupled to the horseradish peroxidase (Sigma-Aldrich, Steinheim, Germany). The antibody signal was detected with the ECL Plus detection system (GE Healthcare, Freiburg, Germany). Comparison of knock down of OCLN reduced entry of Jc1, Jc1/G451R, and Jc1/mCD81 were compared with one-way analysis of variance of hill functions. Similarly, the proportion of infected cells according to their MFI is fitted by nonlinear weighted least squares. IC90 values from neutralization experiments were obtained and compared by nonlinear least squares fit of hill functions. All tests were two-sided and p-values below 5% are considered significant.

The hepatitis C virus (HCV) infects only humans and chimpanzees, which has hampered development of suitable animal models. The inability of HCV to penetrate non-human cells is primarily due to inefficient usage of non-human CD81 and occludin. In this study we adapted HCV to mouse CD81. Efficient utilization of mouse CD81 is conferred by a combination of three mutations in the viral glycoproteins. These changes also permit entry via rat or hamster CD81, and lower viral dependence on additional HCV entry factors. Strikingly, mouse CD81 adapted HCV glycoproteins mediate entry into mouse cells in the absence of human entry factors. The adaptive mutations are not resident in viral domains implicated in direct CD81 binding. Nevertheless, they enhance binding to human CD81, increase susceptibility to different neutralizing antibodies and facilitate induction of viral cell fusion by low pH. This suggests that structural changes accompanied by exposure of the CD81 binding site and neutralizing epitopes have “unlocked” the viral envelope protein complex facilitating infection through non-human entry factors. These results highlight mechanisms of HCV receptor usage and tropism. They also demonstrate that HCV can be adapted to using non-human host factors, which may ultimately facilitate the development of small animal models.

Abstract Introduction Results Discussion Materials and Methods

virology/animal models of infection virology/host invasion and cell entry

2010

Adaptation of Hepatitis C Virus to Mouse CD81 Permits Infection of Mouse Cells in the Absence of Human Entry Factors

Hosts are likely to respond to parasitic infections by a combination of resistance (expulsion of pathogens) and tolerance (active mitigation of pathology). Of these strategies, the basis of tolerance in animal hosts is relatively poorly understood, with especially little known about how tolerance is manifested in natural populations. We monitored a natural population of field voles using longitudinal and cross-sectional sampling modes and taking measurements on body condition, infection, immune gene expression, and survival. Using analyses stratified by life history stage, we demonstrate a pattern of tolerance to macroparasites in mature compared to immature males. In comparison to immature males, mature males resisted infection less and instead increased investment in body condition in response to accumulating burdens, but at the expense of reduced reproductive effort. We identified expression of the transcription factor Gata3 (a mediator of Th2 immunity) as an immunological biomarker of this tolerance response. Time series data for individual animals suggested that macroparasite infections gave rise to increased expression of Gata3, which gave rise to improved body condition and enhanced survival as hosts aged. These findings provide a clear and unexpected insight into tolerance responses (and their life history sequelae) in a natural vertebrate population. The demonstration that such responses (potentially promoting parasite transmission) can move from resistance to tolerance through the course of an individual' s lifetime emphasises the need to incorporate them into our understanding of the dynamics and risk of infection in the natural environment. Moreover, the identification of Gata3 as a marker of tolerance to macroparasites raises important new questions regarding the role of Th2 immunity and the mechanistic nature of the tolerance response itself. A more manipulative, experimental approach is likely to be valuable in elaborating this further. There is increasing recognition that the ability to tolerate a parasite' s presence, where the host accepts infection but actively limits the damage caused, may be a crucial host defence strategy [1], [2]. This is distinct from resistance, where the focus is on limiting the infection burden itself [3], [4], though in reality hosts often combine tolerance and resistance. Adopting a tolerance strategy may have major implications for the epidemiology and co-evolutionary dynamics of infectious disease through its effects both on individual hosts and on parasite transmission [5], [6]—for example, leading tolerant individuals to become infectious “super spreaders” [7] or promoting the evolution of higher transmissibility in tolerant populations [8]. However, parasite tolerance has often been neglected in animal and human studies [4], [9], and virtually nothing is known of its manifestation in natural systems. Here we focus on the expression of tolerance in a natural population and attempt to identify immunological processes underlying this and their consequences for important fitness components. In doing so, though, we must first address how tolerance can be defined and interpreted in studies of field systems. The term “tolerance” is used differently by immunologists and disease ecologists. In moving towards some unification of meaning, we must distinguish between pattern and process. The pattern associated with tolerance, typically the main focus of ecologists, is a relative insensitivity of host health (fitness) to increases in parasite burden [4]. This can result from various host responses (processes) that protect the host from the parasite without reducing parasite fitness. Examples include the up-regulation of wound healing or the down-regulation of immunopathology [9]. The latter would often involve the process conventionally termed tolerance by immunologists: the limitation of T- or B-cell responsiveness to cognate antigens. The pattern of tolerance may also result from host responses that ameliorate pathogen virulence (not tolerance in an immunological vocabulary), for example, by up-regulating lipoproteins that protect against endotoxins or serpins that target bacterial proteases [8]. In using the term tolerance, therefore, it is important to distinguish whether it is the pattern or the process that is being referred to, or both. In natural populations, and also in ecological experiments [4], the pattern will often be known, but the process will not. But equally, in laboratory immunology studies, details of the process may be available, but its consequences in a natural setting (the pattern) may not. The study of tolerance in field systems, then, is an indispensible complement to studies in the laboratory, and there have been calls for the greater integration of data from natural populations on host fitness, immunological variation, and infection pressure [10], [11]. However, field studies prove challenging in separating pertinent patterns from background noise with other causes and in distinguishing causal relationships from simple associations. Hence, it is necessary to predefine putative markers (“signatures”) of tolerance (patterns) and to assess these within a study framework aimed at disentangling alternative causalities. To achieve this, we draw on the perspective provided by an emphasis on “phase space” [12], where the important distinction is drawn between tolerance curves and phase or disease curves. In tolerance curves, the relationship between host health and parasite load is plotted, with each data point being an observation contributed by a different individual at a standardised stage in an infection (e. g. , maximal parasite load). A more tolerant population or subgroup would then be one in which the slope of this relationship was less negative, and a slope of zero (or even more, a positive slope) would be strongly suggestive of tolerance. In phase curves, by contrast, the relationship between an individual host' s health and parasite load is plotted, with each data point being an observation at a different stage in the infection. A more tolerant individual would then be one in which the decline in health with increasing load was less steep, and no decline at all (or even more, a positive slope) would be strongly suggestive of tolerance. This phase space perspective also highlights the absence of any agreed measure to quantify health but favours the use of “gross” (whole organism) measures such as those we have adopted here [12]. We hypothesised, first, that different groups of individuals (e. g. , life-history stages) will differ in their tolerance—their relationship between health and parasite load—and second, that those individuals that maintain good health in relation to parasite burden (i. e. , are more tolerant) will express more of the physiological pattern associated with tolerance. We monitored immunological gene expression, infection, body condition, and survival in replicated natural populations of field voles, Microtus agrestis, using interwoven cross-sectional and longitudinal sampling protocols. This hybrid study design took advantage of the greater range and precision of measurement possible in destructive (cross-sectional) sampling, and of the stronger causal inferences possible in capture–recapture (longitudinal) sequences, where driver variables may be observed to precede the responses they trigger [13]. Our objective was to identify patterns of tolerance, to link these to biomarkers among gene expression variables, and then to examine the life history consequences of the tolerance strategy through the surrogate of biomarker expression. Our overall aim was to place tolerance within a life history context and at the same time suggest possible immunological mechanisms (processes) based on the biomarker (s) identified. Specifically, adopting the phase space perspective (above), plotting our cross-sectional data comes closest to generating a tolerance curve. However, infection stage cannot be standardized in observations on natural populations. Hence, a less negative, zero, or even positive relationship between health and parasite load will be consistent with tolerance (and suggestive of it), but alternative explanations cannot be excluded. By contrast, our longitudinal data generate sequences of time points on a phase curve for a “typical” individual. A pattern in which health fails to decline (or even increases) following an increase in parasite load will therefore be directly indicative of tolerance. The case for demonstrable tolerance will, of course, be strongest if the results from the cross-sectional and longitudinal studies are consistent with one another. In contrast to tolerance, resistance resulting from acquired responses (usually the key component of resistance in co-adapted host–parasite associations; e. g. , [14]) is likely to be characterised by a decelerating or even negative accumulation of parasites over time and by a negative association between resistance markers and infection levels. Our sampling, carried out in natural M. agrestis populations in Kielder Forest, Northumberland, United Kingdom, involved a cross-sectional component (n = 576 destructively sampled voles) and a longitudinal component (n = 920 marked individuals monitored through time, with n = 1,665 sampling points) and was replicated at two different sites in each of 2 y. Biometric and infection data were recorded at all sampling points. In the cross-sectional component, ligands for Toll-like receptors (TLRs) 2 and 7 were used to stimulate splenocyte cultures, from which we measured the expression of immune genes involved in regulatory (IL-10, TGF-β1) and effector (IRF5, IL-1β) pathways recruited during antimicrobial pattern recognition responses. In addition, the expression of genes reflecting different T-helper phenotypes was measured in nonspecifically (mitogen, PHA-L) activated splenocytes. The activity of regulatory T cells was represented by the anti-inflammatory cytokines IL-10 and TGF-β1 and the transcription factor FoxP3. The activity of T-helper cell type 1 (Th1) responses was represented by the pro-inflammatory cytokine IFN-γ and transcription factor Tbet and that of Th2 responses by the transcription factor Gata3. In the longitudinal component, constitutive expression of IFN-γ, IL-10, and Gata3 were measured in peripheral blood. Our overall analytical strategy (initially using standard statistical modelling techniques for single response variables) was to begin with our more detailed cross-sectional data and search first for the patterns of resistance or tolerance described above. We then sought to identify immunological gene expression markers of these patterns and, in turn, to link these to infection and life history variables. Next, we used our independent longitudinal data to corroborate the robustness of the immunological markers, place them within a stronger cause–effect context in relation to the infection and life history variables, and measure their effect on survival. Finally, we returned to the cross-sectional data and used structural equations modelling (SEM) to support the interdependencies between variables indicated by the longitudinal data. In analyses of the cross-sectional data, we primarily represented macroparasite infection using a reduced variable derived as scores from the first principal component of a principal components analysis (PCA) of the most common species that would be expected to be in contact with the host immune system (fleas, ticks, and adult tapeworms) (Table S1). In the search for immunological markers, this variable (PCM main) was tested first against similarly reduced immune gene expression variables (Table S2) and then against individual variables using multiplicity adjustments. Equivalent results were obtained if PCM main was replaced in analyses by a similar reduced variable for all macroparasite species recorded (PCM) or by measures of some influential species (e. g. , ticks). However, the composite measure (PCM main) has the advantage that relationships with host responses will not be diluted when, for example, a low burden for one parasite is accompanied by high burdens for others, such that the response to the low burden is “greater than expected, ” driven by the other parasites. In our initial search for patterns of tolerance or acquired resistance, we used general and generalized linear mixed models (LMMs and GLMMs; details in Methods S1). In the overall population (in models with slopes averaged across life history stages), macroparasites (PCM main) tended to accumulate linearly with individual size and age, with no indication that acquired resistance decelerated the acquisition of infection in older animals (Figure S5). This was not uniform across life history stages, though, with a particularly robust contrast occurring between mature males (those with large descended testes and expanded seminal vesicles) and immature males (Figure 1A). Mature males accumulated macroparasites linearly as age indicators increased (SVL [mm], slope parameter 0. 034±0. 009, p = 4. 5×10−4; eye lens weight [g], 283±79, p = 3. 9×10−4), but immature males supported less infection as age indicators increased (SVL, −0. 035±0. 013, p =. 011; lens weight, −317±97, p =. 002). This pattern is consistent with resistance building up prior to maturity but being absent postmaturity (Figure 1A). By contrast, there was a strong overall pattern consistent with a tolerance response. To examine this, we used as a proxy for individual health, not simple measures of weight but “body condition”: body (or organ) weight normalised for individual size – snout-vent length (SVL) and its quadratic term (see Methods S1). Considering all voles together, those in the best condition tended to be infected with the most macroparasites (body condition, p<5×10−7; Tables S3, S4, S5). Furthermore, stage-specific differences in tolerance were suggested by heterogenous slopes of body condition on macroparasite loads (PCM main) amongst different life history stages. A particularly robust contrast occurred again between mature males, which showed a strong positive relationship between parasite burden and body condition, and immature males, which showed no such relationship (Figure 1B, LMM, Stage×PCM main interaction, p =. 003) (Table S7). Amongst the microparasitic infections that we surveyed, animals displaying overt signs of tuberculosis (TB) (Mycobacterium microti) were also in better condition relative to others (body condition, p =. 018; Tables S3, S4). To identify immunological markers for a tolerance strategy and determine their consequences for host fitness, we then focused primarily on the dichotomy between mature and immature males, as the former showed the strongest increased body condition response to macroparasite infection but no epidemiological evidence of resistance. Patterns in females, which may be obscured by immunosuppression during pregnancy, were less clear cut, as discussed briefly below. We began by asking (using LMM analyses) whether the trend for apparent tolerance rather than resistance in mature versus immature males was linked to variation in immune gene expression. We found that a single parameter, mitogen-stimulated Gata3 expression in cultured splenocytes (Gata3mit-stim), was consistently and strongly associated with macroparasite burden, host condition, and life history in mature males and also showed a different pattern in immature males. First, Gata3mit-stim correlated with macroparasite burden positively in mature males (Figure 1C, p =. 007; Table S11) and negatively in immature males (Figure 1C). Second, Gata3mit-stim was positively correlated with liver and body condition in mature males (Figure 1D, p =. 027 and 7. 1×10−5, respectively; Tables S8 and S9) but not immature males (Figure 1D), even after controlling for the association between macroparasite burden and body condition noted above. Third, Gata3mit-stim had a significant negative association with size-adjusted testis weight (an estimator of male reproductive effort) in mature males (Figure 1E, p =. 014; Table S9). Thus, Gata3 expression was higher in mature males in good condition and infected with many macroparasites (Figure 1F) but with proportionately smaller reproductive investment. Notably, Gata3mit-stim expression was also a strong marker for elevated condition among animals with overt TB (Figure 1G, p<. 001; Table S10). Moreover, because Gata3 would usually be involved where Th2 immunity mediates resistance to macroparasites [15], the differing stage associations of macroparasites with Gata3 expression observed (Figure 1C) are indicative of changes in resistance. Thus, the negative association in immature males contrasted with the positive association in mature males is further evidence (additional to the stage-specific associations of infection with age indicators) (Figure 1A) that immature males may resist macroparasite infections, whereas mature males do not. Gata3 expression, then, may be a useful biomarker for either resistance or tolerance where a predominant pattern of one or the other can be established within groups, as it can in males. Nonpregnant females showed similar tolerance-like accumulations of macroparasites with age indicators and also increases in condition at higher macroparasite burdens. However, post hoc analyses indicated that these trends were not clearly marked by splenic Gata3 expression. This might be because of the re-adjustment of T-helper cell responses often seen during mammalian pregnancy [16] and also because among nonmating nonpregnant females (the majority of the female sample) prereproductives and those returning to a nonmating (imperforate) state following pregnancy could not be distinguished. Females also showed much lower infection levels than males (Figure S5B), and older females in mating condition, which were most heavily parasitized (Figure S5B) and thus most likely to show clear tolerance responses, were poorly represented in our cross-sectional sample (n = 47). Gata3 is a pivotal transcription factor involved in the development of Th2 cells [17]–[19], and its expression in mitogen-stimulated splenocyte cultures is likely to derive from proliferated Th cell populations. Th2 responses are typically triggered by macroparasite infection [20]–[24]. Hence, an initial hypothesis was that macroparasites stimulated higher Gata3 expression in mature males and that either macroparasites or Gata3 expression might then affect host condition and reproductive effort. To further explore possible causal links, we turned to our longitudinal data, focusing on macroparasites. In mature males, exposure to common blood-feeding ectoparasites was strongly associated with increased constitutive Gata3 expression in peripheral blood (Gata3blood) in the following month (LMM, Figure 2A, p =. 010; Table S12). (By contrast, the instantaneous association tended to be negative, perhaps reflecting some level of resistance to ectoparasites due to peripheral blood responses.) In turn, Gata3blood was positively associated with weight gain (adjusted for starting body weight) in the following month (LMM, Figure 2B, p =. 039; Table S13). (By contrast, it was unrelated to weight gain in the preceding month.) Taken together, therefore, the longitudinal data suggest that in mature males at least, macroparasites are the driver of high Gata3 expression, which in turn is part of a response that promotes elevated body condition. That is, as individuals progress along their phase (disease) curve [12], one proxy for health (body condition), far from declining as parasite load increases, appears to increase, indicative of a tolerance response to infection. In applying SEM to our cross-sectional data (Figure 2C; Tables S14 and S15), we specifically sought the SEM that best fitted the data or was indistinguishable from the best in explanatory power, but that was also supported by the details of the longitudinal data analysis above. In fact, the best and one near-best SEM did have this support, though they differed as to whether male reproductive effort had a negative effect on Gata3 expression and body condition or was negatively affected by them (which we take to be more likely biologically). The SEM analysis was thus consistent with a causal path: macroparasites→elevated Gata3→increased somatic condition/reduced reproductive effort. As a final post hoc step in the sequence of analyses in the cross-sectional data, we reiterated the SEM, replacing Gata3mit-stim with each of the other immune gene expression variables in turn. No other variable approached the significant configuration of coefficients seen for Gata3mit-stim, suggesting that this response in particular (and not a wider combination of the expression responses measured) was linked to the tolerance pattern. Constitutive Gata3 expression in peripheral blood (Gata3blood), which decreased with age (using weight as a surrogate), was a significant predictor of survivorship in the longitudinal data (Figure 3; Tables S16 and S17). This effect occurred both for adult males and in the overall dataset including all stages (adult males comprising 27% of records in the longitudinal study). In both the overall and the adult male analyses, younger animals expressing high Gata3blood had lower survival, but older animals expressing high Gata3blood had higher survival (GLMM, Figure 3; Gata3blood×Weight interaction in mature males, p =. 036; in all stages, p =. 011). These analyses suggest that expression of Gata3 (a master transcription factor involved in the differentiation of Th2 cells) can, in some individuals, be a marker for a physiological programme of tolerance to macroparasite infection, with implications for body condition, fecundity, and survival. Tolerance involved apparent overcompensation in body condition following infection, but accompanying that was downwards readjustment of fecundity, and age-dependent changes in survival, with survival improving in older tolerant (Gata3hi) animals. Our longitudinal analyses suggest, moreover, that rather than being a mere correlate, elevated Gata3 expression was triggered by macroparasite infection and preceded the life history readjustments. Some corroborative patterns were also seen for another chronic infectious disease, TB, with animals being in better condition if showing overt signs of infection and especially if also expressing high levels of Gata3. Our ability to infer tolerance in a complex field situation was facilitated by differing responses, in our cross-sectional study, between immature and mature males. These responses indicated a predominant pattern of resistance to macroparasites in the former and tolerance to macroparasites in the latter. Thus, as immature males aged, their infections reduced, consistent with the development of acquired resistance; also, mitogen-stimulated splenocyte Gata3 expression (Gata3mit-stim) was inversely associated with macroparasite infection in these animals. This presents a familiar scenario, consistent with the paradigmatic role of acquired Th2 responses in resistance against macroparasites. Mature males, on the other hand, showed a very different and unexpected pattern. They accumulated macroparasites with age (resulting in very high infection levels) and showed a positive association between Gata3mit-stim and infection. Furthermore, body condition increased in heavily infected animals, whereas Gata3mit-stim was also positively associated with body condition (trends not found in immature males) and negatively associated with male fecundity (adjusted testis weight). Our longitudinal time series data for recaptured adult males indicated that peripheral blood Gata3 expression was triggered by macroparasite exposures and preceded host responses that increased somatic maintenance and ultimately survival (in the oldest Gata3hi animals). This is suggestive of the process, and not just the pattern, of tolerance, as the increase in body condition associated with macroparasite infection and Gata3 expression is indicative of a protective response that does not affect the parasite directly. Gata3 expression in Th2 cells, then, at least in males where there is not the complicating factor of pregnancy, may serve as a biomarker for either resistance or tolerance to macroparasites, depending on whether one of these patterns dominates within a particular group of animals. Further studies are required to determine the relevance of these findings in other vertebrate systems, although the basic data required to establish the context of tolerance or resistance (i. e. , indicators of individual age, condition, and infection levels) are well within the routine scope of ecological researchers focussing on infectious disease. As considered further below, the dual aspect of Gata3 expression (marking either tolerance or resistance in different groups of animals) is biologically plausible given the known role of Th2 responses in some pathways leading to resistance to macroparasites [15] but also in pathways promoting wound-healing and tolerance processes [25]. This suggests the possibility, though, that upstream Th2 signals might drive different response mechanisms during resistance and tolerance. There are alternative possible explanations for some of the cross-sectional associations we observed within mature males. One would be “differential mortality”—that is, high parasite burdens eliciting high Th2 responses, but before our observations are made, animals in poor condition with many macroparasites suffering increased mortality, so that the only animals with high parasite burdens left to observe are those in good condition. A second explanation would be “correlated risk”—that is, increased risk in high condition animals of macroparasite infections, and the Th2 responses they trigger, due, for example, to animals that forage more actively acquiring both more food and more parasites. However, whereas the temporal sequences seen in the longitudinal data directly support the tolerance hypothesis (macroparasites→Gata3→good condition), they support neither differential mortality nor correlated risk, as high Gata3 expression would in neither case precede good condition. Similarly, both differential mortality and correlated risk would predict associations between Gata3 expression and survival that did not change with age, with the former specifically predicting an uncomplicated association of high Gata3 with low survival and poor condition. These predictions, too, are contradicted by our results. Furthermore, differential mortality and correlated risk do not provide explanations for differing cross-sectional associations between Gata3mit-stim, body condition, and macroparasite infection in mature and immature males. Hence, we can exclude these alternatives. Beyond our own system, these results may also provide some general insight into the nature of tolerance, as both pattern and process. The regulatory component of the immune system has often been seen as a potential driver of tolerance, through its role in suppressing immunopathological effector responses. Indeed, helminth parasites have been suggested to increase tolerance and reduce resistance by the elevation of regulatory responses [4]. However, while acknowledging the potential limitations of gene expression data [26], the prominent signature in our study of Gata3 in tolerant animals, rather than of archetypal regulatory genes (IL-10, TGF-β1, FoxP3), suggests that tolerance may in fact be significantly associated with a Th2 mechanism. In apparent opposition to this, Th2 responses have often been considered to mediate resistance to macroparasites (as in immature males in this study). In the laboratory they may do so via a diverse cascade of effector mechanisms [15]. However, not all effectors necessarily produce resistance to all species [15], [27], and vigorous effector responses have been observed without efficient resistance [27], [28]. Recently, moreover, Th2 responses have also been linked [25] with tolerance to macroparasites. A Th2 environment is well known to promote damage repair mechanisms, including the differentiation of alternatively activated macrophages [29], [30]. Also, generic Th2-associated effectors (e. g. , eosinophils, mast cells, antibodies) or Th2-driven fibrotic or granulomatous reactions [25], linked to the expression of Gata3, may mediate the pattern of tolerance by physically containing or more generally limiting the direct pathogenic activities of parasites. Furthermore, Th2 responses in chronic Mycobacterium infections have been thought to facilitate clinical disease progression [31]–[33], but the possibility that they function as a tolerance response has not been considered. And finally, although the responses of nonimmunological cell populations were not measured, Gata3 could also be involved in redirecting resources between life history traits during infection, through its role in the differentiation of cells involved in energy stores (e. g. , adipocytes [34], [35]) and reproductive investment (e. g. , mammary epithelia [36]). It seems possible that increased body condition and reduced fecundity, in mature males but not immature males, might be adaptive by increasing residual reproductive value during macroparasite infection. This is consistent with a positive influence of Gata3 expression on survival in the oldest males. The contrasting patterns seen between mature and immature males suggest these may face very different cost/benefit scenarios when deploying immune responses against macroparasites. At present we can only speculate about the nature of these pressures, although a number of interesting possibilities might be considered. For example, altered exposure (perhaps elevated in mature males due to increased ranging and social contacts) might raise the immunopathological or energetic costs of resisting infection above thresholds of sustainability. Mature males also, due to their greater prior opportunity to develop robust adaptive immunity to common pathogens, may face fewer consequences from secondarily adopting a more tolerant phenotype. Thus, tolerance processes might have more serious indirect consequences for younger individuals through the inhibition of inflammatory responses against primary microbial infections. These are the first results to indicate a mechanism of tolerance in a naturally occurring mammalian population and to place this within a holistic framework of host fitness: context-dependent cost–benefit outcomes for individual condition, reproductive investment, and survival. This work therefore illustrates how immunological studies in natural populations may not only benefit from the many advances made in the laboratory but might also feed insights back into mainstream immunology. The results support the view that Th2 responses may play a crucial role in the tolerance of infection, and they more generally add to the growing body of evidence arguing for an increased focus on tolerance in our quest to understand host responses to parasitic infection. Further progress is likely to be facilitated by adopting approaches that incorporate experimental manipulations. All procedures were carried out under UK Home Office licence regulations. We studied field voles (M. agrestis) in Kielder Forest, Northumberland, United Kingdom, using live-trapping to access individual animals from natural populations. Our study was designed to permit the analysis of individual variation in condition and survival, infection status, and the expression of immune genes (additional details in Methods S1). In order to ensure representativeness, we repeated our field design at two spatially separate sites in 2008–2009 and a further two separate sites in 2009–2010 (Figure S1). The study was divided into longitudinal and cross-sectional components. Each site contained a live-trapping grid (∼0. 375 ha) of 150 (10×15) regularly spaced traps (3–5 m intervals) placed in optimal habitat (Figure S1). Animals from this grid were marked with passive radio frequency transponders (AVID) and monitored over time, as sequences of capture and recaptures, forming the longitudinal component of the study. At each capture, biometric, infection, and immune expression measurements were taken (Figure S2). On each site there were also satellite transects (with traps spaced at ≥5 m intervals) from which 10 animals per month per site were sampled destructively, forming the basis for the cross-sectional component of the study. Animals from this part of the study were returned to the laboratory, where it was possible to collect a more comprehensive and detailed set of biometric, infection, and immune expression measurements (Figure S2). The transects aimed to sample a very large area of the habitat, providing data representative of the entire population at each site, while at the same time avoiding significant demographic readjustments. Each site was monitored by monthly trapping sessions between February (in the 2008–2009 season) or April (in the 2009–2010 season) and November, during which the capture–recapture study was carried out on the grid and destructive samples were retained from the transects (Figure S3). At each site, in November at the end of the field season and again in the following March, larger numbers of animals were destructively sampled both from the transects and from the grid habitats, including some animals previously marked with AVID transponders and processed for small tail-tip blood samples as part of the capture–recapture study. These samples also contributed to the overall cross-sectional component of the study (Figure S3). Measurements of gene expression (IFN-γ, Tbet, IL-2, Gata3, IRF5, IL-10, TGF-β1, FoxP3) in the cross-sectional element of the study focussed on cultured splenocytes and are described in detail in Jackson et al (2011) [20]. Through the combination of stimulatory conditions applied and the genes investigated, these measurements were intended to reflect both innate immune responses (including TLR-mediated responses) and adaptive responses (including Th1, Th2, and T-regulatory responses). In the longitudinal study we measured in vivo expression of a subset (IFN-γ, Gata3, IL-10) of the genes investigated in the cross-sectional study, in peripheral blood samples. Using direct counts or semiquantitative abundance indices, we quantified 25 species of macroparasite in our study animals, some of which were aggregated into ecologically and phylogenetically coherent groups to facilitate analysis (additional details in Methods S1). We also recorded overt symptoms of vole TB caused by M. microti and carried out PCR diagnostics for Bartonella spp. and Babesia microti using previously established methods (see Methods S1, Table S18). Data were analyzed using LMMs or GLMMs to relate individual variables to potential explanatory variables while allowing for nonindependence due to sampling and immunological assaying structure. SEM was used to assess patterns of interdependency among multiple variables simultaneously. Some variables used in these analyses were principal component scores (reduced from larger sets of partially redundant measurements using PCA). In the case of macroparasites, we used the first principal component (PCM main) from a PCA of common taxa likely to be in strong contact with the immune system (i. e. , feeding on, or dwelling in or on blood or internal tissues; >20% prevalence overall) as the main analytical variable (see Methods S1 for more details). This component was dominated by high loadings of the same sign for the main macroparasite groups (fleas, ticks, and adult cestodes). Analyses of vole return rates and survivorship were carried out using GLMMs and Cormack-Jolly-Seber (CJS) methods with individual covariates. In order to control the type 1 error rate, the analyses proceeded according to an a priori sequential strategy, starting with a sparing number of initial main hypotheses and moving onto subsequent main hypotheses conditional upon significant results in previous rounds of testing. This strategy also made use of reduced immunological and parasitological data and multiplicity adjustments within each round of main hypothesis tests. More expansive post hoc analysis of individual variables and subsets of the data, for the purposes of corroboration and describing biological patterns in more detail, was conducted subsequent to each round of main hypothesis tests. More details on the methods and strategy for data analysis are given in Methods S1 (see Figure S4 and Table S19). Data available from the Dryad Digital Repository: http: //dx. doi. org/10. 5061/dryad. bk537 [37].

Hosts do not always resist parasites. And once infection establishes, relatively little is known of how naturally occurring hosts tolerate (mitigate for) adverse effects, or what the life history consequences of this tolerance may be. In this article we demonstrate a pattern of tolerance to parasitic worms and arthropods in wild voles and link this to increased expression of an immunological biomarker. The biomarker, Gata3, is triggered by infection, precedes significant changes in body condition, and impacts on fecundity and survival. These results point to the considerable ecological importance of tolerance in wild vertebrates and to how poorly it is understood, while at the same time giving a new perspective on the natural function of immune response pathways involving Gata3.

Abstract Introduction Results Discussion Materials and Methods

biology and life sciences immunology ecology and environmental sciences

2014

An Immunological Marker of Tolerance to Infection in Wild Rodents

Buruli ulcer (BU) is a chronic necrotizing infectious skin disease caused by Mycobacterium ulcerans. The treatment with BU-specific antibiotics is initiated after clinical suspicion based on the WHO clinical and epidemiological criteria. This study aimed to estimate the predictive values of these criteria and how they could be improved. A total of 224 consecutive patients presenting with skin and soft tissue lesions that could be compatible with BU, including those recognized as unlikely BU by experienced clinicians, were recruited in two BU treatment centers in southern Benin between March 2012 and March 2015. For each participant, the WHO and four additional epidemiological and clinical diagnostic criteria were recorded. For microbiological confirmation, direct smear examination and IS2404 PCR were performed. We fitted a logistic regression model with PCR positivity for BU confirmation as outcome variable. On univariate analysis, most of the clinical and epidemiological WHO criteria were associated with a positive PCR result. However, lesions on the lower limbs and WHO category 3 lesions were rather associated with a negative PCR result (respectively OR: 0. 4,95%CI: 0. 3–0. 8; OR: 0. 5,95%IC: 0. 3–0. 9). Among the additional characteristics studied, the characteristic smell of BU was strongest associated with a positive PCR result (OR = 16. 4; 95%CI = 7. 5–35. 6). The WHO diagnostic criteria could be improved upon by differentiating between lesions on the upper and lower limbs and by including lesion size and the characteristic smell recognized by experienced clinicians. Buruli ulcer (BU) is a chronic necrotizing infectious disease of the skin caused by Mycobacterium ulcerans. After tuberculosis and leprosy, BU is the third most common mycobacterial disease worldwide [1,2]. BU has been reported in over 30 countries, typically in warm and humid intertropical regions where it predominantly affects children among poor and rural populations with difficult access to health care [3–5]. The diagnosis of BU is based on epidemiological and clinical criteria defined by the World Health Organization (WHO) [2,4]. The epidemiological WHO criteria are (i) residence or stay in a known BU endemic area and (ii) age between 0 and 15 years old since over 50% of all BU patients in Africa are children [6–8]. The clinical WHO criteria are (i) lesions on the upper or lower limbs since these represent about 85% of BU cases [4,8–11]; (ii) painless nodules, plaques or edema of the skin that, without treatment, evolve to a necrotic ulceration with [4,8] (iii) undermined and often hyperpigmented edges; (iv) lesions that are generally not accompanied by adenopathy, nor fever, and (v) that may become painful in case of superinfection [2,4]. Among the four tests recommended to confirm a BU diagnosis, two are most often used: direct smear examination (DSE) to detect acid-fast bacilli (AFB) and IS2404 PCR, which is the most sensitive test to date yet with associated delays in the availability of results of more than 10 days [12–14]. Despite having the highest sensitivity of all laboratory tests, PCR does not detect all BU cases. Patients have been described that fulfill the epidemiological and clinical WHO criteria yet repeatedly test negative by PCR [7,15–17]. Given the limited sensitivity of PCR to confirm BU, and to avoid under-treatment of BU if relying on laboratory confirmation, the WHO recommends national BU programs to initiate treatment with BU-specific antibiotics even in the absence of confirmation guided by the WHO clinical and epidemiological criteria described above [4]. However, WHO does recommend to enforce the laboratory confirmation of BU, aiming for at least 70% of notified BU cases to be confirmed by a positive PCR result [12]. In the current context of a declining BU incidence observed in several endemic countries with good surveillance in place, and consequently a proportional increase of non-BU lesions being treated in BU facilities [18,19], we expect waning clinical expertise in the recognition of BU. This can result in diagnostic and therapeutic errors as has been observed for leprosy [20]. Smelling may be among the oldest diagnostic methods, as different pathologies, such as infectious and endogenous metabolic disorders, can affect human body odors [21]. A study in Cameroon found the characteristic smell of BU to be strongly associated with a confirmed BU diagnosis and described it as the smell of rotten fish, cassava or cheese, mixed with that of pyocyanic bacteria [22]. In our clinical experience, this characteristic smell draws our attention to a BU diagnostic, even in atypical presentations. We recently reported on the accuracy of the clinical and microbiological diagnosis of BU and found that clinicians recognized BU with a sensitivity of 92% (95%CI 85%-96%) which was higher than the sensitivity of any of the laboratory tests. However, 14% (95%CI 7%-24%) of patients not suspected to have BU at diagnosis were classified as BU by a clinical expert panel [17]. We have therefore further investigated the WHO clinical and epidemiological criteria of the cohort of patients with lesions compatible with BU from our previous study and explored how these could be improved. The study was approved by the Provisional National Committee for Ethics in Health Research of Benin (registration n°: IRB 00006860), the Institutional Review Board of the ITM (code: 11 25 4 778) and the Committee for Medical Ethics of the Antwerp University Hospital (registration n°: B300201213080). The study also received an administrative authorization of the Benin Ministry of Health Ethics Board (N°IORG 0005695). All patients included in the study provided informed written consent. Parents or guardians provided consent on behalf of their children if participants were under the age of 18. This is an analytical prospective study of 224 consecutive patients presenting with skin and soft tissue lesions that could be compatible with BU (nodules, plaques, edemas, ulcers or osteomyelitis, including those recognized as unlikely BU by experienced clinicians), living in a BU endemic region, who were recruited in the “Centre de Dépistage et de Traitement de l’Ulcère de Buruli” (CDTUB) of Allada and Lalo in southern Benin between March 2012 and March 2015. Traumatic lesions of less than two weeks duration and relapses of BU were excluded from the study. Depending on whether the clinical and epidemiological characteristics of BU were met, the lesions were diagnosed clinically as BU or non-BU by experienced clinicians who had been trained on the WHO BU diagnostic criteria. Two swabs were taken from ulcers, or two fine-needle aspirates from closed lesions for the microbiological confirmation by IS2404 PCR and DSE after auramine staining in the Mycobacteriology Reference Laboratory in Cotonou (Benin), with quality control for molecular analyses performed by the Institute of Tropical Medicine in Antwerp (Belgium). The epidemiological, clinical and microbiological results of the patients were collected using standard WHO forms and entered in a Microsoft Access database by dedicated staff. We collected data related to the WHO diagnostic criteria (age, type and location of the lesion, pain, fever, adenopathy) and additional clinical and epidemiological information (size of lesion, WHO category, gender, and functional limitation). These results were documented for each patient by a team of two clinicians and three nurses, experienced in the diagnosis, treatment and management of BU. The presence of a characteristic smell was discussed systematically between clinicians at the time of sampling, before treatment initiation. The clinical team was not aware of laboratory information (not available yet at the time of clinical examination) but was unblinded to clinical and epidemiological information. The statistical analysis was done using Epi Info 7. 2. 2. 6 (Database and statistics software for public health professionals, Centers for Disease Control and Prevention (CDC), Atlanta, USA) and STATA/SE 11. 0. As a reference standard for BU confirmation we used IS2404 PCR. We performed univariate, bivariate and multivariate analysis using logistic regression with PCR result as dependent variable in order to establish a predictive model for BU. We also tested for interaction and confounding among variables associated with the PCR result using stratified analysis in bivariate analysis. All variables associated with a positive PCR result in the univariate analysis with a p-value <0. 10 were considered in the multivariate model. We used a backward elimination procedure, probability for removal was set at <0. 05. Odds ratios (OR) and their 95% confidence intervals (95%CI) were used as a measure of strength of the associations with a positive PCR result. We studied two predictive models of a positive PCR result and estimated the discriminative ability of both predictive models using the ROC analysis. Among the 224 participants included in this study, 120 (53. 6%) were male and 108 (48. 2%) were ≤15 years old. Median age was 18 years (IQR: 9–42 years). A total of 201 (89. 7%) patients had ulcerated lesions and 201 (89. 7%) had lesions on their limbs. A clinical BU diagnosis was made in 134 (59. 8%) patients. PCR was positive for 98 (43. 7%) participants among whom 9 participants had been clinically diagnosed as non-BU (10. 0% of patients diagnosed as clinically non-BU). DSE was positive for AFB for 37 participants (16. 6%) (Table 1). On bivariate analysis we confirmed that most of the WHO clinical and epidemiological criteria of BU were indeed associated with a positive PCR result (Table 1). Age ≤ 15 years was significantly associated with a positive PCR result with an OR of 5. 8 (95%CI: 3. 2–10. 3). Painless lesions were also significantly associated with a positive PCR result with an OR of 9. 1 (95%CI: 4. 7–17. 8). For factors related to ulcerated lesions such as “necrotic base” and “undermined edge”, the associations were also strong and statistically significant with an OR of respectively 7. 4 (95%CI: 2. 5–22. 0) and 5. 1 (95%CI: 2. 3–11. 3). There was no statistically significant association with the localization of the lesions on the upper or lower limbs (OR: 0. 8,95%CI: 0. 3–2. 0) while a localization on the lower limbs was negatively associated with a PCR confirmed BU (OR: 0. 4,95%CI: 0. 3–0. 8) and a localization on the upper limbs was positively associated with a PCR confirmed BU (OR: 2. 4,95%IC: 1. 3–4. 5). There was no association with “absence of satellite adenopathy” and “absence of fever”. Among the additional characteristics studied, the characteristic smell of ulcerated BU lesions was strongly associated with a positive PCR result with an OR of 16. 4 (95%CI: 7. 5–35. 6), as was the initial clinical BU diagnosis made by the clinicians (OR: 17. 8,95%CI: 8. 2–38. 7), essentially a synthesis of the WHO criteria and clinical experience. A lesion size between 5–15 cm (WHO category 2) had a significant positive association with a positive PCR result with an OR of 2. 35 (95%IC: 1. 30–4. 26). Gender and functional limitation were not associated with a positive PCR result (Table 1). There was no effect modification among the biologically plausible interactions we tested for (characteristic smell/necrotic base and characteristic smell/WHO category 3). In a multivariate predictive logistic regression model exploring the associations between the WHO criteria and a positive PCR result, only “age ≤ 15years”, “absence of pain” and “necrotic base of ulcerative lesion” were retained with ORs of respectively 3. 3 (95%IC: 1. 6–6. 7), 5. 9 (95%IC: 2. 7–12. 8) and 3. 7 (95%IC: 1. 1–12. 2). In a second multivariate predictive model, exploring all variables associated with a positive PCR result at a p <0. 10 on univariate analysis, only the clinical criteria “characteristic smell”, “necrotic base” and “WHO category 2” were retained (Table 2). To estimate the discriminative ability of both predictive models A and B, we estimated the areas under their ROC curves and found that model A (AUC: 0. 82,95%CI: 0. 76–0. 88) discriminated equally well as model B (AUC: 0. 85,95%CI: 0. 79–0. 91) between BU and non-BU patients (p = 0. 1940) (Fig 1). The current BU diagnostic criteria can benefit from revision by differentiating between lesions on the upper and lower limbs and by including lesion size and the characteristic smell recognized by experienced clinicians. Although the characteristic BU smell, strongly associated with a positive PCR result, will be difficult to describe in guidelines that need to be understandable to non-experienced clinicians. Further studies need to clarify if M. ulcerans indeed releases specific volatiles that can serve for the development of Point-of-Care diagnostic tests useful for non-invasive confirmation during active case-finding activities. Reproduced BU volatiles, if safe, may moreover serve for training purposes. Both could be important tools for health care workers, especially in the present context of decreasing BU incidence.

Buruli ulcer (BU) is a neglected necrotizing skin disease caused by Mycobacterium ulcerans. The treatment with BU-specific antibiotics is initiated after clinical suspicion based on WHO diagnostic criteria. In this study we evaluated the WHO diagnostic guidelines for BU and how these criteria could be improved. A total of 224 patients presenting with skin lesions were recruited in two BU treatment centers in southern Benin between March 2012 and March 2015. Most of the clinical and epidemiological WHO criteria were associated with a confirmed BU diagnosis although lesions on the lower limbs were rather associated with a negative PCR result. Among the additional characteristics studied, the characteristic smell of BU was most strongly associated with a positive PCR result. The WHO diagnostic criteria could therefore be improved upon by discriminating between lesions on the upper and lower limbs and by including lesion size and the characteristic smell recognized by experienced clinicians. The volatiles responsible for this smell could serve as a Point-of-Care diagnostic test, useful for non-invasive confirmation during active case-finding activities, and for training of clinicians.

Abstract Introduction Methods Results Discussion

smell medicine and health sciences pathology and laboratory medicine carcinomas tropical diseases cancers and neoplasms social sciences geographical locations neuroscience oncology bacterial diseases mathematics signs and symptoms forecasting statistics (mathematics) benin head and neck tumors neglected tropical diseases africa research and analysis methods head and neck squamous cell carcinoma infectious diseases buruli ulcer epidemiology mathematical and statistical techniques lesions people and places head and neck cancers psychology diagnostic medicine biology and life sciences squamous cell carcinomas sensory perception physical sciences statistical methods

2018

Improving clinical and epidemiological predictors of Buruli ulcer

A molecular diagnostic platform with DANP-anchored hairpin primer was developed and evaluated for the rapid and cost-effective detection of Chikungunya virus (CHIKV) with high sensitivity and specificity. The molecule 2,7-diamino-1,8-naphthyridine (DANP) binds to a cytosine-bulge and emits fluorescence at 450 nm when it is excited by 400 nm light. Thus, by measuring the decline in fluorescence emitted from DANP—primer complexes after PCR reaction, we could monitor the PCR progress. By adapting this property of DANP, we have previously developed the first generation DANP-coupled hairpin RT-PCR assay. In the current study, we improved the assay performance by conjugating the DANP molecule covalently onto the hairpin primer to fix the DANP/primer ratio at 1: 1; and adjusting the excitation emission wavelength to 365/430 nm to minimize the background signal and a ‘turn-on’ system is achieved. After optimizing the PCR cycle number to 30, we not only shortened the total assay turnaround time to 60 minutes, but also further reduced the background fluorescence. The detection limit of our assay was 0. 001 PFU per reaction. The DANP-anchored hairpin primer, targeting nsP2 gene of CHIKV genome, is highly specific to CHIKV, having no cross-reactivity to a panel of other RNA viruses tested. In conclusion, we report here a molecular diagnostic assay that is sensitive, specific, rapid and cost effective for CHIKV detection and can be performed where no real time PCR instrumentation is required. Our results from patient samples indicated 93. 62% sensitivity and 100% specificity of this method, ensuring that it can be a useful tool for rapid detection of CHIKV for outbreaks in many parts of the world. Chikungunya virus (CHIKV) is an arthropod-borne virus transmitted to humans primarily via the bite of an infected [1] Aedes agypti and Aedes albopictus mosquito. [2,3] Currently, there are more than 40 countries including Africa, United States, European countries and Southeast Asian countries affected by chikungunya fever. [2] CHIKV is an enveloped positive-sense single stranded RNA virus belonging to Alphavirus genus of Togaviridae family. [4] The genome is approximately 11. 8 Kb long, encoding four non-structural proteins (nsP1, nsP2, nsP3, nsP4) and five structural proteins (C, E3, E2,6K and E1). [5] The clinical symptoms of chikungunya fever are similar to that of dengue fever which is caused by Dengue virus (DENV), an arthropod virus belonging to Flaviviriridae family transmitted by same vectors as CHIKV. [6] This may result in cases of misdiagnosis in places where both viruses co-exist. As there is no vaccine or specific therapeutic agent available for CHIKV infection, early diagnosis of CHIKV is crucial in preventing the collapse of health care system due to unprecedented number of cases usually encountered during CHIKV epidemics. [7] Virus isolation is classified as the gold standard in detection of CHIKV despite being a time-consuming process requiring 1–2 weeks to determine the presence of virus. The limitations associated with virus isolation resulted in the development of serological and molecular diagnostic methods that are rapid and less labour intensive. Enzyme-linked-immunosorbent assay (ELISA) and Immunochromatographic test (ICT) are examples of serological diagnostic assays which detect IgM and/or IgG antibodies that are specific to CHIKV present in patient sera. ELISA and ICT tests are inexpensive and easy to perform as they do not require handling live viruses. A four-fold increase in antibodies by comparing acute phase and convalescent phase serum samples is usually required to confirm CHIKV infection. IgM is detected on an average of two days after infection and persists for several weeks to three months, while IgG is detected in convalescent samples and may persist for years. [8] The outcome of having antibodies present in serum samples after recovery phase may deduce as false-positive detection. Blacksell and co-workers reported that commercially available antibody-based assays are not suitable for acute diagnosis of CHIKV as the results obtained showed ICT and ELISA kits having sensitivity of 1. 9–3. 9% and 3. 9% respectively. [9] An alternative serological method of anti-CHIKV antibody detection has been reported to be used in commercial ELISA kits, but has shown cross-reactivity with other alphaviruses such as Ross River and O’ nyong-yong viruses as they are closely related serologically. [9] Thus, serological methods for CHIKV detection have been inefficient for acute phase diagnosis. [9–11] Recently, molecular diagnosis has been well established for rapid, highly sensitive and specific detection of CHIKV infection during the acute phase. Viral RNA is extracted from serum samples collected 1–7 days post-infection [8] were detected by primers targeting the conserved regions of Chikungunya genome specifically. In comparison, conventional RT-PCR appears to be a less sensitive and relatively more time-consuming process than TaqMan and SYBR Green I-based real-time RT-PCR assays. However, real-time RT-PCR assays require highly sophisticated instruments with yearly maintenance and calibration, restricting the utilization of such assays in places with poor financial and technical resources. [12] Previously, we have reported a novel diagnostic assay for CHIKV detection by adapting hairpin primers and fluorescent molecule, 2,7-Diamino-1,8-naphthyridine (DANP), into a conventional PCR procedure. [13] In brief, DANP molecule contains a naphthyridine ring which enables it to bind specifically to a cytosine-bulge in a hairpin structure of the PCR primer by hydrogen bonds. [13] The binding of DANP molecule to DNA gives rise to a 400 nm excitation and 450 nm emission property to the bound DANP molecule. As PCR proceeds, the primer is incorporated into double stranded DNA and the hairpin is opened, causing the release of DNAP molecule and thereby decreasing the fluorescence intensity. [13] The utilization of DANP coupled hairpin PCR has also been demonstrated in a single-nucleotide polymorphism study of the cytochrome P450 gene 2C9*3 by Takei and colleagues. [14] However, the binding of DANP molecule to the hairpin-primer is in an equilibrium manner, so that excess DANP molecules must be added to ensure a detectable fluorescence intensity. Therefore, the background signal given off by unbound DANP molecules limits the sensitivity and consistency of the assay. In the present study, DANP molecule was covalently immobilized on the hairpin PCR primers containing C-G base-pairs directly after the C-bulge to quench the fluorescence emission, as shown in Fig 1A. As PCR progresses, the hairpin structure is opened up and the DANP molecule is moved to the outer surface of the double-stranded DNA molecule, away from cytosine-bulge, resulting in an increase in fluorescence emission at 430 nm when it is subjected to UV-light at 365 nm. Increments in fluorescence intensity can be picked up only if the viral RNA template is present in the reaction with negligible background signal as no excess DANP molecules were added to the reaction. The method is highly effective as it uses a conventional RT-PCR protocol followed by measurement of fluorescence signal intensity using a spectrophotometer. The assay is more rapid and cost-effective as compared to real-time PCR methods. The assay was also validated with CHIKV infected patient serum samples and healthy individual serum samples for its sensitivity and specificity. CHIKV (GenBank accession No. FJ445502) was isolated from an infected patient during the CHIKV outbreak in Singapore in 2008. The virus was propagated in Aedes albopictus C6/36 cells. Briefly, cells were grown to about 80% confluency in T75 tissue culture flasks. Following removal of the growth media, virus inoculum was added to give a multiplicity of infection (MOI) of 0. 1 PFU/cell. Flasks were incubated at 28°C for 1 hours with constant agitation at every 15 min interval. After the incubation, Rosewell Park Memorial Institute (RPMI) 1640 growth medium (Sigma-Aldrich Corp) supplemented with 2% FBS (Hyclone) was added and flasks were maintained at 28°C for about 3–5 days or until cells showed 80% cytophatic effects (CPE). The viral titers were determined by plaque forming assay. [13] Ross River virus (RRV), Sindbis virus (SINV), Kunjin virus (KUNV, MRM 61C strain), West Nile virus (WNV, Sarafend strain), Zika virus (ZIKV, MR 766 strain), DENV-1 (S144 strain), DENV-2 (New Guinea C strain), DENV-3 (Eden 130/05 strain), DENV-4 (S8976 strain), Influenza A virus subtype H1N1, H3N2, Poliovirus type 1 (PV1, Sabin strain), type 2 (PV2, Sabin strain), type 3 (PV3, Sabin strain), Human enterovirus 71 (HEV71, AF316321 strain), Coxsackie B2 virus (CB2), Coxsackie A16 virus (CA16, WHO strain) and Enteric cytopathic human orphan virus 7 (Echo7) were also used to examine the cross-reactivity of this assay. The ZIKV, DENV1-4, PV1-3, HEV71, CB2, CA16 and Echo7 viruses were maintained in the laboratory. The RRV, KUNV and WNV were kindly provided by Professor Mary Mah-Lee Ng, Department of Microbiology, National University of Singapore. The Influenza A viruses were kindly provided by Associate Professor Tan Yee Joo, Department of Microbiology, National University of Singapore. A set of 22 serum samples from CHIKV-infected patients, and 30 from uninfected individuals were collected at the National University Hospital, Singapore, with informed consent, to evaluate the clinical sensitivity and specificity of the DANP-anchored assay. All of the sera were confirmed as febrile illness associated with a positive result from the real-time RT-PCR. [15] This part of the study was performed in accordance with the National University of Singapore Institutional Review Board approved protocol (No. 10–234). Environmental Health Institute (EHI), National Environmental Agency of Singapore kindly provided a set of 25 serum samples from clinically-suspected patients in which the presence of CHIKV was confirmed by a real-time RT-PCR assay. [16] Written informed consent was given for all samples involved in this study. CHIKV RNA was extracted from 140 μL of infected cell culture supernatants (3. 6 X 10^7 PFU/mL) and serum samples using the QIAamp viral RNA mini kit (Qiagen, Hilden, Germany) according to the manufacturer’s instructions. The RNA was eluted in a final volume of 50 μL of nuclease-free water and stored at −80°C until use. The full genomes of multiple geographically different strains of CHIKV from recent outbreaks were retrieved from GenBank and aligned using the ClustalX (version 2. 1) [17] sequence alignment software. Primers were designed to target the highly-conserved nsP2 regions of CHIKV genome, as shown in Table 1. The primers were designed with hairpin (underlined sequences in Table 1) at the 5’ end to accommodate DANP molecule which is covalently conjugated to the thymine nucleotide (bolded sequences in Table 1) of the primers. RT-PCR reactions were performed in C1000 thermal cycler (Bio-Rad, Hercules, CA). Reactions were optimized with a One Step RT-PCR kit (Biotech Rabbit, Hannover, Germany). Each reaction was performed in 25 μL total reaction mixture containing 12. 5 μL of 2x reaction buffer, 0. 2 μmol/L of each forward and reverse DANP hairpin primers, 1. 25 μL of 20x RT-RI blend (reverse transcriptase and RNAse Inhibitor) and 1 μL of viral RNA. 10 μL from each total reaction volume was set aside while the remaining 15 uL of reaction was subjected to RT-PCR. The thermal profile was optimized as follows; reverse transcription step at 45°C for 20 minutes, activation of Taq polymerase at 95°C for 2 minutes, followed by 30 cycles of PCR cycling steps consisting of 95°C for 10 seconds, 60°C for 10 seconds and 72°C for 15 seconds. In order to determine the fluorescence intensity, 10 μL of each reaction was diluted with 90 μL of nuclease-free water in each well of a white opaque flat-bottom 96-well plate. The fluorescence intensity from each well was scanned by Infinite® 200 PRO microplate reader (Tecan Trading AG, Switzerland) with an excitation filter at 365-nm and an emission filter at 430-nm. A sample positive for CHIKV infection was determined as the increment in fluorescence intensity after PCR was more than 100 arbitrary units (AU) as compared to background fluorescence in pre-PCR reaction mixture. For assay validation, all PCR products were analysed using 8% native polyacrylamide gel electrophoresis (PAGE), followed by ethidium bromide staining for two minutes. Gel images were captured using the GeneSnap software version 7. 02 (Syngene, Cambridge, UK). In order to determine the number of PCR cycles that gives off the most significant increment in fluorescence intensity after PCR, the fluorescence intensity level was measured and compared after every five PCR cycles using CHIKV genomic RNA as positive control and nuclease-free water as negative control (NTC). CHIKV RNA was extracted from 140 μL of infected cell culture supernatants with viral titre of 3. 6 X 10^7 PFU/mL, using the QIAamp viral RNA mini kit (Qiagen, Hilden, Germany) according to the manufacturer’s instructions. The viral RNA was eluted in a final volume of 50 μL of nuclease-free water and then serial diluted logarithmically until a final concentration of 1 X 10−5 PFU/μL. 1 μL of each of the serial diluted viral RNA samples was subjected to the DANP-anchored RT-PCR assay to determine the limit of detection of the assay. The RNA concentration ranges tested were 1. 0 X 101 to 1. 0 X 10−5 PFU/reaction (3. 6 X 103 to 3. 6 X 10−3 PFU/mL). RRV, SINV, KUNV, WNV, ZIKV, DENV1-4, Influenza H1N1, H3N2, PV1-3, HEV71, CB2, CA16 and Echo7 were used to examine the cross-reactivity of this assay. Viral RNA was extracted from 140 μL of each of the viruses and eluted in 50 μL of nuclease-free water. The RNA concentration were measured using NanoDrop ND2000 Spectrophotometer (NanoDrop Technologies, Inc. , Wilmington, DE, USA) and only samples with at least 50 ng/μL were proceeded to cross reactivity study of the assay. In order to determine the number of PCR cycles that gives off the most significant increment in fluorescence intensity after PCR, the fluorescence intensity level was measured and compared after every five PCR cycles using CHIKV genomic RNA as positive control and nuclease-free water as negative control (NTC). As shown in Fig 2, the difference in fluorescence intensity between before PCR and after PCR samples reached the maximum at 30 cycles. Due to the formation of primer dimer and non-specific PCR products in the NTC, the difference in fluorescence intensity began to narrow down after 30 cycles. Therefore, 30 cycles of PCR reaction was used in the rest of the study. To validate the suitability of hairpin primers and to verify the initial fluorescence intensity level, DANP-anchored hairpin RT-PCR procedure was carried out with and without CHIKV genomic RNA. All PCR products were analyzed by PAGE to determine the assay specificity. As indicated in Fig 3A, the specific PCR product of 296 bps can only be seen when CHIKV RNA is present. There was no significant change in fluorescence intensity between before/after PCR in NTC reactions (Fig 3B). In contrast, an increment of more than 2000 AU of fluorescence intensity was observed after 30 cycles of PCR in the presence of CHIKV RNA. The detection limit of the DANP-anchored RT-PCR assay was determined through replicates of reactions, using serial logarithmic dilutions of the control CHIKV genomic RNA. Fig 4 shows the change in fluorescence intensity before and after 30 cycles of PCR reaction. A statistically significant increase of 120 AU was observed in 0. 001 PFU per reaction, the lowest level of detection by the assay. The cross-reactivity of the assay was determined by using a panel of other RNA viruses. RRV and SINV were used as representative members of Alphavirus; KUNV, WNV, ZIKV and DENV-1, DENV-2, DENV-3, and DENV-4 were used as representative members of Flavivirus in the cross-reactivity study. The remaining RNA viruses included H1N1, H3N2, Polio 1, Polio 2, Polio 3, HEV71, CB2, CA16 and ECHO7. Only CHIKV RNA samples demonstrated positive results, indicating the lack of cross-reactivity with other viruses tested (Fig 5). In order to evaluate the sensitivity and specificity of the present assay for clinical diagnosis, 47 serum samples obtained from patients with confirmed CHIKV infection during the acute phase and 30 serum samples from uninfected individuals were tested. The present assay demonstrated high sensitivity by picking up 44 of the 47 CHIKV cases (93. 62% sensitivity; 95% CI, 81. 44% to 98. 37%). None of the 30 serum samples from uninfected individuals was false diagnosed as positive (100% detection specificity; 95% CI, 85. 87% to 100%) (Table 2). CHIKV has been relatively understudied as it was restricted to Africa and Asia countries. [18] Since 2005, CHIKV started to spread to countries in Indian Ocean and then globally. [19] It is estimated that > 1. 5 million people were infected in India during the 2006 outbreak alone, [20] and, currently, Chikungunya fever has been documented in more than 40 countries. [21] The unprecedented worldwide spread of CHIKV was driven by international travel and the A226V mutation on the envelope protein 1, which better adapts the virus to Aedes albopictus. [3] Due to increased globalization and mosquito vectors expand to new areas, early diagnosis of CHIKV is critical in the absence of any licensed antiviral therapy and prophylaxis, especially in developing countries. Currently, the diagnosis of CHIKV largely relies on virus isolation, detection of specific antibody and nucleic acid. Virus isolation in tissue culture is time-consuming and technically complex that is limited in developing countries. Because of extensive cross-reaction between alphaviruses due to common antigens, serological assays often face the difficulty in differentiating commonly occurring alphaviruses. These drawbacks have made molecular assays the method of choice for diagnosis during acute phase of chikungunya fever. Molecular techniques based on the detection of genomic sequences by RT-PCR, nested RT-PCR, and real-time RT-PCR are rapid and sensitive and have replaced virus isolation as the new standard method for the detection of CHIKV in acute-phase serum samples, but the reagents and equipment are too costly for widespread use. In this regard, this DANP-anchored RT-PCR assay reported in this study is advantageous, because of its simplicity, rapidity, and cost-effectiveness. Only a standard conventional PCR procedure, with DANP hairpin primer used, and a fluorescence reading procedure is required without involving of sophisticated instrument or costly reagent. In comparison, previously, we have reported a novel DANP-coupled hairpin RT-PCR for rapid detection of CHIKV in the acute phase serum samples. PCR primers were designed specifically to target nsP2 gene of CHIKV with hairpin tag containing a cytosine-bulge. The DANP molecule binds to the C-bulge in its protonated form (DANPH+) before PCR reaction starts, resulting in fluorescence emission. During PCR amplification, the hairpin primer opens up and releases the DANP molecule resulting in a drop in fluorescence emission giving rise to a ‘turn-off’ system. CHIKV positive samples are determined by comparing the fluorescence intensity recorded before and after PCR process by subjecting the PCR products to UV-light and detecting the emitted fluorescence at 430 nm. Despite it was a rapid, sensitive, specific and cost effective assay; the optimization of DANP concentration due to background signal restricted its usage. In order to overcome the issue, in the present study, we covalently conjugated the DANP molecule onto the hairpin structure of the PCR primer. Therefore, the ratio between DANP molecule and primer is fixed at 1: 1 and this standardization simplifies the optimization of the assay. In addition, by changing the reading spectrum from 400 nm excitation and 450 nm emission to 365 nm excitation and 430 nm emission, we managed not only to minimize the background signal but also give rise to a ‘turn-on’ system. In addition, we have also shortened the assay turnaround time from 90 minutes to 60 minutes including fluorescence reading, by cutting down the reverse transcription step duration and optimizing the PCR cycle number from 40 to 30. The detection limit of the assay was 0. 001 PFU per reaction that is lower than that of the previous DANP coupled assay [13] and is comparable to real time RT-PCR assays developed by other groups [22,23]. A side-by-side comparison of our assay with the abTES DEN 5 qPCR I Kit (Cat: 300152) from AIT biotech, a Taqman probe-based multiplex real time RT-PCR for DENV/CHIKV detection. Comparable limit of detection was noted. Despite the fact that the viremia load is usually above 4 log10 during the acute phase of CHIKV infection, [24] low detection limit of the assay enables us to detect CHIKV RNA even during late acute phase when the viral titers start to decline rapidly. More importantly, the present assay is not cross-reactive with a panel of RNA viruses that are co-circulating in endemic regions. Given that CHIKV is commonly misdiagnosed as DENV and vice versa, the outstanding specificity of our assay could benefit both clinicians and patients at the point-of-care by providing accurate diagnosis.

Chikungunya has reemerged as an important mosquito-borne infection with global health significance. Rapid diagnosis plays an important role in early clinical management of patients due to lack of a vaccine and effective treatment. Laboratory diagnosis is generally accomplished by blood tests to detect virus-specific antibodies but these antibodies are usually developed one week after infection, which misses the window of effective clinical management. On the other hand, although detecting the viral genome can be done in early stage of infection by real-time polymerase chain reaction (PCR) but it is costly to the patients. Here we utilized a fluorescent compound to improve the cost-efficiency of the molecular assay for diagnosis of Chikungunya virus infection. By testing on 77 serum samples, this improved assay has proven to be highly sensitive and specific towards Chikungunya virus. We believe that this research could benefit both clinicians and patients by providing early and accurate diagnosis.

Abstract Introduction Materials and Methods Results Discussion

reverse transcriptase-polymerase chain reaction medicine and health sciences enzyme-linked immunoassays fluorescence pathology and laboratory medicine togaviruses chikungunya infection pathogens immunology tropical diseases microbiology rna extraction alphaviruses electromagnetic radiation viruses chikungunya virus rna viruses neglected tropical diseases molecular biology techniques immunologic techniques cross reactivity extraction techniques research and analysis methods infectious diseases artificial gene amplification and extension medical microbiology microbial pathogens immunoassays luminescence molecular biology physics polymerase chain reaction viral pathogens biology and life sciences viral diseases physical sciences organisms

2016

Development of 2, 7-Diamino-1, 8-Naphthyridine (DANP) Anchored Hairpin Primers for RT-PCR Detection of Chikungunya Virus Infection

The analysis of microcircuitry (the connectivity at the level of individual neuronal processes and synapses), which is indispensable for our understanding of brain function, is based on serial transmission electron microscopy (TEM) or one of its modern variants. Due to technical limitations, most previous studies that used serial TEM recorded relatively small stacks of individual neurons. As a result, our knowledge of microcircuitry in any nervous system is very limited. We applied the software package TrakEM2 to reconstruct neuronal microcircuitry from TEM sections of a small brain, the early larval brain of Drosophila melanogaster. TrakEM2 enables us to embed the analysis of the TEM image volumes at the microcircuit level into a light microscopically derived neuro-anatomical framework, by registering confocal stacks containing sparsely labeled neural structures with the TEM image volume. We imaged two sets of serial TEM sections of the Drosophila first instar larval brain neuropile and one ventral nerve cord segment, and here report our first results pertaining to Drosophila brain microcircuitry. Terminal neurites fall into a small number of generic classes termed globular, varicose, axiform, and dendritiform. Globular and varicose neurites have large diameter segments that carry almost exclusively presynaptic sites. Dendritiform neurites are thin, highly branched processes that are almost exclusively postsynaptic. Due to the high branching density of dendritiform fibers and the fact that synapses are polyadic, neurites are highly interconnected even within small neuropile volumes. We describe the network motifs most frequently encountered in the Drosophila neuropile. Our study introduces an approach towards a comprehensive anatomical reconstruction of neuronal microcircuitry and delivers microcircuitry comparisons between vertebrate and insect neuropile. The brain of all higher animals is formed by a large number of interconnected neurons. Typically, neurons are grouped into larger assemblies (“brain compartments”), such as brain stem nuclei or cortical layers in the vertebrate brain, or neural lineages in the insect brain [1], [2]. The analysis of the structure, development, and function of the brain can therefore proceed at two levels: the level of individual neurons and synapses, and the level of brain compartments. Compartments represent structural and functional modules; interconnected by bundles of axons, they form “macro-circuits” that control certain aspects of behavior. Unraveling macro-circuits has been the mainstay of classical vertebrate neuroanatomy and physiology. Present-day studies employing functional imaging (e. g. , [3], [4]) walk in the foot steps of this approach, given that the signals registered by MRI or PET scanners (for these and all other abbreviations, see Table 1) reflect the activity of large numbers of contiguous cells [5]. The study of macrocircuitry informs us of how the brain is built, which “packets of information” may interact, where in the brain this interaction takes place, and what output channels are activated to elicit a behavior that is correlated with the observed macroscopic brain activity. Addressing macrocircuitry leaves the question of how nervous tissue operates in processing information unanswered. To tackle this problem, an approach is required that considers the structure and connectedness of the building blocks of the brain—i. e. , the neurons, neurites, and synapses (“microcircuitry”). The way in which a given neuron is tuned to a specific input stimulus, or the pattern of activity triggered in this neuron when providing a specific input, depends on the distribution of excitatory and inhibitory synapses that connect the neuron with its neighbors [6]–[9]. Given the small size of neurites and synapses, the number and density of connections is immense. Calculations based on both light- and electron microscopic preparations point out that in mammalian brain, 1 mm3 contains more than 105 neurons, more than 4 km of axon and 500 m of dendrite, and more than 700 million synapses [10]. The goal of the analysis of microcircuitry is to elucidate the geometric algorithms that describe the connectivity within small volumes of brains; based on these algorithms, one may hope we will be eventually able to model the neuronal activity and information flow pervading brain tissue while controlling sensory and motor activity of an organism. Due to the small size of synapses and terminal neurites (0. 1–0. 5 µm), structural aspects of microcircuitry can be conclusively analyzed only electron microscopically. Traditionally, the acquisition and analysis of complete series of TEM sections required a considerable effort; as a consequence, studies of microcircuitry have mostly been restricted to small parts of neurons or neuropile compartments in (e. g. , [11]–[14]). The problem is now becoming solvable, at least for small brains (or small volumes of large brains) with digital image recording and specialized software for both image acquisition and post-processing [15]–[18]. The resulting stacks of registered digital sections can be segmented and analyzed in their entirety. For the implementation of such a neuronal reconstruction pipeline, aimed at the analysis of microcircuitry of the Drosophila larval brain, we used the TrakEM2 open source software. Drosophila serves as a favorable model system in which molecular pathways involved in a wide range of events, from neural stem cell proliferation, cell fate determination, neurite pathfinding, and neurite connectivity, can be studied (e. g. , [19]–[22]). With the help of sophisticated transgenic constructs (e. g. , Gal4 lines; [23]), one can target specific cell types, labeling these cells, or manipulating them genetically. As a result, Drosophila also represents a great model system to study systems-level questions, such as how neural circuits develop or control behavior. Finally, Drosophila (and insects in general) offers the advantage that its nervous system is formed by a relatively small number of genetically and structurally defined modules, the neural lineages [1], [24]–[25]. Early in development, a set of dedicated neural progenitors, called neuroblasts, segregate from the ectoderm and subsequently proliferate to form the neurons and glial cells of the CNS. Each neuroblast produces an invariant set (“lineage”) of neurons; these neurons form a genetic as well as a structural “module” of the brain (Figure 1). Thus, axons of neurons belonging to the same lineage form a cohesive tract and arborize within discrete compartments of the brain. The axon tracts formed by neural lineages, and recognized from early larval stages into the adult brain, represent the structural/developmental units of brain macrocircuitry. Lineage tracts represent an invariant, easily recognizable system of structural landmarks onto which smaller elements—i. e. , microcircuits or individual neurons—can be mapped. We argue that for the efficient analysis of microcircuit data, it is advantageous to follow an approach that integrates light microscopic landmark structures, such as lineage tracts (the “macrocircuitry”), with the analysis of TEM datasets. The reconstruction of microcircuitry from serial TEM material involves two main operations. The first is to segment the profiles of neurites through many consecutive sections to establish branch points and synaptic contacts. Given the amount of time involved in manual segmentation (or proofreading of automatically segmented material; [18]), there are currently two different strategies of serial TEM data analysis. One (“sparse strategy”) is to focus on a defined, relatively small set of neurons, such as sensory afferents, or motor neurons, and trace their profiles and synaptic outputs/inputs in their entirety. The other strategy, introduced in this article, is to break down the overall TEM stack volume into smaller “microvolumes, ” in the range of 2×2×1. 5 µm up to 5×5×5 µm, and reconstruct all profiles contained within these microvolumes (“dense strategy”). From these small volumes, structural parameters of terminal neurite connectivity (e. g. , diameter, orientation, and density of terminal neurites; structure, size, and density of synapses) can be extracted. Irrespective of the strategy (sparse versus dense segmentation) chosen, an additional step is to put the connectivity data resulting from the segmentation into the context of brain macrocircuitry: what brain compartment is the microcircuit located in, what are its input and output channels, how does it compare to microcircuits located at other positions in the brain. To be able to make these connections, the operator has to be able to repeatedly switch between the EM and LM level and to rapidly navigate from one location/compartment to another. The TrakEM2 software used in this study combines all the components required for data acquisition, registration, navigation, and segmentation ([26], [27]; http: //www. ini. uzh. ch/~acardona/trakem2. html). TrakEM2 enables the analysis of a brain volume at the microcircuit level (synapses, individual neurite branches) along with the macrocircuit level (brain compartments, axon fascicles). We illustrate the use of TrakEM2 towards this end by applying it to the brain of the first instar (L1) larval Drosophila brain. As raw data we use two stacks of stitched and registered EM sections, one containing 500 sections that include most of the neuropile of one brain hemisphere, and another one of 250 sections including one complete abdominal neuromere of the ventral nerve cord. We have cropped out several small volumes and evaluated the structure and connectivity of the terminal neurites contained within these volumes. Furthermore, we have registered a confocal stack with the brain TEM stack and evaluated the accuracy with which we can translate light microscopic features from the confocal stack to the TEM stack. Our study is intended to develop an approach towards a comprehensive anatomical reconstruction of neuronal microcircuitry. Freshly dissected first instar fly brains were fixed in 4% paraformaldehyde and 2. 5% glutaraldehyde in 0. 1 M freshly made phosphate buffer (PB; ph 7. 3) for 24 h at 4°C. Brains were then rinsed 5×10 min in 0. 1 M PB, postfixed in 1% osmium tetraoxyde in 0. 2 M PB for 60 min at 4°C, and rinsed 4×10 min in distilled water. Dehydration was done through an acetone series (10 min 50%, 10 min 70%, 10 min 96%, 3×10 min 100%). Preparations were embedded in Epon resin at room temperature as follows: 2 h in 1∶3 Epon: acetone, 3 h 2∶2, overnight 3∶1, overnight pure Epon. Blocks were cured for 16 h at 60°C. 60 nm serial sections were cut on a Leica UC8 ultratome and collected in ribbons onto pioloform-coated single slot copper grids. Grids were contrasted in 8% uranyl acetate (30 min at 60°C) and in Reynold' s lead citrate [28]. Larval brains of a line in which the neurons expressing pigment dispersing factor (PDF) were labeled by a Gal4-driven GFP reporter were dissected in PBS and fixed in PBT (PBS with 0. 1% Triton X-100) containing 4% formaldehyde for 30 min at room temperature. A mouse anti-Neurotactin antibody ([29]; 1∶10; Hybridoma Bank) was used to label neurons and axon tracts. Secondary antibodies were purchased from Jackson Laboratory and used at the manufacturer' s recommended concentrations. Stained brains were mounted in Vectashield (Vector Laboratory; H-1000). Confocal images were taken on a Biorad MRC1024ES microscope using Laser sharp version 3. 2 software. Complete series of optical sections were taken with a 40× oil immersion lens at 1 µm intervals. Images were analyzed using the ImageJ software. The imaging of hundreds of sections at sufficient resolution to resolve details like synapses used to pose a major problem in high-throughput transmission electron microscopy (TEM). The software package Leginon [30] automates TEM image acquisition of multiple sections and large tissue areas. We used an FEI electron microscope equipped with a Tietz camera and a goniometer-powered mobile grid stage. Leginon software automatically controls every component of the electron microscope and attached camera. The acquisition starts by inserting a grid with about 10 serial sections into the microscope. Leginon images the entire slot of the grid at low resolution and offers a grid atlas for manually or semi-automatically picking the tissue areas of interest in all sections. Then Leginon automatically adjusts the stage, power, magnification, and camera to acquire the necessary sets of high-resolution image tiles that cover all areas of interest. For our larval neuropile sections, we chose a magnification of 5,600× binned at 2, delivering a resolution of 4 nm/pixel. Acquired image tiles carry associated stage position coordinates, but these alone would result in suboptimal montages. First, we correct for lens deformations, which are constant to all tiles. For the purpose, we extract scale-invariant-feature-transform (SIFT) features [31] from nine heavily overlapping images and then estimate and apply the correcting transformation to all tiles [32]. The correction of lens deformation greatly facilitates the montaging of image tiles. Each section presents independent non-linear deformations, generated during sectioning and also induced by heat while imaging. No ground-truth is available for the correction of these deformations. Our observations indicate that, generally, section-wide gross deformations in consecutive sections are mostly independent and that local non-linear deformations contain translations smaller than the dimensions of the image tiles. An approach that registers image tiles all-to-all within and across sections would, to a considerable extent, cancel out all independent deformations. We approximate this ideal registration with an as-rigid-as-possible tile-wise registration method. We extract SIFT features from all image tiles and search for correspondences with their neighboring tiles within the section and with tiles in the previous and subsequent sections in the series. We estimate a rigid transformation model for each tile, simultaneously from all tile-to-tile feature correspondences. An iterative optimizer relaxes all tile-to-tile correspondences until the sum of their square inter-distances is minimal. From this configuration, we estimate a non-linear transformation for each individual section using the Moving Least Squared method [33] with tile centers as control points. With our approach, imaged objects (such as sectioned neural arbors) that span over multiple tiles within one tissue section are as smoothly continuous over tile boundaries as possible, while preserving maximum continuity across sections with minimal deformation applied to each individual image [27]. The storage, retrieval, and navigation of a terabyte-sized dataset of TEM images by a human operator in real time is a difficult and time-consuming task, even when considering small objects like the Drosophila L1 brain. We have found a successful approach in the combination of two main factors: (1) the means to browse through large areas and volumes at sufficiently high speed, and (2) the ability to zoom in and out at very high speed, for the purpose of obtaining a positional reference. We approach browsing with mipmaps, that is, precomputed multiresolution images, which match the current view magnification and are thus optimal for data transfer and display [28], [34]. Our immediate goal in analyzing the TEM stack was 2-fold: on the one hand, we want to crop out small volumes, in the range of 5 µm across, for which we manually dense-segment every process and synapse. Secondly, we wanted to identify and segment larger structures, such as axon bundles and compartment boundaries, which together form a framework of landmarks to which the microvolumes could be related. To efficiently segment these different structures by hand, we employed a small set of segmentation data types: an “area list” (a list of 2D multi-area objects, one per section, can represent individual neurites but also large objects like compartments), a “ball” (a list of x, y points with a radius, sitting on any section, can represent a specific landmark, like point of intersection of tracts), a “pipe” (a serial sequence of points, each point with an x, y coordinate on a specific section, and with a radius that defines a tube passing through all points, can represent axon tract), and a series of other convenience objects like floating text labels (with associated x, y coordinates and on a specific section). Densely segmented volumes, even at the scale of microvolumes, contain a large number of separate objects that stand in specific relationships to each other. We call the objects of a given brain the “object hierarchy” of this brain. The structural elements of the object hierarchy are user-provided; the L1 brain, for example, contains, among others, the high-order elements “neuropile compartments, ” “neuropile tracts, ” “lineages, ” and “neurons. ” High-order elements include further, lower tiers of elements; a neuron, for example, forms primary branches A, B; these in turn have secondary branches A1, A2, … and B1, B2 … etc. Each branch forms a set of presynaptic and/or postsynaptic sites, through which a given branch of a given neuron communicates with a certain branch of another neuron. In other words, the effective manipulation of the multitude of objects included in a segmented brain (microvolume) demands a strategy that groups these objects into recursively smaller and smaller groups, effectively collapsing the complexity to an arbitrary level at which high-order elements (like “neurons” or “branches”) become the elements to operate on: to measure, to hide/show, to visualize in 3D, to color, to highlight, etc. The result is a hierarchal tree of abstract objects: a “neuron” is represented as a composition of lower-level objects like “soma” and “arbor, ” each of which in turn is composed of further lower-level objects like “nucleus, ” “cytoplasm, ” “axon, ” “dendrite, ” and so on, until reaching the level of primitive object instances (whose terminology follows a controlled vocabulary, such as “area list” or “pipe”). An “area list, ” for example, would represent the series of 2D sections of a nucleus or synapse, which has been created by manually or semi-automatically painting with mouse movements over TEM images. The object hierarchy window is set up in a manner that will ultimately allow one to export the data in “sif” format into programs designed to perform network analysis. Thus, tools like “Neuron” (http: //www. neuron. yale. edu/neuron/) are designed to accept data sets composed of large numbers of individual elements connected via defined synapses and compute neural circuit simulations using Hodgkin-Huxley models (integrate-and-fire models). TrakEM2 simultaneously presents two ways of browsing and manipulating the data: on the one hand, the tree hierarchy of abstract groupings, with the actual segmentation object instances at the tips of the tree (“object hierarchy window”), and on the other hand, a 2D display (“canvas”) for the manipulation of the data of the segmentation objects (to fill in an area, to point-and-click to add spheres, or lines, etc.) The 2D canvas is a view of the list of serial sections, which can be any series of images such as image stacks from confocal microscopy or registered histological or TEM serial sections. Via the 2D display, the images themselves are manipulated. The key operations for serial section TEM are batch-importing, image stitching within a section, registration of adjacent sections, and contrast adjustment. Our image registration approach preserves as much as possible the dimensions of each individual image, avoiding the introduction of artifactual image deformations [27]. Both image volumes and segmented objects may be visualized in 3D for spatial analysis, using the ImageJ “3D Viewer” [35]. One of the requirements for an efficient serial section TEM analysis is the recognition of structures that have already been characterized at the light microscopic level. These structures provide the context for the analysis of microstructures like synapses and statistics of arborizations, etc. Currently, no algorithms exist for generic automatic cross-modal image registration; that is, no unsupervised algorithm can recognize structures both at the TEM and at confocal images. TrakEM2 offers a simple 3D nonlinear image registration approach based on user-picked fiduciary marks, common between TEM and confocal images. From the fiduciary marks, a nonlinear transformation is estimated using the Moving Least Squares method [33] for 3D affine transformation. The confocal image stack is thus brought into register with the TEM sections. Then a re-sliced 50 nm section of the confocal image stack is overlaid on top of the TEM section, using color composites. We prepared an uninterrupted series of TEM sections of a small brain, the Drosophila first instar (L1) brain, recorded digital images of a complete brain hemisphere at a high enough resolution to reconstruct synapses and fine processes, and registered these images so they can be navigated like a confocal stack. The Drosophila L1 brain hemisphere has a diameter of approximately 50 µm. It consists of an outer cortex of neuronal cell bodies and a central neuropile containing the branched neurites and synapses (Figure 1A). The diameter of the neuropile measures 30 µm. In a first series of TEM sections we focused on the neuropile; 500 sections of 60 nm thickness included the entire neuropile of one brain hemisphere. In addition, we sectioned transverse slice of the ventral nerve cord, containing 250 sections. For image capturing, we used the software Leginon (Automated Molecular Imaging group at the Scripps Institute, San Diego, CA). To clearly resolve ultrastructural details such as neurotubules (about 25 nm diameter) and synaptic vesicles (30–40 nm diameter), we aimed at a resolution of approximately 3–4 nm per pixel, which can be achieved at a magnification of 3,000–5,000× (Figure 1). At that resolution, the complete data stack including the neuropile of one brain hemisphere amounts to approximately 1 terabyte; for the entire L1 brain, the size would be approximately 5 terabyte. Our strategy of “dense neuropile analysis” was to extract from the TEM image volume multiple smaller “microvolumes, ” in the range of 2×2×1. 5 µm up to 5×5×5 µm (Figure 2A), and reconstruct all profiles contained within these microvolumes. TrakEM2 allows us to efficiently crop out such microvolumes and re-register them automatically. Contained within a microvolume are the contiguous short segments and terminal branches of many neurons and their synaptic contacts (Figure 2F–H; Figures S1, S2). As shown below, the dense analysis of these objects can shed light on a number of fundamental parameters of microcircuitry. Furthermore, we argue that the reconstruction and analysis of objects from stacked TEM images should be guided by known “macroarchitectural” brain landmark structures. In other words, the interpretation of the pattern of neurites contained within a given microvolume will be made easier by establishing position, input, and output relationship of that volume (Figure 2C). TrakEM2 embeds the analysis of the TEM stack at the microcircuit level into a light microscopically derived three-dimensional framework of landmark structures. This framework is provided by the invariant pattern of compartment boundaries and lineage-related axon tracts, many of which have been identified in brains of all developmental stages ([36]–[40]; Figure 2D, E). Lineage related primary axon tracts (PATs) in a L1 brain contain 5–20 tightly bundled, straight axons that originate and terminate at defined positions. In many cases, PATs of several contiguous lineages converge and form large fascicles, such as the antenno-cerebral tract, peduncle, medial and lateral equatorial fascicle, or posterior-lateral fascicle. Fascicles are all associated with high concentrations of glial processes, which makes it easier to identify them in the TEM stack (Figure 2B). Glial densities also accompany many of the compartment boundaries [41]. Fascicles and segment boundaries (“macrolandmarks”) are segmented and become the objects of an “intrinsic macro-model” of the brain contained within the TEM stack. TrakEM2 allowed us to seamlessly transition from the light-microscopic (“macrocircuitry”) to ultrastructural level (“microcircuitry”) of brain analysis. The graphic user interface of TrakEM2 features three simultaneously active windows: the object hierarchy window, the 2D (raw data) canvas, and the 3D viewer (Figure 3). The digitized TEM stack of the larval brain is opened as a “project” in TrakEM2 and appears in the 2D canvas, where it can be navigated in the manner of a three-dimensional map [26]. Landmark structures that are segmented within the TEM stack are represented as nodes of a hierarchical tree in the object hierarchy window (Figure 3A). From within the object hierarchy window, individual objects can be activated or inactivated, and properties of their digital rendering, such as color or transparency, can be changed. An object (or any number of objects) activated from within in the object hierarchy window will appear as an outline overlaid upon the electron microscopic images visible in the 2D canvas window (Figure 3C). At the same time, a surface rendered 3D digital model of the object can be displayed in the 3D viewer (Figure 3D). Alternatively, an object can be activated from within the 2D canvas and is then identified in the object hierarchy window. As a result, we were able to focus at the ultrastructural level on any part of the neuropile in the 2D canvas and at the same time get orienting cues from the intrinsic macro-model (displayed in the 3D viewer) about the exact position relative to compartment boundaries and axon fascicles. TrakEM2 also offers the possibility of adding additional macrostructures imaged from other (“extrinsic”) larval brains. Thus, a confocal stack of a L1 brain in which a certain brain object (e. g. , cell type, or lineage), aside from the landmark structures, is labeled by antibody or reporter construct is opened as an additional, separate image volume. The landmark structures define a set of fiduciary marks that allow one to register the extrinsic confocal stack with the TEM stack. Following the registration, all structures that are labeled, or manually segmented as objects, in the extrinsic confocal project are overlaid with the TEM project. In the example shown in Figure 3E–J, the confocal stack of a brain in which the four neurons expressing the peptide PDF (pigment dispersing factor) were labeled was merged with the TEM stack. We used the points of intersection of 14 axon fascicles, with each other or compartment boundaries, as fiduciary marks for registering the confocal stack with the TEM stack (Figure 3E, F). To assess the accuracy of the registration, we first determined how closely the position of corresponding axon fascicles segmented from the TEM stack and the confocal stack matched. Shown in Figure 3G are the profiles of the MEF (medial equatorial fascicle) and dePF (descending protocerebral fascicle) as they appear in the TEM stack (outlined in magenta). Overlaid in blue (peduncle) and red (MEF) are the profiles of the corresponding fascicles that were segmented in the confocal stack and, following registration, overlaid upon the TEM stack. In most sections, as the one shown in Figure 3G, the profiles (which measure less than 2 µm in diameter) overlap. Profiles of the mushroom body peduncle (diameter: 5 µm) segmented from the confocal section overlap more than 50% with the peduncle derived from the TEM stack throughout all sections. The accuracy of registration can be further improved by refining and adding more fiduciary marks. Of course, one cannot expect that, employing the approach introduced here, a single neurite (whose diameter averages around 0. 25 µm) segmented from a confocal stack will “fall” precisely over the corresponding counterpart within the TEM stack. For that not to happen, variability between brains (at the level of individual neurons) in itself is reason enough. However, the import of labeled objects from confocal stacks will assist significantly in identifying specific neurons, as shown in the following for a small group of neurons, the PDF neurons. A GFP reporter expressed under the control of the PDF gene promoter [42] labels four neurons located in the dorso-lateral L1 brain hemisphere. Figure 3H and 3I illustrate the PDF neurons as they appear in a confocal section (3H) and overlaid upon the EM stack (3I). The main axons of the four neurons converge and fasciculate, extending as a thin transverse bundle along the dorso-lateral neuropile boundary. Dendritic branches of the PDF neurons branch off towards the larval optic neuropile (outside the plane of section); numerous short secondary and tertiary axonal branches are formed along the length of the PDF bundle into the CPLd compartment. PDF neurons belong to the class of peptidergic neurons, all of which are characterized ultrastructurally by prominent dense core vesicles filling the entire neurite tree. Peptidergic neurites innervate the neuropile quite sparsely; the overall number of peptidergic neurons in the L1 brain is in the order of 50 per hemisphere (reviewed in [43]), out of 1,500 neurons. Therefore, the profiles with dense core vesicles (Figure 3J) that we identify in the TEM stack in the domain occupied by the overlaid profiles of PDF neurons imported from the confocal stack most likely belong to the PDF neurons inherent to the TEM stack. We have segmented and reconstructed neurites in five microvolumes located in different compartments (calyx and spur of mushroom body, dorso-lateral protocerebrum, and dorso-lateral domain of ventral nerve cord). As outlined above, the dense reconstruction of microvolumes yields information regarding structural parameters such as neurite diameters, directionality, branching, and synapse placement. Our data show that brain ultrastructure is conserved in several aspects in all regions sampled. We were able to identify four classes of neurites in each of the microvolumes (Figure 4). The first class, termed “axiform neurites, ” is comprised of straight, unbranched processes of even diameter, ranging between 0. 2 and 0. 4 µm (Figure 4B, D). Axiform neurites form bundles originating in the cortex; they correspond to the primary axon tracts (PATs) emitted by the neurons belonging to one lineage. Within the neuropile, groups of PATs typically coalesce to form larger assemblies (fascicles; see examples of MEF or APT in Figure 3C). The second class of neurites (“varicose neurites”) consists of branched processes that alternately decrease and increase in diameter along their trajectory (Figure 4A, D). Thick segments (swellings or “varicosities”) measure between 0. 5 and 1. 5 µm; thin segments 0. 15–0. 4 µm. Varicose neurites account for most of the volume of the neuropile. A variant of the varicose neurites are “globular neurites”; principally similar in shape than the former, they have swellings (“globules” or “boutons”) that are more voluminous than varicosities, exceeding 1. 5 µm in diameter. The globules of these central nerve fibers resemble the endplates, or “boutons, ” of peripheral motor axons. The fourth class of neurites (“dendritiform neurites”) is formed by highly branched, thin (less than 0. 2 µm) fibers that frequently change direction (Figure 4C, D). Synapses are ultrastructurally defined by their characteristic presynaptic membrane specializations, consisting of an electron dense membrane thickening bordered by synaptic vesicles and the T-bar (also called synaptic ribbon), a cytoplasmic specialization involved in tethering and docking of synaptic vesicles ([44], [45]; Figure 4F, G). We observed that, independent of location within the brain, presynaptic sites are highly uniform in size, ranging from 0. 15–0. 3 µm, and are always found on the swellings of varicose and globular neurites (Figure 4E). These then constitute the terminal axons of the brain. In several cases, multiple presynaptic sites (individually defined by the synaptic ribbon) were confluent and formed band-like synaptic conglomerates. Postsynaptic sites, characterized by (relatively inconspicuous) membrane densities lacking T-bars or synaptic vesicles, are found almost exclusively on dendritiform neurites (Figure 4E) and (occasionally) on thin side branches of varicose neurites, implying that this class of fibers represents (terminal branches of) dendrites. Note that in insect neurons, the relatively strict distinction between dendrites and axons as two different types of neurites (a distinction that is quite typical for vertebrate brains) does not exist. Thus, the neuronal cell body emits a single neurite, which in the neuropile forms numerous branches that may be dendritic (i. e. , postsynaptic) or axonal (i. e. , presynaptic). Our data show that at the level of terminal branches, dendritic and axonal processes are mostly separate (pre- and postsynaptic sites rarely occur on the same preterminal process) and structurally distinct (presynaptic sites on large diameter neurites, postsynaptic sites on small diameter neurites). The disparity in size between terminal axons and dendrites also accounts for the fact that most, if not all, synapses of the Drosophila brain are of the polyadic type, where a single (large) presynaptic profile contacts multiple postsynaptic sites. The pre- to postsynaptic ratio in most synapses ranges between three and five (see examples shown in Figure 4F, G). Although the types of neurites depicted above are encountered in every region of the brain and ventral nerve cord, their relative numbers, directionality, branching density, and placement of synapses differ from one microvolume to the next. Two microvolumes, taken from the calyx of the mushroom body and the dorso-lateral column of the ventral nerve cord, are presented as examples (Figures 2,5, S1). The calyx represents the input region of the mushroom body that receives the axons of antennal projection neurons [46]. The microvolume extracted from the center of the calyx is characterized by a largely parallel array of varicose neurites carrying the majority of presynaptic sites. These neurites then most likely represent segments of the terminal axonal branches of antennal projection neurons, which project, via the antenno-protocerebral tract, to the calyx (Figure 2B). In total, the 20 µm3 calyx microvolume contained 33 varicose and two globular neurite segments. These neurites have an average diameter (reducing the shape of neurites to that of a smooth cylinder) of 0. 38 µm. The overall “cable length” (adding together all fragments of varicose/globular neurites within the microvolume) amounted to 48 µm. There were 16 branch points, corresponding to a density of branch points of one every 3 µm of axonal cable. We find a density of about 1. 8 presynaptic sites per 1 µm3. Dendritiform neurites form bundles of 5–10 thin fibers winding around the varicosities of preterminal axons to whom they are postsynaptic. Synapses are mostly dyadic-tetradic. Frequently, an individual dendritiform neurite participates in two or more synaptic contact made with the same axon (see also below for the VNC microvolume). Axonal cable length, branch density, and synapse density were similar in two other brain microvolumes, one in the spur (an output region of the mushroom body; [47]) and the other in the CPLd compartment of the protocerebrum. In the spur, we counted one branch point for every 4 µm of axonal cable; in the CPLd, a branching occurred every 3. 2 µm. The synapse density in the spur and CPLd was 3. 3 and 1. 2, respectively. The patterning of presynaptic (varicose/globular) neurites in the dorso-lateral domain of the ventral nerve cord is characterized by a lower density of branch points (approximately one branch every 7. 5 µm) and synapses per volume unit (0. 8/µm3) when compared to the brain. Cable length and average diameter of neurites, on the other hand, equals that in the brain. All types of neurites (varicose/globular, axiform, and dendritiform) are oriented preferentially along the longitudinal axis of the ventral nerve cord (VNC; Figure 5A–F; Figure 1S). The high quality of the VNC stack (only two out of 300 sections missing) made it possible to reliably reconstruct the pattern of thin dendritiform neurites. Similar to what is seen for the brain microvolumes, these neurites have a higher branch density than varicose/globular neurites (one branch every 4. 1 µm). Also, the overall dendritic cable length exceeds that of axonal elements by a factor of almost 2. Thus, the 85 µm3 VNC microvolume contained 247 µm of axiform/varicose/globular neurites (“axons”) and 420 µm of dendritiform neurites (“dendrites”). When following individual dendritiform processes throughout the microvolume, an interesting convergence-divergence pattern becomes apparent (Figure 4H–J). Thus, at any given level, dendritiform processes form aggregates of 5–10 processes each. However, aggregates, when following the dendritiform processes along the z-axis, do not translate into bundles: processes forming an aggregate at a given level stay together only for a short interval (<1 µm), after which they diverge and redistribute. This pattern reflects the fact (mentioned above) that dendritiform neurites do not follow a straight course but change direction constantly and abruptly (Figure 4C, Figure S2). By contrast, axonal processes have relatively straight trajectories: axiform neurites (long axons without synapses) form tight bundles where neighborhood relationships between neurites is maintained over manyµm (Figures 4D, H–J, 5A–C); terminal varicose/globular axons are more loosely packed but, like axiform processes, extend more or less parallel, maintaining their position relative to each other (Figures 4B, D, H–J, 5A–C). If we take the microvolume-based measurements of cable length, branch density, and synapse density, complement them with light microscopic findings, and extrapolate to the brain as a whole, we arrive at conclusions that are both interesting and helpful for further functional/developmental analyses. The L1 brain hemisphere has a volume of approximately 20,000 µm3 [25]. Differentiated primary neurons whose processes make up the volume of the brain neuropile number approximately 1,500 per hemisphere [25]. The average values of presynapse density, axonal branch point density, and axonal cable length (taken from the four microvolumes) are 2/µm3,0. 8/µm3, and 2. 9/µm3. Extrapolated to an entire brain hemisphere, this would amount to a total of 40,000 presynaptic sites, 16,000 axonal branch points, and 58,000 µm axonal cable length. For the average neuron, that means approximately 40 µm axonal length, 11 axonal branches, and 27 presynaptic sites. These estimates go along well with light microscopic data based on DiI fills (or other labelings) of L1 brain neurons. We randomly sampled L1 neuron shapes by injecting single cells with DiI and imaging neurons by fluorescence microscopy (unpublished data). It is possible to roughly estimate neurite length, and numbers of varicosities and branches from fluorescent images of labeled neurons, and obtain numbers that match closely our extrapolations from the microvolume analysis. An important aspect of Drosophila neuronal architecture is its “ultradense” neuropile architecture, for which three major structural features are responsible. First, neuronal cell bodies, in terms of volume a considerable part of the brain, are located outside of the neuropile and therefore do not form part of circuits (i. e. , do not carry synapses). Second, dendritiform processes are extremely thin and highly branched. Third, synapses are polyadic, with each presynaptic site attaching to an average of four postsynaptic sites. As a result of these factors, neurites within a microvolume are highly interconnected, and one can detect in high numbers certain types of network motifs [48], [49] that may determine the function of the corresponding neuropile microvolume. We will in the following summarize our first data concerning connectivity and network motifs for the microvolume generated for the dorso-lateral VNC. From the 85 µm3 VNC microvolume we reconstructed 170 elements over 1 µm length (Figures S1, S2). Of these, 39 were varicose neurites; one was globular, 25 axiform, and 105 dendritiform. We counted 68 presynaptic sites, all concentrated on the large diameter segments of the globular neurite and of 24 varicose neurites. Only a single presynaptic site was found on an axiform process. Among the 24 neurites with synapses, the synapse number per neurite ranged between one and nine (average 2. 7). The dendritiform processes formed 256 postsynaptic sites, yielding an average ratio of four postsynaptic sites to every one presynaptic site. In turn, each dendritiform process contacted an average of 2. 1 presynaptic neurites. Contacts between neurites of the VNC microvolume are highly concentrated along the longitudinal (z) axis. In other words, postsynaptic partners of a given, longitudinally oriented presynaptic process are clustered closely around this process (Figures 5M, 6). To quantify the correlation between the likelihood of forming a synaptic contact and distance between two processes, we visualized the envelopes that included a given axon and the dendritiform neurites (Figure 6A) and then estimated in a pairwise fashion the amount of overlap between axonal and dendritic envelope (complete overlap, more than 50%, 10%–50%, less than 10%). For each class we determined the frequency at which a synaptic contact is made (Figure 6E). This frequency is a function of overlap (which in turn reflects distance): if the overlap exceeds 50%, more than half of the dendritic processes form synapses with the corresponding axon; this drops to about a quarter at an overlap between 50% and 10% and to 17% at less than 10% overlap (Figure 6B). Most of the interconnected pre- and postsynaptic elements of the VNC microvolume engaged in what is called a “dense overlapping regulon motif” [49], defined as a network where a given input element (axon) diverges onto multiple targets, and at the same time, each target element (dendrite) receives input from multiple presynaptic elements (Figure 5M, O). In the example shown in Figure 5G–I, one varicose neurite possesses three presynaptic sites on two varicosities. Contacting these sites are 12 postsynaptic dendritiform processes. Each dendrite in turn receives input from up to five presynaptic partners. Most of the 24 axonal processes within the VNC microvolume formed part of such a dense overlapping regulon network. A second type of network, encountered much less frequently, is the feed forward motif. Here, a neurite, “A, ” receives input from a second neurite, “B, ” and both “A” and “B” provide output to a common target, “C” (Figure 5M, N). Thus, feed forward motifs involve instances of neurites that serve as presynaptic partners to some and postsynaptic partners to other fibers. In the VNC microvolume, we encountered five cases where varicose neurites, which are mostly presynaptic, also formed a postsynaptic site. In all cases, this postsynaptic site was located not on the varicosity itself, but the thin segment or a finger-like side branch of the neurite (Figure 5J–L). In two cases, axiform neurites had postsynaptic sites. Interestingly, fibers that, based on the presence of characteristic dense core vesicles, belong to peptidergic neurons all form part of axo-axonal contacts. In three cases, they served as presynaptic partners, and in two cases as postsynaptic partners. Statistical analyses of light microscopic preparations (e. g. , Golgi stained preparations) and representative EM sections of mammalian brains yielded estimates for parameters like synapse density, cable length, and branch point density (e. g. , [10]) that can be compared to the data we present here for the fly larval brain. Surprisingly, when considering the enormous difference in overall size of individual neurons and the brain as a whole, a number of parameters are very similar. For example, the average diameter of axon shafts and synapse bearing varicosities, as well as presynaptic sites themselves, is almost the same in Drosophila brain and mammalian cortex. As a result, the overall axonal cable length per neuropile volume is also very similar. One mm3 of mouse cortex contains an estimated 4,100 m of axon; translating this number to a smaller cube of 10 µm length, more appropriate when dealing with miniature brains, would yield 4. 1 mm of axon. In the microvolumes analyzed for the larval brain, axonal cable length ranged from 1. 5 mm to 4. 0 mm in a cube of 10 µm length. A conspicuous difference between the mammalian cortex and fly brain lies in the size of dendrites. In mammals, dendrite shafts have an average diameter of close to 1 micrometer [10], and many dendrites are considerably thicker and less branched. As a result, dendritic cable length per volume unit appears to be much lower than axonal cable length in mammalian brain (0. 5 versus 4. 1 mm in a 10×10×10 µm cube), whereas the opposite is true in fly brain. Thus, when extrapolated to a 10×10×10 µm volume, the overall length of varicose/globular neurites totaled 2. 9 mm; that of dendritiform neurites was 5. 4 mm. Another distinguishing characteristic of the Drosophila brain appears to be the higher density of branch points, in particular for dendrites. Statistical analyses in mammalian cortex yielded average distances of 10 µm and higher between branch points [10]. Measured here for Drosophila, terminal axons had average branch point intervals of 4 µm (mushroom body calyx), 2. 8 µm (mushroom body, spur), or 7. 5 µm (VNC), respectively. The density of branch points in dendrites appears to be even higher. Thus, for the VNC cube where thin dendrites could be reliably reconstructed, we measured an average interbranch point distance of 4. 9 µm. In light of the higher branch point density, one might also expect a higher number of synaptic contacts per volume unit in Drosophila, compared to the mammalian brain. In mammalian cortex, synaptic density has been estimated at 0. 72/µm3 [10]. Synapses are predominantly of the monadic type, where one presynaptic site contacts a single postsynaptic site. By contrast, synapses are polyadic in the Drosophila nervous system. For each presynaptic site, one observes multiple postsynaptic sites. In terms of total number of synaptic contacts (where one would count a tetrad with three postsynaptic partners as three contacts), the Drosophila brain does have a much higher synapse density than the mammalian brain; when only counting presynaptic sites, numbers are comparable. Thus, in the material analyzed for this study, presynapse density falls within a range from 0. 8/µm3 (VNC) to about 4/µm3 (input region of mushroom body). We expect that for mammalian brain, additional direct measurements will yield values of synapse density that may considerably vary between different neuropile compartments. The higher density of branches and synaptic contacts in the Drosophila brain, compared to mammalian brain, is correlated with a larger degree of “connectedness” between neurites. As shown in this study, the large majority of neurites contained within a microvolume of less than 100 µm3 are engaged in networks of the type of dense overlapping regulon or feed forward motifs, which simply reflects the fact that in average, terminal axons and dendrites in such a small volume have multiple branches that engage in synaptic contacts. This is not the case in mammalian cortex. When considering the low average branch point density (one every 10 µm; see above), it is unlikely that many neurite segments included within a 100 µm3 volume will have a branch, and thereby more than one contact. This estimate is confirmed in a recent microvolume reconstruction of rat hippocampus [18], [56]. Here, the only type of connectivity observed consists in the convergence of multiple axonal segments onto isolated dendritic segments. However, very few of the (unbranched) axonal segments give synaptic input to more than one (unbranched) dendritic segment. As a result, network motifs like the dense overlapping regulon motif or feed forward motif are not observed within microvolumes of 5 µm diameter. In other words, the volumes of mammalian brain that “contain” microcircuits are considerably larger than in Drosophila: connectivity occupies more space. The number and size of vertebrate neurons are typically much larger than in insects. Other fundamental architectural features of vertebrate brains are the inclusion of neuronal somata into the neuropile (somata carry many postsynaptic sites), the large diameter of dendrites, and concomitantly, the absence of polyadic synapses. One may speculate that one of the prime driving forces of the evolution of vertebrate brain architecture was the increasing number of neurons. Consecutively, the average vertebrate neuron also had to grow individually in axonal and dendritic length, in order to accommodate the higher number of synapses needed to connect a given neuron to a certain fraction of the (increasing) pool of neurons. Finally, as a result of the increase in cell number, cable length, and synapse number, branches of dendrites and axons became spaced further apart. Typical “microcircuits” in the mammalian brain, formed by multiple interconnected elements providing for stimulus convergence and divergence and inhibitory/excitatory feed forward and feed back loops, will occupy volumes of (300–1,000 µm) 3 diameter [7], [57], [58]. In the Drosophila brain, these volumes would be two orders of magnitude smaller, which constitutes a significant advantage when reconstructing circuitry from densely segmented serial EM sections. The dense segmentation of serially sectioned neuropile plays a pivotal role for reconstructing microcircuitry. However, the ability to reconstruct ultrastructural features of neurites in all parts of the neuropile of a small brain, such as the Drosophila larval brain, will prove to be of great value for in vivo cell biological and genetic studies of neuronal development as well. The complex shape and connectivity of a neuron is a reflection of an intracellular molecular machinery that places membrane proteins (e. g. , adhesion molecules, receptors, channels) and cytoskeletal proteins into the right position, such that pre- and postsynaptic sites, branch points, and specific connections with targets are formed in the right pattern. The analysis of these developmental phenomena is in its infancy. We know of many proteins that are differentially expressed in specific membrane domains [59], [60] and that the directional protein transport in axons and dendrites is controlled by different mechanisms. However, the exact mechanisms of targeted protein expression are unknown. Models such as Drosophila offer the opportunity of a genetic approach, that is, study the phenotype following genetic manipulation. This approach has been very successful to elucidate the development of several types of (peripheral) sensory neurons and motor neurons [61]–[63]. However, up until now, virtually no mutant phenotypes have been established for central brain neurons or circuits on the ultrastructural level, simply because the basic parameters of wild type neuronal ultrastructure were not available, and the necessary technology, serial EM, was too time-consuming. With the computer-aided reconstruction of EM stacks, such an analysis is now within reach. Thus, once parameters of neurite size, branching, and connectivity have been established for a number of compartments, it will be possible to generate stacks of EM sections that contain specific brain compartments (such as the calyx or lobes of the mushroom body) of genetically manipulated animals, and carry out detailed, quantitative comparisons with the wild-type. Such stacks (for the early larval brain) would contain as little as 100 contiguous sections; their preparation, image capturing, and analysis (when following the microvolume approach) may require only a matter of weeks. It is therefore realistic to generate EM stacks numerous specimens' brains of larvae carrying specific mutations, to then establish changes in basic neurite parameters, such as diameters, branch and synapse density, and synapse architecture. Serial TEM is the classical approach for the reconstruction of microcircuitry. In the days before digital photography and computer-assisted image processing, this approach was extremely labor intensive and was therefore utilized mostly for small parts of individual neurons (“sparse segmentation”). The only exception was the reconstruction of the C. elegans central nervous system [64], [65]. We predict that given the speed of imaging that is now possible (and will certainly further increase), serial EM will experience a renaissance as an approach for the reconstruction of microcircuitry. Two recent technological developments have reduced the difficulty of large-scale serial section electron microscopy and improved its reliability. Serial block-face scanning electron microscopy [66] has proven useful in imaging relatively large volumes of neural tissue at an isotropic resolution of about 20 nm/pixel. The smaller dimensions of Drosophila neural tissue components, compared to the vertebrate equivalents, require a resolution of at least 8 nm/pixel (ideally 4 nm/pixel) for the reconstruction of small terminal dendrites, which is necessary for the conclusive elucidation of synaptic partners. The second novel technique, focused ion beam (FIB) milling combined with block-face scanning electron microscopy [67], delivers up to 5 nm/pixel resolution. While FIB delivers images with the necessary resolution for the reconstruction of Drosophila neuropile, its imaging field of view is currently limited in practice to a window of 20×20 microns (Graham Knott, personal communication), which does not enclose a transverse section of the entire nerve cord neuropile of Drosophila larva. Traditional serial section transmission electron microscopy, as employed in this article, delivers the required high imaging resolution on the plane (4 nm/pixel or better) but reduced resolution in the z-axis (50 nm, the thickness of the section). However, our reconstructions indicate that, given sufficient resolution in the XY plane, the 50 nm/pixel resolution of the z-axis does not prevent the full reconstruction of even the smallest terminal dendrites, measuring only about 60 nm in diameter. We anticipate that for the immediate future, serial TEM, BF/SEM, and FIB/SEM will coexist as more or less equally valuable techniques for the electron microscopic reconstruction of microcircuitry. The generation of large EM image data sets has made imperative the development of novel specialized software for its processing, visualization, and analysis [15]. The sheer size of the data sets has prompted the development of novel algorithms for the automatic segmentation and reconstruction of neural arbors [18]. Thus, given the current conditions, to manually segment an entire L1 Drosophila brain would take one person in the order of 50 years, which means that the development of tools for automatic segmentation is of high priority. It should be noted that the 50-year projection does not take into account the fact that neuropile ultrastructure, to some extent, is most likely modular. As a result, the network diagram extracted from a given microvolume can be extrapolated to neighboring volumes, as long as they fall within the same compartment. In other words, the hope and anticipation (in particular in regard to “big brains” of vertebrates) is that one does not dense-segment every single “voxel” of a given compartment but focus on a certain set of samples that are (manually or automatically) densely segmented; then, using algorithms that need to take into account data from many microvolumes, circuitry in the samples can be extrapolated to a compartment as a whole. Until recently, two software packages were primarily used for visualization and analysis of TEM serial sections: IMOD [17] and Reconstruct [16]. IMOD is the current gold standard in EM software for image composition and processing. Reconstruct provides an efficient GUI for manual and semiautomatic image stitching and is particularly noted for its manual image segmentation toolkit. Neither of these programs handles data sets that are considerably larger than RAM and have limited support for large-scale image registration with nonlinear transformations. The ir-tools and associated visualization applications [15], while providing effective image registration and a comprehensive image analysis toolkit, depend on RAM and lack ease of customization for highly specialized tasks. We have found that TrakEM2, while not yet fully independent of RAM, provides the means to register hundreds of thousands of images, overcoming numerous limitations imposed by computer hardware. TrakEM2 eases the concatenation of image transformations, including polynomial models for lens deformation correction [32]. In contrast to the ir-tools, TrakEM2 operates on image tiles that correspond to the original acquired images and not on stitched large images. Precomputed image pyramids for each tile enable enhanced performance. We highlight the robustness of the image registration library associated with TrakEM2, and the ability to manually correct errors when these inevitably occur, an ability facilitated by the tile-oriented approach. Finally, TrakEM2 is a (large) component of Fiji [68], an ImageJ-based image processing environment in active development and with thousands of image analysis plugins readily available. We believe that TrakEM2 represents a useful addition to existing software packages that can handle special tasks, in particular the segmentation and subsequent analysis of large data sets with high numbers of individual elements. In summary, TrakEM2 improves over existing software packages in being less dependent on scarce computer resources and by bundling numerous image segmentation and analysis tools within a unique graphical interface. TrakEM2 acknowledges that any automatic procedure (such as image registration and image segmentation) will eventually fail partially or fully and will require manual correction by a human operator. The combination of both manual and automatic procedures for neuronal reconstruction makes TrakEM2 a practical application for the reconstruction of large volumes of brain neuropiles.

Brains contain a vast number of connections between neurons, termed synapses. The precise patterns of these synaptic contacts form the structural underpinning of electrical microcircuits responsible for animal behavior. Due to their small size, synaptic contacts can be conclusively shown using only high-resolution electron microscopy (EM). Therefore, complete series of ultrathin sections are required to reconstruct neuronal microcircuitry. The acquisition and analysis of EM sections (with 15,000 sections per millimeter of tissue) is practical only by computer-assisted means. In this article, we demonstrate the utility of the software package TrakEM2 to model interconnections of nerve fibers from consecutive EM sections and to efficiently reconstruct the neural networks encountered in different parts of a small brain, the early larval brain of the fruit fly Drosophila melanogaster. Neuronal networks are composed of patterns of axons and dendrites (neuronal extensions that transmit and receive signals, respectively), and using TrakEM2, we describe the most common motifs they form. Our study introduces an approach towards a comprehensive anatomical reconstruction of neuronal microcircuitry and delivers microcircuitry comparisons between vertebrate and insect brains.

Abstract Introduction Materials and Methods Results Discussion

neuroscience/neurodevelopment

2010

An Integrated Micro- and Macroarchitectural Analysis of the Drosophila Brain by Computer-Assisted Serial Section Electron Microscopy

Treponema pallidum is a highly invasive pathogen that undergoes rapid dissemination to establish widespread infection. Previous investigations identified the T. pallidum adhesin, pallilysin, as an HEXXH-containing metalloprotease that undergoes autocatalytic cleavage and degrades laminin and fibrinogen. In the current study we characterized pallilysin' s active site, activation requirements, cellular location, and fibrin clot degradation capacity through both in vitro assays and heterologous treponemal expression and degradation studies. Site-directed mutagenesis showed the pallilysin HEXXH motif comprises at least part of the active site, as introduction of three independent mutations (AEXXH [H198A], HAXXH [E199A], and HEXXA [H202A]) abolished pallilysin-mediated fibrinogenolysis but did not adversely affect host component binding. Attainment of full pallilysin proteolytic activity was dependent upon autocatalytic cleavage of an N-terminal pro-domain, a process which could not occur in the HEXXH mutants. Pallilysin was shown to possess a thrombin cleavage site within its N-terminal pro-domain, and in vitro studies confirmed cleavage of pallilysin with thrombin generates a truncated pallilysin fragment that has enhanced proteolytic activity, suggesting pallilysin can also exploit the host coagulation process to facilitate protease activation. Opsonophagocytosis assays performed with viable T. pallidum demonstrated pallilysin is a target of opsonic antibodies, consistent with a host component-interacting, surface-exposed cellular location. Wild-type pallilysin, but not the HEXXA mutant, degraded fibrin clots, and similarly heterologous expression of pallilysin in the non-invasive spirochete Treponema phagedenis facilitated fibrin clot degradation. Collectively these results identify pallilysin as a surface-exposed metalloprotease within T. pallidum that possesses an HEXXH active site motif and requires autocatalytic or host-mediated cleavage of a pro-domain to attain full host component-directed proteolytic activity. Furthermore, our finding that expression of pallilysin confers upon T. phagedenis the capacity to degrade fibrin clots suggests this capability may contribute to the dissemination potential of T. pallidum. The expression of host-interacting proteases has been shown to contribute to the pathogenesis of bacteria of medical interest by promoting host colonization and immune evasion, acquisition of nutrients, tissue invasion and dissemination of infection. Several pathogenic bacteria, including Streptococcus pneumoniae [1], Yersinia pestis [2], Vibrio cholerae [3], and Clostridium perfringens [4], express bacterial proteases and toxins which play a central role in the infection process through facilitation of bacterial dissemination and tissue invasion by proteolytic degradation of host proteins. Bacterial proteases and toxins have been shown to target a wide array of host molecules, including the extracellular matrix (ECM) components collagen [4], elastin [5], fibrinogen [6], fibronectin [7] and laminin [8]. Although host component degradation is common to many disseminating pathogens, such a mechanism has yet to be confirmed within the highly invasive causative agent of syphilis, Treponema pallidum subsp. pallidum. Numerous studies have demonstrated that T. pallidum is capable of gaining rapid entry to the circulatory system following infection, with subsequent dissemination to distant host sites [9]–[14]. The highly invasive nature of the pathogen is further emphasized by the diverse clinical manifestations that can occur in untreated syphilis infections, including skin rashes, meningitis, ocular disease, and cardiovascular and neurological complications, and by the fact that T. pallidum can cause bone destruction in congenital and tertiary stage syphilis [15]. Furthermore, T. pallidum is one of only a few pathogens that can traverse the placental and blood-brain barriers. Previously our laboratory identified the T. pallidum laminin-binding adhesin Tp0751 (GenBank accession number, AAC65720; also referred to as ‘pallilysin’) [16], [17]. Since laminin is an abundant glycoprotein component of the blood-brain barrier and basement membranes underlying endothelial cell layers, barriers which T. pallidum must traverse during the course of infection, pallilysin was proposed to contribute to the T. pallidum infection process [16]. Pallilysin-specific antibodies have been detected in serum from both natural and experimental T. pallidum infections [16], indicating that the protein is expressed during the course of infection. Additionally, heterologous expression of pallilysin on the surface of the culturable non-adherent spirochete, Treponema phagedenis, conferred upon the bacterium the ability to bind laminin, a specific interaction which was inhibited by the presence of pallilysin-specific antibodies [18]. Recently we have shown that pallilysin is also capable of specific binding to human fibrinogen [19], a key structural protein in blood coagulation. A second central component of the coagulation cascade is thrombin, an abundant serine protease that catalyzes the conversion of soluble fibrinogen to insoluble fibrin, the major structural component of haemostatic clots and abscesses [20]. Expression of both fibrinogen and thrombin is significantly up-regulated as a host response to inflammation and infection, and the coagulation process is critical for pathogen containment during infection [21]. Several pathogenic bacteria have developed strategies to overcome this host defence mechanism, including expression and activation of fibrinogenolytic proteases [22]–[24]. Previously we showed that pallilysin degrades human fibrinogen and laminin, and consistent with this finding we identified a putative metalloprotease motif (HEXXH) within the C-terminus of pallilysin [19]. Prior investigations also indicated pallilysin is capable of being activated via autocatalysis [19], a common activation mechanism for ECM-degrading matrix metalloproteases (MMPs) [25]. Bacterial metalloproteases whose activity is regulated by autocatalysis, such as thermolysin-like metalloproteases [26], [27], are initially synthesized as an inactive pre-pro-protease consisting of a signal (pre-) sequence, pro-domain, and mature active protease domain (s). The major role of the pro-domain, which is cleaved to form the active enzyme, is to mediate correct protein folding and to maintain the protease in an inactive form to prevent premature activation prior to attainment of its target location [28]. Numerous eukaryotic pro-proteases, such as the pro-matrix metalloproteases (pro-MMPs) -2 and -9, are cleaved into their active mature form by the action of another protease, including other MMPs [29]. To our knowledge, there have not been any prior reports of inter-molecular activation of a bacterial metalloprotease via a host protease-mediated processing event. In the current study we elucidate pallilysin' s mechanisms of activation and proteolysis via identification of key active site residues using a site-directed mutagenesis-based approach. We demonstrate that, in contrast to wild-type pallilysin, alanine-substituted HEXXH mutant forms of the protease fail to degrade fibrinogen and insoluble fibrin clots, indicating that the C-terminal HEXXH motif comprises, at least in part, the metalloprotease active site. We identify contiguous thrombin and autocatalytic cleavage sites in the N-terminus of pallilysin and demonstrate that mature active pallilysin is formed by removal of an N-terminal pro-domain via either inter-molecular autocatalytic cleavage or thrombin cleavage. Furthermore, we show that heterologous expression of pallilysin on the surface of T. phagedenis confers upon the transformed bacteria the capacity to degrade insoluble fibrin clots. Finally, we demonstrate that pallilysin is a target of opsonic antibodies, suggesting that the protease is also surface-exposed in T. pallidum and thus capable of directly interacting with host components during infection. In order to investigate the potential activation and proteolytic mechanisms of pallilysin, several soluble recombinant proteins were expressed and purified in E. coli; (1) wild-type full-length (C24-P237) pallilysin purified in the presence and absence of zinc and calcium and (2) three full-length (C24-P237) pallilysin active site mutants (AEXXH [H198A], HAXXH [E199A], and HEXXA [H202A]) purified in the presence of zinc and calcium. In agreement with previous findings [19], and as shown in Figure 1, wild-type pallilysin migrated as multiple bands via SDS-PAGE analysis. Major bands were consistently detected at approximately 32 kDa, 26 kDa, and 18 kDa (wild-type pallilysin in presence of zinc and calcium only), with the 26 kDa protein band being the dominant wild-type pallilysin form detected following purification in the presence of zinc and calcium. In contrast, the 32 kDa protein band was the dominant form detected in the pallilysin mutants (Figure 1). Mass spectrometry analyses (data not shown) and Edman sequencing, as described below, identified all bands as pallilysin, indicating that all recombinant proteins were purified to homogeneity. The major band corresponding to 32 kDa was identified as full-length pallilysin and the 26 kDa and 18 kDa peptides as truncated forms of pallilysin. To determine whether the previously identified pallilysin C-terminal HEXXH motif [19] is required for zinc coordination, one of the potential pallilysin zinc-coordinating mutants (AEXXH [H198A]) and the potential pallilysin catalytic general acid/base mutant (HAXXH [E199A]) were analyzed for their ability to bind divalent zinc. This was accomplished via quantitative inductively coupled plasma-mass spectrometry (ICP-MS), a highly sensitive mass spectrometry technique that determines the concentration of trace metals in a given sample at concentrations as low as one part per trillion. As expected, these analyses indicate wild-type pallilysin and the HAXXH (E199A) mutant, but not the AEXXH (H198A) mutant, bind zinc (Table 1). For HEXXH-containing proteins that bind zinc, stoichiometries of approximately 1. 0 would be expected. In the current study, the lower than expected stoichiometries were likely due to the fact that zinc could not be added during pallilysin expression as this resulted in high levels of pallilysin adherence to gel filtration columns and very low pallilysin yields. Furthermore, non-His-tagged pallilysin binds to nickel affinity columns (data not shown), presumably through the HEXXH motif, suggesting nickel may competitively inhibit coordination of zinc to the pallilysin HEXXH motif. In the current study, lower zinc levels were detected in wild-type pallilysin compared to previous findings [19]. This is likely due to the variable levels of zinc available for protease coordination during protein expression, primarily resulting from variable zinc concentrations in the bacterial growth media. An ELISA-based approach was used to assess the binding specificity of soluble wild-type pallilysin (C24-P237) to immobilized fibrinogen and laminin as a function of varying pallilysin concentrations (1–20 µg). As shown in Figure 2A, increasing concentrations of pallilysin bound to both fibrinogen and laminin in a dose-dependent manner. Furthermore, all concentrations of wild-type pallilysin (1–20 µg) exhibited statistically significant levels of binding to both fibrinogen and laminin (p<0. 0001) when compared to the level of binding exhibited by the negative control, Tp0453 [30] (20 µg) (GenBank accession number; AAC65443). The ability of the three pallilysin HEXXH active site mutants to exhibit attachment to fibrinogen and laminin was investigated using the ELISA-based assay. As shown in Figure 2B, wild-type pallilysin and all three pallilysin HEXXH mutants exhibited similar, statistically significant levels of binding to fibrinogen and laminin (p<0. 0001) when compared to the level of binding exhibited by Tp0453. These results indicate that binding of pallilysin to fibrinogen and laminin is HEXXH-independent, and therefore cannot be solely attributed to a transient metalloprotease-substrate interaction. To determine the effect of the introduced HEXXH mutations on pallilysin-mediated host protein degradation, wild-type pallilysin and the three HEXXH mutants were incubated with fibrinogen for 24 h at 37°C. Samples were removed at hourly intervals and the fibrinogenolytic activity of each protein was determined by gel-based analyses. Substitution of each of the three predicted active site residues with alanine abolished the fibrinogenolytic activity of the corresponding mutants (Figure 3). Wild-type pallilysin initiated fibrinogen degradation within 1 h of incubation, with complete degradation of the fibrinogen α-, β-, and γ-chains observed by 23 h post-incubation, whereas the corresponding mutants failed to degrade any of the three fibrinogen chains over the 24 h duration of the experiment (Figure 3). The negative control, Tp0453, did not exhibit fibrinogen degradation (data not shown). These results indicate that the pallilysin C-terminal HEXXH motif resides within the protease active site. As previously reported [19], SDS-PAGE analysis of purified recombinant pallilysin consistently results in the detection of multiple protein bands suggesting that pallilysin undergoes autocatalytic cleavage. Since we have demonstrated the HEXXH motif constitutes, at least in part, the metalloprotease active site, we predicted that substitution of the potential zinc binding histidines and catalytic glutamate with alanine would prevent autocatalytic cleavage of the mutant proteins. In order to test this prediction, wild-type and the three pallilysin HEXXH mutants were incubated independently for 24 h at 37°C, samples were removed at hourly intervals, and autocatalytic activity was assessed via SDS-PAGE analysis. Compared to wild-type pallilysin, the ability to undergo autocatalytic cleavage was abolished in each of the three pallilysin HEXXH mutants (Figure 4A). Incubation of wild-type pallilysin resulted in complete degradation of the major pallilysin bands (32 kDa and 26 kDa) by 21–22 h post-incubation with a concurrent increase in the intensity of the 18 kDa protein band (Figure 4A). Interestingly this timing correlates with maximal fibrinogen degradation (Figure 3). Conversely, the 32 kDa and 26 kDa protein bands from each of the three pallilysin HEXXH mutants remained stable throughout the course of the 24 h incubation (Figure 4A). In order to identify the pallilysin autocatalytic cleavage sites, the 32,26, and 18 kDa wild-type pallilysin protein bands generated during the 24 h incubation (Figure 4A) were analysed by Edman sequencing. As indicated in Table 2, the first 10 N-terminal amino acids identified from the 32 kDa protein band corresponded to the histidine tag originating from the expression vector, indicating that this represents full-length pallilysin (C24-P237). The 26 and 18 kDa protein bands matched the pallilysin amino acid sequence that would be expected following cleavage between residues T46-A47 and Q92-T93, respectively (Table 2). In silico analyses predict cleavage of this presumed “pro-domain” between T46-A47 and Q92-T93 generates peptides with approximate molecular masses of 21 kDa (A47-P237) and 16 kDa (T93-P237). However, all three pallilysin forms migrate on SDS-PAGE with slower-than-predicted mobility, a result consistent with previous findings demonstrating that pallilysin exhibits anomalous SDS-PAGE migration [19]. To test whether pallilysin undergoes intra- and/or inter-molecular autocatalysis, mature active wild-type pallilysin was incubated with an equal concentration of the AEXXH-mutated protease for 22 h at 37°C. As shown in Figure 4B, wild-type pallilysin was capable of cleaving mutant pallilysin resulting in complete degradation by 22 h post-incubation. Mutant pallilysin (AEXXH) alone did not undergo autocatalytic cleavage, whereas complete autocatalytic proteolysis of wild-type pallilysin was observed following 22 h incubation at 37°C (Figure 4B). Degradation of the wild-type pallilysin 18 kDa protein band following 22 h incubation (Figure 4B, lane 3) is likely due to enhanced autocatalytic activity arising from the variable zinc concentrations that are present in the bacterial growth medium used during protein purifications. These results demonstrate that in vitro pallilysin autocatalysis occurs via two sequential cleavage events whereby residues C24–T46 are initially removed via inter-molecular cleavage, followed by a second inter-molecular cleavage event which removes residues A47–Q92. To determine if wild-type pallilysin is in a more proteolytically active form following incubation and autocatalytic cleavage, wild-type pallilysin was pre-incubated at 37°C for 21 h, which generated the final autocatalytically cleaved mature protease (T93-P237). This version of the protease was incubated with fibrinogen for 5 h at 37°C, samples were removed at hourly intervals and the fibrinogenolytic activity was compared to that of a similarly pre-incubated negative control (Tp0453) and wild-type pallilysin (C24-P237) that had not undergone the pre-incubation step. As shown in Figure 5A, pre-incubated pallilysin (T93-P237) exhibited enhanced proteolytic activity over its non-pre-incubated (C24-P237) counterpart, as evidenced by accelerated fibrinogen α-chain degradation and complete β-chain degradation by 5 h post-incubation, a process which normally takes up to 20 h in the absence of a pre-incubation step for pallilysin (Figure 3). Pre-incubated Tp0453 failed to degrade fibrinogen (Figure 5A). Our previously established FRET-based fibrinogen degradation assay [19] was also performed to further compare the fibrinogenolytic activity of pre-incubated (T93-P237) and non-pre-incubated (C24-P237) pallilysin. As shown in Figure 5B, pre-incubated pallilysin (T93-P237) exhibited a statistically significant level of fibrinogenolysis at 8 h post-incubation (p = 0. 0270) and beyond (p<0. 0001 at 48 h), compared to the level of fibrinogen degradation exhibited by the negative control Tp0453. Non-pre-incubated pallilysin (C24-P237) exhibited a statistically significant level of fibrinogenolysis at 10 h post-incubation (p = 0. 0344) and beyond (p<0. 0001 at 48 h), compared to the level of fibrinogen degradation exhibited by the negative control. From this degradation curve it was evident that the majority of pre-incubated (T93-P237) pallilysin-mediated fibrinogenolysis occurred between 10–20 h post-incubation, whereas most non-pre-incubated (C24-P237) pallilysin-mediated fibrinogenolysis occurred approximately 20 h later during the 30–40 h post-incubation period. These results demonstrate that cleavage of pallilysin at Q92-T93 results in more rapid pallilysin-mediated degradation of fibrinogen. Analysis of the pallilysin amino acid sequence for the presence of predicted protease cleavage sites using the ‘peptide cutter program’ [31] failed to identify any single potential cleavage site within pallilysin that could be targeted by commonly known proteases (data not shown). However, manual analysis of the pallilysin amino acid sequence by comparison to the thrombin cleavage site specificity matrix from MEROPS [32], a database which indicates the most common amino acid residues to be located in protease cleavage subsites, identified a predicted thrombin cleavage site between pallilysin residues R77 and S78 (Figure 6A). According to the MEROPS analysis, which at the time of writing was based on 184 thrombin cleavage reactions, the most common residues located in thrombin substrate subsites P2, P1, and P1′ are proline, arginine, and serine, respectively. All three residues were identified within the N-terminus of pallilysin (P76, R77, S78). In order to experimentally confirm the predicted pallilysin thrombin cleavage site, full-length mature pallilysin (C24-P237) and the negative control, Tp0453, were incubated with thrombin at 20°C for 24 h. Protein samples were removed at 0,1, 2,4, 6, and 24 h post-incubation and analyzed for thrombin cleavage using SDS-PAGE. Thrombin cleaved mature recombinant pallilysin into two major peptides with molecular masses of approximately 18 kDa and 14 kDa (Figure 6B). Thrombin cleavage of the recombinant control, Tp0453, was not observed (data not shown). Given the fact that cleavage of pallilysin at the predicted thrombin cleavage site (R77-S78) would result in the generation of an approximately 18 kDa peptide, the protein band corresponding to this size was subjected to N-terminal amino acid sequence analysis using Edman degradation. As indicated in Figure 6B, the first 10 N-terminal amino acids identified matched the pallilysin sequence expected following thrombin cleavage at the R77-S78 site (SHGNAPPAPV). Substitution of the highly conserved arginine from the pallilysin thrombin cleavage P1 subsite with glycine (R77G) abolished the ability of thrombin to cleave pallilysin at the mutated internal thrombin site (data not shown). These results confirm the presence of an N-terminal thrombin cleavage site within pallilysin at residues R77-S78, a location which is just upstream of the final autocatalytic cleavage site (Q92-T93). To test if the thrombin cleavage site within the N-terminus of pallilysin could also be involved in pallilysin activation, recombinant non-tagged pallilysin comprising residues C24-P237 (full-length pallilysin, includes N-terminal pro-domain) and S78-P237 (lacking N-terminal pro-domain due to truncation at internal thrombin cleavage site) were produced and their rate of fibrinogenolysis was compared using the SDS-PAGE-based fibrinogen degradation assay. As shown in Figure 7, truncation of pallilysin at the internal thrombin site did not adversely affect pallilysin proteolytic activity, in that the S78-P237 recombinant protein retained the capability to degrade the fibrinogen α- and β-chains. The negative controls comprising fibrinogen incubated in the absence of recombinant protein and in the presence of recombinant Tp0453 remained stable throughout the duration of the incubation period (Figure 7). Of particular interest, non-His-tagged full-length pallilysin (C24-P237) exhibited enhanced fibrinogenolysis, with complete α- and partial β-chain degradation by 5 h. This result is consistent with the concept that pro-domains are required to mediate correct protein folding [28] and our previous finding that removal of the N-terminal His-tag resulted in enhanced pallilysin-mediated fibrinogenolysis [19]. Given the fact that pallilysin has been shown to degrade human fibrinogen [19], we predicted that it may also mediate degradation of insoluble fibrin, the major structural constituent of blood clots. To test this prediction, various forms of purified pallilysin were incubated with fibrin clots and the level of clot degradation was visually analyzed at selected time points. As shown in Figure 8A, wild-type pallilysin (C24-P237) that had not been pre-activated via removal of the pro-domain was capable of mediating complete degradation of fibrin clots between 24 and 48 h post-exposure. Fibrin clots incubated with autocatalytically activated wild-type pallilysin (T93-P237) showed more rapid degradation, with complete degradation observed between 0 and 24 h following exposure. In contrast, both water and the pallilysin active site mutant HEXXA (H202A) had no detectable effect on fibrin clot stability over the course of 48 h. To quantitate the fibrinolytic activity of recombinant pallilysin, the percent degradation of insoluble fibrin clots incubated with pallilysin over the course of 48 h was investigated. As shown in Figure 8B, wild-type pallilysin (C24-P237) and activated wild-type pallilysin (T93-P237) exhibited complete fibrin clot degradation by 48 h post-incubation, compared to minimal levels of clot degradation observed in the presence of water and the pallilysin active site mutant HEXXA (H202A) (p<0. 0001). To investigate the fibrinolytic potential of pallilysin within the context of a viable treponeme, an in vitro growth assay was performed to compare the capacity of T. phagedenis heterologously expressing pallilysin (T. phagedenis_tp0751/pKMR) to degrade fibrin clots with that of T. phagedenis transformed with the shuttle expression plasmid alone (T. phagedenis_pKMR). Treponemes were introduced to fibrin clots, grown until late exponential phase, and the level of clot degradation was visually analyzed at selected time points. As shown in Figure 8C (two right side panels), degradation of fibrin clots by T. phagedenis_tp0751/pKMR was clearly apparent following 48 h growth (late exponential phase; 1. 25×109±2. 9×107 cells ml−1 [mean ± standard error]). Fibrin clots incubated with T. phagedenis_tp0751/pKMR in the presence of thrombin (Figure 8C, second panel from right) exhibited increased levels of degradation compared to clots incubated with T. phagedenis_tp0751/pKMR in the absence of thrombin (Figure 8C, right panel). In contrast, when fibrin clots were incubated with T. phagedenis_pKMR in the presence of thrombin for 48 h (late exponential phase; 1. 25×109±5. 0×107 cells ml−1 [mean ± standard error]), minimal levels of fibrin clot instability were observed which were comparable to the levels observed when fibrin clots were incubated in treponemal growth medium alone (Figure 8C, two left side panels). To quantitate the fibrinolytic activity of pallilysin-expressing T. phagedenis, the percent degradation of insoluble fibrin clots mediated by T. phagedenis_pKMR and T. phagedenis_tp0751/pKMR after 48 h growth was determined. As shown in Figure 8D, T. phagedenis_tp0751/pKMR grown in both the presence and absence of thrombin exhibited statistically significant levels of insoluble fibrin clot degradation (61%±5. 8% and 54%±4. 9% [mean ± standard error], respectively; p<0. 005) when compared to the level of clot degradation observed in the presence of media alone or T. phagedenis_pKMR. In order to determine if pallilysin could be surface-exposed in T. pallidum, standard opsonophagocytosis assays were performed to determine the ability of pallilysin antiserum to opsonize T. pallidum. As shown in Figure 9, pallilysin-specific polyclonal antiserum exhibited significant opsonic activity compared with normal rabbit serum (p<0. 0001). The level of opsonic activity observed for anti-pallilysin was similar to that of immune rabbit serum and consistently higher than the positive control BamA-specific antiserum (BamA/Tp0326; GenBank accession number, AAC65313, has been shown to be at least partially surface-exposed in T. pallidum [33], [34]). Importantly, the level of opsonic activity observed for antiserum generated against a periplasmic protein, FlaA (Tp0249; GenBank accession number, AAC65235), and a cytoplasmic membrane lipoprotein, TpN47 (Tp0574; GenBank accession number, AAC65545), was similar to the level of opsonic activity observed with normal rabbit serum (Figure 9), indicating the treponemal outer membranes remained intact throughout the course of the assays. These results suggest pallilysin is located on the surface of T. pallidum. The highly invasive and disseminating nature of T. pallidum likely represents a multi-factorial process. Several mechanisms have been proposed to explain the invasiveness of T. pallidum, including treponemal adhesion [16], [19], [35], motility [36], [37], chemotaxis [15], unusual ultrastructure/low outer-membrane protein content [38], [39], host-inflammatory and immune responses [15], and antigenic variation [15], [40]. However, the factors mediating this striking pathogenic trait remain to be definitively identified. The current study further advances our understanding of T. pallidum pathogenesis by providing novel insight into the mechanism of pallilysin-mediated host component degradation, a pathogenic strategy we predict may be used by T. pallidum to facilitate bacterial dissemination and tissue invasion during the course of infection. In the current study we further characterize the attachment and proteolytic capabilities of pallilysin, definitively establish that pallilysin is a zinc-binding metalloprotease, and demonstrate that the C-terminal HEXXH motif is located within the active site. ICP-MS analysis showed that H198, but not E199, is required for pallilysin zinc coordination, a finding that is consistent with the typical metalloprotease HEXXH motif whereby the glutamate functions as an essential general acid-base catalyst but is not directly involved in coordination of zinc [41], [42]. Additionally, the finding that mutation of the HEXXH active site residues abolishes host component proteolysis but does not adversely affect host component binding confirms the previously reported bifunctionality of pallilysin [19], in that the host component binding and proteolytic functionalities of pallilysin appear to be physically separate and functionally independent. In the current study we demonstrate that each of the three pallilysin HEXXH active site mutants, which were purified under conditions identical to that of wild-type pallilysin, was stable following 24 h incubation at 37°C, whereas wild-type pallilysin exhibited characteristic self-cleavage. These results confirm that pallilysin, as opposed to contaminating proteases derived from the protein expression systems, solely mediates the observed post-purification pallilysin cleavage. Using Edman sequencing, we demonstrated that removal of the pallilysin N-terminal pro-domain occurs via autocatalytic cleavage between residues T46–A47 and Q92–T93. These results indicate that pallilysin' s active site substrate subsites are able to accommodate very different amino acid types, which agrees with our evidence that the protease is capable of targeting multiple host proteins [19]. Using an approach similar to that used by Marie-Claire and colleagues [43], which directly tests the potential of a wild-type protease to process an active site mutant unable to undergo autocatalysis, the current study demonstrates that in vitro pallilysin maturation occurs via a sequential cleavage mechanism involving inter-molecular autocatalysis. It should be noted that the current in vitro study does not rule out the possibility of intra-molecular autocatalytic pallilysin activation occurring both in vitro and in vivo. Intra-molecular processing may be advantageous when pallilysin is present at very low concentrations. Our results also demonstrate that autocatalytically-cleaved pallilysin (T93-P237) exhibits enhanced rates of fibrinogenolysis compared to its full-length counterpart (C24-P237), indicating that it is in a more proteolytically active conformation. Eukaryotic convertases, such as the ubiquitously expressed furin, have been shown to activate bacterial toxins in the intracellular environment through limited proteolytic cleavage. Examples include the Shiga toxin, anthrax toxin, Pseudomonas exotoxin A, diphtheria toxin, botulinum toxin and tetanus toxin [44]–[46]. Furthermore, many bacterial proteases, such as the aspartic protease Pla from Yersinia pestis, the metalloprotease aureolysin from Staphylococcus aureus, and the LasB metalloprotease of P. aeruginosa, have been shown to cleave and activate host proteins, in particular components of the host plasminogen activation system [47]–[49]. To our knowledge the reverse mechanism, that is activation of bacterial proteases through limited cleavage by host proteases, has not been previously reported. In the current study we have experimentally confirmed the presence of an N-terminal thrombin cleavage site between residues R77-S78 of pallilysin. Further, we have demonstrated that pallilysin truncated at the N-terminal thrombin cleavage site retains the ability to degrade human fibrinogen. The finding in the current study that non-His-tagged full-length pallilysin (C24-P237) exhibited enhanced fibrinogenolysis compared to truncated pallilysin (S78-P237) is consistent with the concept that pro-domains often function as intra-molecular chaperones essential for mediating correct protein folding and formation of the mature active protease [28], [50]. In light of this fact, it is possible that during recombinant expression truncated pallilysin (S78-P237) is incorrectly folded in the absence of this important molecular chaperone, which in turn results in sub-optimal proteolytic activity. To our knowledge, this study represents the first published report describing a bacterial protease that may be activated via limited proteolysis mediated by host-expressed thrombin and thus may represent a novel paradigm in bacterial pathogenesis. The fact that T. pallidum encodes one of the smallest bacterial genomes identified to date [51] is consistent with the concept that the bacterium has evolved virulence mechanisms that exploit the biological function of host-encoded proteins for the promotion of pathogenicity during the course of infection. The functional redundancy of host-dependent (thrombin cleavage) and host-independent (autocatalysis) mechanisms of pallilysin activation may enable treponemal dissemination within the diverse host environments the bacterium encounters throughout the course of long-term infection. For example, thrombin levels within the skeletal and nervous systems, which T. pallidum is capable of infecting [15], [52], are significantly lower than levels found within the circulatory system [53]. Localized coagulation is immediately induced in the host following inflammation and infection, functioning as one of the first lines of defense to limit tissue damage and promote healing [54]–[56]. Furthermore, recent studies have clearly demonstrated that fibrin clots are involved in the entrapment, localization, and killing of invasive strains of bacteria, including Escherichia coli, Staphylococcus aureus, and Streptococcus pyogenes [57], [58]. In the current study we demonstrate that purified recombinant pallilysin directly mediates fibrin clot dissolution. Furthermore, heterologous expression of pallilysin on the surface of T. phagedenis was shown to confer fibrinolytic activity upon this non-invasive model treponeme. Assuming this proteolytic capacity can be extended to T. pallidum, these results suggest this highly invasive pathogen may employ this rare bacterial fibrinolytic capability to inhibit and overcome the initial localized host-defense response raised against invading treponemes immediately following infection. The definitive identification of surface-exposed proteins in T. pallidum remains highly controversial, primarily due to the technical difficulties associated with working with T. pallidum. The current study employed opsonophagocytosis assays to demonstrate that pallilysin is a target of opsonic antibodies, which is consistent with localization of this protein to the outer leaflet of the outer membrane. Evidence that our laboratory has amassed from previous studies also strongly supports surface exposure of pallilysin in T. pallidum. Specifically, previous studies indicated that: (i) pallilysin is a lipoprotein that is exposed on the surface of our model treponeme, T. phagedenis [18], [19]; (ii) pallilysin-specific antibodies and pallilysin-specific peptides prevent attachment of viable T. pallidum to laminin-coated surfaces [17]; and (iii) viable, pallilysin-expressing T. phagedenis gains the ability to attach to host component-coated surfaces, an interaction that is inhibited by the presence of pallilysin-specific serum [18]. Furthermore, our contention that pallilysin is present on the surface of T. pallidum is consistent with the propensity for pallilysin to be cleaved by host-originating thrombin, to mediate host component binding and degradation, and by our finding that viable, pallilysin-expressing T. phagedenis gains the ability to degrade fibrin, as described herein. It has been shown that pallilysin is expressed during experimental and natural syphilis [16], [59]–[61], however, a prior transcriptome study that analyzed T. pallidum gene expression during experimental rabbit infection also indicated that pallilysin is transcribed at low-to-moderate levels during rabbit infection [62]. This suggests pallilysin' s contribution to treponemal dissemination may result from localized host component degradation in the immediate vicinity of the invading treponemes, a concept supported by our results showing fibrin clot dissolution by pallilysin-expressing T. phagedenis. Collectively, the findings in the current study and previous studies [17]–[19] allow for the generation of a model to explain the potential role of pallilysin in T. pallidum dissemination and invasion (Figure 10). In this model, pallilysin would be produced as a lipoprotein and exported to the treponemal surface through a currently unknown mechanism. The adhesive functionality of pallilysin would facilitate attachment to host ECM components such as the laminin-rich basement membranes lining blood vessels and fibrinogen/fibrin clots. Previously, our laboratory identified specific pallilysin amino acid residues located between P98 and S185 that are critical for laminin-binding [17], suggesting that, at least for laminin, host component attachment is mediated via pallilysin residues C-terminal to the pallilysin thrombin and final autocatalytic cleavage sites. Therefore, cleavage of the pallilysin pro-domain is anticipated to result in release of the host component-bound activated protease from the treponemal surface. In this model, released pallilysin would promote localized host laminin and fibrinogen/fibrin degradation, which would facilitate basement membrane degradation/treponemal entry into the circulation and fibrin clot degradation, respectively, which would in turn promote bacterial dissemination and invasion. All animal studies were approved by the local institutional review boards at the University of Victoria and the University of Washington, and were conducted in strict accordance with standard accepted principles as set forth by the Canadian Council on Animal Care (CCAC), National Institutes of Health and the United States Department of Agriculture in facilities accredited by the American Association for the Accreditation of Laboratory Animal Care and the CCAC. T. pallidum subsp. pallidum (Nichols strain) was propagated in, and extracted from, New Zealand White rabbits as described elsewhere [63]. T. phagedenis biotype Kazan, which had been previously transformed with either the shuttle plasmid pKMR4PEMCS (pKMR) or the full length pallilysin-expressing construct tp0751/pKMR4PEMCS (tp0751/pKMR [18]), were grown in an atmosphere of 97. 5% N2 and 2. 5% H2 in a Coy Laboratory Products anaerobic chamber (Mandel Scientific Company Inc. , Guelph, ON) at 37°C in TYGVS medium [64] supplemented with 20% heat-inactivated rabbit serum (Sigma) in the presence of 10 µg/ml rifampicin and 40 µg/ml erythromycin. Plasminogen-depleted human fibrinogen (Calbiochem) was purchased from VWR International (Mississauga, ON). Thrombin isolated from human plasma and laminin isolated from Engelbreth-Holm-Swarm murine sarcoma basement membrane were purchased from Sigma-Aldrich Canada Ltd. (Oakville, ON). Pallilysin (tp0751) DNA fragments encoding amino acid residues C24-P237 and S78-P237 and tp0453 DNA encoding amino acid residues A32-S287 were PCR amplified from T. pallidum subsp. pallidum (Nichols strain) genomic DNA using the primers listed in Table S1. Pallilysin amplicons were cloned into the T7 promoter destination expression vectors, pDEST-14 and pDEST-17 (Gateway technology, Invitrogen), according to the manufacturer' s instructions. The pDEST-14 and pDEST-17 expression vectors allow for the generation of tagless and N-terminally fused hexahistidine-tagged recombinant proteins, respectively. The negative control, Tp0453, was cloned as previously described [19]. For site-directed mutagenesis of the pallilysin HEXXH motif, the pDEST-17 pallilysin (C24-P237) construct described above was subjected to three independent amino acid substitutions, AEXXH (H198A), HAXXH (E199A), and HEXXA (H202A) using the primers listed in Table S1 and the QuikChange site-directed mutagenesis kit (Stratagene; purchased from VWR International, Mississauga, ON), according to the manufacturer' s instructions. All constructs were confirmed as correct by DNA sequencing. In order to ensure consistent levels of protein purity between wild-type and mutant forms of pallilysin, all recombinant constructs (wild-type and mutant) were expressed and purified under identical conditions, as previously described [19], with the following modifications. FPLC purification of soluble tagless and histidine-tagged recombinant proteins was performed using immobilized metal ion affinity chromatography. Following loading of the recombinant proteins onto 1 ml HisTrap FF affinity columns (GE healthcare, Baie D' Urfe, QC) pre-packed with precharged nickel sepharose 6 fast flow resin, the columns were washed with buffer containing zinc chloride and calcium chloride at pH 7. 0 (20 mM HEPES pH 7. 0,500 mM sodium chloride, 20 mM imidazole, 10 µM zinc chloride, 25 mM calcium chloride, 1% glycerol). Bound recombinant proteins were eluted with a zinc/calcium-containing elution buffer (20 mM HEPES pH 7. 0,500 mM sodium chloride, 250 mM imidazole, 10 µM zinc chloride, 25 mM calcium chloride, 1% glycerol). In order to further ensure that all proteins were free of potential contaminants originating from the E. coli expression strain, concentrated recombinant proteins were loaded onto, and eluted from, a gel filtration column (HiLoad 16/60 Superdex 75; GE Healthcare) using calcium-containing gel filtration buffer (20 mM HEPES pH 7. 0,150 mM sodium chloride, 25 mM calcium chloride, and 1% glycerol). Identical gel filtration fraction ranges were collected from each of the wild-type and mutant pallilysin purifications. All recombinant proteins were deemed to be purified to homogeneity as indicated by SDS-PAGE analysis (Figure 1), mass spectrometry and Edman degradation (as described below). The following recombinant proteins were purified and subsequently used for rabbit immunizations: Pallilysin (amino acids 54–173) [17], BamA (amino acids 25–837) [33], TpN47 (amino acids 34–459) [65], FlaA (amino acids 33–416) [65], and Tp0453 (amino acids 30–287) [61]. Polyclonal antiserum specific for pallilysin (Tp0751), BamA (Tp0326), FlaA (flagellar protein, Tp0249), TpN47 (cytoplasmic membrane lipoprotein, Tp0574), and Tp0453 were raised in New Zealand rabbits as previously described [17], [33]. Normal rabbit serum (NRS) was collected prior to the initial immunizations. Immune rabbit serum (IRS) was collected from rabbits infected with T. pallidum for >90 days. Zinc ion content in wild-type and mutant pallilysin was analyzed at Cantest Ltd. (Burnaby, BC) using Inductively Coupled Plasma-Mass Spectrometry (ICP-MS). Briefly, wild-type pallilysin, the active site mutants (AEXXH [H198A] and HAXXH [E199A]), as well as gel filtration buffer (20 mM HEPES pH 7. 0,150 mM sodium chloride, 25 mM calcium chloride, and 1% glycerol) were dispensed into individual 15 ml polypropylene tubes. The four samples were each reacted with 200 µl of nitric acid, and heated in a 95°C water bath for 1 h. The digested solutions were made up to a final volume of 5 ml with de-ionized water. The samples were analyzed by conventional ICP-MS at a dilution of 5x. The metal element Zn was fully quantified against a certified standard using single point calibration. Sample analysis and operation of the ICP-MS was done according to Cantest' s in-house standard operating procedures. To test for adherence of recombinant wild-type pallilysin (C24-P237), pallilysin active site mutants (AEXXH [H198A], HAXXH [E199A], and HEXXA [H202A]), and Tp0453 (negative control) to fibrinogen and laminin, ELISA-based assays were performed as previously described [16] with the following modifications. All ELISA-based assays were performed using Greiner plates (VWR International). Recombinant proteins were diluted in TBS (Tris-Buffered Saline) and wells were washed with TBS-0. 1% Tween 20 (TBS-T). Primary antibodies were pre-absorbed overnight in 5% skim milk powder in TBS-T using an E. coli lysate expressing an irrelevant His-tagged recombinant protein. Adherent recombinant proteins were detected using a 1∶2,500 dilution of anti-pallilysin or anti-Tp0453 serum (diluted in TBS-T). The secondary antibody (goat anti-rabbit IgG conjugated to HRP [Sigma-Aldrich Canada Ltd. ]) was used at a 1∶10,000 dilution. Wells were developed using the o-Phenylenediamine dihydrochloride (OPD) substrate system (Sigma-Aldrich Canada Ltd.) and the reaction was stopped with a 1∶10 dilution of H2S04. Plates were read at 492 nm with a Spectra Max 5 ELISA plate reader (Molecular Devices, Sunnydale, CA). Statistical analyses were performed using the Student' s two-tailed t test. Fibrinogen degradation assays were performed as previously described [19] with the following modifications. In vitro fibrinogen degradation assays (SDS-PAGE-based) were performed using recombinant pallilysin (C24-P237) (tagless and His-tagged wild-type and His-tagged mutants), pallilysin (S78-P237) (tagless wild-type), pallilysin pre-incubated at 37°C for 22 h (T93-P237) (His-tagged), and the negative control, Tp0453 (His-tagged). Recombinant proteins were incubated with plasminogen-free human fibrinogen in a HEPES-based protease activation buffer at pH 7. 0 (20 mM HEPES, pH 7. 0,25 mM CaCl2). For the fibrinogen degradation fluorescence resonance energy transfer (FRET) assays, recombinant proteins were diluted in HEPES-based activation buffer at pH 7. 0 (20 mM HEPES pH 7. 0,25 mM CaCl2) and added in replicates of between four and eight to sterile Corning Costar 96-well plates (Fisher Scientific). Blank sample wells consisted of HEPES activation buffer in the absence of recombinant proteins. Pallilysin was analysed for the presence of predicted protease cleavage sites using the peptide cutter program [31] (http: //web. expasy. org/peptide_cutter/) and manually analysed for the presence of potential thrombin cleavage sites by comparison to the MEROPS [32] thrombin cleavage site specificity matrix (http: //merops. sanger. ac. uk/cgi-bin/pepsum? id=S01. 217type=P). MEROPS is a freely available online database for proteases and protease inhibitors developed at the Wellcome Trust Sanger Institute. Pallilysin autocatalytic activity was monitored by incubating recombinant wild-type (500 µg) and pallilysin HEXXH active site mutants (500 µg) at 37°C for 24 h. Protein samples (10 µg) were removed hourly throughout the 24 h incubation, mixed with 10 µl of sample buffer (20% glycerol [w/v], 0. 125 M Tris-HCl pH 6. 8,10% β-mercaptoethanol [v/v], 8% SDS [w/v], and 0. 02% bromophenol blue [w/v]), heated at 95°C for 10 min and protein self-lysis analysed by electrophoresis in 15% polyacrylamide gels at a constant voltage of 200 Volts for 1 h. Gels were stained in Coomassie Brilliant Blue R-250 solution (0. 25% [w/v] Coomassie Brilliant Blue R-250,7. 0% [v/v] acetic acid, and 40. 0% [v/v] methanol) and destained in 5% (v/v) acetic acid, 20% (v/v) methanol, 75% (v/v) water. To determine whether pallilysin undergoes intra- or inter-molecular autocatalysis, wild-type full-length pallilysin (C24–P237) (100 µg) was incubated with an equal amount of pallilysin active site mutant (AEXXH [H198A]) at 37°C for 22 h. Protein samples (10 µg) were removed at 0 and 22 h post-incubation and analyzed for autocatalytic processing as described above. Autocatalytic cleavage products were compared to the autocatalytic cleavage products generated from wild-type pallilysin and AEXXH (H198A) mutant pallilysin incubations. Cleavage of recombinant proteins with thrombin was performed using the Thrombin CleanCleave kit (Sigma) according to the manufacturer' s instructions. Briefly, 100 µg of wild-type pallilysin (C24-P237) and the negative control, Tp0453, was incubated at 20°C for 24 h with 30 µl of thrombin-agarose resin (4. 5 µg of thrombin) in the presence of 10 mM calcium chloride. Protein samples (10 µg) were removed at 0,1, 2,4, 6, and 24 h post-incubation and analyzed for thrombin cleavage using SDS-PAGE as detailed above. For Edman sequencing, autocatalytic peptides (20 µg total protein loaded per lane) and thrombin-cleavage fragments (10 µg total protein loaded per lane) were electrophoresed on 16% and 12% precast Novex Tris-glycine gels (Invitrogen), respectively. Proteins were transferred to Millipore Immobilon-P polyvinylidene fluoride (PVDF) membranes (Fisher scientific) at 200 mA for 1 h. Membranes were washed extensively in ultrapure water to remove SDS and glycine and stained in 0. 1% Coomassie Brilliant Blue R250 solution (0. 1% Coomassie Brilliant Blue R250,40% methanol, 1% acetic acid, 58. 9% ultrapure water) for 60 seconds. Membranes were destained in ultrapure water for 1 h at 20°C, followed by a 15 min destain in 50% methanol and 3×10 min destain steps in ultrapure water. Membranes were air-dried for 16 h and pallilysin autocatalytic protein bands and the thrombin-cleavage peptide corresponding to approximately 18 kDa were excised from the PVDF membranes. N-terminal sequence analysis of excised protein bands was performed using Edman chemistry on an Applied Biosystems (AB) Procise liquid-pulse protein sequenator at the Protein and Nucleic Acid (PAN) facility, Stanford University (CA). Briefly, PTH (phenylthiohydantoin) -amino acids were separated on a Brownlee C-18 reverse phase column (2. 1 mm×22 cm) at 55°C using a linear gradient of Buffer A (3. 5% tetrahydrofuran and 2% proprietary AB premix buffer concentrate [approximately 10–20% acetic acid, <10% sodium acetate, <10% sodium hexanesulfonate, 65–75% water]) and 14–44% buffer B (12% isopropanol in acetonitrile) over 17. 6 min. PTH-amino acid analysis was performed using model 610A software, version 2. 1. In order to determine if recombinant pallilysin is capable of degrading insoluble fibrin clots, 60 µl plasminogen-free human fibrinogen (5. 0 mg/ml) was incubated with 50 µl human thrombin (5 U/ml in 50 mM sodium chloride, 0. 1% bovine serum albumin [BSA], 5 mM calcium chloride) for 30 min at 37°C. Excess buffer was removed, the resulting clot weighed (time 0 h), and then the clots were incubated with water (negative control) or 40 µg of each recombinant protein (pallilysin HEXXA [H202A] active site mutant, wild-type pallilysin [C24-P237] and autocatalytically activated wild-type pallilysin [T93-P237]) for 48 h at 37°C. At 0,24, and 48 h post-incubation digital pictures were taken to record fibrin clot size. The assay was repeated seven times. To determine if pallilysin heterologously expressed in the culturable spirochete T. phagedenis is capable of mediating fibrin clot degradation, T. phagedenis transformed with the pallilysin expression construct tp0751/pKMR was compared with T. phagedenis transformed with the shuttle vector alone (pKMR) for the ability to degrade fibrin clots. Transformation of T. phagedenis and demonstration of pallilysin surface expression was previously performed by Cameron and co-workers [18]. Fibrin clots were prepared as described above and incubated anaerobically for 48 h with either TYGVS medium alone, T. phagedenis transformed with pKMR, or T. phagedenis transformed with tp0751/pKMR. When present, human thrombin (Sigma) was added to a final concentration of 2 µg/ml. At 0 h (lag phase; 2. 4×107±7. 6×105 cells ml−1 [mean ± standard error]) and 48 h (late exponential phase; 1. 25×109±2. 2×107 cells ml−1 [mean ± standard error]) post-incubation, digital pictures were taken to record fibrin clot size. Darkfield microscopy with a Nikon Eclipse E600 microscope (Nikon Canada, Mississauga, ON) was used for cell counts and to ensure treponemes remained viable and motile throughout the experiment. The assay was repeated six times. A decrease in clot weight following the 48 h incubations (expressed as percent decrease) was used to quantitate the level of clot degradation. Following 48 h incubation, all buffers, recombinant proteins or cultures were carefully removed from the fibrin clots by pipette aspiration and low speed centrifugation (3 seconds at 100× g) and clot weight was measured and compared with the clot weight at time 0 h. All measurements were performed using a Denver instrument S-114 analytical balance (Fisher Scientific, Ottawa, ON). Statistical analyses were performed using the Student two-tailed t test. Clot quantitation assays were repeated three times. NRS, IRS, and polyclonal antisera generated against recombinant pallilysin, BamA (Tp0326; positive control), TpN47 (Tp0574; negative control), and FlaA (Tp0249; negative control), as described above, were tested for their ability to opsonize T. pallidum using standard opsonophagocytosis assays as previously described [33], [66]. Opsonophagocytosis assays were repeated at least twice. Tp0249 (FlaA); AAC65235, Tp0326 (BamA); AAC65313, Tp0453; AAC65443, Tp0574 (TpN47); AAC65545, Tp0751 (Pallilysin); AAC65720

Syphilis, caused by the spirochete Treponema pallidum, is a chronic sexually transmitted disease which infects 12 million people annually. Treponema pallidum is highly invasive and undergoes widespread dissemination via the circulatory system. Similar to other invasive pathogens, T. pallidum has been shown to express a host-component-degrading protease, pallilysin, that binds and degrades human fibrinogen and laminin, suggesting a role for pallilysin in bacterial dissemination. Here we identify pallilysin active site residues using mutagenesis and show that, unlike wild-type, mutants fail to degrade fibrinogen. We show that pallilysin is converted into a highly proteolytically active form via truncation of a pro-domain through either autocatalytic cleavage or host-derived, thrombin-mediated cleavage. We also demonstrate that recombinant pallilysin enables clot dissolution and that pallilysin expressed on the surface of the non-invasive spirochete Treponema phagedenis confers the ability to degrade fibrin clots. Further, we show that pallilysin is present on the surface of T. pallidum and thus resides in a cellular location that facilitates direct contact with host components. Our study provides insight into the mechanism of interaction between pallilysin and two important coagulation system proteins, fibrinogen and thrombin, and suggests a novel mechanism that T. pallidum may utilize for dissemination during infection.

Abstract Introduction Results Discussion Materials and Methods

biochemistry basement membrane molecular cell biology cell biology proteins extracellular matrix microbial pathogens biology microbiology host-pathogen interaction extracellular matrix adhesions bacterial pathogens pathogenesis

2012

Activation and Proteolytic Activity of the Treponema pallidum Metalloprotease, Pallilysin

Tumors develop through multiple stages, implicating multiple effectors, but the tools to assess how candidate genes contribute to stepwise tumor progression have been limited. We have developed a novel system in which progression of phenotypes in a mouse model of pancreatic islet cell tumorigenesis can be used to measure the effects of genes introduced by cell-type-specific infection with retroviral vectors. In this system, bitransgenic mice, in which the rat insulin promoter (RIP) drives expression of both the SV40 T antigen (RIP-Tag) and the receptor for subgroup A avian leukosis virus (RIP-tva), are infected with avian viral vectors carrying cDNAs encoding candidate progression factors. Like RIP-Tag mice, RIP-Tag; RIP-tva bitransgenic mice develop isolated carcinomas by ∼14 wk of age, after progression through well-defined stages that are similar to aspects of human tumor progression, including hyperplasia, angiogenesis, adenoma, and invasive carcinoma. When avian retroviral vectors carrying a green fluorescent protein marker were introduced into RIP-Tag; RIP-tva mice by intra-cardiac injection at the hyperplastic or early dysplastic stage of tumorigenesis, approximately 20% of the TVA-positive cells were infected and expressed green fluorescent proteins as measured by flow cytometry. Similar infection with vectors carrying cDNA encoding either of two progression factors, a dominant-negative version of cadherin 1 (dnE-cad) or Bcl-xL, accelerated the formation of islet tumors with invasive properties and pancreatic lymph node metastasis. To begin studying the mechanism by which Bcl-xL, an anti-apoptotic protein, promotes invasion and metastasis, RIP-Tag; RIP-tva pancreatic islet tumor cells were infected in vitro with RCASBP-Bcl-xL. Although no changes were observed in rates of proliferation or apoptosis, Bcl-xL altered cell morphology, remodeled the actin cytoskeleton, and down-regulated cadherin 1; it also induced cell migration and invasion, as evaluated using two-chamber transwell assays. In addition, myosin Va was identified as a novel Bcl-xL-interacting protein that might mediate the effects of Bcl-xL on tumor cell migration and invasion. Cancer typically arises by a multi-step process during which cells acquire activating mutations in oncogenes and inactivating mutations in tumor suppressor genes [1]. By comparing cancer tissues with their normal counterparts, it is possible to identify many differences in cellular, molecular, and biological properties. For example, transcriptional profiling and genome sequencing methods have documented alterations in gene expression and somatic mutations in cancers [2,3]. Several types of cancers share common somatic changes or display changes in the same functional pathways, revealing the underlying principles that drive the transformation of normal cells into highly malignant derivatives [4]. Mouse models of human cancers provide important experimental systems for understanding the complexities of human cancer pathogenesis [5,6]. However, in most mouse models, mutant genes have been more clearly implicated in the initiation than in the progression of cancer, even when newer methods, such as Cre/loxP-mediated recombination or tetracycline- or estrogen-based gene regulation, are used to produce a particular gene product conditionally in a tissue-specific and time-controlled manner [7,8]. Moreover, it is time-consuming and expensive to generate alterations in the mouse germ line for each gene of interest. Seeking to improve means for studying later stages in tumorigenesis, we have developed a strategy for assessing causal links between individual genes and specific cellular changes during tumor progression. The strategy is based on the use of a transgene that encodes the receptor for subgroup A avian leukosis virus, TVA, allowing cell-specific infection with avian retroviral vectors during tumor development [9,10]. The RIP1-Tag2 (or RIP-Tag) transgenic mouse [11] was chosen as a platform to study the effects of candidate genes in tumor progression. In this well-characterized model, SV40 T antigen is expressed under the control of the Rat insulin promoter (RIP) in the β cells of pancreatic islets, providing the driving force for tumor initiation by blocking the activities of the Rb and p53 tumor suppressors. The transgenic mice develop multifocal pancreatic β-cell tumors through well-defined stages that resemble lesions of human tumor development and progression. The natural focal distribution of the target tissue, into ∼400 focal nodules, has facilitated identification and quantification of the distinguishable stages in islet tumorigenesis, which include hyperplasia or low-grade dysplasias (“hyperplastic islets”), high-grade angiogenic dysplasias (“angiogenic islets”), encapsulated, noninvasive islet tumors (IT, adenomas), and distinctive grades of invasive carcinoma [12]. Using a RIP-tva transgene to express the TVA receptor in β cells of the pancreas, we demonstrate that candidate genes can be introduced somatically into developing neoplastic β- cell lesions in RIP-Tag; RIP-tva bitransgenic mice by infection with a subgroup A avian retroviral vector, termed RCASBP (subgroup A replication-competent avian leukosis virus with a splice acceptor and the Bryan-RSV pol gene) [13,14], thereby modulating tumorigenesis. By doing so, we have established the functional roles of two candidate progression factors, a dominant-negative form of cadherin 1 (dnE-cad) and Bcl-xL, when introduced stochastically in incipient neoplasias. Our study also reveals an unexpected activity of Bcl-xL in altering the actin cytoskeleton and promoting invasiveness, findings substantiated by analysis of cultured tumor cells. Our results suggest that targeted somatic delivery of candidate progression factors via retrovirus vectors will be useful for identifying factors and dissecting mechanisms that underlie advanced stages of tumorigenesis. The general strategy employed in this work involves equipping pancreatic β cells with TVA, the cell surface receptor for subgroup A avian leukosis viruses [9,10], so that the cells are susceptible to the RCASBP vector [15] derived from those viruses. To that end, we generated transgenic mice that express TVA under the control of the RIP exclusively in the β cells of pancreatic islets (Figure 1A–1C). Three independent lines of RIP-tva mice transmitted the transgene in a simple Mendelian pattern of inheritance. To detect TVA on the surface of pancreatic cells in RIP-tva newborn and adult mice, we performed immunofluorescent staining with a rabbit polyclonal anti-TVA antibody on pancreatic sections. To identify pancreatic islets, we co-stained the sections with goat antisera against PDX1, a transcription factor expressed in pancreatic precursor cells and islet β cells. In both newborn and adult animals, TVA was observed in cells within the islets in all three RIP-tva transgenic lines with similar expression levels (Figure 1B and data not shown). No TVA staining was detected in either exocrine cells from the RIP-tva mice (Figure 1B) or pancreases from non-transgenic animals (Figure 1C). In addition, expression of TVA in other organs was not detected by immunofluorescent staining (data not shown). We conclude that TVA is produced specifically in the pancreatic β islets of RIP-tva transgenic mice, although we cannot rule out the possibility that other cell types may express the transgene at levels below those detectable with the method employed. Examination of pancreatic tissues from newborn and adult animals did not reveal any significant differences in islet morphology between RIP-tva animals and non-transgenic littermates (data not shown). The distribution and presence of cells expressing insulin II (known as insulin), glucagon, somatostatin, and pancreatic polypeptide were also similar in RIP-tva mice and non-transgenic littermates (data not shown), indicating that the viral receptor, TVA, itself did not perturb any of the properties of islet cells that might be relevant to this study. To prepare to study the effects of candidate progression factors, introduced by virus infection, on the course of tumor development, one of the three RIP-tva lines was crossed with the RIP-Tag line to generate bitransgenic RIP-Tag; RIP-tva mice (Figure 2A). It has been shown that the strain backgrounds influence the tumorigenesis of RIP-Tag mice, and the course of tumorigenesis is more consistent in a C57BL/6 background [16,17]. We therefore backcrossed RIP-tva mice from a mixed genetic background for ten generations onto a pure C57BL/6 background with RIP-Tag. We initially verified the co-production of TVA and SV40 T antigen in the islets of 7-wk-old bitransgenic mice (Figure 1D and 1E). We then determined whether the RIP-tva transgene conferred susceptibility to infection by RCASBP viruses in islet cells derived from the bitransgenic mice. Cell proliferation is required for successful infection by RCASBP retroviruses, but adult islet cells have low proliferation rates [18]. To increase the fraction of proliferating cells, cultures were established from β-cell tumors from RIP-Tag; RIP-tva bitransgenic mice, and the tumor cells were infected with a RCASBP vector carrying the green fluorescent protein (GFP) marker (RCASBP-GFP). Analysis by fluorescence microscopy at 1 wk after the second of two rounds of infection revealed that more than 70% of the tumor cells expressed GFP, demonstrating that cells derived from the β-cell tumors were readily infectable by RCASBP in vitro (Figure 2B and 2C). No green fluorescence was detectable in uninfected cells (data not shown). We next evaluated gene transfer in vivo. We chose to infect RIP-Tag; RIP-tva bitransgenic animals at 7 wk after birth, an age when hyperplastic islets develop in RIP-Tag mice. A concentrated preparation of RCASBP-GFP viruses (0. 1 ml; >108 infectious units per milliliter) was delivered by intra-cardiac injection to ensure distribution of viruses to all organs fed by the arterial circulation. Two weeks after infection, the percentage of GFP-expressing TVA-positive cells in the islets was determined by fluorescence-activated cell sorting (FACS) (Figure 2D). Suspensions of single pancreatic cells were prepared as described in Materials and Methods, using a protocol modified from Shih et al. [19] and Dor et al. [20]. Approximately 11%–18% of the cells in these preparations were TVA-positive when isolated from bitransgenic animals, and about 20% of the TVA-positive cells expressed GFP (Figure 2D, lower right panel), implying that the infection efficiency in vivo is around 20% in TVA-positive cells. The green fluorescence was not due to autofluorescence, because almost no fluorescence was observed in pancreatic cells from wild-type or RIP-tva monotransgenic animals injected with RCASBP-GFP or from bitransgenic mice injected with RCASBP viruses encoding human placental alkaline phosphatase (RCASBP-ALPP; previously known as RCASBP-AP) (Figure 2D). The lack of detectable infection in adult RIP-tva monotransgenic animals injected with RCASBP-GFP confirmed that proliferation of the TVA-expressing cells is necessary for successful infection. To ascertain whether the development of β-cell tumours could be accelerated by the somatic delivery of factors previously reported to promote tumor progression in RIP-Tag mice, we infected RIP-Tag; RIP-tva bitransgenic mice with RCASBP vectors that carry cDNA encoding dnE-cad, a mutant protein lacking the extracellular, adhesion-mediating domain of cadherin 1 [21]. In previous studies involving RIP1-Tag2; RIP1-dnE-cad bitransgenic mice, suppression of cadherin 1 function by expression of a dnE-cad cDNA enhanced and accelerated the progression from noninvasive ITs (adenomas) to invasive carcinomas and increased the frequency of otherwise rare lymph node metastasis [22]. We injected high titer stocks (>108 infectious units per milliliter) of RCASBP viruses carrying cDNA encoding dnE-cad into 7-wk-old RIP-Tag; RIP-tva bitransgenic animals by the intra-cardiac route (Figure 3A). Nine weeks later, pancreases from 16-wk-old mice were harvested for histological staging and grading of the lesions according to established criteria [23]. In brief, tumors were scored as IT (noninvasive tumor; >1 mm in diameter; Figure 3B); invasive carcinoma type 1 (IC-1; >1 mm in diameter; focal regions of invasion; Figure 3C); or invasive carcinoma type 2 (IC-2; variable sizes; widespread invasion; Figure 3D). Pancreatic sections were also subjected to immunohistochemical tests for the expression of synaptophysin, a neuroendocrine marker found in all types of islet cells, and for insulin II, a β-cell marker. By counting and classifying pancreatic lesions (Table 1), we found that (i) the average tumor incidence in bitransgenic mice infected with RCASBP-dnE-cad (n = 11) was 1. 5-fold higher than that in bitransgenic mice injected with the control RCASBP-ALPP viruses (n = 6); (ii) mice injected with RCASBP-dnE-cad (n = 11) exhibited a higher incidence of invasive carcinomas than mice injected with the control viruses (n = 6); and (iii) mice injected with RCASBP-dnE-cad had a higher tumor burden than controls (p = 0. 003, n = 6 for each group, Wilcoxon rank sum test). By staining pancreatic lymph nodes with antisera against insulin II and synaptophysin to search for metastatic β-cell tumors, we found that six of the 15 mice infected with RCASBP-dnE-cad developed metastases to the pancreatic lymph nodes (Table 1; Figure 4A, upper panel); metastases were not detected in monotransgenic RIP-Tag mice of the same age or in RIP-Tag; RIP-tva bitransgenic mice infected with RCASBP-ALPP. No metastases to other distant organs (liver, kidney, lung, heart, or thymus) were observed in the animals infected with RCASBP-dnE-cad or RCASBP-ALPP, consistent with the findings in RIP1-Tag2; RIP1-dnE-cad bitransgenic mice. Taken together, these data indicate that the RIP-Tag; RIP-tva mice can be used to assess the effects of genes of interest on tumor progression in vivo, with targeted somatic delivery of dnE-cad, largely recapitulating the effects seen with a traditional transgenic approach that constitutively expresses the candidate progression factor in all of the target cells throughout ontogeny [22]. We next investigated a second progression factor, Bcl-xL, a member of the anti-apoptotic subgroup of the BCL2 family of apoptotic regulators [24] via somatic gene transfer. In previous studies, constitutive expression of Bcl-xL via a RIP7-Bcl-xL transgene was found to suppress apoptosis and alternatively accelerate or enable islet tumorigenesis in mouse models of cancer evoked by SV40 T antigen [25] or MYC [26], respectively. We sought to assess the effects of expressing exogenous Bcl-xL in sporadic cells, beginning after neoplastic development had ensued, as contrasted to its constitutive expression throughout tumorigenesis in the bitransgenic RIP1-Tag2; RIP7-Bcl-xL mice. We injected high titer stocks of RCASBP viruses carrying cDNA encoding Bcl-xL into 7-wk-old RIP-Tag; RIP-tva bitransgenic mice, and subsequently harvested pancreases at 16 wk of age for histopathology (Table 1). We observed that (i) the average tumor incidence in bitransgenic mice infected with RCASBP-Bcl-xL (n = 10) was 2-fold higher than that in bitransgenic mice injected with the control RCASBP-ALPP viruses (n = 6); (ii) mice infected with RCASBP-Bcl-xL (n = 10) exhibited a higher incidence of invasive carcinomas than did mice that received the control viruses (n = 6); and (iii) mice infected with RCASBP-Bcl-xL displayed a higher tumor burden than controls (p = 0. 003, n = 6 for each group, Wilcoxon rank sum test). We also found that seven of the 15 mice infected with RCASBP-Bcl-xL developed metastases to the pancreatic lymph nodes, but no metastases to other organs (Table 1; Figure 4A, lower panel). Therefore, overexpression of Bcl-xL, beginning in hyperplastic or early dysplastic lesions, after tumorigenesis had been initiated, increased tumor burden, incidence, and invasion, and facilitated lymph node metastasis (p = 0. 05, Fisher' s exact test). Histological analysis of the tumors revealed that insulin II expression was not detectable in a small subset (∼4. 5%) of IC-2 in RIP-Tag; RIP-tva mice infected with RCASBP-Bcl-xL or RCASBP-dnE-cad (Figure 4B, right panels). All tumors stained with antisera recognizing the neuroendocrine marker synaptophysin (Figure 4B, left panels), suggesting a process of de-differentiation or transformation of less mature cells into an “anaplastic” islet carcinoma, a phenotype not observed in bitransgenic animals that were uninfected or infected with control vectors. Bcl-xL is known to protect cells from apoptosis [24]. We therefore investigated whether the increased invasive and metastatic behavior of the tumors from mice infected with RCASBP-Bcl-xL was due to increased cell survival, producing greater cell numbers and a higher probability of tumor spread, or due to some other properties of Bcl-xL. We observed no significant differences in proliferative or apoptotic indices by staining tumors from 16-wk-old mice previously infected with RCASBP-ALPP, RCASBP-Bcl-xL, or RCASBP-dnE-cad with antisera against a proliferation marker, MKI67 (antigen identified by monoclonal antibody Ki67), or against activated caspase 3 (Kruskal-Wallis nonparametric test, three-way test, p = 0. 79 and 0. 69, respectively) (Figure 5A). These results suggest that the increased invasive and metastatic properties were not due to the effects of Bcl-xL on proliferation or survival of the cancer cells at the stages of adenomas and carcinomas. However, it remains possible that Bcl-xL provides some protection from apoptosis at the metastatic stage or earlier stages in tumorigenesis, as cell death is greatly reduced in both hyperplastic and angiogenic islets in RIP1-Tag2; RIP7-Bcl-xL bitransgenic mice that constitutively express high levels of Bcl-xL in all β cells in normal islets and at all stages of tumorigenesis [25]. Since only a small fraction of the β cells in premalignant lesions became somatically infected in our experiments, such effects would likely be masked by the preponderance of uninfected cells in those lesions. To determine whether the effects of Bcl-xL occurred early or late in the neoplastic process, we examined sections of pancreases from RIP-Tag; RIP-tva mice at 2 and 5 wk after RCASBP-Bcl-xL delivery (9 and 12 wk of age, respectively), as well as 9 wk after infection (Figure 5B and 5C). This experiment not only confirmed the findings presented in Table 1 at late times after infection, but also showed that the effects of Bcl-xL could be observed as early as 2 wk after viral delivery. RIP-Tag; RIP-tva mice infected with RCASBP-Bcl-xL exhibited higher tumor burdens (for three ages p ≤ 0. 009, Wilcoxon rank sum test) and a greater tumor incidence at all three ages (9,12, and 16 wk of age) than did the bitransgenic control mice infected with RCASBP-ALPP (Figure 5B). In addition, at all ages the numbers of IC-1 and IC-2 were higher in mice infected with RCASBP-Bcl-xL than in mice that received RCASBP-ALPP; furthermore, the small invasive carcinomas appeared before ITs in RIP-Tag; RIP-tva mice at 2 wk after RCASBP-Bcl-xL delivery (Figure 5C). These findings suggest that some of the hyperplastic or angiogenic islets develop highly invasive carcinomas (variable sizes) without progressing through a noninvasive IT stage (>1 mm in diameter), a phenotype first observed in bitransgenic mice that express both SV40 T antigen and a receptor tyrosine kinase, the insulin-like growth factor (IGF) 1 receptor (IGF1R), in the pancreatic islets [23]. No significant effects of Bcl-xL on proliferative or apoptotic indices in the tumors were observed at 2 and 5 wk after viral delivery (data not shown). Thus, introducing RCASBP-Bcl-xL into RIP-Tag; RIP-tva mice with islet cell hyperplasia has a substantial impact on tumor burden, tumor incidence, and invasive potential within 2 wk after infection. Since infection of bitransgenic mice with RCASBP-Bcl-xL increases the number of tumors by about 2-fold at 16 wk of age (Table 1; Figure 5), we anticipated that about half of the tumors would contain avian proviral DNA as evidence of infection that confers a selective growth advantage. To test this prediction, we isolated genomic DNA from freshly microdissected tumors or from tumors in paraffin-embedded sections, and we then performed PCR using primers specific to RCASBP sequences. Using materials from the experiment shown in Figure 5, we detected RCASBP DNA in eight of 11 tumors (72%), six of 12 tumors (50%), and ten of 25 tumors (40%) harvested 2,5, and 9 wk after viral delivery, respectively (Figure 6A and data not shown). In contrast, RCASBP DNA was detectable in only two of 20 tumors from mice sacrificed 9 wk after infection with the control RCASBP-ALPP viruses (Figure 6B), which is consistent with an in vivo infection efficiency of at least 10%. These results support the hypothesis that infection leading to production of Bcl-xL, but not the control infection with RCASBP-ALPP, provides a selective growth advantage to islet cells during tumorigenesis in RIP-Tag; RIP-tva mice. Because only small amounts of DNA were obtained from these small tumors, we were unable to assess the clonality of tumors containing RCASBP-Bcl-xL by restriction mapping and thereby determine if they arose from the expansion of single infected cells. In addition, no suitable anti-Bcl-xL antibodies were found to identify tumors expressing Bcl-xL by immunohistochemistry. We observed lymph node metastases only in mice infected with RCASBP-Bcl-xL or RCASBP-dnE-cad, not in mice infected with RCASBP-ALPP, so we considered the possibility that metastases had arose from tumor cells expressing Bcl-xL or dnE-cad. We confirmed the presence of RCASBP proviral DNA in all nine lymph node metastases from mice infected with RCASBP-Bcl-xL or RCASBP-dnE-cad (Figure 6C). We cannot, however, claim that all the metastatic cells were derived from RCASBPP-Bcl-xL- or RCASBP-dnE-cad-infected cells. Since tumor invasion and metastasis often involve modulation of the actin cytoskeleton to increase cell motility [27,28], we evaluated the actin cytoskeleton in tumors from RIP-Tag; RIP-tva mice infected with RCASBP-Bcl-xL, using an immunofluorescence assay with rhodamine-phalloidin to display filamentous actin (F-actin). We observed a dramatic rearrangement of the actin cytoskeleton in invasive carcinomas at 2 and 5 wk after infection with RCASBP-Bcl-xL, but not in tumors from age-matched animals infected with RCASBP-ALPP (Figure 7A), suggesting that Bcl-xL remodels the actin cytoskeleton to increase cell motility. To further understand the effects of Bcl-xL on tumor cells, we infected a β-cell tumor cell line (βTC-N134) derived from an uninfected RIP-Tag; RIP-tva mouse with RCASBP-Bcl-xL. In addition, we infected the tumor cells with RCASBP-dnE-cad or RCASBP-GFP, to control for any effects of retroviral infection. Two weeks after infection with RCASBP-Bcl-xL, the tumor cells became more elongated compared to the uninfected parental cells and to the cells infected with either RCASBP-GFP or RCASBP-dnE-cad (Figure 7B). Staining with rhodamine-phalloidin demonstrated decreased cortical actin and a less organized epithelial sheet in tumor cells infected with RCASBP-Bcl-xL, but not in the other cultures (Figure 7C). Taken together, these results suggest that Bcl-xL promotes remodeling of the actin cytoskeleton, affecting cell shape and adhesion. To examine whether Bcl-xL increased the ability of morphologically altered cells to migrate or invade, we performed two-chamber migration and invasion assays. In the migration assay, uninfected tumor cells (βTC-N134) or tumor cells infected with RCASBP-GFP, RCASBP-Bcl-xL, or RCASBP-dnE-cad were seeded in the upper chambers of transwell inserts with 8-μm porous polycarbonate membranes. We then measured cell migration along a serum gradient through the membrane to the bottom of the chambers. After a 72-h incubation, the number of migratory tumor cells infected with RCASBP-Bcl-xL or RCASBP-dnE-cad was approximately 1,000-fold higher than that of uninfected cells or cells infected with RCASBP-GFP (Figure 8A). In the invasion assay, 8-μm porous polycarbonate membranes were replaced with polyester membranes coated with basement membrane matrix, Matrigel. The tumor cells infected with RCASBP-Bcl-xL invaded through Matrigel about 20-fold more efficiently than those infected with RCASBP-dnE-cad, and none of the uninfected parental cells or the cells infected with RCASBP-GFP had the ability to invade (Figure 8B). The results reveal that Bcl-xL has profound effects on cell migration and invasion in vitro. To ensure that any increase in cell migration and invasion in vitro was not simply due to an increase of cell number by the anti-apoptotic effect of Bcl-xL, immunocytochemistry using antibodies for Ki67 and activated caspase 3 was performed to assay cell proliferation and cell survival. The frequency of Ki67-positive cells and activated-caspase-3-positive cells remained similar in all three of the infected tumor cell lines compared to the uninfected parental cell line (Figure 8C and 8D). These results, consistent with our in vivo data (Figure 5A), suggest that Bcl-xL is able to promote cell migration and invasion without changing cell proliferation or survival. In addition to the cytoskeleton rearrangement, loss of cadherin 1 can also confer invasive properties on tumor cells [22]. To examine if the expression of cadherin 1 was altered in the tumor cells infected with RCASBP-Bcl-xL, we measured cadherin 1 levels by Western blot analysis using an anti–cadherin 1 antibody (clone 36) (Figure 8E). Cadherin 1 protein was much less abundant in the tumor cells infected with RCASBP-Bcl-xL, suggesting that overexpression of Bcl-xL down-regulates cadherin 1. We also observed moderately increased amounts of snail homolog 1 protein (Figure 8E), a transcriptional repressor of cadherin 1 [29,30]. We also performed immunoprecipitation (IP) experiments using two anti–cadherin 1 antibodies (clones ECCD-2 and 36) and whole cell extracts prepared from the uninfected tumor cells or from the tumor cells infected with RCASBP-dnE-cad or RCASBP-Bcl-xL. The immuoprecipitates were subjected to Western blotting using the clone 36 antibody, which was raised against the cytoplasmic domain of human cadherin 1 and cross-reacts with the mouse homolog. ECCD-2 was raised against a mouse liver cadherin 1 fragment [31], and it did not recognize dnE-cad (Figure 8F, lane 4). The levels of endogenous cadherin 1 were not affected by overexpression of dnE-cad in the tumor cells infected with RCASBP-dnE-cad (Figure 8F, lane 1). In marked contrast, cadherin 1 was considerably reduced in the tumor cells infected with RCASBP-Bcl-xL (Figure 8F, lanes 3,6, and 10). To determine whether regulation of cadherin 1 levels was a result of lowered mRNA levels, we surveyed tumor cell RNAs with oligonucleotide arrays (Affymetrix). We observed that the levels of cadherin 1 RNA were 36-fold lower in RIP-Tag; RIP-tva tumor cells infected with RCASBP-Bcl-xL than in uninfected parental cells (data not shown), suggesting that Bcl-xL is likely to lower expression of cadherin 1 at the transcriptional level. Surprisingly, we did not detect a significant reduction of cadherin 1 on the tumor cells infected with RCASBP-Bcl-xL by immunostaining assays, using the two anti–cadherin 1 antibodies (data not shown), raising questions about the specificity of the antibodies used in the assays. Taken together, overexpression of Bcl-xL reduces the levels of cadherin 1 RNA and protein, and loss of cadherin 1 is likely to contribute to cell migration and invasion. However, suppression of cadherin 1 function by dnE-cad is not as efficient as Bcl-xL in promoting invasion of β-cell tumor cells (Figure 8B). The assembly and organization of the actin cytoskeleton can be regulated by small guanosine triphosphatases (GTPases) [32]. We evaluated whether the activities of four members of the small GTPase family (RHO, CDC42, RAC1, and RAS) were changed in the tumor cells infected with RCASBP-Bcl-xL, by measuring the signaling activities of these four small GTPases as reflected in the binding of GTP. Their specific effector proteins were expressed as glutathione S-transferase (GST) –fusion proteins to pull down GTP-bound GTPases, which were detected by Western blotting using antibodies against specific GTPases. However, no significant changes in the signaling strength of RHO, CDC42, RAC1, or RAS were detected in the tumor cells infected with RCASBP-Bcl-xL, compared with the uninfected parental cells (data not shown), indicating that altered activities of these four proteins were not responsible for the remodeling of the actin skeleton in the tumor cells infected with RCASBP-Bcl-xL. We cannot exclude the possibilities that small GTPase activities vary spatially within the cells or that other members of the family are responsible for the reorganization of the actin cytoskeleton when Bcl-xL is overexpressed. To further investigate the mechanism by which Bcl-xL increases cell motility, we sought to identify Bcl-xL-interacting proteins by mass spectrometry. Bcl-xL is located both in the cytosol and in mitochondrial membranes [33,34], whereas BCL2, a related anti-apoptotic protein, is exclusively membrane-bound [35]. We postulated that the cytosolic fraction of Bcl-xL might contribute to its effects on cell shape, motility, and invasiveness, so we prepared whole cell extracts from tumor cells infected with RCASBP-Bcl-xL in two buffer conditions, and used an anti-Bcl-xL antibody to co-precipitate interacting proteins. The antibody-bound complexes were eluted from beads by boiling, separated by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE), and visualized by silver staining. Although many protein bands were present in the IP, we observed that the intensities of several protein bands, including Bcl-xL itself, were enhanced with increasing amounts of an anti-Bcl-xL antibody (Figure 9A). These putative Bcl-xL-interacting bands were excised and subjected to tryptic digestion, and peptides were analyzed by mass spectrometry. The identities of the proteins were BAX, myosin Va, and myosin, heavy polypeptide 9, non-muscle isoform 1. The current view of apoptotic regulation is that Bcl-xL and its relatives bind to and thereby inhibit the pro-apoptotic activity of BAX and BAK1, which are otherwise poised to initiate the apoptotic cell death cascade [36,37]. Activated BH3 sensors, such as BAD, bind to Bcl-xL-related proteins, relieving suppression of BAX and BAK1. We confirmed the interaction between BAX and Bcl-xL in the β-cell tumor cells infected with RCASBP-Bcl-xL by IP-Western blotting (Figure 9B). We also observed an interaction between Bcl-xL and BAD, but not phosphorylated BAD (Figure 9B). The significance of the reduction of phosphorylated BAD in the tumor cells infected with RCASBP-Bcl-xL remains to be determined. We obtained an antibody against myosin Va and confirmed the presence of myosin Va in the anti-Bcl-xL IP (Figure 9B), suggesting that the interaction between Bcl-xL and myosin Va was specific and significant. No suitable antibodies against the other implicated myosin were found, so Western blotting to verify its specificity by IP from the cells infected with RCASBP-Bcl-xL was not possible. Furthermore, the association between Bcl-xL and myosin Va was unlikely to be mediated by the actin cytoskeleton, because actin was not present in Bcl-xL IP (Figure 9B). Myosins have been implicated in cell movement and migration [38–40], and therefore it is possible that Bcl-xL promotes cell motility through its interaction with myosins. The studies presented here demonstrate a system to transfer cDNAs into premalignant lesions in mice to analyze their effects on tumor progression. This system utilizes RCASBP avian retroviral vectors to deliver cDNAs into pancreatic β cells in RIP-Tag; RIP-tva bitransgenic mice during tumorigenesis. This approach has the advantage of introducing somatic genetic changes in a tissue-specific and time-controlled fashion, which more faithfully mimics sporadic tumor development. It also avoids any potential perturbation of normal tissue formation due to the ectopic expression of the gene of interest during development. It is much faster to generate vectors carrying genes of interest than to generate transgenic mice. In addition, using cell lines derived from β-cell tumors in RIP-Tag; RIP-tva bitransgenic mice, further biochemical and phenotypic analysis can be performed, taking advantage of the high efficiency of viral infection in vitro. The low efficiency of in vivo infection has been a limitation of using the RCAS-TVA system [13,14]. Here, we were able to achieve 10%–20% in vivo infection efficiency in hyperplastic pancreatic islet β-cell lesions using high viral titer (>108 infectious units per milliliter) and the intra-cardiac route to deliver the RCASBP viruses into mice. Since not all the premalignant lesions are infected with the RCASBP vectors and 2%–4% of islets develop into tumors in the RIP-Tag; RIP-tva bitransgenic mice without viral infection, only the factors that confer a selective advantage over the natural course of tumorigenesis in RIP-Tag; RIP-tva mice can be identified. Moreover, the percentage of RCASBP-Bcl-xL-positive tumors decreases over time as the uninfected RIP-Tag; RIP-tva cells gradually develop into tumors. In particular, RCASBP vectors carrying genes that promote metastasis will be most easily recognized, because metastasis to pancreatic lymph nodes or other organs does not normally occur in RIP-Tag; RIP-tva mice. Although tumors are likely to be clonally expanded from single cells infected with RCASBP vectors carrying progression factors, we were not able to show whether the tumors were clonal because of the limited DNA materials isolated from the small tumors in this study. We delivered RCASBP viruses carrying tumor progression factors into RIP-Tag; RIP-tva bitransgenic animals at 7 wk of age, when hyperplastic islets develop. Analogous to the results obtained from RIP1-Tag2; RIP1-dnE-cad bitransgenic mice [22], we demonstrated that introduction of RCASBP-dnE-cad into the hyperplastic islets of RIP-Tag; RIP-tva animals promoted invasive tumor formation and metastases to the pancreatic lymph nodes. Mice at other ages could also be used to study whether the time at which the candidate factors are introduced influences the outcome, as long as islet cells undergo cell division to allow infection. For example, the transforming growth factor β1 (TGFB1) receptor can act as a tumor suppressor at early stages of tumor development, but at later stages TGFB1 responsiveness promotes invasion and metastasis [41,42]. Moreover, a combination of several RCASBP viruses encoding different factors of interest could be used for infection simultaneously or sequentially, and the proviral DNA could be isolated from especially metastatic malignant tumors to identify single factors or combinations of factors that may contribute to the malignancy. However, this approach is limited by the inability of the RCASBP vector to accommodate cDNAs greater than 2. 5 kb in size. The effects of overproducing Bcl-xL via somatic gene transfer were provocative and unexpected. First, no significant protection against apoptosis was found either in the tumors from mice infected with RCASBP-Bcl-xL or in the tumor cells infected with RCASBP-Bcl-xL in vitro. These findings are quite distinct from the demonstrable anti-apoptotic effect seen when Bcl-xL was expressed via a transgene in all islet β cells throughout tumorigenic ontogeny in the RIP1-Tag2 model [25], as well as in MYC transgenic mice [26]. It remains possible that Bcl-xL contributes to cell survival at earlier stages (hyperplasia/angiogenesis) or a later stage (metastasis) in mice infected with RCASBP-Bcl-xL. Additionally, we cannot exclude episodic effects on apoptosis when tumors were forming, as opposed to chronic effects detectable at the later time points of our analysis. This apparent insensitivity of β-cell tumors to the anti-apoptotic effect of Bcl-xL deserves future investigation. Notably, a similar phenotype of increased invasiveness without suppression of apoptosis has been observed when another apoptotic modulator was overexpressed in the RIP1-Tag2 mice. Exogenous production of IGF1R for the IGF1/2 survival factors in all islet β cells caused increased invasiveness and lymph node metastasis, with increased (not decreased) apoptosis in the premalignant stages, while having no impact on the tumor stages [23]. Yet elimination of the gene encoding IGF2 produced highly apoptotic tumors with reduced growth, demonstrating the importance of anti-apoptotic signaling for islet tumorigenesis [43]. The similar consequences of ectopically expressing Bcl-xL and IGF1R may imply mechanistic similarity that should be further explored. Second, we observed that the actin cytoskeleton was rearranged in invasive tumors infected with RCASBP-Bcl-xL, presumably contributing to increased motility. In addition, tumor cells infected with RCASBP-Bcl-xL in vitro displayed altered morphology, abnormal cortical distribution of the actin cytoskeleton, and elongated cell shape. The morphological changes in tumor cells infected with RCASBP-Bcl-xL may contribute to enhanced cell migration and invasion as observed in the two-chamber assays. Third, as measured by Western blotting and cDNA microarray expression profiling, the epithelial cell–cell junction protein, cadherin 1, was down-regulated and a transcriptional repressor of cadherin 1, snail homolog 1, was up-regulated in the tumor cells infected with RCASBP-Bcl-xL. These changes may contribute to the rearrangement of the actin cytoskeleton and diminish the strength of cellular adhesion. Indeed, tumor cells infected with RCASBP-Bcl-xL exhibited a strong capability to migrate through transwell membrane and to invade through Matrigel, but the mechanism by which Bcl-xL regulates cadherin 1 remains to be determined. Finally, the identification of a novel Bcl-xL-interacting protein, myosin Va, may help to explain the pro-invasive activity of Bcl-xL. Myosins belong to a superfamily of actin-based motor proteins comprising at least 15 classes [39]. There are two main groups of myosins: the conventional, well-characterized myosin II class of muscle and non-muscle cells and the unconventional myosins. The myosin II class is referred to as “conventional myosin” because for many years this was the only class of myosin known; members of this class are able to form filaments and are involved in muscle contraction, cell migration, and cytokinesis. The functions of most of the unconventional myosins are not known, but some have been shown to participate in the extension of processes at the leading edge of crawling cells. For example, myosin I is required for formation of pseudopod extensions in the amoeba Dictyostelium, and myosin V for extension of filopodia in neurons [44,45]. By comparing the effects of suppressing the function of cadherin 1 with dnE-cad and overproducing Bcl-xL, we have highlighted the provocative pro-invasive activity of Bcl-xL. Cadherin 1 is a well-established barrier to invasive growth in many other epithelial cancers [46], and its suppression enables invasive growth of pancreatic β-cell tumors in transgenic mice [22]. Yet Bcl-xL produced a more highly invasive phenotype in a Matrigel invasion assay of cultured tumor cells than did dnE-cad, supporting the proposition that Bcl-xL has roles other than suppression of cadherin 1 that enhance the invasive growth phenotype. The levels of endogenous cadherin 1 in these two types of cells could contribute to the differences. Whereas endogenous cadherin 1 protein levels were not affected when the dominant-negative form was introduced into the cells, cadherin 1 was down-regulated in the tumor cells infected with RCASBP-Bcl-xL. This effect could reduce cell–cell adhesion and also affect the organization of the actin cytoskeleton. On the other hand, the frequency of lymph node metastasis was similar in RIP-Tag; RIP-tva mice infected with RCASBP-Bcl-xL and RCASBP-dnE-cad. It is possible that the β-cell tumor microenvironment in vivo is less discriminating than Matrigel, a solubulized basement membrane preparation extracted from a mouse sarcoma [47]. The BCL2 family proteins function in carcinogenesis by preventing apoptosis of tumor cells, instead of promoting cell proliferation [48]. Of the BCL2 family members, BCL2 and Bcl-xL are most closely related to each other, and repress cell death through common mechanisms [49,50]. However, several lines of evidence indicate that they are not functionally equivalent in tumorigenesis. In primary breast cancer, overexpression of Bcl-xL is associated with higher tumor grade and nodal metastasis, and overexpression of Bcl2 is correlated with lower tumor grade and smaller tumor size [51]. Moreover, nude mice with orthotopic implants of human breast cancer cells transfected with Bcl-xL, but not BCL2, develop lymph node metastasis [52]. Our findings suggest that Bcl-xL enhances cell motility, remodels the actin cytoskeleton, down-regulates cadherin 1, and interacts with myosin Va without affecting cell proliferation or apoptosis. It remains to be investigated whether these properties are unique to Bcl-xL among the BCL2 family members. In conclusion, we have developed a mouse model to assess the effect of candidate genes in tumor progression, without the need to generate conditional transgenic lines in which genes of interest are expressed at different stages of tumorigenesis. Thus far, we have employed this approach to improve our knowledge of the contribution of Bcl-xL to tumor invasion and metastasis. We postulate that Bcl-xL has a pro-invasive function other than its anti-apoptotic activity. Given that many somatic gene mutations and altered expression of thousands of genes have been discovered in cancers, it will not be a simple task to verify that each candidate gene is important during carcinogenesis. By delivering libraries of retroviruses encoding candidate factors or inhibitory RNAs into transgenic mice, it may be possible to screen many of the genes for effects during tumorigenesis in a variety of tissues. The elastase-tva transgene construct [53] was digested with BamHI and HindIII to release a 800-bp tva cDNA fragment encoding the glycosylphosphatidylinisotol-anchored form of the receptor. This cDNA fragment was cloned into a BglII- and HindIII-digested pSG5 vector with an expanded multiple cloning site (gift of E. Emison, Johns Hopkins University). The tva cDNA was then released by XbaI/HindIII digestion of the pSG5-tva plasmid and subcloned into a XbaI/HindIII-digested RIP-DIPA plasmid that contains the RIP 5′ to the XbaI site and the SV40 small t intron and terminator sequences 3′ to the HindIII site. The resulting RIP-tva transgene construct was released from the pBR322 vector backbone by digestion with BamHI, purified, and resuspended in TE for pronuclear injection. RIP-tva transgenic founders on a CBA/CAJ × C57BL/6 mixed genetic background were backcrossed for ten generations onto a pure C57BL/6 background. The RIP-tva mice were crossed with the previously described RIP1-Tag2 mice [11] to obtain RIP-Tag; RIP-tva mice. All mice were housed in accordance with institutional guidelines. Genotypes were determined by PCR using tail DNA. PCR primers for TVA were 5′-GCCCTGGGGAAGGTCCTGCCC-3′ and 5′-CTGCTGCCCGGTAACGTGACCGG-3′. A pancreatic β-cell tumor cell line (βTC-N134) was derived from a tumor from a RIP-Tag; RIP-tva bitransgenic animal at 16 wk of age, and maintained in Dulbecco' s Modified Eagle Medium (DMEM) supplemented with 10% fetal bovine serum (FBS), 0. 2 mM L-glutamine, and 1% penicillin/streptomycin in a humidified 37 °C incubator under 5% CO2; method modified from [54]. RCASBP is a replication-competent avian leukosis virus with a splice acceptor and the Bryan-RSV pol gene. RCASBP-ALPP has been described previously [55]. RCASBP-GFP was a gift from Maureen Peters and Constance Cepko, Harvard Medical School. Human Bcl-xL cDNA was a gift from Stanley Korsmeyer, Dana-Farber Cancer Institute. RCASBP-dnE-cad (myc-tagged, mouse origin) and RCASBP-Bcl-xL were generated by Yi Li and William Pao, respectively, in the Varmus laboratory. The production of the dnE-cad and Bcl-xL was verified by Western blot analysis. Chicken fibroblasts DF-1 [56,57] transfected with RCASBP vectors were maintained in DMEM supplemented with 10% FBS, 0. 2 mM L-glutamine, and 1% penicillin/streptomycin in a humidified 37 °C incubator under 5% CO2. For in vitro infection, RCASBP viral supernatant was passed through a 0. 45-μm filter to obtain cell-free viruses and used as growth medium for the RIP-Tag; RIP-tva pancreatic β-cell tumor cell line once a day for 2–3 d. For in vivo infection, viral supernatant was passed through 0. 45-μm filters and was concentrated by high-speed ultracentrifugation at 23,000 rpm for 1. 5 h before intra-cardiac injection into mice. Intra-cardiac injection was performed as described [58]. Viral titer was determined by end-point dilution of DF-1 producer cells. PCR primers for RCASBP were 5′-ACCGGGGGATGCGTAGGCTTCA-3′ and 5′-CCGCAACACCCACTGGCATTACC-3′. Tissues were removed and either fixed in 10% buffered formalin overnight at room temperature or immediately placed in OCT and frozen on dry ice. Fixed and frozen tissues were processed and cut into 5-μm and 10-μm sections, respectively, at Histoserv (http: //www. histoservinc. com/index. php). Formalin-fixed/paraffin-embedded sections were deparaffinized and rehydrated by passage through a graded xylene/ethanol series before staining. Cells were cultured on glass chamber slides for 3 d before fixation in 4% paraformaldehyde in PBS for 20 min, and were permeabilized by 0. 1% Triton X-100 (in PBS) for 4 min. Immunochemistry with VECTASTAIN Elite ABC Kit (Vector Laboratories, http: //www. vectorlabs. com/) was performed according to the manufacturer' s instructions. Primary antibodies used were mouse anti–SV40 T antigen (1: 400; Calbiochem, http: //www. emdbiosciences. com/html/CBC/home. html), rabbit anti-insulin (1: 600; Immunostar, http: //www. immunostar. com/), rabbit anti-synaptophysin (1: 100; Dako, http: //www. dako. com/), rabbit anti-Ki67 (1: 1,000; Novocastra, Vision BioSystems, http: //www. leica-microsystems. com/biosystems. html), and rabbit anti–activated caspase 3 (1: 100; Cell Signaling Technology, http: //www. cellsignal. com/). Frozen sections were air-dried overnight at room temperature and fixed in cold acetone for 10 min. Formalin-fixed/paraffin-embedded sections were deparaffinized and rehydrated by passage through a graded xylene/ethanol series. Cells were cultured on glass chamber slides for 3 d before fixation in 4% paraformaldehyde in PBS for 20 min, and permeabilized by 0. 1% Triton X-100 (in PBS) for 4 min. After blocking, slides were incubated with primary antibodies overnight at 4 °C or for 1 h at room temperature. After three washes of TBS-T, slides were incubated with secondary antibodies for 1 h at room temperature. Primary antibodies used were rabbit anti-TVA (1: 200; gift of Andrew Leavitt, University of California San Francisco), goat anti-mouse PDX1 (1: 2,000; gift of Chris Wright, Vanderbilt University), and rhodamine-phalloidin (1: 40; Invitrogen, http: //www. invitrogen. com/). TRITC-donkey anti-rabbit IgG (1: 200) and FITC-donkey anti-goat IgG (1: 200) were purchased from Jackson ImmunoResearch Laboratories (http: //www. jacksonimmuno. com/). To stain DNA, slides were incubated with DAPI (4′, 6-diamidino-2-phenylindole; 5 μg/ml in PBS) for 15 min at room temperature. Pancreases were perfused and digested into small pieces with collagenase (Sigma-Aldrich, http: //www. sigmaaldrich. com/) at 37 °C for 15 min, and were centrifuged in a Ficoll gradient (11%, 20. 5%, 23%, and 25% [w/v]) as described [19]. Partially digested pancreatic tissues in the top two layers of the gradient were further digested and dispersed into single-cell suspension as described [20]. After blocking with 3% donkey serum at 4 °C for 30 min, cells were incubated with rabbit anti-TVA antibodies at 4 °C for 30 min, incubated with donkey Cy5-conjugated anti-rabbit antibodies at 4 °C for 30 min, and analyzed for Cy5 and GFP by FACS using a BD Biosciences (http: //www. bdbiosciences. com/) FACS-DiVa Cell Sorter or FACSCalibur flow cytometer. To prepare whole cell extracts, cells were lysed in buffer containing 100 mM NaCl, 100 mM Tris (pH 8. 2), 0. 5% NP-40, and protein inhibitor cocktail at 4 °C. Then 20 μg of total proteins were separated by SDS-PAGE and electrotransferred onto nitrocellulose membranes. The membranes were stained with Ponceau S and immunoblotted with rabbit anti-Bcl-xL antibody (1: 1,000; Cell Signaling Technology), mouse anti–cadherin 1 antibody (clone 36,1: 1,000; BD Biosciences), rabbit anti-actin antibody (1: 5,000; Sigma-Aldrich), rabbit anti–snail homolog 1 (1: 1,000; Abcam, http: //www. abcam. com/), rabbit anti-BAX antibody (1: 1,000; Cell Signaling Technology), rabbit anti-BAD antibody (1: 1,000; Cell Signaling Technology), mouse anti-phospho-BAD (Ser112) antibody (1: 1,000; Cell Signaling Technology), rabbit anti–myosin Va antibody (1: 1,000; Sigma-Aldrich), and appropriate secondary antibodies. The proteins were detected with ECL chemiluminescent substrate (GE Healthcare, http: //www. gehealthcare. com/) on Kodak (http: //www. kodak. com/) BioMax MR films. Transwell chambers with 8-μm porous polycarbonate membranes (Corning, http: //www. corning. com/) and invasion chambers with 8-μm porous polyester membranes and coated with Matrigel basement membrane matrix (BD Biosciences) were used. One million cells were plated in the upper chambers in DMEM containing 2% FBS, 0. 2 mM L-glutamine, and 1% penicillin/streptomycin. The lower chambers were filled with DMEM containing 10% FBS, 0. 2 mM L-glutamine, and 1% penicillin/streptomycin. After a 72-h incubation in a humidified 37 °C incubator under 5% CO2, cells migrating or invading into the bottom chambers were fixed, stained with hematoxylin, and counted in five fields under 200× magnification. To prepare whole cell extracts for anti-Bcl-xL IP, cells were lysed in cell lysis buffer containing 150 mM NaCl according to the manufacturer' s instructions from Cell Signaling Technology or in L100 buffer [59] containing protein inhibitor cocktail at 4°C for 30 min. The extracts were precleared with protein A Sepharose beads for 1 h at 4 °C. Rabbit anti-Bcl-xL antibody (Cell Signaling Technology) was mixed with the extracts for at least 4 h at 4 °C, and then protein A Sepharose beads were added for 1 h at 4 °C. The precipitates were washed extensively and were eluted from the beads by boiling in Laemmli buffer. Recombinant human Bcl-xL and BCL2 proteins were purchased from R&D Systems (http: //www. rndsystems. com/). Anti–cadherin 1 IP using two anti–cadherin 1 antibodies—clone ECCD-2 (Zymed, http: //www. invitrogen. com) and clone 36 (BD Biosciences) —was performed as described [60] except that protein A/G plus-agarose (Santa Cruz Biotechnology, http: //www. scbt. com/) was used instead of protein A Sepharose beads. The Entrez Gene (http: //www. ncbi. nlm. nih. gov/sites/entrez? db=gene) accession numbers for gene products discussed in this paper are BAD (12015), BAX (12028), cadherin 1 (12550), myosin Va (17918), myosin, heavy polypeptide 9, non-muscle isoform 1 (17886), PDX1 (18609), snail homolog 1 (20613), and TVA (420066).

Cancer cells accumulate multiple genetic alterations. Some of these contribute to tumor development while others are a mere by-product of genomic instability. To determine whether a candidate gene can promote tumor development, we have developed a novel experimental system using engineered viruses to deliver genes into premalignant lesions. We used genetically engineered mice in which both an oncogene (SV40 T antigen) and a specific docking molecule for the virus are produced in β cells in the pancreatic islets of Langerhans. Tumors form in only a subset of the islets expressing this oncogene, indicating that tumor development requires other events. Since these precancerous cells also express the virus docking molecule, we could deliver candidate progression genes via the virus to see whether they contributed to tumor progression. We show that genes encoding two proteins (a survival factor, Bcl-xL, and an inhibitory form of the cell adhesion molecule cadherin 1) can be delivered into premalignant β cells and thereby enhance tumorigenesis. Both of these proteins were previously implicated in tumor progression, confirming that our technique can identify such progression genes. Moreover, we find that Bcl-xL promotes tumor cell migration and invasion by a mechanism distinct from its known role in cell survival.

Abstract Introduction Results Discussion Materials and Methods Supporting Information

oncology biochemistry

2007

Assessing Tumor Progression Factors by Somatic Gene Transfer into a Mouse Model: Bcl-xL Promotes Islet Tumor Cell Invasion

The development of vaccines against fungi and other intracellular microbes is impeded in part by a lack of suitable adjuvants. While most current vaccines against infectious diseases preferentially induce production of antibodies, cellular immunity is essential for the resolution of fungal infections. Microbes such as fungi and Mycobacterium tuberculosis require Th17 and Th1 cells for resistance, and engage the C-type lectin receptors including Dectin-2. Herein, we discovered a novel Dectin-2 ligand, the glycoprotein Blastomyces Eng2 (Bl-Eng2). Bl-Eng2 triggers robust signaling in Dectin-2 reporter cells and induces IL-6 in human PBMC and BMDC from wild type but not Dectin-2-/- and Card9-/- mice. The addition of Bl-Eng2 to a pan-fungal subunit vaccine primed large numbers of Ag-specific Th17 and Th1 cells, augmented activation and killing of fungi by myeloid effector cells, and protected mice from lethal fungal challenge, revealing Bl-Eng2’s potency as a vaccine adjuvant. Thus, ligation of Dectin-2 by Bl-Eng-2 could be harnessed as a novel adjuvant strategy to protect against infectious diseases requiring cellular immunity. Fungal disease remains a challenging clinical and public health problem. Despite medical advances, invasive fungal infections have skyrocketed over the last decade and pose a mounting health threat in immune-competent and -deficient hosts with worldwide mortality rates ranking 7th, even ahead of tuberculosis [1,2]. The development of safe, effective vaccines remains a major hurdle for fungi. Critical barriers to progress include the lack of defined fungal antigens (Ags) and suitable adjuvants that together exert protective immunity. Recent strides in our understanding of fungal immunity and discovery of fungal Ags have raised the prospect that vaccines against fungi can be developed to elicit lasting protective immunity if suitable adjuvants are available. Adaptive immunity is critical for the prevention and resolution of fungal infections. The contribution of antibodies to host defense is debated [3,4]. In contrast, Ag-specific CD4+ T cells play the major role in fungal resistance [4,5], as evidenced by the high incidence of life-threatening fungal infections in patients with impaired CD4+ T cells. CD4+ T cells confer resistance by secretion of T-helper 1 (Th1) and Th17 cytokines such as IFN-γ, TNF-α, GM-CSF, and IL-17A, which activate neutrophils, monocytes, macrophages and DCs for fungal clearance [3,4, 6,7]. Since CD4+ T cells are germane to host defense against fungi, the challenge is how best to elicit these T cells. The transition from innate to adaptive immunity is fostered by dendritic cells (DCs). These cells process and present Ag to naïve CD4+ T cells in the context of co-stimulatory factors (e. g. cell surface ligands and cytokines) that provide the combination of signals necessary to induce naive T cell activation and proliferation. During their interactions with DCs, naive T cells also become functionally specialized. Helper T cell polarization occurs as a result of the cytokines produced by DCs: Th1 polarization is associated with DC production of high levels of IL-12p70, and Th17 polarization is associated with DC production of IL-1β and IL-6. While vaccine Ags typically have little impact on the nature of the cytokines produced by DCs, the adjuvant can have a dramatic effect. Alum (aluminum hydroxide), which is the most commonly used adjuvant in current vaccine formulations, suppresses DC production of pro-inflammatory cytokines such as IL-12p70 [8], creating an environment that polarizes T cells towards a Th2 phenotype. Thus, a major weakness and central challenge in the field of vaccinology is the lack of adjuvants that drive Th1 and/or Th17 polarization and stimulate DCs to produce the appropriate cytokines. Pathways that can differentially activate DC cytokine profiles include toll-like receptors (TLRs), C-type lectin receptors (CLRs), co-stimulatory ligands such as CD40, and cytokine receptors. C-type lectins are important in fungal recognition by DCs and in inducing anti-fungal Th1 and Th17 responses [9,10,11]. Dectin-1 and Dectin-2 induce Th1/Th17 cells in response to Candida albicans [12,13,14,15] and Aspergillus fumigatus [16,17,18] infection. While Dectin-1 is dispensable, Dectin-2 is requisite for the development of protective Th1 and Th17 cells and vaccine resistance against dimorphic fungi [19]. Crude fractions of mannoproteins isolated from Malassezia pachydermatis as well as a lipoglycan (Man-LAM) of Mycobacterium tuberculosis [20] have been shown to trigger Dectin-2 signaling, however they have not been evaluated as vaccine adjuvants, and glycans and lipids may be difficult to express and scale. Here, we report the identification of a novel fungal Dectin-2 ligand from an attenuated vaccine strain of Blastomyces dermatitidis, Bl-Eng2. We tested whether the ligation of Dectin-2 effectively vaccinates mice against fungi. Our vaccination strategy was to ligate Dectin-2 with Bl-Eng2 and assess the adjuvant activity by combining it with the recently reported pan-fungal vaccine calnexin [21]. Fungal recombinant Bl-Eng2 was expressed and scaled efficiently, it stimulated IL-6 and IL-1β production in vitro and Th1 and Th17 cells in vivo and, when used as an adjuvant in combination with calnexin, it protected mice against pneumonia in a model of lethal pulmonary fungal infection. To evaluate whether Bl-Eng2 is recognized by Dectin-2, we cloned and expressed the recombinant protein in Pichia pastoris. This eukaryotic expression system modifies recombinant proteins with both O- and N-linked glycosylation. Full-length Bl-Eng2 was fused to a N-terminal α-factor secretion signal and a C-terminal Myc-6×His tag (Fig 2B). Ni-NTA purified Bl-Eng2 showed a band of ~120 kDa on SDS-PAGE gel (Fig 2C), which falls within the size range determined in S1A and S1B Fig. Periodic acid-Schiff (PAS) based glyco-stain of Bl-Eng2 showed strong glycosylation (Fig 2C), which likely accounts for the discrepancy between predicted Mr of 57 kDa and apparent Mr of ~120 kDa. Gas chromatography (GC) analysis indicated that mannose is the major monosaccharide, and constitutes 82. 8% in glycan mass of Pichia-expressed Bl-Eng2 (Fig 2D). To verify Bl-Eng2 ligand activity, B3Z reporter cells expressing Dectin-2 or other distinct CLRs were incubated with recombinant Bl-Eng2. Bl-Eng2 elicited strong NFAT-lacZ signalling from Dectin-2 reporter cells, but not from the other CLR-expressing cells (Fig 3A), indicating a specific interaction between Dectin-2 and Bl-Eng2. Since Aspergillus Eng2 (Asp-Eng2) exhibits a high degree of similarity to Bl-Eng2 and contains a Ser/Thr-rich C terminus, we also tested whether Asp-Eng2 is recognized by Dectin-2. Asp-Eng2 and Bl-Eng2 were similarly recognized by Dectin-2 expressing reporter cells (S3A and S3B Fig), hence Eng2 from both fungal species are Dectin-2 ligands. To investigate whether Bl-Eng2 stimulates primary cells, we examined pro-inflammatory cytokine production from bone marrow–derived dendritic cells (BMDCs). BMDCs from wild type mice, but not Dectin-2-/- or Card9-/- mice, produced a strong IL-6 response when stimulated with recombinant Bl-Eng2, but not PDIA1 (Fig 3B), indicating ligand specificity for Dectin-2. Lack of stimulation of BMDCs from knockout mice also excludes the possibility of endotoxin contamination as the stimulus of IL-6 in wild type cells. Thus, Pichia-expressed Bl-Eng2 triggers a cytokine response in vitro that requires Dectin-2 and downstream Card9. These results together indicate that Bl-Eng2 appears to be a selective Dectin-2 ligand. Dectin-2 recognizes several fungi including C. albicans, A. fumigatus and Malassezia, which possess N- and O-linked mannan on their surface [17,18,23,24,25,26]. Thus, not surprisingly, there are two other Dectin-2 ligands described in the literature. They are Furfurman from Malassezia spp. [25] and Man-LAM from M. tuberculosis [20]. In addition to these ligands, by using B3Z reporter cells in the work here, we observed that MP98 from Cryptococcus neoformans [27] is also recognized by Dectin-2 (S3C Fig). MP98 also triggers IL-6 by BMDC in a Dectin-2- and concentration-dependent manner (S3D Fig). MP98 is a mannoprotein of Mr of 98 kDa with 103 Ser/Thr residues at the C-terminus that serve as potential O-linked glycosylation sites, and 12 putative N-linked glycosylation sites [27]. To begin to evaluate the relative potency of Dectin-2 ligands, we compared the ability of Bl-Eng2 and the other three Dectin-2 ligands to induce cytokine production by BMDCs. Bl-Eng2 induced the strongest IL-6 production by BMDCs when compared at equal molar and mass ratios to the other ligands (Fig 3C). These results suggest that Bl-Eng2 is relatively potent for triggering IL-6 and might be used as an adjuvant for vaccination to boost the development of Ag-specific T cell responses. A suitable adjuvant for vaccine formulation should ideally stimulate human accessory cells. To test this capacity, we assessed the effect of Bl-Eng2 on the function of human PBMCs. After stimulation with plate-coated Bl-Eng2, human PBMCs from five healthy subjects produced up to 17 ng/ml IL-6 and 9 ng/ml IL-1β as measured in the cell culture supernatants by ELISA (Fig 3D). These data suggest that recombinant Bl-Eng2 has the capacity to induce the production of Th17 cell priming cytokines by human antigen-presenting cells (APC) in vitro. To investigate whether Bl-Eng2 could be harnessed as a vaccine adjuvant, we performed preclinical studies in mice. We first tested whether Bl-Eng2 augments the development of vaccine Ag-specific T cells. To assess these T cell responses in vivo, we vaccinated mice with the pan-fungal Ag calnexin [21] and enumerated CD4+ T cell responses by TCR Tg 1807 cells, which are specific for calnexin [28]. Calnexin was suspended with incomplete freund’s adjuvant (mineral oil) and injected subcutaneously. The addition of Bl-Eng2 into the formulation sharply increased the frequency of IL-17 producing 1807 T cells (Fig 4A) and the number of activated (CD44+) and IL-17 and IFN-γ producing 1807 T cells, as measured by ex vivo stimulation with anti-CD3 and anti-CD28 mAb (Fig 4B and S4A Fig). Ex vivo stimulation with the vaccine Ag calnexin also yielded sharp increases in the amount of IL-17 produced by T cells from the draining lymph nodes (Fig 4D). Thus, Bl-Eng2 promoted the development of Th17 cells more so than Th1 cells. Addition of Bl-Eng2 to the vaccine also reduced lung CFU as early as four days after mice received a lethal experimental challenge, and did so in a concentration-dependent manner (S4B Fig). In a parallel group, at the time unvaccinated control mice were moribund (day18 post-infection), the addition of Bl-Eng2 to the vaccine reduced lung CFU by more than two logs (Fig 4C). Combining the vaccine with commercial alum as an adjuvant did not increase the frequency and numbers of cytokine producing T cells or reduce the fungal burden (Fig 4A–4D). However, combining Bl-Eng-2 together with Alum increased the adjuvancy of Alum as measured by the number of activated (CD44+), IL-17 and IFN-γ producing 1807 T cells and the reduction in lung CFU (S5 Fig). These results suggest that Bl-Eng-2 can work in concert with other (commercially available and FDA approved) adjuvants and augment vaccine efficacy. Bl-Eng2 failed to increase the development of Th17 and Th1 cells, the production ex vivo of IL-17 and IFN-γ, or reduce lung CFU in Dectin-2-/- mice, verifying that the adjuvant effect is Dectin-2-dependent in vivo (S4C–S4E Fig). Thus, Bl-Eng2 exhibits adjuvant-like properties by increasing the development of Ag-specific (1807) Th17 and Th1 cells and protecting mice from lethal pulmonary infection with B. dermatitidis. The studies above exploited TCR Tg T cells to sensitively report the ability of Bl-Eng2 to enhance development of calnexin Ag-specific Th17 and Th1 cells upon vaccination. However, adoptive transfer of these cells into mice artificially enhances the number of CD4+ T cell precursors in the animal. To investigate whether Bl-Eng2 also has the capacity to induce the development of endogenous calnexin-Ag specific CD4+ T cells and similarly protect animals, we vaccinated wild type mice in the absence of adoptive transfer. The formulation of Bl-Eng2 with the calnexin subunit vaccine again reduced lung CFU by over two logs vs. control mice vaccinated with calnexin in mineral oil alone (IFA), and by over 3 logs vs. mice that got IFA alone (Fig 4E). The addition of Bl-Eng2 to the calnexin vaccine formulation also increased survival significantly vs. control mice vaccinated with calnexin alone (Fig 4E). This is remarkable since the number of Ag-specific T cell precursors before vaccination was far lower in the absence than in the presence of transferred of naïve 1807 cells, indicating that Bl-Eng2 is a powerful adjuvant that drives the development of protective endogenous calnexin-specific CD4+ T cells (Fig 4F). To investigate the downstream myeloid effector mechanisms of Bl-Eng-2 adjuvancy we used red fluorescent B. dermatitidis yeast to report phagocytic uptake and fungal viability during cellular interactions with the murine leukocytes. The concept of using fluorescence to monitor microbial fate and investigate functional outcomes of individual microbial cell-host cell encounters has been introduced recently [29] and provides a powerful strategy to measure effector mechanisms in vivo. At day 4 post-infection, mice vaccinated with calnexin+Bl-Eng-2 and calnexin+Alum+Bl-Eng-2 showed increased activation and killing by neutrophils and alveolar macrophages vs. calnexin and calnexin+ Alum controls (Fig 5A–5C and S6 Fig). The increase in in vivo fungal killing by neutrophils and macrophages correlated with reduced numbers of DsRed+ yeast in the lung (Figs 5D and S6D) and CFU by plating (Figs 5E and S6D). Bl-Eng-2 mediated effects were observed in the presence of adoptively transferred 1807 T cells (Fig 5) and by endogenous CD4+ T cells without adoptive transfer (S6C and S6D Fig). Thus, the addition of Bl-Eng-2 augments the activation and killing by myeloid effector cells such as the neutrophils and alveolar macrophages in the lung. We describe a novel ligand for Dectin-2: Bl-Eng2. Discovery of a potent CLR ligand may address a limitation of current vaccines: the lack of adjuvants that elicit protective cell-mediated immunity. The approach we took to identify Bl-Eng2 was based on prior work from our group and other laboratories. Dectin-2 recognizes and mediates host defense against several fungi including C. albicans, C. glabrata, A. fumigatus, Malassezia spp. , Coccidiodes posadasii, Histoplasma capsulatum and B. dermatitidis [17,18,19,23,24,25,26,30]. Additionally, Dectin-2-/- mice vaccinated with attenuated B. dermatitidis yeast fail to prime Ag-specific Th1 and Th17 cells or acquire vaccine resistance to pulmonary infection. Thus, Dectin-2 regulates innate recognition of the fungal vaccine, and the development of a protective cellular immune response [19]. Hence, we sought to identify the Dectin-2 ligand from the vaccine strain. We hypothesized that the ligand would prime APC to produce cytokines (e. g. IL-6) that are known to foster the development of Th17 cells that protect against lethal fungal challenge [6]. By using Dectin-2 reporter cells as a probe, we enriched and identified Bl-Eng2 by ConA binding, gel filtration and Mass spectrometry. The identification of Bl-Eng2 also led us to unveil the unappreciated role of Asp-Eng2 in binding Dectin-2. Both Eng2 proteins are bona fide Dectin-2 ligands since they trigger NFAT signaling in Dectin-2 reporter cells. Bl-Eng2 features a 45. 2% overall and 60. 1% GH16 domain sequence similarity to Eng2 from A. fumigatus (Asp-Eng2) and contains a Ser/Thr-rich C-terminus that both proteins have in common. Bl-Eng2 and Asp-Eng2 respectively harbor 68 and 74 potential O-linked glycosylation sites within their respective 134-aa and 234-aa long Ser/Thr-rich C-terminus, but display no consensus sites for N-linked glycosylation (Asn-X-Ser/Thr). In addition to the Eng2 glycoproteins, we now also establish here that MP98 from C. neoformans serves as a ligand for Dectin-2. Dectin-2 has been reported to recognize high mannose structures of fungi [23], such as α-1,2-mannan from C. albicans [13,14] and furfurman, which is a mannoprotein from Malassezia spp. [25]. Man-Lam from M. tuberculosis consists of four components: a mannosyl-phophatidyl-myo-inositol (MPI) anchor, a mannose backbone, an arabinan domain, and a α1,2-mannose cap [20]. MP98 from C. neoformans is a mannoprotein with a Mr of 98 kDa; it contains 12 possible N-linked glycosylation sites, and 103 Ser/Thr residues at the C-terminus that serve as potential O-linked glycosylation sites [27]. The minimal unit of Bl-Eng2 that confers ligand activity is uncertain. Since both mannosidase and proteinase K digestion of CWE starting material reduced Dectin-2 ligand activity, both the protein and glycan moieties of Bl-Eng2 may contribute to its action, perhaps explaining its superior stimulation of cytokine responses compared to the other ligands. We found that recombinant Bl-Eng2 elicits potent downstream functions. It induces the production of IL-6 by BMDC in a Dectin-2- and Card9-dependent manner. In addition, Bl-Eng2 induces the production of IL-6 and IL-1β by human PBMC, which may have strong implications for the translational aspect of our discovery. In comparison to previously described Dectin-2 ligands, Bl-Eng2 triggers superior cytokine production by murine BMDC. Ligand induced IL-6 production was >100 fold higher for Bl-Eng2 than the other Dectin-2 ligands: Furfurman from Malassezia spp. [25] and Mannose-capped lipoarabinomannan (Man-Lam) from M. tuberculosis [20] and MP98 from C. neoformans [27]. Bl-Eng2 induction of T cell priming cytokines by APCs efficiently promoted the development of calnexin Ag-specific Th17 cells (more so than Th1 cells), and recall of these cells to the lung upon fungal challenge of vaccinated mice. The large numbers of pro-inflammatory T cells sharply reduced lung CFU and increased survival after infection of Bl-Eng2 vaccinated vs. control mice. In comparison, combining commercial Alum with the calnexin subunit vaccine did not show an adjuvant effect. However, Bl-Eng-2 combined with Alum augmented its adjuvancy indicating that Bl-Eng-2 has the potential to improve T cell priming by the commercially available and FDA approved Alum. Thus, in our subunit vaccine model, Bl-Eng2-induced Dectin-2 signaling was associated with cellular immune responses that protected mice against lethal pulmonary fungal infection. Although not experimentally addressed in this manuscript, it is conceivable that Bl-Eng-2 can also augment the induction of CD4+ T cell-dependent antibody responses that promote host protection against fungi, especially when combined with Alum since the latter is known to stimulate both T and B cell immune responses [31]. It remains to be investigated whether antibody will be protective in our vaccine setting [32]. We previously reported that mice vaccinated with calnexin and other adjuvants (glucan particles engaging Dectin-1, Adjuplex, or the combination) were optimally protected when we adoptive transferred naïve 1807 cells to increase the pool of Ag-experienced CD4+ T cells [21]. Here, the addition of Bl-Eng2 to the same calnexin vaccine reduced lung CFU by more than two to three logs vs. control mice even without adoptive transfer of large numbers of naïve 1807 T cell precursors. These results imply that engagement of Dectin-2 by Bl-Eng2 may be better than engagement of Dectin-1 by glucan particles and other previously used adjuvants at expanding the pool of endogenous calnexin-specific CD4+ T cell precursors or that Bl-Eng2 induced individual Ag-experienced cells to produce larger amounts of effector cytokines. Thus, Bl-Eng2 may be a powerful vaccine adjuvant in situations where T cell precursors are low in number and adoptive transfer of naïve T cell precursors is either not feasible or too costly. In contrast to the protective effects of Bl-Eng2 vaccination, Man-Lam induced Dectin-2 responses that caused Th17 cell-mediated autoimmune disease pathology and EAE [20]. Man-Lam stimulation of Dectin-2 led to the development of MOG35-55 peptide-specific T cells that produced IL-17, IFN-γ and GM-CSF upon ex vivo stimulation. This could simply relate to model selection rather than adjuvant efficiency. Thus, it is unclear whether Man-Lam is capable of inducing protective T cell immunity in an infectious disease setting. Although C. neoformans MP98 and its glycan modifications also promoted T cell activation, the T-helper phenotype and functional role in resistance by primed T cells were not investigated [33,34]. In conclusion, among the few Dectin-2 ligands reported to date, or newly discovered here, Bl-Eng2 is the most potent at stimulating murine and human cells to produce cytokines known to foster the development of protective Th17 and Th1 cells e. g. IL-6 and IL-1β. The production of IL-17 and IFN-γ by Th17 and Th1 cells then promotes the activation and killing of fungi by myeloid effector cells such as neutrophils and alveolar macrophages [6]. Since Bl-Eng2 also greatly augments protective immunity mediated by a subunit vaccine, Bl-Eng2 could potentially be harnessed as an adjuvant for vaccination against infectious disease that requires cellular immunity for host defense. The structural basis underpinning Bl-Eng2 potency as an adjuvant will be important to investigate and understand so that those features can be harnessed for vaccine development in the fight against infectious disease due to intracellular pathogens. Strains used were wild-type, virulent B. dermatitidis ATCC strain 26199, DsRed26199 [35] and strain #55, the isogenic, attenuated mutant lacking BAD1 [36]. B. dermatitidis was grown as yeast on Middlebrook 7H10 agar with oleic acid-albumin complex (Sigma) at 39°C. Inbred wild type C57BL/6 and congenic B6. PL-Thy1a/Cy (stock #00406) mice carrying the Thy 1. 1 allele were obtained from Jackson Laboratories, Bar Harbor, ME. Blastomyces-specific TCR Tg 1807 mice were generated in our lab and were backcrossed to congenic Thy1. 1+ mice as described elsewhere [28]. Dectin-2-/- [14] mice were bred at our facility. Mice were 7–8 weeks old at the time of these experiments. Mice were housed and cared for as per guidelines of the University of Wisconsin Animal Care Committee who approved all aspects of this work. Blastomyces dermatitidis yeast were harvested from 7H10 agar, washed with H2O, and sonicated for 3 min on ice. After centrifuging, the soluble extract was collected, passed through a 0. 45-μm pore-size filter and used as CWE. The protein level was measured with the Pierce BCA assay (Thermo Fisher Scientific). To enrich the mannosylated proteins, CWE was incubated with Concanavalin A (ConA) Sepharose resin (Sigma-Aldrich), and the bound fraction was eluted with methyl-α-D-mannopyranoside as described previously [21]. The ConA-enriched proteins were then applied to a size exclusion column of Ultragel AcA 44 resin (Pall) and eluted with PBS. The ConA enrichment and size exclusion fractions were assessed using SDS-PAGE and silver staining. Size exclusion fractions that contained Dectin-2 ligand activity were analyzed by mass spectrometry as previously described [21] at the Mass Spectrometry Facility, University of Wisconsin-Madison. Briefly, peptides were analyzed by nano-LC-MS/MS using the Agilent 1100 nanoflow system (Agilent Technologies) connected to a hybrid linear ion trap-orbitrap mass spectrometer (LTQ-Orbitrap XL, Thermo Fisher Scientific) equipped with a nanoelectrospray ion source. Bl-Eng2 was cloned and expressed in P. pastoris using standard recombinant techniques. Total RNA was extracted from B. dermatitidis yeast and transcribed to cDNA as previously described [37]. Using the cDNA as a template, the Bl-ENG2 coding sequence was amplified using KOD Hot Start DNA Polymerase (Toyobo) with primers 5′-GGCTCGAGAAAAGAGAGGCTGAAGCTAGGGCTACCAAGCTCGCGTT and 5′-GTTTCTAGACCGTACTTGTCATTTGTGGGGTATCCCG, and inserted in-frame into the XhoI/XbaI sites of the pPICZαA vector (Invitrogen). The resulting expression vector was then linearized with PmeI and transformed into Pichia pastoris strain X-33 (Invitrogen) by electroporation. Yeast colonies were screened for Bl-Eng2 protein expression by Western blot analysis using an anti-His antibody (Cell Signaling Technology). Bl-Eng2 protein secreted from methanol-induced yeast cells was purified using Ni-NTA agarose (Qiagen) according to the manufacturer' s protocol, and dialyzed against PBS. Purity of recombinant Bl-Eng2 was assessed by SDS-PAGE and silver staining. Bl-Eng2 protein glycosylation was assessed using the Pierce Glycoprotein Staining Kit (Thermo Fisher Scientific). Monosaccharide composition was determined by gas chromatography as described elsewhere [38]. B3Z/BWZ reporter cells expressing Dectin-2, Mincle, MCL and Dectin-1 have been described previously [19,39]. For B3Z/BWZ cell stimulation, 105 B3Z/BWZ cells per well in a 96-well plate were incubated for 18 h with heat-killed fungal cells or plate-coated ligands. β-galactosidase (lacZ) activity was measured in total cell lysates using CPRG (Roche) as a substrate. OD 560 was measured using OD 620 as a reference. Generation of bone marrow–derived dendritic cells (BMDCs) has been described previously [19]. Peripheral blood mononuclear cells (PBMCs) were isolated from heparinized whole blood collected over Ficoll-Paque Plus (GE). 1–2 × 105 BMDCs or 5 × 105 PBMCs per well in a 96-well plate were incubated with plate-bound Bl-Eng2. After 24 h, supernatants were collected and cytokine levels were measured by ELISA (R&D Systems or Biolegend) according to the manufacturer’s specifications. Prior to vaccination, mice received adoptively transferred naïve 1807 T cells [28] or not. Mice were vaccinated twice subcutaneously with 10μg recombinant calnexin and 10μg Bl-Eng2 formulated in incomplete Freund’s adjuvant (IFA), two weeks apart. Two weeks after the boost, mice were challenged with 2x104 26199 yeast and analyzed for lung T cell responses (at day 4 post-infection) and lung CFU (at day 4 or two weeks post-infection). 1807 T cell responses were detected with the congenic Thy1. 1 marker and endogenous, calnexin-specific T cells by tetramer [28]. T cells were detected using the following antibodies: tetramer-PE, CD4-BUV395, CD8-PeCy7, CD3-BV421, CD90. 2-BV785, CD44-BV650, Live-dead Near IR, IFN-γ-A488 and IL-17-A647. Lung cells were harvested at day 4 post-infection. Cells (0. 5 × 106 cells/ml) were stimulated for 5 hours with anti-CD3 (clone 145-2C11; 0. 1μg/ml) and anti-CD28 (clone 37. 51; 1μg/ml) in the presence of Golgi-Stop (BD Biosciences). Stimulation with fungal ligands yielded comparable cytokine production by transgenic T-cells compared to CD3/CD28 stimulation. After cells were washed and stained for surface CD4 and CD8 using anti-CD4 BV395, anti-CD8 PeCy7, and anti-CD44-FITC mAbs (Pharmingen), they were fixed and permeabilized in Cytofix/Cytoperm at 4°C overnight. Permeabilized cells were stained with anti-IL-17A PE and anti-IFN-γ Alexa 700 (clone XMG1. 2) conjugated mAbs (Pharmingen) in FACS buffer for 30 min at 4°C, washed, and analyzed by FACS. Cells were gated on CD4 and cytokine expression in each gate analyzed. The number of cytokine positive CD4+ T cells per lung was calculated by multiplying the percent of cytokine- producing cells by the number of CD4+ T cells in the lung. Bone marrow-derived dendritic cells (BMDCs) were obtained from the femurs and tibias of individual mice. Each bone was flushed with 10 ml of 1% FBS in RPMI through a 22G needle. Red blood cells were lysed followed by wash and re-suspension of cells in 10% FBS in RPMI medium. In a petri dish, 2 × 106 bone marrow cells were plated in 10 ml of RPMI containing 10% FBS plus penicillin-streptomycin (P/S) (HyClone), 2-mercaptoethanol and 20 ng/ml of rGM-CSF. The culture media were refreshed every three days and BMDCs were harvested after 10 days for in vitro co-culture assays. Ex vivo cell culture supernatants were generated using the brachial and inguinal draining lymph nodes harvested from mice 28 days post-vaccination and at day 4 post-infection, washed and resuspended in complete RPMI containing 10 μg/ml recombinant calnexin [21,40], and plated in 96-well plates at a concentration of 5 × 105 cells/well. Supernatants were collected from ex vivo co-cultures after three days of incubation at 37°C and 5% CO2 [6]. IFN-γ and IL-17 (R&D System) were measured by ELISA according to manufacturer specifications (detection limits, 0. 05 ng/ml and 0. 02 ng/ml, respectively). Mice were euthanized three days after challenge i. t. with 105 DsRed yeast and hearts were perfused with PBS to remove blood from the lungs to improve staining. Lungs were dissociated, digested and stained as described previously [35]. In summary, lungs were dissociated and digested in buffer containing collagenase D and DNase I. After erythrocyte lysis, cells were stained for myeloid cell markers and then fixed in Cytofix/Cytoperm (BD Biosciences, San Jose, CA). Cells were stained for 30 minutes at room temperature with 1 μg/ml Uvitex-2B (Polysciences, Warrington, PA) diluted in BD perm/wash buffer and then subsequently washed with BD perm/wash buffer and fixed with 2% paraformaldehyde. Differences in the number of cells and lung CFU were analyzed using Wilcoxon rank and Mann Whitney test for non-parametric data or a T-test if data were normally distributed. A Bonferroni adjustment was used to correct for multiple tests. A value of P < 0. 05 is considered significant. Studies with human peripheral blood mononuclear cells were approved by University of Wisconsin-Madison IRB (protocol 2014–1167 CR002) and patients provided informed written consent. The animal studies performed were governed by protocols M00969 as approved by the IACUC committees of the University of Wisconsin-Madison Medical School. Animal studies were compliant with all applicable provisions established by the Animal Welfare Act and the Public Health Services (PHS) Policy on the Humane Care and Use of Laboratory Animals.

Despite several million new systemic fungal infections annually worldwide, there are no commercial vaccines available. The lack of appropriate adjuvants is one major impediment to developing safe and effective vaccines against infections with fungal pathogens. Current vaccines against infectious diseases preferentially induce protective antibodies, driven by adjuvants such as alum. While clonally-derived and adoptively transferred monoclonal antibodies may confer protection against fungi, the contribution of antibody to host defense is likely to be limited. Animal studies and clinical observations implicate cellular immunity as an essential component of the resolution of fungal infections. We found a promising adjuvant that augments cell mediated immune responses and vaccine-induced protection against fungal infection. We anticipate that our discovery will be a useful adjuvant for vaccination with non-replicating and safe subunit vaccines against many microbial pathogens that require protective cell mediated immune responses.

Abstract Introduction Results Discussion Material and methods

blood cells graduates innate immune system medicine and health sciences immune cells immune physiology cytokines pathology and laboratory medicine pathogens immunology microbiology alumni vaccines preventive medicine developmental biology fungi immunologic adjuvants experimental organism systems molecular development educational status infectious disease control vaccination and immunization fungal pathogens research and analysis methods public and occupational health infectious diseases mycology white blood cells animal cells medical microbiology t cells microbial pathogens immune system yeast people and places candida cell biology physiology biology and life sciences cellular types yeast and fungal models population groupings organisms candida albicans

2017

Ligation of Dectin-2 with a novel microbial ligand promotes adjuvant activity for vaccination

Human adenoviruses from multiple species bind to coagulation factor X (FX), yet the importance of this interaction in adenovirus dissemination is unknown. Upon contact with blood, vectors based on adenovirus serotype 5 (Ad5) binds to FX via the hexon protein with nanomolar affinity, leading to selective uptake of the complex into the liver and spleen. The Ad5: FX complex putatively targets heparan sulfate proteoglycans (HSPGs). The aim of this study was to elucidate the specific requirements for Ad5: FX-mediated cellular uptake in this high-affinity pathway, specifically the HSPG receptor requirements as well as the role of penton base-mediated integrin engagement in subsequent internalisation. Removal of HS sidechains by enzymatic digestion or competition with highly-sulfated heparins/heparan sulfates significantly decreased FX-mediated Ad5 cell binding in vitro and ex vivo. Removal of N-linked and, in particular, O-linked sulfate groups significantly attenuated the inhibitory capabilities of heparin, while the chemical inhibition of endogenous HSPG sulfation dose-dependently reduced FX-mediated Ad5 cellular uptake. Unlike native heparin, modified heparins lacking O- or N-linked sulfate groups were unable to inhibit Ad5 accumulation in the liver 1h after intravascular administration of adenovirus. Similar results were observed in vitro using Ad5 vectors possessing mutations ablating CAR- and/or αv integrin binding, demonstrating that attachment of the Ad5: FX complex to the cell surface involves HSPG sulfation. Interestingly, Ad5 vectors ablated for αv integrin binding showed markedly delayed cell entry, highlighting the need for an efficient post-attachment internalisation signal for optimal Ad5 uptake and transport following surface binding mediated through FX. This study therefore integrates the established model of αv integrin-dependent adenoviral infection with the high-affinity FX-mediated pathway. This has important implications for mechanisms that define organ targeting following contact of human adenoviruses with blood. Adenoviruses are non-enveloped, icosahedral double-stranded DNA viruses of 70–90nm diameter. 54 different human serotypes have been identified to date and are classified into species based on their ability to agglutinate human, monkey or rat erythrocytes [1]. Adenoviruses cause a range of illnesses depending on the route of initial infection, largely dictated by inherent adenoviral tropism. These illnesses are usually self-limiting but can become potentially life-threatening in certain circumstances. For example, species C adenoviruses 1,2 and 5 initially cause respiratory tract infections after inhaled droplet transmission [2] but are associated with fulminant hepatitis in bone marrow transplant patients [3], [4]. Invasive adenovirus infection following liver transplant is relatively common, occurring in approximately 10% of paediatric and 6% of adult liver transplantion recipients (reviewed in [5]) and may be due to latent donor-associated infection of the transplanted organ. Adenovirus has also been detected in peripheral blood from immunocompromised patients [6], [7], [8], [9] a significant proportion of whom then go on to develop potentially fatal disseminated adenoviral disease. Taken together, these studies underline the clinical significance of these common human pathogens. The primary and secondary receptor systems used by adenoviruses for cellular uptake following contact with different environments in vivo are thus of particular interest and importance. The species C adenovirus Ad5 can efficiently infect a wide variety of cell types. In vitro studies have demonstrated that cell tethering is mediated by a primary interaction of the Ad5 fiber knob domain with the coxsackievirus and adenovirus receptor (CAR; [10], reviewed in [11]), while the subsequent internalisation of Ad5 particles is dependent on binding of αvβ3/αvβ5 integrins to an RGD motif in the adenovirus penton base [12], [13]. Several in vivo studies, however, have shown that direct interaction with CAR is not required for uptake of Ad5 into the liver [14], [15], [16], [17], which is the primary target organ after contact of adenovirus with the bloodstream in rodent models and in non-human primates (reviewed in [18]). Moreover, CAR is now thought to be localised primarily to tight junctions in intact epithelium, rendering it inaccessible to viral particles (reviewed in [19]). Instead, recent studies have demonstrated that uptake of Ad5 into the liver is mediated by a high affinity interaction with blood coagulation factor X (FX), which putatively ‘bridges’ the hexon protein in the adenovirus capsid to heparan sulfate proteoglycans (HSPGs) expressed on the surface of hepatocytes [17], [20], [21]. Ad5 utilises the host FX protein, which circulates at approximately 8–10 µg/ml in the bloodstream, for cell binding since the cell surface interaction of the Ad5: FX complex is mediated through the serine protease domain of FX and not through a direct interaction of the virus with the cell surface [21]. This is of particular significance in the context of disseminated adenoviral disease affecting immunocompromised patients, since typing studies have found a predominance of species C adenoviruses in peripheral blood samples from these individuals [6], [22], [23]. Interestingly, recent surface plasmon resonance (SPR) studies have demonstrated that of 22 Ad species tested, from species A, B, C and D, 14 can bind FX [21] indicating that the interaction of Ad5 with FX may be highly conserved. Only adenoviruses from species D lacked the capacity to bind FX. HSPGs are widely-expressed molecules composed of a core protein to which one or more heparan sulfate (HS) glycosaminoglycan (GAG) sidechains are covalently linked (reviewed in [24]). Their core protein diversity, structural heterogeneity and high negative charge (imparted by the HS-GAG sidechains, which consist of highly-sulfated disaccharide repeats of N-acetylglucosamine and glucuronic/iduronic acid) ensure that HSPGs play important roles in many biological processes [25]. Furthermore, several viral pathogens including the human immunodeficiency virus-1 (HIV) [26], human papilloma virus (HPV) [27], adeno-associated virus (AAV) [28] and herpes simplex virus (HSV) [29] exploit HSPGs as primary attachment receptors in different tissues and cell types. Although in vitro and in vivo studies suggest that the Ad5: FX complex interacts with membrane HSPGs [17], [30] the specific receptor requirements underlying FX-mediated adenoviral uptake have not been characterised. Interestingly, liver HS have been shown to possess a specialised structure, with much higher levels of N- and O-sulfation than HS from other tissues [31], [32], [33]. Viral interactions with HS sidechains at the cell surface are frequently associated with the presentation of a particular ‘sulfation signature’ [34], [35]. The substantial liver specificity of systemically disseminating Ad5 may therefore be due to the preferential interaction of Ad5: FX complexes with highly-sulfated liver HS. Here, we investigated the receptor requirements for FX-mediated Ad5 cellular uptake in vitro and in vivo. We first showed in a number of cell lines that the interaction of the Ad5: FX complex with the cell surface was entirely independent of CAR and αv integrins. By analysing Ad5 binding and uptake into enzymatically-pretreated cells we demonstrated that the primary attachment of the Ad5: FX complex was specifically mediated by HS sidechains. Detailed immunocytochemical analysis in vitro revealed delayed FX-mediated cell entry and cytosolic transport to the microtubule organising centre (MTOC) of a fluorescently-labelled Ad5 mutant lacking the penton base RGD motif, showing that rapid and efficient post-attachment kinetics is dependent on engagement of αv integrins. Chemical inhibition or genetic ablation of endogenous HS sulfation completely abrogated FX-mediated Ad attachment and cell uptake, indicating that the Ad5: FX complex interacts with a specific HS sulfation pattern, while heparin-mediated inhibition of adenoviral gene transfer in vitro and Ad attachment to liver slices ex vivo was significantly attenuated in the absence of heparin N- and, in particular, O-linked sulfate groups. In vivo, Ad5 liver accumulation 1 h after intravenous administration to mice was significantly and dose-dependently inhibited by pre-inoculation with unmodified heparin but not by de-sulfated heparins. Immunohistochemical analysis of liver sections from mice intravenously injected with fluorescently-labelled Ad5 revealed localisation of Ad particles in CD31+ hepatic sinusoids and surrounding hepatocytes. We have thus integrated the established model of αv integrin-dependent adenoviral infection with the FX-mediated pathway leading to liver uptake of Ad5. We utilised Ad5CTL (vector based on wild type Ad5 capsid), Ad5KO1 (CAR binding mutated), Ad5PD1 (integrin-binding mutant) or Ad5KP (both mutations) – see Methods for details. To assess the importance of HSPG heparan sulfate (HS) sidechains in Ad5 cell binding and uptake mediated by interaction with human FX we pre-treated human HepG2 hepatoma and SKOV3 ovarian carcinoma cells with heparinase III prior to performing Ad5 cell attachment and gene transfer experiments in the presence or absence of FX. Both HepG2 and SKOV3 cells express HSPGs, however unlike HepG2 cells, SKOV3 cells express very low levels of CAR [36]. Heparinase III is a heparin lyase that specifically cleaves N-acetylated (NAc) and transition domains of HS and can be used in vitro and in vivo to inhibit HS-mediated viral attachment [27], [37]. Cleavage of HS at NAc and transition domains was assayed by FACS. Heparinase III digestion significantly and dose-dependently reduced the percentage of cells positive for antibody 10E4, while significantly increasing the percentage of cells positive for antibody 3G10, confirming the substrate activity of heparinase III treatment (Fig. S1). To verify that heparinase III treatment had no effect on the widely-expressed GAG chondroitin sulfate, which is primarily composed of acetylgalactosamine hexosamine groups, cell surface chondroitin sulfate was quantified by FACS using the CS-56 mouse monoclonal antibody. While treatment of SKOV3 cells with chondroitinase ABC dose-dependently reduced the percentage of CS-56 positive cells, heparinase III treatment had no effect (Fig. S1), indicating that heparinase III treatment has no effect on chondroitin sulfate. Ad5 attachment to cells was substantially increased in the presence of FX in both HepG2 and SKOV3 cells (Fig. 1A and 1B; p<0. 01). Similar results were observed when Ad5 mutants ablated for CAR-binding and/or αv integrin binding (Ad5KO1, Ad5PD1 or Ad5KP; Fig. 1A) were used, in agreement with previous studies showing that the FX-mediated increase in virus cell attachment and gene transfer is CAR- and αv integrin- independent [17], [30]. Similar FX-mediated enhancement of virus uptake was also observed in a panel of CARhigh and CARlow cell lines when Ad5CTL and the αv integrin-binding mutant Ad5PD1 were compared (Fig. S2). As expected levels of FX-mediated enhancement in gene transfer were lower in CARhigh cells compared to CARlow cells (Fig. S1) since the CAR and FX pathways are both efficient in vitro pathways in the former. The FX-mediated increase in virus cell attachment was significantly attenuated by heparinase III pretreatment of HepG2 (p<0. 01) or SKOV3 (p<0. 05) cells (Fig. 1B). FX-mediated gene transfer was also significantly decreased after heparinase III pretreatment (Fig. 1C) of HepG2 and SKOV3 cells (p<0. 01). We next carried out virus transport experiments in vitro using an Alexa488-labelled virus in the presence or absence of heparin. Heparin and HS sidechains have very similar structures, although heparin displays higher general sulfation than HS [25]. Heparin is therefore frequently used as a competitive inhibitor for HS binding [38], [39]. In the absence of heparin, Ad5: FX complexes were rapidly and efficiently transported to the MTOC in SKOV3 cells within an hour of adding fluorescently-tagged Ad5CTL: FX complexes to cells, as demonstrated by colocalisation with the MTOC marker pericentrin (Fig. 1D, upper panel). In the presence of heparin, however, FX-mediated attachment of fluorescently-tagged Ad5CTL to the cell surface was completely abrogated (Fig. 1D, lower panel), confirming that HS sidechains mediate Ad5CTL: FX attachment to the cell surface. Similar results were observed in SKOV3 and A549 human lung adenocarcinoma cells (Fig. 1E). Taken together, these results indicate that FX-mediated Ad5CTL cell attachment in vitro is dependent on the presence of HS sidechains. Although in vitro studies have shown that an intact penton base RGD motif is required for efficient endosome escape after CAR-mediated attachment of Ad5 to the cell surface [40] several in vivo studies indicate that ablation of the penton base RGD motif does not significantly affect liver uptake after systemic administration of Ad5 [14], [15], [16], [17], [41]. Having established that FX-mediated attachment of Ad5CTL to the cell surface requires HSPGs, we next assessed the role of αv integrins during FX-mediated Ad5 cell uptake. In vitro cell tracking experiments were carried out in CARlow SKOV3 [36] and CARhigh A549 cells [42] using Ad5 vectors with a mutated penton base RGD motif (Ad5PD1). Both Ad5CTL and Ad5PD1 efficiently bound the cell membrane in the presence of FX (Fig. 2A, B). In SKOV3 cells at 15 and 60 minutes post-internalisation, both Ad5CTL and Ad5PD1 particles had entered endosomal compartments as demonstrated by partial colocalisation with the early endosomal marker EEA1 and Rab5 (Fig. 2A, B). Similar results were observed in A549 cells (data not shown). MTOC colocalisation was quantified by assessing the proportion of cells in a 40× microscope field with colocalisation of fluorescently-labelled Ad5 particles (green) with the MTOC marker pericentrin (red; see upper panel in Fig. 3A, quantification Fig. 3B–C). Cell entry and cytosolic transport kinetics of the CAR binding-ablated vector Ad5KO1 closely resembled Ad5CTL, confirming that FX-mediated cell uptake does not require CAR (Fig. S3). Conversely, only 20–25% MTOC colocalisation was observed for Ad5PD1 in SKOV3 and A549 cells at the same timepoint (Fig. 3B and Fig. 3C, respectively). These data suggest that an integrin-mediated post-internalisation signal is required for optimal transport of Ad5CTL to the MTOC after FX-mediated cell surface attachment of Ad5CTL to HSPGs. To confirm the role of integrins we next performed a short hairpin (sh) RNA approach to knockdown αv integrin expression in SKOV3 cells and assessed the effect of this on intracellular transport of Ad5CTL. Target knockdown was confirmed using TaqMan and flow cytometric analysis compared to mock-transfected and scrambled control shRNA cells (Fig. 4A, B). Knockdown led to a significant reduction in the localisation of Ad5CTL to the peri-nuclear compartment (Fig. 4B–C) thus confirming the importance of integrin engagement for transport via the FX-mediated pathway. We next determined the effect of kinase inhibitors that are known to affect cell entry and intracellular transport of adenovirus [43]. We used H89 dihydrochloride (an inhibitor of PKA), L Y294002 hydrochloride (an inhibitor of PI3K) and SB 203580 hydrochloride (an inhibitor of p38 MAPK). We observed that co-incubation of Ad5CTL-transduced cells in the presence of inhibitors of either PKA, PI3K or p38 MAPK was able to significantly reduce Ad5CTL-mediated transport to the MTOC in the presence of FX (Fig. S4). Since these molecules are linked to activation of cellular integrins during Ad internalisation or intracellular transport this further suggests that integrins are a component of the viral entry cycle in the presence of FX. Having established that FX-mediated Ad5CTL cell surface attachment required HS sidechains, we next investigated whether HS sidechain sulfation affects FX-mediated Ad5CTL cell uptake. In vitro experiments were therefore carried out in HepG2 and SKOV3 cells pretreated with increasing concentrations of sodium chlorate, a selective inhibitor of sulfation [44]. We confirmed that sodium chlorate treatment inhibited sulfation in a dose-dependent manner (Fig. S5A), reduced binding of Ad5CTL to cells in the presence of FX (Fig. S5B) in the absence of cellular toxicity (Fig. S5C). Gene transfer was used as a marker of successful cellular internalisation, cytosolic transport and nuclear uptake. FX-mediated Ad5CTL gene transfer was inhibited in a dose-dependent manner by pretreatment with sodium chlorate in both HepG2 and SKOV3 cells (Fig. 5A), suggesting that FX-mediated Ad5 uptake in vitro may depend on HS sidechain sulfation. No effect on basal levels of Ad5CTL uptake was observed. To assess the importance of HS sulfation for FX-mediated Ad5CTL transduction, cell attachment and transduction assays were carried out in CHO cell lines deficient in HS biosynthesis enzymes. Although FX induced a 60-fold increase in Ad5 cell attachment and uptake in parental CHO-K1 (CAR-) cells, no FX-mediated increase was observed in CHO-pgsA745 cells, which do not express xylosyltransferase-1 (XT1) and are consequently defective in HS-GAG synthesis [45] (Fig. 5B–C). The FX-mediated increase in Ad5 cell attachment and transduction was also significantly attenuated in N-deacetylase/N-sulfotransferase-1 (NDST1) -deficient CHO-pgsE606 cells, which synthesise HS chains with significantly reduced O- and, in particular, N sulfate groups [46] (Fig. 5B–C). Interestingly, FX was unable to increase Ad5CTL cell attachment and uptake in CHO-pgsF17 cells, which are 2-O-sulfotransferase-deficient and therefore lack 2-O-sulfated residues [47] (Fig. 5B–C). Similar results were obtained in all cell lines using the αv integrin-binding mutant AdPD1 (Fig. 5B and 5C). Taken together, these data suggest that attachment of the Ad5CTL: FX complex to cell surface HSPGs in vitro requires HS sidechain sulfation. To confirm the importance of HS sidechain sulfation for FX-mediated Ad5CTL cell attachment and uptake, in vitro gene transfer and ex vivo attachment experiments were carried out in the presence of heparan sulfates or heparins with biosynthetically-modified sulfation. Bovine intestinal heparin and porcine intestinal heparan sulfate are very highly-sulfated [39]. In contrast, porcine kidney heparan sulfate possesses fewer N- and O- sulfate groups, while de-N-sulfated heparin lacks N-sulfated glucosamine residues [39] and de-O-sulfated heparin lacks O-sulfate groups [39], [48]. Competitive inhibition experiments in vitro and ex vivo were carried out in SKOV3 cells or mouse liver slices, respectively, in the presence or absence of FX. Dose-dependent inhibition of FX-enhanced Ad5CTL gene transfer into SKOV3 cells was observed in the presence of all heparins and heparan sulfates (Fig. 6A). However the IC50 values for the highly-sulfated bovine intestinal heparin and porcine intestinal heparan sulfate (5. 1 µg/ml and 5. 2 µg/ml respectively) were lower than the IC50 values for the de-sulfated heparins (39. 1 µg/ml and 73. 4 µg/ml for de-N-sulfated or de-O-sulfated heparins respectively) or heparan sulfate with reduced sulfation (porcine kidney heparan sulfate, 17. 7 µg/ml) (Fig. 6A). Similar results were obtained when CAR- or αv integrin binding-ablated viruses were used (Table S1). Next, the attachment of fluorescently-labelled Ad5CTL (green) to mouse liver slices ex vivo was analysed. Adherent Ad5CTL particles were quantified by analysing captured images of 60× microscope fields using the ImageJ automated cell counting function. Co-incubation with FX significantly increased the attachment of fluorescently-labelled Ad5CTL to liver sections (Fig. 6B and Fig. 5C). At the concentrations used, heparin significantly inhibited FX-mediated attachment of Ad5CTL to liver sections (p<0. 01; Fig. 6B–C). However neither de-O-sulfated nor de-N-sulfated heparin inhibited the attachment of Ad5CTL to liver sections in the presence of FX. These results indicate that binding of the Ad5CTL-FX complex to heparin/HS in vitro requires the presence of highly anionic sulfate groups, supporting the hypothesis that FX-mediated Ad5CTL cell uptake is dependent on HS sidechain sulfation. To identify whether the sulfation status of hepatocyte HS contributes to the liver uptake of Ad5CTL from the circulation, competitive inhibition experiments were carried out in vivo in the presence of heparins with biosynthetically-modified sulfation. Mice were inoculated with increasing concentrations of heparins prior to intravascular administration of Ad5CTL. Ad5CTL liver uptake at an early timepoint post-inoculation was then quantified by assessing viral genome accumulation 1 h post-administration. Cellular localisation of Ad5CTL in mouse livers was analysed by staining liver sections with CD31 to visualise the vasculature in conjunction with fluorescently labelled Ad5CTL. We have previously shown that the FX-pathway is hepatocyte-specific and Kupffer cell uptake is unaffected by pharmacological modulation or genetic approaches to modify FX binding as Kupffer cell uptake is a scavenging process [49], [50], [51], [52]. We therefore used macrophage-depleted mice to allow selective visualisation of HSPG uptake mechanisms via the FX pathway in vivo at an early time point post injection. Heparin pre-inoculation at both 20 mg/kg and 50 mg/kg significantly and dose-dependently inhibited Ad5CTL accumulation in the liver 1 h post-inoculation (p<0. 01) (Fig. 7A). Although no effect was observed at either dose of de-O-sulfated heparin (20 mg/kg or 50 mg/kg), significantly fewer Ad5CTL genomes were detected in livers of mice pre-treated with 50 mg/kg de-N-sulfated heparin (p<0. 05) (Fig. 7A). Immunofluorescence staining for CD31 was performed to facilitate identification of endothelial sinusoids in the liver architecture. Sections from control mice inoculated with fluorescently-labelled Ad5CTL (green) showed accumulation of Ad5CTL particles in liver sinusoids and on the surface of hepatocytes (Fig. 7B, upper panel). Administration of heparin (both 20 mg/kg and 50 mg/kg doses) or high-dose de-N-sulfated heparin clearly reduced accumulation of labelled Ad5 while de-O-sulfated heparin had no effect at either dose (Fig. 7B, lower panel). Taken together, our data indicate that the hepatic uptake observed after intravenous administration of Ad5CTL is dependent on HS sidechain sulfation, with a particular requirement for O-sulfation. In conjunction with previous studies showing that liver HS is highly enriched in 2-O-sulfated residues [31] our in vitro, ex vivo and in vivo results further suggest that the hepatic tropism of Ad5CTL may be due to preferential binding of the Ad5: FX complex to HS sidechains. Previous studies have shown that the liver uptake of Ad5 after exposure to the circulation is dependent on FX binding directly to the hexon protein in the Ad5 capsid, putatively via a FX-mediated interaction with hepatocyte membrane HSPGs [17], [30]. In the present study we investigated the functional receptor requirements for the Ad5CTL: FX complex using a variety of in vitro, ex vivo and in vivo experimental approaches. We have demonstrated that FX-mediated Ad5CTL cell attachment in vitro requires the presence of HS sidechains but not CAR or αv integrins. FX-mediated Ad5CTL binding to CARhigh HepG2 and CARlow SKOV3 cells was not affected by fiber knob or penton base mutations ablating CAR- or αv integrin-interacting motifs, respectively, indicating that neither CAR nor αv integrins were required for the FX-mediated primary interaction with the cell surface. Conversely, cleavage of HS sidechains by heparinase III pretreatment significantly inhibited FX-mediated Ad5 attachment to and uptake into HepG2 and SKOV3 cells. Heparin, a highly-sulfated HS analogue, also abrogated FX-mediated adenoviral cell surface binding. Moreover, no FX-mediated enhancement of Ad5CTL cell binding or gene transfer was observed in CHO-pgsA745 cells which do not display HS sidechains. Taken together, these results clearly demonstrate that the primary interaction of the Ad5CTL: FX complex with the cell surface is mediated via HS sidechains. Although our data indicate that CAR- or αv integrins are not required for FX-mediated attachment of Ad5CTL to the cell surface, intracellular transport experiments using a mutant with an ablated penton base RGD motif (Ad5PD1) and a knockdown shRNA approach revealed that efficient and rapid post-internalisation transport of virus particles to the nucleus requires engagement of RGD-interacting integrins. A similar delay in intracellular transport of Ad5PD1 was observed in CARhigh A549 and CARlow SKOV3 cells, suggesting that the altered transport was not affected by differences in CAR expression. Interestingly, a previous study investigating the cytoplasmic transport of Ad5 after CAR-mediated cell surface attachment has demonstrated a similar reliance on RGD-integrin interactions [40]. In conjunction with this study, our results indicate that integrin engagement is required for rapid and efficient intracellular transport of Ad5CTL regardless of the primary attachment receptor used. Furthermore, the potential HS sidechain dependence of FX-mediated cell surface attachment suggests that the Ad5CTL: FX complex may utilise HSPGs as attachment factors in a similar manner to other hepatotropic viruses such as HSV [53] hepatitis [34], [54] and AAV-2 [55], [56]. A central aim of the present study was to establish whether FX-mediated Ad5CTL cell uptake is dependent on the degree or type of HS sidechain sulfation. Blocking HS sulfation by pre-incubating cells with increasing concentrations of sodium chlorate dose-dependently inhibited FX-mediated Ad5CTL gene transfer in HepG2 and SKOV3 cells. Moreover, FX-mediated enhancement of Ad5CTL cell attachment and uptake was significantly attenuated in CHO-pgsE606 cells, which have reduced overall sulfation due to a deficiency in the N-deacetylase/N-sulfotransferase-1 (NDST1) gene [46]. In addition, IC50 values for porcine kidney heparan sulfate, which possesses fewer sulfate groups than heparin, were approximately 3-fold higher than IC50 values for highly-sulfated bovine intestinal heparin or porcine intestinal heparan sulfate. These data show that FX-mediated Ad5 cell attachment and uptake is dependent on the degree of HS sidechain sulfation. Interestingly, removal of N- or O- sulfate groups significantly attenuated the inhibitory capabilities of heparin on Ad5CTL uptake in vitro, increasing IC50 values approximately 8- or 14-fold respectively. Furthermore, unlike native heparin, de-O-sulfated and de-N-sulfated heparins were unable to inhibit FX-mediated attachment of fluorescently-labelled Ad5CTL to liver slices ex vivo. Finally, no FX-mediated enhancement of Ad5CTL cell attachment or uptake was observed in CHO-pgsF17 cells, which lack 2-O-sulfate groups due to a deficiency in the 2-O-sulfotransferase gene [47]. Taken together, our results suggest that while the degree of sulfation modulates the FX-mediated uptake of Ad5CTL in vitro, the Ad5CTL: FX complex may also preferentially interact with specific sulfate moieties. A number of previous studies have examined the biochemical composition of heparan sulfate from different tissues and have shown that liver heparan sulfate is enriched in sulfated moieties, in particular 2-O sulfate groups [31], [33]. Interestingly, hepatic clearance of intravenously-administered very low density lipoprotein (VLDL) is significantly reduced in mice with liver-specific knockout of the heparan sulfate 2-O sulfotransferase enzyme [57], thereby adding mechanistic insight into previously published work documenting increased levels of systemic VLDL in mice with reduced overall liver HS sulfation [32]. These studies show that the specialised structure of liver HS can contribute to the hepatic accumulation and uptake of circulating particles, and indicate how liver HS sulfation may contribute to the accumulation of systemically-disseminated, FX-interacting adenoviruses such as Ad5. The final aim of this study was therefore to investigate the importance of HS sidechain sulfation in FX-mediated Ad5CTL interactions with liver cells in vivo. While pre-injection of native heparin or high-dose de-N-sulfated heparin significantly attenuated Ad5CTL genome accumulation in the livers of macrophage-depleted mice 1 h after intravenous delivery, administration of de-O-sulfated or low-dose de-N-sulfated heparin had no effect. Immunohistochemical analysis of fluorescently-labelled Ad5CTL in liver sections from these mice clearly showed a significant reduction in Ad5CTL accumulation around CD31+ hepatic sinusoids after pre-injection of high-dose native and de-N-sulfated heparin, but not de-O-sulfated heparin. These results are consistent with our in vitro and ex vivo data, suggesting that HS sidechain sulfation (in particular O-sulfation) may contribute to the accumulation of Ad5CTL in the liver at this timepoint after intravenous administration. Taken together, our data indicate that the FX-mediated interaction of Ad5CTL with HSPGs both in vitro and after intravascular administration in vivo involves the presentation of a ‘sulfation signature’. The domains of FX responsible for mediating Ad5 transduction of hepatocytes have been demonstrated. The Gla domain of FX docks in the cup at the centre of each hexon trimer and the virus: FX complex is then delivered to the hepatocyte surface via a heparin binding exosite in the FX serine protease domain which tethers to HSPGs at the cell surface [21]. While activated FX (FXa) has previously been shown to bind to the cell surface of hepatocytes and tumour cell lines, this interaction was not observed with FX [58]. The cell surface receptors that mediate FXa interactions with hepatocytes were later shown to be tissue factor pathway inhibitor and nexin-1 and required a functional FXa active site [59]. Previously, it was also shown that a Ca2+-mediated interaction between the Gla domain of FX and phospholipid components of the cell membrane mediated cell surface interactions [60], [61]. A similar phospholipid-mediated interaction between FX and platelets has also been reported [62]. FX has also been previously shown to mediate interactions with the cell membrane of other cell types via other identified receptors. For example, in whole human blood FX has been shown to bind monocytes via the αMβ2 integrin (CD11b/CD18), resulting in its activation via cathepsin G-mediated cleavage, although the domain of FX that binds CD11b was not identified [63]. Previous studies have shown that binding of Ad3 fiber knob to HSPGs in vitro is also dependent on HS sidechain sulfation [64]. A putative HSPG-binding region has been identified in the Ad5 fiber shaft (91KKTK94) [41]. However while reduced hepatic uptake was observed after intravascular administration of a virus harbouring a mutation of the KKTK motif (91KKTK94→GAGA), in vitro assays showed that this virus was deficient in intracellular transport [65]. A recent study has clearly demonstrated that fiber is not involved in binding of the Ad5: FX complex to the cell surface, since a fiberless Ad5 mutant showed no significant reduction in FX-mediated cell surface attachment [21]. This study also showed that the Gla domain of FX binds to hypervariable regions in the Ad5 hexon [21], while positively-charged residues in the FX serine protease domain putatively interact with HSPGs [21], [66], [67], [68]. It is therefore likely that the HS sidechain-dependent interaction of Ad5 with the cell surface is mediated by FX ‘bridging’ to hexon capsid proteins rather than by a direct interaction with fiber. In the past two decades there has been significant interest in the potential use of sulfated polysaccharides such as heparin and heparan sulfates in antiviral therapy (reviewed in [69]). For example, the polyanionic compound PRO 2000 competitively inhibits attachment of the HIV-1 envelope protein gp120 to HSPGs on CD4+ T cells and is currently under development as a topical antiviral gel to prevent cervical HIV-1 transmission [70], [71]. However undesirable side-effects such as anticoagulation limit the use of certain highly-sulfated, high molecular weight polysaccharides, including heparin. As stated previously, viral interactions with HS sidechains at the cell surface are often associated with the presentation of a particular ‘sulfation signature’. For instance, hepatitis E cell binding is thought to be dependent on 6-O sulfation [34] while the interaction of HSV-1 with the surface of target cells during infection in vivo is mediated by 3-O sulfate moieties [35]. Knowledge of the specific positioning and number of sulfate groups required for optimal virucidal activity has facilitated the development of targeted antiviral polyanions such as carrageenan and cellulose sulfate, which have significantly fewer side-effects [69]. This underlines the therapeutic relevance of fully understanding the sulfation requirements for Ad5: FX attachment to host cell HSPGs. This is of particular importance in the context of disseminated adenoviral disease in immunocompromised patients, as several studies have identified FX-binding species C adenoviruses in peripheral blood samples from these individuals [6], [7], [8], [9]. Further detailed studies on the receptor-mediated interactions of Ad5 in circulation are now required to fully characterise the factors underlying the clinical pathogenicity of this virus. All animal experiments were approved by the University of Glasgow Animal Procedures and Ethics Committee and performed under UK Home Office licence (PPL 60/3752) in strict accordance with UK Home Office guidelines. Purified human blood coagulation factor X (FX) was purchased from Cambridge Biosciences (Cambridge, UK). Heparinase III, chondroitinase ABC, bovine intestinal heparin, porcine intestinal heparan sulfate, porcine kidney heparan sulfate, de-N-sulfated heparin, de-O-sulfated heparin and sodium chlorate were obtained from Sigma (Sigma-Aldrich, Gillingham, UK). Primary antibodies raised against EEA1, Rab5, pericentrin or α-tubulin were obtained from Abcam (Cambridge, UK). Primary antibodies raised against intact heparan sulfate (clone 10E4) or heparinase III-digested heparan sulphate (clone 3G10) were obtained from AMS Biotechnology (Oxford, UK). The primary antibody raised against chondroitin sulfate (clone CS-56) was obtained from Sigma (Sigma-Aldrich, Gillingham, UK). All secondary antibodies were obtained from Invitrogen (Paisley, Scotland, UK). The kinase inhibitors LY 294002 hydrochloride, H 89 hydrochloride and SB 202190 hydrochloride were obtained from Tocris Bioscience (Bristol, UK). A549 (human lung carcinoma ATCC CCL-185), HT29 (human colorectal adenocarcinoma: ATCC HTB-38), MDA-MB-231 (human breast adenocarcinoma: ATCC HTB-26) NCI-H522 (human lung adenocarcinoma: ATCC CRL-5810), SKOV3 (human ovarian carcinoma: ATCC HTB-77) and SNB19 cells (human glioblastoma: ATCC CRL-2219) were grown in RPMI 1640 medium supplemented with 10% fetal calf serum, 2 mM L-glutamine and 1% penicillin-streptomycin (Invitrogen, Paisley, UK). HepG2 (human hepatocellular carcinoma: ATCC CRL-11997) and 293 (Human Embryonic Kidney: ATCC CRL-1573) cells were grown in Dulbecco' s Modified Eagle' s Medium (DMEM; Invitrogen, Paisley, UK) supplemented with 10% fetal calf serum, 2 mM L-glutamine and 1% penicillin-streptomycin. CHO-pgsA745 (ATCC: CRL-2242), CHO-pgsE606 (ATCC: CRL-2246) and CHO-pgsF17 cells [47] were grown in Ham' s F-12 medium (Invitrogen, Paisley, UK) supplemented with 10% fetal calf serum, 2 mM L-glutamine and 1% penicillin-streptomycin. The E1/E3-deleted Ad5CTL adenovirus encodes a Rous sarcoma virus (RSV) promoter-driven LacZ expression cassette as described previously [66]. Ad5KO1 is based on Ad5CTL and contains a two-amino acid substitution in the fiber knob (S408E, P409A) that ablates CAR binding. Ad5PD1 is also based on Ad5 and contains a substitution of amino acids 337–344 of the penton base gene (HAIRGDTF) with amino acids SRGYPYDVPDYAGTS, ablating RGD-mediated αv integrin-binding. Ad5KP contains both fiber knob (KO1) and penton base (PD1) mutations. Viruses were propagated in 293 cells and purified by CsCl gradient centrifugation. Vector genomes were quantified by SYBR green quantitative polymerase reaction (qPCR) on an Applied Biosystems ABI Prism 7700 sequence detection system using primers directed against the LacZ transgene (forward 5′-ATCTGACCACCAGCGAAATGG-3′ and reverse 5′-CATCAGCAGGTGTATCTGCCG-3′). Viral particles were determined by micro bicinchoninic-acid assay (Perbio Science, Cramlington, UK) using the formula 1 µg protein = 4×109 VP [72]. Adenoviruses were fluorescently labelled using an Alexa Fluor-488 (green) protein labelling kit according to the manufacturer' s instructions (Invitrogen, Paisley, UK). Free label was dialysed from labelled Ad5 using 10,000 molecular weight cut off slide-a-lyzer cassettes (Perbio Science, Cramlington, UK) overnight in 100 mM Tris, 50 mM EDTA. A dye: virus particle ratio of 3∶1 was used for all labelling reactions. Fluorescent dye labelling efficiency was assessed using the ‘proteins and labels’ function on a Nanodrop-1000 spectrophotometer. Infectivity of labelled adenoviruses was verified by in vitro gene transfer assay as described below. FACS was performed on SKOV3 cells cultured under the conditions described above. Cells were detached from culture vessels using a 1× citric saline solution and counted using trypan blue exclusion. Cells were then resuspended in serum-free DMEM (SFDMEM) at a concentration of 4×106 cells/ml. Heparinase III or chondroitinase ABC was then added to 50 µl of this cell suspension at the required concentrations (0 U/ml, 0. 5 U/ml, 1 U/ml, 5 U/ml) and incubated with cells in a shaking waterbath at 37°C for 1 h. Cells were washed twice in SFDMEM) then incubated with primary antibodies (10E4,3G10 or CS-56; all mouse monoclonal antibodies) or a matching isotype control in SFDMEM containing 0. 1% BSA for 30 min on ice. Cells were then washed twice and incubated with a FITC-labelled secondary antibody in SFDMEM for 30 minutes on ice. Cells were washed twice and cell labelling was then detected on a FACS Canto II flow cytometer (Beckton Dickinson, Oxford, UK) using FACS DIVA software. Viable cells were gated by their FSC/SSC profiles, with a minimum of 5000 gated events analysed per sample. Results are expressed as percentage positively-stained cells per sample, from 3 independent samples analysed in triplicate. Total sulfated glycan content was measured in cell lysates using the Blyscan sulfated GAG assay kit (Biocolour, Newtonabbey, Northern Ireland, UK) according to manufacturer' s instructions. Briefly, cultured cells were lysed in RIPA buffer (50 mM Tris, 150 mM NaCl, 0. 1% SDS, 0. 5% sodium deoxycholate, 1% NP40) then lysates were incubated with a molar excess of the cationic, sulfate-binding dye 1,9 dimethylmethylene blue. Lysates were pelleted and unbound dye was removed. Soluble GAG content was measured by determining the quantity of bound dye by spectrophotometric standard curve analysis at 656 nm. Protein concentrations were measured by Bicinchoninic Acid Assay (Perbio Science, Cramlington, UK) as described above. Data are expressed as µg sGAG/mg protein. Cells were seeded in 4-well chamber slides at 1×105 cells/well 24 h prior to assay. Cells were gently washed with PBS then incubated with 1×104 vp/cell in 300 µl SFDMEM for 1 h on ice. Factor X and bovine intestinal heparin were both used at a concentration of 10 µg/ml. Cells were then gently washed with PBS and incubated at 37°C for 15,30,60 or 180 minutes prior to fixation. Localisation of Ad particles at the MTOC was characterised by staining cells using a polyclonal rabbit pericentrin antibody (1∶200 dilution: Abcam, Cambridge, UK) while localisation of Ad particles in early endosomes was characterised by staining cells using a polyclonal rabbit EEA1 antibody (Early Endosome Antigen-1) or a polyclonal rabbit Rab5 antibody at a 1∶200 dilution (Abcam, Cambridge, UK). Cell morphology was assessed using a polyclonal mouse α-tubulin antibody at a 1∶500 dilution (Abcam, Cambridge, UK). Specific binding of primary antibodies was visualised using a goat anti-rat Alexa Fluor 546 (red) secondary antibody in PBS at a dilution of 1∶500. Cells were imaged using a Zeiss confocal imaging system (LSM500). Colocalisation of Ad5 with the MTOC was quantified by visually assessing the percentage of cells with Alexa488-virus and pericentrin co-staining. Data were averaged from 5 40× microscope fields per experimental condition. Cells were seeded in 8-well chamber slides at 5×104 cells/well 24 h prior to assay. Cells were gently washed with PBS then incubated with 100 µM LY 294002 hydrochloride, 40 µM H 89 hydrochloride or 10 µM SB 202190 hydrochloride (Tocris Bioscience, UK) for 30 minutes at 37°C. Cells were gently washed with PBS then incubated with 1×104 vp/cell of Alexa Fluor-488 labelled virus in the presence of 100 µM LY 294002 hydrochloride, 40 µM H 89 hydrochloride or 10 µM SB 202190 hydrochloride in 150 µl SFDMEM for 1 h on ice. Factor X was used at a concentration of 10 µg/ml. Cells were incubated at 37°C for 180 minutes prior to fixation. Localisation of Ad particles at the MTOC was characterised and quantified as previously described. To evaluate the effect of depletion of cellular αv integrins on FX mediated Ad5 trafficking, SKOV3 cells (5×104 cells/well in either 24 well plates or 8 well chamber slides) were transfected with shRNA targeting αv integrin, “off-target” control shRNA, or liposomes only (mock transfected), according to manufacturer' s instructions. Briefly, 2. 5 µl/well of 5 µM shRNA was diluted 50 µl in serum free media before being mixed with 50 µl/well of SFDMEM containing 2 µl Dharmafect transfection reagent. The lipid and shRNA solution were mixed and allowed to stand at room temperature for 30 minutes before the addition of 400 µl/well of complete media. Cells were washed with PBS and 500 µl/well of lipid/shRNA solution was added and allowed to transfect cells for 24 hours prior to analysis of knockdown. We confirmed specific knockdown of αv integrin by detection of αv integrin mRNA by RT-qPCR and by flow cytometry for surface levels of the αv subunit. Total cellular mRNA was harvested using RNeasy mini kit (Qiagen), quantified, and 300 ng of mRNA was converted to cDNA by in vitro reverse transcription. Levels of αv integrin mRNA in 2. 5 µl of cDNA were subsequently quantified by TAQman qPCR and normalised to total levels of the housekeeper GAPDH. For analysis of αv integrin knockdown by flow cytometry, cells were detached 48 h post-transfection and incubated with an anti-αv antibody (mAb mouse IgG1 clone L230) for 1 h at 4°C at a final concentration of 10 µg/ml. Cells were washed with SFDMEM, incubated with goat anti-mouse Alexa488-secondary (1∶125 dilution) for a further hour at 4°C, washed again with SFDMEM and resuspended in a final volume of 350µl. Surface levels of the αv integrin subunit were detected using a BD FACS CANTO II flow cytometer, acquiring >10,000 gated events. For cell transport studies, cells were transfected as above in 8-well chamber slides for 24 hours. Cells were subsequently washed and cooled to 4°C, before the addition of fluorescently labelled Ad5 (10,000 vp/cell) in serum free media containing physiological levels of FX. Cells were then placed at 37°C for the stated times, washed, fixed using 4% paraformaldehyde in PBS for 10 minutes before counterstaining with 4′, 6-diamidino-2-phenylindole (DAPI) and mounting in Prolong Gold for analysis as previously described. Six µm frozen liver sections from macrophage-depleted male MF1 mice were incubated with 1×109 vp of Alexa488-labelled Ad5CTL in SFDMEM in the presence or absence of 10 µg/ml FX and/or increasing concentrations of heparins for 1 h on ice. Sections were then washed twice with PBS and mounted using ProLong Gold antifade reagent with DAPI. Sections were imaged using an Olympus Cell∧M imaging system. To quantify adherent Alexa488-Ad5CTL particles, 40× images captured using an Olympus imaging system and were processed using PaintShop Pro and ImageJ. Viral particles were counted using the semi-automated cell counting tool from ImageJ. An average of 5 captured images were analysed per experimental condition. All animal experiments were approved by the UK Home Office. Male MF1 outbred mice aged between 8–10 weeks (weight approximately 35g) and housed in secure barrier facilities were used for all in vivo experiments. Macrophage depletion was carried out by clodronate liposome pretreatment as described previously [21], [66]. Kupffer cell depletion was confirmed by staining frozen liver sections with a rat anti-mouse F4/80 primary antibody at a 1∶50 dilution (Abcam, Cambridge, UK) and a goat anti-rat Alexa Fluor 546 (red) secondary antibody at a 1∶500 dilution and all sections in macrophage depleted mice showed a complete absence of Kupffer cells to confirm the efficiency of depletion (data not shown). For the analysis of virus genome accumulation in the liver 1 h after intravenous virus administration, 1×1011 vp Ad5CTL in 100 µl PBS was injected into the tail-vein of macrophage-depleted mice 5 minutes after intravenous administration of 20 mg/kg or 50 mg/kg heparins in 100 µl PBS. Mice were sacrificed and perfused with PBS 1h post-inoculation. Livers were harvested and total DNA was purified using the QiaQuick Spin DNA Extraction Kit as described above. To characterise Ad localisation in vivo, 1×1011 vp Alexa-labelled Ad5CTL in 100 µl PBS was injected into the tail vein 5 minutes after intravenous administration of 20 mg/kg or 5 0mg/kg heparins in 100 µl PBS. Livers were flushed by cardiac PBS perfusion 1 h later to remove non-adherent virus particles and the largest lobe was then embedded and immediately frozen in OCT Tissue-Tek embedding compound. Frozen liver sections (4 µm) were fixed and stained with rat anti-mouse CD31 antibody at a 1∶50 dilution (BD Pharmingen, Oxford, UK) to detect endothelial cells. Specific binding of primary antibodies was visualised using a goat anti-rat Alexa Fluor 546 (red) secondary antibody in PBS at a dilution of 1∶500. Sections were imaged using an Olympus imaging system. Statistical significance was calculated using 2-sample, 2-tailed student' s t-tests. P-values of <0. 05 or over were considered statistically significant. Results presented are representative data from a minimum of two separate experiments with at least 3 experimental replicates per group. All virus binding and transduction experiments were performed in biological triplicates and on at least three independent occasions. All error bars represent SEM.

Adenoviruses can infect many cell types and cause a range of illnesses in humans, including respiratory, ocular and gastrointestinal disorders. These illnesses are rarely fatal; however, in immunocompromised individuals, especially young children, disseminated adenovirus infections can cause serious and life-threatening complications. Studies have shown that several adenoviruses including vectors based on adenovirus serotype 5 (Ad5) bind to coagulation factor X (FX) in the bloodstream. Ad5 uses the high-affinity interaction with FX to putatively bind to heparan sulfate proteoglycans (HSPGs). However, very little is known about this infection pathway. Here we demonstrate that interaction of Ad5: FX with HSPGs is solely via the HS sidechains of these ubiquitously-expressed molecules. We further show that this interaction is dependent on HS sulfation, in particular O-sulfation. Although attachment of Ad5: FX to HSPGs is independent of the coxsackievirus and adenovirus receptor (CAR) or αv integrins, efficient and rapid intracellular transport of Ad5 retains a dependence on engagement of αv integrins via the penton base protein. This is the first study to characterise the receptor requirements for cell uptake via the recently-identified, FX-mediated infection pathway, which may be of significance for the development of therapies against disseminated adenoviral disease.

Abstract Introduction Results Discussion Materials and Methods

virology/virion structure, assembly, and egress virology/host invasion and cell entry

2010

Requirements for Receptor Engagement during Infection by Adenovirus Complexed with Blood Coagulation Factor X

Aggregation of misfolded proteins or peptides is a common feature of neurodegenerative diseases including Alzheimer’s, Parkinson’s, Huntington’s, prion and other diseases. Recent years have witnessed a growing number of reports of overlap in neuropathological features that were once thought to be unique to only one neurodegenerative disorder. However, the origin for the overlap remains unclear. One possibility is that diseases with mixed brain pathologies might arise from cross-seeding of one amyloidogenic protein by aggregated states of unrelated proteins. In the current study we examined whether prion replication can be induced by cross-seeding by α-synuclein or Aβ peptide. We found that α-synuclein aggregates formed in cultured cells or in vitro display cross-seeding activity and trigger misfolding of the prion protein (PrPC) in serial Protein Misfolding Cyclic Amplification reactions, producing self-replicating PrP states characterized by a short C-terminal proteinase K (PK) -resistant region referred to as PrPres. Non-fibrillar α-synuclein or fibrillar Aβ failed to cross-seed misfolding of PrPC. Remarkably, PrPres triggered by aggregated α-synuclein in vitro propagated in animals and, upon serial transmission, produced PrPSc and clinical prion disease characterized by spongiosis and astrocytic gliosis. The current study demonstrates that aggregated α-synuclein is potent in cross-seeding of prion protein misfolding and aggregation in vitro, producing self-replicating states that can lead to transmissible prion diseases upon serial passaging in wild type animals. In summary, the current work documents direct cross-seeding between unrelated amyloidogenic proteins associated with different neurodegenerative diseases. This study suggests that early interaction between unrelated amyloidogenic proteins might underlie the etiology of mixed neurodegenerative proteinopathies. Aggregation of misfolded proteins or peptides is a common feature of neurodegenerative diseases including Alzheimer’s, Parkinson’s, Huntington’s, prion and other diseases [1,2]. According to traditional view, each neurodegenerative disease is characterized by aggregation of one or two disease-specific proteins or peptides, for instance, Aβ and tau in Alzheimer’s disease, α-synuclein in Parkinson’s disease, or prion protein in prion diseases. In recent years, however, an increasing number of studies have revealed that some individuals show co-occurrence of neuropathological features characteristic of more than one neurodegenerative disease [3–8] (reviewed in [9]). For example, α-synuclein pathology (Lewy bodies) can be detected in Creutzfeldt-Jakob disease (CJD) patients including sporadic and genetic forms [7,10,11]. Moreover, histological examination demonstrated that aggregates of unrelated amyloidogenic proteins or peptides, including prion protein, tau, Aβ peptides, α-synuclein, immunoglobulin light chain λ, and β2-microglobulin, could be observed within the same plaques or in close proximity [12–17]. However, the mechanisms responsible for co-aggregation of different amyloidogenic proteins are not understood. Could cross-seeding between unrelated amyloidogenic proteins contribute to the etiology of neurodegenerative diseases? Two different mechanisms have been proposed to explain how diseases with mixed brain pathologies or symptoms may arise. According to one mechanism, a general decline in proteostasis including ER stress, impairment of proteasome-, lysosome-, or autophagosome-dependent degradation during normal ageing or disease conditions, could trigger co-aggregation of multiple amyloidogenic proteins that are prone to misfolding. Such a mechanism assumes that imposing a stress on proteostasis results in misfolding and aggregation of multiple proteins independently or semi-independently, i. e. in the absence of direct cross-seeding. Consistent with this hypothesis is a study in which shorter incubation times to prion disease were observed in transgenic mice that overexpress human A53T α-synuclein compared to non-transgenic controls inoculated with three prion strains [18]. Another study reported that inoculation of prions into aged transgenic mice overexpressing human wild type α-synuclein resulted in more extensive and abundant intraneuronal and synaptic accumulation of α-synuclein relative to non-transgenic control mice [19]. In addition, in agreement with the latter mechanism is the observation that CJD patients exhibit impairments of the nigrostriatal pathway, which is a hallmark of Parkinson’s disease [7]. An alternative mechanism proposes that co-occurrence of mixed brain pathologies arises by direct cross-seeding of protein aggregates of one disease-related protein by fibrils or oligomers of an unrelated protein. A few studies have provided experimental evidence in support of this idea. Examples of cross-talk between different yeast prion proteins in a cell were documented more than a decade ago [20,21]. In addition, reactive protein A amyloidosis or senile apolipoprotein A-II amyloidosis were found to develop in mice as a result of cross-seeding by fibrils of apolipoprotein A-II or protein A, respectively [22]. In recent studies, examples of cross-seeding of Tau by Aβ fibrils or α-synuclein fibrils were illustrated using in vitro, cellular and animal models [23–25]. In this study we examined whether prion replication can be induced by cross-seeding with aggregated α-synuclein or Aβ peptide. We found that α-synuclein aggregates formed either in cultured cells or in vitro triggered misfolding of PrPC in serial Protein Misfolding Cyclic Amplification reactions, producing self-replicating PrP states characterized by a short C-terminal proteinase K (PK) resistant region referred to as PrPres. Non-fibrillar α-synuclein or fibrillar Aβ failed to cross-seed misfolding of PrPC. Remarkably, PrPres triggered by aggregated α-synuclein in vitro propagated in animals and, upon serial transmission, produced PrPSc and clinical prion disease characterized by spongiosis and astrocytic gliosis. The current study demonstrates that aggregated α-synuclein is potent in cross-seeding misfolding and aggregation of the prion protein in vitro, producing self-replicating states that can lead to transmissible prion diseases upon serial passage in wild type animals. In summary, this work documents direct cross-seeding between unrelated amyloidogenic proteins associated with different neurodegenerative diseases. To examine the ability of α-synuclein to cross-seed aggregation of PrPC, a Protein Misfolding Cyclic Amplification with beads that employs partially deglycosylated Syrian hamster PrPC as a substrate (dgPMCAb) was used [26,27]. Partial deglycosylation by treatment with PNGase F altered the ratio of the three PrPC glycoforms in favor of mono- and unglycosylated PrPC at the expense of diglycosylated PrPC (Fig 1a). Previously we showed that partial deglycosylation of PrPC removes spatial constraints imposed by bulky N-linked carbohydrates, expanding the range of possible folding patterns that PrPC can acquire upon conversion into self-replicating, PrPSc-like states [26–28]. As an illustration of this effect, amyloid fibrils prepared from recombinant hamster PrP in vitro failed to seed prion replication in Protein Misfolding Cyclic Amplification with beads (PMCAb) that used non-treated PrPC, but displayed consistent seeding activity in dgPMCAb with partially deglycosylated PrPC [26,28]. To examine cross-seeding activity, lysates of HeLa cells transfected with either GFP-tagged wild type (WT) α-synuclein or with the Parkinson’s disease (PD) A30P mutant α-synuclein were assessed for their ability to seed PrPC conversion in serial dgPMCAb reactions consisted of seven serial rounds. Preliminary studies revealed that, in contrast to WT α-synuclein, the A30P mutant formed aggregates in HeLa cells (S2a and S2b Fig). Remarkably, misfolded PK-resistant PrP states were formed in serial dgPMCAb reactions seeded with lysates of cells expressing the A30P mutant, but not WT α-synuclein (Fig 1b). The dgPMCAb-derived PK-resistant PrP products will be referred to as PrPres. The dgPMCAb-derived PrPres consisting of di-, mono-, and unglycosylated bands were detectable by the SAF-84 antibody, which recognizes the C-terminal epitope 160–170. Notably, the PK resistance pattern of dgPMCAb-derived PrPres was similar to PrPres formed in dgPMCAb reactions seeded with Syrian hamster (Ha) full-length recombinant PrP (rPrP) fibrils described in previous studies (Fig 1b) [26]. To test whether seeding activity is attributable to aggregated forms of α-synuclein, recombinant α-synuclein obtained from two independent sources described in Materials and Methods was converted in vitro into amyloid fibrils and tested in serial dgPMCAb (S1 Table, S2c Fig). Fibrillar preparations of α-synuclein from both sources consistently showed cross-seeding activity in serial dgPMCAb in three independent experiments, whereas all reactions seeded with non-fibrillar α-synuclein were negative (Fig 1b and 1c). All experiments were performed using equipment and laboratory space that have never been exposed to prions. Nevertheless, as negative controls, non-seeded dgPMCAb reactions were conducted in parallel in each experiment and were always negative (Fig 1b). To examine the specificity of PrPC cross-seeding by α-synuclein, serial dgPMCAb was seeded with Aβ amyloid fibrils prepared in vitro under six different solvent conditions using Aβ (1–40) peptide (S1 Table, S2d Fig). All dgPMCAb reactions seeded with Aβ fibrils were negative (Fig 1c). To further establish the specificity of cross-seeding, dgPMCAb reactions were seeded with amyloid fibrils prepared in vitro under four different solvent conditions using mouse (Mo) full-length rPrP rather than Ha rPrP (S1 Table). While the sequences of Mo and Ha full-length PrPs are 94% identical (S3 Fig), our previous studies established that the fibrils formed by two rPrP variants are structurally different [29,30]. Serial dgPMCAb reactions seeded with Mo rPrP fibrils were all negative (Fig 1b). In summary, in addition to Ha rPrP fibrils, only α-synuclein in aggregated states (produced in vitro or by cultured cells, respectively) showed cross-seeding activity in the serial dgPMCAb assay. The relative efficiency of cross-seeding or the amounts of active seeds could be estimated from a time-point or a round number at which the first PrPres could be detected by Western blot. When seeded with Ha rPrP fibrils, the conversion products were detected by the third or fourth dgPMCAb rounds (Fig 1d and [26]). In the reactions seeded by fibrillary α-synuclein, the first PrPres was visible by the fifth round. As expected, this result illustrates that the cross-seeding of PrPC aggregation by α-synuclein fibrils was less efficient than seeding by Ha rPrP fibrils. To test whether dgPMCAb-derived PrPres induced prion disease in animals, Syrian hamsters were inoculated with the products of serial dgPMCAb reactions seeded with (i) the lysates of HeLa cells expressing α-synuclein A30P or (ii) WT α-synuclein fibrils produced in vitro (Table 1). The products of non-seeded serial dgPMCAb reactions were inoculated as a negative control. In addition, to examine whether α-synuclein cross-seeded prion replication directly in animals, two animal groups were inoculated with either fibrillar or non-fibrillar WT α-synuclein (Table 1). No obvious clinical signs were detected in any animal groups, and all animals were euthanized at 561 days after inoculation (Table 1). Nevertheless, all animals inoculated with PrPres products from dgPMCAb reactions seeded with fibrillar WT α-synuclein fibrils or lysates of HeLa cells expressing α-synuclein A30P mutant showed PrPres (Fig 2a, Table 1). Remarkably, the PK-digestion pattern of the animal-derived PrPres was very similar, but not identical, to that of dgPMCAb-derived PrPres (Fig 2b). The animal-derived PrPres consisted of predominant monoglycosylated and smaller amounts of di- and unglycosylated bands that can be detected by C-terminal SAF-84 antibody (epitope 160–170) (Fig 2a). The animal-derived PrPres was not detectable by 3F4 antibody immunoreactive to the epitope 109–112 (Fig 2a). The total amount of brain-derived PrPres exceeded more than 103-fold the amount of PrPres in the inocula arguing that PrPres was able to replicate effectively in animals. In addition to PrPres, very small amounts of PrPSc detectible by 3F4 were found in animals in both groups (Fig 2a). None of the animals inoculated with fibrillar or non-fibrillar WT α-synuclein or dgPMCAb-derived material from non-seeded reactions displayed any PK-resistant products, as judged by SAF-84 or 3F4 staining (Fig 2a, Table 1). In summary, PrPres generated in vitro under conditions with altered PrPC glycoform ratios were able to propagate in animals despite unfavorable ratios of di-, mono- and unglycosylated PrPC resulting in accumulation of animal-derived PrPres and small amounts of PrPSc. Histopathological evaluation of two animal groups inoculated with dgPMCAb-derived PrPres revealed mild focal reactive astrogliosis in all examined animals mostly in the hippocampus and variably in other regions such as frontal cortex or thalamus (Fig 3, S4 Fig). Typical spongiform changes associated with prion diseases were not evident, although single small vacuoles were found in certain regions of the brain. Immunostaining for PrP using SAF-84 revealed granular or fine diffuse synaptic deposits, but no plaques or amorphous deposits (Fig 3, S4 Fig). PrP immunoreactivity was seen predominantly in the cortex and hippocampus and variably in subcortical areas. Consistent with very minor amounts of 3F4-positive PrPSc detectable by Western blot in animals of these groups, immunostaining for PrP using 3F4 did not reveal unequivocal pathological deposits (S5 Fig). Overall, histopathological analysis showed similar changes in animals of both groups that involved fine diffuse/synaptic SAF-84-positive PrP immunoreactivity, minor astrocytic gliosis, minimal if any spongiosis and lack of microgliosis (Fig 3, S6 Fig). Control groups including animals inoculated with products of non-seeded dgPMCAb reactions, WT α-synuclein fibrils or non-fibrillar α-synuclein also showed only minor GFAP staining, no significant vacuolization, and lack of microgliosis as well as lack of PrP deposits, as probed by SAF-84 (S7 Fig). In summary, both animal groups inoculated with dgPMCAb-derived PrPres showed limited prion pathology as judged by Western blot and histopathological analysis, but no clinical disease. To test whether animals inoculated with dgPMCAb-derived PrPres might develop α-synucleinopathy due to possible amplification of α-synuclein aggregates in dgPMCAb, the animal group injected with dgPMCAb-derived PrPres seeded with fibrillary WT α-synuclein and un-inoculated age-matched control group were also examined by staining with anti-α-synuclein antibodies. Immunostaining for α-synuclein with antibody 4D6 revealed prominent synaptic α-synuclein immunoreactivity in animals of both groups (S8 Fig), which is consistent with its normal physiologic distribution. Synaptic immunoreactivity was more prominent in areas with larger synaptic boutons (e. g. cerebellum granular layer or hippocampus) (S8a–S8f Fig). In addition, coarse α-synuclein deposits and very rare diffuse neuronal cytoplasmic staining were detected in all examined animals of both groups, however, no unequivocal Lewy-body like inclusions were found (S8 Fig). While we cannot exclude the possibility that these coarser deposits represent early aggregates, these findings should be interpreted with great caution, as it is currently unclear whether Syrian hamsters can develop any human-like α-synuclein pathology. To examine whether serial transmission of brain material containing PrPres leads to clinical prion disease, brain material from animals inoculated with dgPMCAb-derived PrPres induced either by α-synuclein WT fibrils, the lysates of HeLa cells expressing α-synuclein A30P, or non-seeded dgPMCAb-derived material were used for the second passage (Table 1). The animal groups that were injected with brain material containing PrPres developed clinical symptoms including hyperreactivity, dry skin, rough patchy coat and dry eyes (Table 1). All animals from these two groups showed substantial amounts of PrPres detectible by SAF-84 and PrPSc that was detected by both SAF-84 and 3F4 antibody (Fig 2c). Assessment of the dynamics revealed that PrPres continued to propagate during serial transmission, whereas the amounts of PrPSc increased substantially in the course of the second passage relative to those found in the first passage (Fig 2b). The control group of animals, which is the second passage of non-seeded dgPMCAb-derived material, did not display any clinical signs (Table 1). Brain material from this group did not contain any PK-resistant products (Fig 2c). Histopathological examinations of animals with clinical disease from the second passage of dgPMCAb-derived PrPres induced either by α-synuclein WT fibrils or the lysates of HeLa cells expressing α-synuclein A30P revealed features typical for TSE including spongiform degeneration, both reactive astrogliosis and microgliosis, and PrPSc deposition (Figs 4 and 5). Deposition of PrP immunoreactive with 3F4 was detected in multiple brain regions including cortex, cerebellum, thalamus, hippocampus, and caudate putamen (Figs 4 and 5). Several types of 3F4-positive PrP aggregates were observed, including pronounced plaques and amorphous deposits in the subventricular zones; notable perivascular aggregates, perineuronal deposits and small diffuse deposits in deeper layers of the cortex (Figs 4 and 5). Typically, the areas with PrP deposition showed considerable overlap with the areas of reactive astrogliosis suggesting that astrocytes are activated in the region characterized by PrP replication and/or accumulation (Fig 4b and 4c). Although less pronounced than reactive astrogliosis, the reactive microgliosis was also noticeable (Fig 4d, 4e, 4g and 4h). Similarly to the control animal groups from the first passage, the control group from the 2nd passage of non-seeded dgPMCAb-derived material showed only minor GFAP staining, no significant vacuolization, lack of reactive microgliosis, and lack of PrPSc deposits immunoreactive with 3F4 (S9 Fig). The current work demonstrates that aggregated forms of α-synuclein can cross-seed aggregation of the prion protein. Cross-seeding gave rise to PK-resistant, self-replicating PrP states referred to as PrPres that can lead to transmissible prion disease when inoculated and serially passaged in wild type animals. We do not know whether a similar cross-seeding mechanism might take place in vivo. When animals were inoculated with fibrillar α-synuclein directly, no clinical, histopathological or biochemical signs of prion disease such as presence of PrPres or PrPSc as judged by Western blot were observed. The failure to induce prion disease in animals directly by fibrillar α-synuclein could be due to the very low efficiency of cross-seeding in vivo, efficient seed clearance and/or prolonged incubation times that exceed the animal’s life expectancy. If this is the case, a much larger cohort of animals would have to be tested to detect possible rare cases of cross-seeding by α-synuclein than the small groups used in the current study. The format of dgPMCAb assay employed in the current study consisted of seven serial rounds. This amplification protocol was designed to identify fibril preparations capable of seeding prion replication regardless of the amounts of potent seeds present. This format does not intend to compare the relative potency of different fibril preparations, because even miniscule amounts of PrPres could be amplified to the levels detectible by Western blot in seven serial rounds [27,31]. The fact that PrPres was observed only by the fifth dgPMCAb round suggests that the number of active seeds of aggregated α-synuclein capable of initiating PrPres replication is very small and/or that the efficiency of the cross-seeding process is relatively low. Additionally, because α-synuclein forms a variety of aggregated states, including structurally diverse oligomers and fibrils, we do not know the specific α-synuclein states involved in cross-seeding of PrPres [32–34]. We also do not know whether successful seeding by the lysates of cells expressing A30P variant was due to quantitative differences in the amount of aggregates or qualitative differences in the type of aggregates formed in the cells expressing A30P variant versus WT α-synuclein. As judged from experiments performed in vitro, while the A30P variant is less fibrillogenic than WT α-synuclein, it is prone to form oligomers and fibrils structurally different from those of WT α-synuclein [34–36]. Moreover, the possibility that misfolded monomers of the A30P variant are capable of cross-seeding should not be completely excluded. The dgPMCAb substrate used in the current study was PrPC that was treated with PNGase F, which partially removed N-linked glycans from PrPC molecules, changing the ratios of glycoforms in favor of mono- and unglycosylated PrPC at the expense of diglycosylated PrPC. Because of this PNGase F treatment, dgPMCAb should be considered as a very artificial in vitro system. Previously we showed that changing the glycoform ratios in favor of mono- and unglysocylated PrPC releases structural constraints imposed by N-linked glycans and opens up multiple misfolding pathways, resulting in alternative self-propagating structures including PrPres [27,37]. Even a modest change in the glycoform ratio in favor of mono- and unglysocylated PrPC was sufficient to expand the range of plausible self-replicating PrP structures. As such, in contrast to PMCA, dgPMCAb conditions offer advantages in the search of self-replicating states of amyloidogenic proteins capable of successful cross-seeding of PrP. PrPres characterized by a short, C-terminal, PK-resistant region was found to be the first product of PrPC misfolding initiated by cross-seeding. While the conditions of dgPMCAb are considered artificial, dgPMCAb-derived PrPres was very similar to the C-terminal, PK-resistant fragments observed in the majority of patients with sporadic CJD [38] or in atypical bovine spongiform encephalopathy (H-BSE), which is believed to be sporadic in origin [39]. Analysis of PK-resistant species in brains of sporadic CJD-affected individuals identified PrP-derived fragments corresponding to the C-terminal regions encompassing residues ~154/156-231 and 162/167-231, in addition to PrPSc [38]. The relationship between bona fide PrPSc and the C-terminal PK-resistant fragments is not clear [40]. Nevertheless, considering that in PrPSc the C-terminal region is the one that is the most resistant to solvent-induced denaturation [41], the C-terminal PK-resistant fragments found in sporadic CJD might represent an intermediate state toward PrPSc. While dgPMCAb conditions might be considered non-physiological, the products might still be relevant since the relative expression of PrPC glycoforms in a brain varies in a region-specific manner [42]. Therefore, PNGase F-treated PrPC might represent PrPC in those brain regions that display higher proportions of mono- and un- versus diglycosylated PrPC. Remarkably, the PrPres generated in dgPMCAb under conditions with altered PrPC glycoform ratios was capable of propagating in animals despite the high proportion of di- versus mono- and unglycosylated PrPC in brain (Fig 2). This result, along with a fact that PrPres was not amplifiable in PMCAb with non-treated substrate [27], suggests that in brain, PrPres might specifically target only those brain regions or cell types that have a higher proportion of un- and mono-glycosylated PrPC. Nevertheless, overall unfavorable ratios of PrPC glycoforms in brain were likely to be responsible for the slow rate of PrPres replication as well as a modest shift in glycoform ratios observed in animal-derived PrPres relative to dgPMCAb-derived PrPres (Fig 2b). Previous studies on the evolution of synthetic strains helped to establish a relationship between PrPres and PrPSc [26,28,43,44]. The dynamics between PrPres and PrPSc described in the current study resembled the changes in self-propagating PrP states during evolution of prion strains of synthetic origin [26,28,43,44]. In previous studies, transmissible prion disease could be produced in wild type animals by inoculating with either rPrP amyloid fibrils or dgPMCAb-derived PrPres formed upon seeding with rPrP amyloid fibrils, and subsequent serial passaging [26,28,43]. PrPres was found to be the first product of PrPC misfolding in animals inoculated with either rPrP amyloid fibrils or dgPMCAb-derived PrPres [26,28,43,44]. In the course of serial transmission, PrPres gave rise to bona fide PrPSc and was replaced by PrPSc by the end of 2nd or 3d passages [26,28,43,44]. As judged from biochemical assays, PrPres and PrPSc were structurally different [26,27,40]. Nevertheless, a detailed analysis of the dynamics of the PrPres-to-PrPSc transition suggested a mechanism, in which PrPSc forms as a result of rare deformed templating events during replication of PrPres. Once the first PrPSc particles are generated, PrPSc replicates independently of PrPres and replaces PrPres because of its faster replication rate [26,43,44]. Similarities in the dynamics between PrPres and PrPSc in the current work and those observed during evolution of synthetic prions suggest that PrPSc evolved from transmissible, self-replicating PrPres, which was the first misfolded state triggered by cross-seeding. Animals from both groups inoculated with dgPMCAb-derived PrPres showed fine diffuse synaptic PrP deposits immunoreactive with SAF-84 antibody in the cortex and hippocampus, but relatively minor astrocytic gliosis and minimal if any spongiosis or microgliosis. Lack of substantial lesions in the first passage despite deposition of considerable amounts of PrPres correlated well with the lack of clinical symptoms and suggested that PrPres is not toxic per se and does not lead to inflammation of glia. These results are in good agreement with the previous studies that documented lack of clinical symptoms and neuronal toxicity in animals that had deposits of self-propagating C-terminal PrPres states in the absence of PrPSc [43–45]. Animals of the second passage displayed clinical symptoms and pronounced lesions including spongiosis, astrocytic gliosis and microgliosis, the major histopathological hallmarks of TSEs. The presence of clinical symptoms and TSE-specific lesions correlated well with accumulation of PrPSc detectible by immunostaining of brain slices and in Western blots. Several types of 3F4-positive PrP deposits were observed including large plaques in subventricular zones, consistent perivascular aggregates, perineuronal deposits and small diffuse deposits. One of the most intriguing findings of the current work is that self-replicating PrP states that lead to transmissible prion diseases could arise via cross-seeding by α-synuclein that has no sequence homology with the mammalian prion protein (S3 Fig). Surprisingly, fibrils prepared from mouse rPrP, which is 94% identical to Syrian hamster PrP, did not have any detectible seeding effects with respect to hamster PrP as a substrate in this assay. Moreover, our previous studies demonstrated that Syrian hamster rPrP fibrils (referred to as S fibrils), which had a cross-β folding pattern different from that of the hamster fibrils used in the current work [46,47], also failed to seed PrPres in dgPMCAb conducted in Syrian hamster brain homogenates [26]. Together, these results suggest that commonality in folding patterns of fibrils rather than high sequence identity or homology between seeds and a substrate might be critical for successful cross-seeding. In the current work, human α-synuclein aggregates were found to cross-seed hamster PrP. Because human α-synuclein and PrP have no sequence homology, regardless of the species-specific sequence of PrP, it is likely that seeding specificity by human α-synuclein is not limited to hamster PrP. Nevertheless, the question of whether the cross-seeding is due to an idiosyncrasy of these two specific players or whether this effect could be generalized is of great clinical importance and should be addressed in future studies. Notably, studies on cross-species prion transmission demonstrated that differences in PrP primary structures between host and donor do not always guarantee a strong species barrier [48]. For instance, prions from a variety of species can be transmitted very effectively to the bank vole despite differences in amino acid sequences, showing very little if any species barrier and suggesting that the bank vole is a universal host [48,49]. On other hand, in certain lines of transgenic mice expressing human PrPC the transmission of a new variant CJD showed significant barrier, as judged from long incubation times, incomplete attack rates or lack of clinical diseases, despite identity in amino acid sequences of the host PrPC and donor PrPSc [50,51]. The current study, along with the observations that spontaneous non-seeded conversion or conversions seeded with Mo rPrP or Aβ fibrils have not been observed, argues that in dgPMCAb conditions conversion of PrPC into self-propagating PrPres displays a high energy activation barrier. What characteristics of seeds are important for successful cross-seeding? Lack of detectable cross-seeding effects by non-fibrillar α-synuclein or fibrillar Aβ suggests that having amyloid-specific, cross-β folding patterns might be important but is not sufficient for cross-seeding. Consistent with the mechanism postulated by deformed templating [52,53], we propose that successful cross-seeding requires a partial overlap between the cross-β folding pattern of the seed and the folding pattern favored by the primary structure of the substrate, whereas high sequence homology between two proteins is not as important. As bulky N-linked glycans limit the range of possible self-replicating states accessible to PrPC due to spatial interference, their partial cleavage makes the states, that are otherwise prohibited, accessible to PrPC. Although Aβ fibrils formed in vitro did not seed formation of PrPres in dgPMCAb, the possibility of prion cross-seeding by Aβ cannot be completely dismissed. Several conformationally distinct strains of Aβ fibrils have been identified in Alzheimer’s disease patients and mouse models of Alzheimer’s disease [54–57]. Moreover, unless seeded with brain extracts from individuals with Alzheimer’s disease, the fibrils produced in vitro were found to be structurally different from those formed in human brain [58]. Since it is highly unlikely that our in vitro preparations of Aβ fibrils contain structures similar to those formed in human brain, the possibility of cross-seeding of prion formation by naturally occurring Aβ fibrils needs to be examined. Recent studies demonstrated cross-seeding activity of Tau aggregation by Aβ fibrils in cellular assays, producing potent Tau seeds that induced Tau pathology in vivo [23]. In support of the hypothesis that the structure of seeds is important in determining the effectiveness of the cross-seeding, two distinct strains of recombinant α-synuclein fibrils produced in vitro were shown to cross-seed tau aggregation in primary neurons and transgenic animals with strikingly different efficiency [24]. Nevertheless, the experiment using Aβ fibrils illustrates that not every fibrillar state possesses cross-seeding activity, documenting the high selectivity of cross-seeding under dgPMCAb conditions. The possibility of cross-seeding raises questions regarding the etiology of prion diseases that are considered to be sporadic in origin. As shown in the current study, cross-seeding might give rise to self-replicating PrP states, which are not toxic per se but lead to PrPSc and prion disease upon serial transmission. One might hypothesize that in patients with α-synuclein pathology, cross-seeding might trigger PrPC misfolding leading to a progression and combined Parkinson’s and prion diseases. Keeping in mind that Parkinson’s disease is more prevalent than sCJD and that the progression of the clinical stage of Parkinson’s disease is slower than that of sCJD, one cannot exclude the possibility that cross-seeding of prions by pathogenic states of α-synuclein might be responsible for a small fraction of sCJD cases. Vice versa, misfolding of PrPC variants associated with genetic prion disease might cross-seed α-synuclein aggregates resulting in mixed brain pathologies. In agreement with this hypothesis, previous studies have described the coexistence of clinical symptoms of CJD and Parkinson’s disease or deposition of both prion and α-synuclein aggregates in brain [3–6,59]. Indeed, 15% of individuals with genetic CJD associated with PrP mutation E200K were found to exhibit Lewy-type α-synuclein pathology [10]. Preclinical multiple system atrophy, which is characterized by inclusions of α-synuclein deposits in glia, was found in a patient who succumbed to sCJD [6]. α-synuclein-immunoreactive deposits have also been found in the central nervous system of patients with various prion diseases, including sCJD, variably protease-sensitive prionopathy, in natural scrapie in sheep and goats, and in hamsters infected with scrapie [4,5, 7,12,13]. Which cellular sites are involved in cross-seeding? While the majority of aggregated α-synuclein including α-synuclein A30P mutant is deposited intracellularly in the form of Lewy bodies [60,61] (reviewed in [62]), a series of recent studies described extracellular α-synuclein, which is believed to be responsible for cell-to-cell spread of α-synuclein aggregates [63–67]. Extracellular α-synuclein oligomers were found either in association with exosomes or free [66]. Recent studies suggested that PrPC might be involved in spreading extracellular α-synuclein and that the charged PrPC region encompassing residues 95–110 is responsible for the interaction with α-synuclein [68]. Fibrillar α-synuclein was found to bind strongly to PrPC-expressing cells and spread faster in PrPC-overexpressing mice in comparison to the wild type or knockout controls [68]. Lysosomes might serve as alternative cellular sites of cross-seeding, as both oligomeric α-synuclein and PrPC are processed through the endo-lysosomal system [67,69–72]. It will be important to establish in future studies whether cross-seeding of prions by α-synuclein occurs in vivo and whether it can be induced by extracts of pathological α-synuclein derived from Parkinson’s disease patients. In addition, future studies of the interaction among proteins associated with different neurodegenerative diseases should establish whether the concept of cross-seeding can be generalized. This study was carried out in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. The animal protocol was approved by the Institutional Animal Care and Use Committee of the University of Maryland, Baltimore (Assurance Number A32000-01; Permit Number: 0215002). Full-length recombinant mouse or Syrian hamster PrPs (rPrP) were expressed and purified according to a previously described procedure [73]. Lyophilized preparations of rPrPs were dissolved in 5 mM HEPES (pH 7. 0) immediately before use. To form mouse rPrP fibrils, the rPrP stock solution was supplemented with 50 mM MES (pH 6. 0), three 2/32” Teflon beads (McMaster-Carr, Robbinsville, NJ), and either 0,0. 1,0. 5 or 2. 0 M GdnHCl, and incubated at 37°C under continuous agitation. Fibrillation was performed with 0. 25 mg/ml rPrP under 600 rpm horizontal shaking using Wallac 1296–004 Delfia Plateshake in a total volume 0. 6 ml. Amyloid formation was confirmed by the Thioflavin T (ThT) fluorescence assay as previously described [74] (S1 Table). Syrian hamster rPrP fibrils were formed in 0. 5 M GdnHCl as previously described [26]. Human wild type (WT) α-synuclein was from two sources: (#1) purified as previously described [75] and (#2) purchased (cat # S-1001-2, rPeptide, Bogart, GA). The purity of purified α-synuclein was confirmed by electrophoresis on precast 12% SDS-PAGE (S1 Fig). Lyophilized α-synuclein was resuspended in PBS to a final concentration of 280 μM, supplied with 3 2/32” Teflon beads (McMaster-Carr) in a total volume 0. 5 ml and incubated at 37°C under 600 rpm horizontal shaking using Wallac 1296–004 Delfia Plateshake. Formation of fibrils was confirmed using the ThT fluorescence assay as described [74]. The fibrils were then used for seeding protein misfolding cyclic amplification with partially deglycosylated substrate (dgPMCAb) or inoculation into Syrian hamsters. Human Aβ (residues 1–40) was purchased (Pepnome Limited, Hong Kong, China) and subjected to fibrillation in a total volume 0. 5 ml under six solvent conditions: (#1) 0. 5 mM Aβ, 150 mM HEPES pH 7. 4,150 mM NaCl, 24 h at room temperature; (#2) 0. 1 mM Aβ, 10 mM PBS pH 7. 4,7 days at 4°C; (#3): 25 μM Aβ, PBS pH 7. 4,7 days at 25°C; (#4) 50 μM Aβ, 100 mM NaCl, PBS pH 7. 4,24 h at 37°C; (#5): 40 μM Aβ, 0. 5 M Tris pH 7. 5,7 days at 37°C with 3 beads (McMaster-Carr); and (#6) 200 μM Aβ, 0. 5 M Tris pH 7. 5,7 days at 37°C with 3 beads (McMaster-Carr). For conditions #5 and #6, the reactions were conducted under 600 rpm horizontal shaking using Wallac 1296–004 Delfia Plateshake. Formation of fibrils was confirmed using the ThT fluorescence assay as described [74] (S1 Table). The fibrils were then used as seeds in dgPMCAb (S1 Table). HeLa cells (American Type Culture Collection, Manassas, VA) were transfected with GFP-tagged human α-synuclein WT or the A30P mutant, and stable cell lines expressing the proteins were isolated. Cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM, Invitrogen) supplemented with 10% fetal bovine serum in the absence of sodium arsenate (cat # 10437, Life Technologies, condition #1), or in the presence of 5 μM sodium arsenate (Sigma-Aldrich) added 1 hour prior to cell harvesting (condition #2). Fluorescent images of live cells were captured using an inverted microscope (Nikon Eclipse TE2000-U) equipped with an illumination system X-cite 120 (EXFO Photonics Solutions Inc, Exton, PA, USA) and a cooled 12-bit CoolSmap HQ CCD camera (Photometrics, Tucson, AZ, USA). Images were processed using WCIF ImageJ software (National Institute of Health, Bethesda, MD, USA). 10% normal brain homogenate (NBH) from healthy hamsters was prepared as described previously [28]. To produce substrate for dgPMCAb, 10% NBH from healthy hamsters was treated with peptide-N-glycosidase F (PNGase F) (cat # P0705S, New England BioLabs, Ipswich, MA) as follows. After preclearance of NBH at 500 × g for 2 min, 1500 U/ml PNGase F was added to the supernatant, and the reaction was incubated on a rotator at 37°C for 5 h. For serial dgPMCAb reactions, 90 ul PNGase F-treated NBH aliquots were placed in thin-wall PCR tubes, supplied with two 2/32” Teflon beads (McMaster-Carr) [76], and seeded with 10 ul of one of the following: (i) in vitro-produced α-synuclein fibrils; (ii) non-fibrillar α-synuclein; (iii) in vitro-produced Aβ fibrils; (iv) in vitro-produced mouse rPrP fibrils, (v) lysates of HeLa cells expressing wild type (WT) human α-synuclein, or (vi) lysates of HeLa cells expressing A30P variant α-synuclein. As negative controls in each experiment non-seeded dgPMCAb reactions were placed onto the same microplate horn and sonicated/incubated together with seeded dgPMCAb reactions. dgPMCAb reactions were carried out using a Misonix S-4000 microplate horn (Qsonica LLC, Newtown, CT) for seven rounds. Each round consisted of 30 sec sonication pulses delivered at 170W energy output applied every 30 min during a 24 hour period. For each subsequent round, 10 μl of the reaction from the previous round were added to 90 μl of fresh substrate. Ten μl of dgPMCAb-derived samples were supplemented with 5 μl SDS and 5 μl PK to a final concentration of 0. 25% SDS and 25 μg/ml PK and incubation for 1 hour at 37°C. The digestion was terminated by addition of SDS sample buffer and heating the samples for 10 min in a boiling water bath. Samples were loaded onto NuPAGE 12% BisTris gels, transferred to PVDF membrane, and probed with anti-PrP SAF-84 antibodies (Cayman Chemical, Ann Arbor, MI). Four to five week old Golden Syrian hamsters (all males, Harlan Laboratories, Indianapolis, IN) were inoculated intracranically (IC) into the left hemisphere, ~3 mm to the left of the midline and ~3 mm anterior to a line drawn between the ears under 2% isoflurane anesthesia. Each animal received 50 μl of dgPMCAb-derived materials diluted 10-fold in 1% BSA/PBS or 50 μl of α-synuclein (280 μM in a fibrillar or non-fibrillar state) diluted 10-fold in PBS (final concentration in inocula 28 μM). After inoculation, hamsters were observed daily for disease using a ‘blind’ scoring protocol. Animals from the first passage did not develop any clinical symptoms and were euthanized at 561 days post inoculation by asphyxiation with CO2 (Table 1). For the second passage, 10% brain homogenates (BH) in PBS were dispersed by 30 sec of sonication immediately before inoculation. Each hamster received 50 μl of 10% BH inoculum IC under 2% isoflurane anesthesia and was observed daily for disease using a ‘blind’ scoring protocol. The following symptoms were observed: a non-habituating startle response to sound and touch; an agitated, fidgeting behavior; dry skin, patchy and shedding hair. Animals were sacrificed at 638–642 days post inoculation by asphyxiation with CO2 (Table 1). Animals that did not develop clinical signs of disease were euthanized at 642 days post inoculation (Table 1). Animals were not perfused. 10% (wt/vol) brain homogenate (BH) was prepared in PBS, pH 7. 4, using glass/Teflon homogenizers attached to a cordless 12 V compact drill (Ryobi) as previously described [77]. 10% BH was briefly sonicated and mixed with an equal volume of 4% sarcosyl in PBS, supplemented with 50 mM Tris, pH 7. 5, and digested with 20 μg/ml PK (cat # P8107S, New England BioLabs, Ipswich, MA) for 30 min at 37°C with shaking. PK digestion was stopped by adding SDS sample buffer and heating the samples for 10 min in a boiling water bath. Samples were loaded onto NuPAGE 12% Bis-Tris gels, transferred to PVDF membranes, and probed with 3F4 or SAF-84 antibodies. Histopathological studies were performed on three animals per group. Formalin fixed brain halves were divided at the midline. Right hemisphere was frozen, and left hemisphere was fixed in 10% neutral buffered formalin solution. Formalin-fixed hemispheres were paraffin embedded, sliced into 4 μm sections and processed for hematoxylin-eosin stain as well as for immunohistochemistry for PrP using the mouse monoclonal anti-PrP antibody SAF-84 (1: 1000, Cayman Chemical) and 3F4 (1: 1000, Covance, Berkeley, CA, USA), anti-glial fibrillar acidic protein (GFAP; 1: 3000, Dako, Glostrup, Denmark), or anti-Iba1 antibody (1: 500, Wako, Richmond, VA, USA). Horse radish peroxidase-labeled goat anti-rabbit and anti-mouse antibody (KPL, Milford, MA) were used as secondary antibody for GFAP and Iba1 (rabbit), 3F4 (mouse) and SAF-84 (mouse). Detection was performed using DAB Quanto chromogen and substrate (VWR, Radnor, PA). Brains were treated in formic acid (96%) prior to embedding in paraffin to deactivate prion infectivity. For detection of disease-associated PrP, we applied a pretreatment of 30 minutes hydrated autoclaving at 121°C followed by 5 minutes in 96% formic acid. We evaluated the brain for the presence of inflammatory infiltrates, spongiform changes, degree of gliosis, and PrP immunoreactivity. For staining of α-synuclein, brain sections were pretreated by microwaving for 10 minutes in citrate buffer (pH 6), followed by treatment in 98% formic acid for 1 minute. Immunostainings was performed using anti-α-synuclein antibodies: 4D6 (1: 10,000, immunogen: human alpha-Synuclein; Covance, Emeryville, CA, USA).

Aggregation of misfolded proteins or peptides is a common feature of neurodegenerative diseases. Recent years have witnessed a growing number of reports of overlap in neuropathological features specific to two or more neurodegenerative diseases in individual patients. However, the origin for the overlap remains unclear. One possibility is that disease that have mixed brain pathologies might arise from cross-seeding of one amyloidogenic protein by fibrillar states of unrelated proteins. The current study examined whether prion replication can be induced by cross-seeding by α-synuclein or Aβ peptide in their aggregated states. We found that α-synuclein aggregates display cross-seeding activity and trigger misfolding of the prion protein (PrPC) in vitro, producing self-replicating PrP states. Non-fibrillar α-synuclein or fibrillar Aβ failed to cross-seed misfolding of PrPC. Remarkably, misfolded PrP triggered by fibrillar α-synuclein in vitro propagated in animals and, upon serial transmission, produced clinical prion diseases. In summary, the current work documents direct cross-seeding between unrelated amyloidogenic proteins associated with different neurodegenerative diseases. This study suggests that early interaction between unrelated amyloidogenic proteins might underlie the etiology of mixed neurodegenerative proteinopathies.

Abstract Introduction Results Discussion Materials and methods

medicine and health sciences hela cells prions biological cultures brain vertebrates animals mammals cell cultures research and analysis methods hamsters infectious diseases specimen preparation and treatment staining zoonoses proteins cell lines brain diseases biochemistry rodents hippocampus immunostaining anatomy neurology biology and life sciences cultured tumor cells amniotes organisms amyloid proteins prion diseases

2017

Cross-seeding of prions by aggregated α-synuclein leads to transmissible spongiform encephalopathy

Sample-induced image-degradation remains an intricate wave-optical problem in light-sheet microscopy. Here we present biobeam, an open-source software package that enables simulation of operational light-sheet microscopes by combining data from 105–106 multiplexed and GPU-accelerated point-spread-function calculations. The wave-optical nature of these simulations leads to the faithful reproduction of spatially varying aberrations, diffraction artifacts, geometric image distortions, adaptive optics, and emergent wave-optical phenomena, and renders image-formation in light-sheet microscopy computationally tractable. Light-sheet fluorescence microscopy is a popular tool for the volumetric imaging of developing organisms [1–5]. As light-sheet microscopes continue to be developed for progressively bigger biological samples [6,7], there is an increasing need for computers to process gigantic sets of imaging data and extract biologically relevant information [8]. With sample size, however, also light-scattering induced imaging artifacts become increasingly prevalent. During data acquisition as well as post-processing, these imaging artifacts are mostly dealt with on a purely phenomenological basis. A faithful forward model of the wave-optical imaging process, however, would i) enable rigorous benchmarks of deconvolution and segmentation strategies against ground-truth data [9], ii) serve as training platforms for machine learning approaches for image restoration and information extraction and iii) leverage the efficient use of adaptive optics [10,11] to prevent sample-induced image degradation during the acquisition process. Predicting light-tissue interactions is particularly demanding when leaving the single scattering regime [12] or strictly diffusive transport [13]. And despite significant computational advances [14], generally applicable solutions [15] remain computationally costly, effectively prohibiting the simulation of image formation in microscopy. Even when constraining simulations to the biological relevant case of predominantly forward scattering tissue [10,16], individual point spread function (PSF) calculations still require multiple seconds [17]. As aberrations deep inside tissues are unique for virtually each point in the sample [10,12] a realistically large biological specimen, i. e. an embryo with a volume ∼ 100 μm3 would require 105–106 volumetric PSF calculations in order to faithfully mimic the wave-optical imaging process, which with current methods would take several weeks. As a consequence, attempts to simulate microscopic imaging have been limited to ray optics [18] or convolution with a constant PSF [19], approaches that do not reflect the wave-optical nature of light interaction with optically heterogeneous biological samples. Here, we report on biobeam, a software package that enables the first rigorous simulations of wave-optical image formation in light-sheet microscopes by i) a novel multiplexing scheme for PSF calculations and ii) efficient GPU parallelization. The pipeline underlying our software is based on the observation that the beam-propagation model (BPM) for fiber optics [20,21] can also be used to mimic scattering biological cells [22]. To guarantee good accuracy of BPM beyond strictly paraxial wave propagation, we use the exact propagator together with a locally adapted expansion of refractive indices (see S1 Text, Notes 1 and 4 for details and validation against analytically tractable scattering models). To massively reduce the computational cost for ∼106 wave-optical PSF calculations, we introduce a novel multiplexing scheme. In this we exploit the fact that for typical imaging scenarios, a single camera image can be constructed from sets of 100s to 1000s of mutually independent, spatially varying, non-overlapping PSFs. Such sets can be calculated within single, highly multiplexed simulations (see S3 Fig and S1 Text, Note 8), which is similar to the operational principle of highly multiplexed confocal measurements of spinning disc microscopes [23,24]. Our software is further accelerated ∼20 fold by the efficient use of GPU implementations for all low-level calculations. In this way, the propagation of an arbitrary light field such as the multiplexed set of ∼1000 PSFs through a typical volume of 10243 voxels takes less than 500ms on a single graphics card (cf. S1 Text, Note 5, S1 Table and S4 Video), corresponding to a 20. 000 fold acceleration compared to the sequential calculation of PSFs. Implementation was done as an open source Python software package. All computationally heavy parts are lifted to the GPU via OpenCL [25], thereby keeping all the advantages of Python as a dynamically typed high level language that is vastly used in the scientific community without compromising on performance. Apart from its technical focus on speed, biobeam is specifically designed to make wave-optical experiments in-silico as easy as possible, e. g. by providing a simple API that makes it easy to apply different input fields as well as PSF/aberration calculations via diffraction-limited point source propagation through tissue (see S1 Text, Note 3 for example listings). We first demonstrate the power of our software package by presenting the first wave-optical simulation of the volumetric image-formation process in a light-sheet microscope [5,9, 18]. To start, a cylindrical sheet of light is propagated at a specific axial position through a fluorescently labelled embryo model with refractive indices of cytoplasm, nuclei and organelles as previously reported [12] (Fig 1a and S1 & S7 Videos). From this, one obtains the fluorescence excitation at every point in the volume for every axial position of the incident light field. Next, for each focal plane a full set of detection PSFs is obtained by propagating light from multiplexed, diffraction-limited point-sources orthogonally through an idealized, refocusing lens towards the camera (Fig 1c, upper left, S2 Video). This way, we obtain a quasi-continuum of spatially-dependent, volumetric PSFs with position-dependent aberrations that stem from distortions and scattering in both the illumination and the detection paths (Fig 1c left, see also S4 Fig for calculated aberration maps). Convolving the exhaustively sampled sets of spatially-varying PSFs with the fluorescent object finally yields the wave-optical image as seen by the camera (Fig 1c, right). These simulations are particularly demanding due to the large grid size (i. e. 1024 × 2048 × 1024 voxels for Fig 1) and sampling density requirements (typically 0. 5–5μm), here resulting in a total of 106 PSFs that were calculated in well under 10 minutes on a single graphics card (see also S1 and S2 Videos). In contrast, the non-multiplexed calculations would have required more than two months on a single CPU. As a result of this computational pipeline, biobeam generates faithfully calculated 3D microscopy data sets that account for both refraction and diffraction based imaging artifacts. These include image blur and contrast loss, spatially varying PSFs, speckle artifacts, image granularity, as well as sample-induced geometric distortions such as lensing, image displacements and split-screen type double images (cf. Fig 2a and 2b, S3 Video and S1 Fig for benchmarks against analytically tractable models). We additionally validated our approach by comparing simulation results for several experimental situations, such as the diffraction pattern behind an edge knife (S5 Fig), the scattering of a light-sheet at agarose embedded micro-spheres (S6 Fig) and the image distortions obtained when projecting test images through a refracting sphere (S7 and S8 Figs). Beyond the theoretical validation of the core propagation model, these successful comparisons of simulations against experiments further demonstrate that our high-level implementation works correctly in the biological relevant case of low refractive index contrasts. Being a particularly flexible software package, biobeam provides further pre-implemented illumination modalities, including Bessel lattices (Fig 1d, right, see S2 Fig, S5 Video), and allows for practical extraction of sample-induced aberrations as spatially resolved Zernike maps (see S4 Fig and S1 Text, Note 9). Next, we demonstrate biobeam’s capability to accurately simulate wave-optical effects relevant to adaptive optics (AO) imaging [10]. It is well-established and exploited in imaging that perfect imaging foci can be created behind strongly scattering screens [26] and within biological samples by appropriately shaping the wavefront before entry into the scattering medium. We recapitulated this finding by explicit simulations of light propagating through a spatially extended synthetic tissue sample represented by a Perlin-noise refractive index distribution (n = 1. 36 ± 0. 03). As expected, we find also in simulations that conjugation of the wavefront at the surface of the sample allows one to generate diffraction-limited foci inside these scattering tissues (Fig 2c, S6 Video and S9 Fig). Furthermore, we show that biobeam is capable of reproducing the shift-shift memory effect, an emergent wave-optical phenomenon responsible for the significant robustness of adaptive imaging against lateral focus displacements. According to this wave-optical phenomenon, the radius of the iso-planatic patch (distance over which an AO correction pattern works) is determined by the statistics of a tissue generated speckle field inside this medium. Based on a total of 22500 PSF calculations, we show that our simulations faithfully capture this emergent wave-optical effect. In agreement with recent analytical arguments and their empirical confirmation [10], we show by our computational microscopy experiments that the average persistence of the laterally shifted focus is precisely limited by the autocorrelation length of a speckle pattern that would result from an incident plane wave. This phenomenon is accurately reproduced even four mean-free-path lengths inside the tissue (Fig 2d). While currently using a variant of the beam propagation method, our PSF multiplexing scheme is in principle also compatible with other low-level field stepping algorithms, including those that iteratively account for multiple back-reflections [14,15], when even higher precision or penetration depths are required. We summarize that biobeam enables faithful whole-tissue wave-optical simulations of light-sheet microscopes due to i) a novel multiplexing scheme of PSF calculations and ii) efficient GPU parallelization. Biobeam renders the biological imaging process computationally tractable, thus providing the link between wave-optically recorded image material, and the ground truth object. Given the modular nature of our software package, these simulations are easy to implement and can be flexibly adapted to custom imaging scenarios and microscopes. Beyond the reproduction and identification of commonly occurring imaging artifacts (see Supplement for comparison with experimental data), we demonstrated that biobeam is compatible with imaging scenarios in which sample-induced PSFs degradation is overcome by the use of adaptive optics. Furthermore, we showcase the optical capabilities and accuracy of our software by explicitly demonstrating that emergent wave-optical phenomena such as the shift-shift memory effect are quantitatively reproduced deep inside tissues. While we chose a variant of BPM as a low-level field stepping routine, the here presented strategy of multiplexing PSF calculation is more general in nature, and may also be used in combination with other light propagation algorithms, e. g. in scenarios where higher accuracy at large angles and/or more isotropic scattering is of relevance [14]. We conclude that biobeam is a flexible and particularly powerful platform to systematically study wave-optical image-formation by microscopes in scattering biological tissues. Prospectively, we see biobeam helping to improve microscope design, enhancing deconvolution and segmentation strategies by providing realistic imaging data-sets along with ground truth data, and paving the way for a new generation of smart, adaptive microscopes that learn to treat the sample as a part of the optical path. Of great help in this will be the rapidly increasing knowledge of stereotypical refractive index distributions in embryos, tissues, cells, subcellular compartments and their constituents as derived from tomographic phase microscopy [12,27–29] and complemented by electron microscopy morphological data [30]. Beyond the simulation of light-sheet microscopes, our software can also be used to simulate other imaging modalities such as wide-field, laser-scanning confocal, and light-field [31] microscopes, as well as for novel micro-lens concepts [32] and the emerging field of soft photonics [33]. Especially, we emphasize the potential to improve the understanding of physiological image-formation inside the eye [34], taking into account the optics of the lens [35] as well as the retina [36–39]. Biobeam is available as open-source (BSD-3 license) python package at https: //maweigert. github. io/biobeam. Datasets can be found at https: //publications. mpi-cbg. de/6874-data/.

Modern microscopes permit to acquire high quality images of large fields of view, which is the result of a decade-long development of computer aided optical design. However, this high image quality can only be obtained at the very surface of biological specimens: when trying to penetrate deeper into biological tissues, light scattering by cells rapidly leads to severe image blur and computers have so far been unable to model the process by which light forms images in such turbid optical environments. We developed a software that allows one to simulate how microscopes record images deep inside scattering biological samples. Our software reproduces a wide range of optical effects that underlie image blur in tissues. Hence strategies to improve image quality within three-dimensional samples can now be systematically tested by computers. Specifically, our software reproduces intricate wave-optical effects that have recently been proposed as strategies to gain perfect images even in the most turbid environments. This provides the chance for a new generation of microscopes, in which computer models guide the imaging process to enable highest possible resolution even deep inside biological specimens.

Abstract Introduction Design and implementation Results Discussion Availability and future directions

fluorescence imaging chemical characterization diffraction digital video imaging microscopy medicine and health sciences engineering and technology light microscopy light electromagnetic radiation simulation and modeling adaptive optics light scattering optical analysis microscopy optical equipment pharmacology waves research and analysis methods tissue distribution imaging techniques pharmacokinetics physics biochemistry scattering equipment biology and life sciences physical sciences refractive index video microscopy

2018

Biobeam—Multiplexed wave-optical simulations of light-sheet microscopy

Spike-timing-dependent plasticity (STDP), a form of Hebbian plasticity, is inherently stabilizing. Whether and how GABAergic inhibition influences STDP is not well understood. Using a model neuron driven by converging inputs modifiable by STDP, we determined that a sufficient level of inhibition was critical to ensure that temporal coherence (correlation among presynaptic spike times) of synaptic inputs, rather than initial strength or number of inputs within a pathway, controlled postsynaptic spike timing. Inhibition exerted this effect by preferentially reducing synaptic efficacy, the ability of inputs to evoke postsynaptic action potentials, of the less coherent inputs. In visual cortical slices, inhibition potently reduced synaptic efficacy at ages during but not before the critical period of ocular dominance (OD) plasticity. Whole-cell recordings revealed that the amplitude of unitary IPSCs from parvalbumin positive (Pv+) interneurons to pyramidal neurons increased during the critical period, while the synaptic decay time-constant decreased. In addition, intrinsic properties of Pv+ interneurons matured, resulting in an increase in instantaneous firing rate. Our results suggest that maturation of inhibition in visual cortex ensures that the temporally coherent inputs (e. g. those from the open eye during monocular deprivation) control postsynaptic spike times of binocular neurons, a prerequisite for Hebbian mechanisms to induce OD plasticity. Association-based Hebbian plasticity is a powerful form of activity-dependent synaptic modification capable of shaping the response properties of neurons during development, and is a proposed substrate for experience-dependent learning [1]–[3]. However, associative forms of plasticity by themselves are destabilizing and must be constrained for circuit activity to remain balanced [4]–[6]. Studies from various preparations demonstrate that both the magnitude and direction of synaptic modification is dependent on the relative timing between pre- and postsynaptic spike events [7]–[13]. Modeling studies indicate that such spike-timing-dependent plasticity (STDP) is inherently stabilizing and competitive because presynaptic inputs that consistently drive postsynaptic spike events ultimately dominate to control postsynaptic spike timing, at the expense of those inputs that are less effective in bringing the postsynaptic cells to threshold. Thus, the net excitatory input onto a given postsynaptic neuron is constant due to re-organization of the synaptic weight distribution [14]. It has been demonstrated that cortical synapses can be modified by STDP rules in vivo in response to visual [15]–[18] and whisker [19] stimulation. Importantly, these studies demonstrate that the temporal precision of spike times required for STDP can be propagated from the periphery to the level of the cortex. Thus, STDP may contribute to maintaining circuit stability during experience-dependent plasticity. In the present study we consider how the stabilizing properties of STDP influence experience-dependent plasticity during postnatal development in the visual cortex. According to the STDP rule, factors that enhance the probability of a given synapse to evoke a postsynaptic action potential, defined as synaptic efficacy [20], will lead to its strengthening. For example, temporal clustering of different inputs is an effective means of increasing synaptic efficacy because inputs that arrive in a temporally coherent group are more likely to summate and bring the postsynaptic neuron to spike threshold. Through such cooperation, a cluster of synapses can grow stronger, while weakening other synapses that are not part of the cluster [21], [22]. Here we use the term temporal coherence to refer to the degree of temporal clustering among presynaptic spike times and operationally define it as the width of the cross-correlogram peak among pairs of spike trains within a pathway. In addition to temporal coherence, the initial synaptic strength (ISS) of a synapse is also a major determinant of its impact on postsynaptic spiking [23], [24]. Strong synapses have an advantage among converging inputs, because they are more likely to drive postsynaptic spiking and thus their spike times are likely to fall within the potentiation window of STDP. Indeed, in the developing visual system of the tadpole, Zhang et. al. (1998) demonstrated that synaptic strength is capable of conferring a competitive advantage during STDP-mediated synaptic re-organization. This raises the question of whether and how initially strong synapses can be weakened, and which parameter, temporal coherence or initial synaptic strength, determines the outcome of synaptic competition. This issue is relevant to ocular dominance (OD) plasticity in the developing primary visual cortex (V1). During a postnatal critical period, monocular deprivation (MD) causes binocular neurons in the primary visual cortex to shift their responsiveness towards open-eye inputs via a Hebbian-based process [1], [25]. Importantly, monocular blurring using an overcorrecting contact lens, which distorts but does not eliminate vision, is equally effective in inducing a shift in ocular dominance [26]. These results suggest that it is the pattern of visual input, likely manifested as the temporal correlation of retinal afferent activity, that drives plasticity [27], [28]. In rodents, the majority of neurons in the binocular visual cortex are normally dominated by inputs from the contralateral eye, yet closure of this eye during the critical period results in the weakening of its inputs and strengthening of the ipsilateral, open eye inputs. Therefore, temporally coherent inputs are able to overcome the initially strong but less coherent inputs and eventually dominate in driving binocular neuron responses. However, the conditions and cellular mechanisms that confer an advantage to ipsilateral inputs are not well understood. GABAergic inhibition potently influences input summation, which is required for spike generation, by restricting the temporal window over which inputs are able to effectively cooperate [29]–[31]. Intuitively, increasing the strength of GABAergic inhibition would seem a good candidate for shifting the control of postsynaptic spiking to inputs with a higher temporal coherence versus inputs with a higher synaptic strength. In addition, GABAergic inhibition develops in a protracted postnatal period [32]–[34], and this protracted development was shown to regulate the timing of the critical period for OD plasticity [35]–[40]. We therefore hypothesize that the developmental increase of inhibition in V1, by biasing the control of postsynaptic spiking, ensures that Hebbian plasticity mechanisms are engaged during MD to strengthen the temporally coherent inputs over those with higher initial synaptic strength. Here we combine modeling and experimental approaches to examine the role of GABAergic inhibition in promoting the selective strengthening of temporally coherent inputs in the context of OD plasticity. We characterized the maturation of a major class of GABAergic interneuron in rodent visual cortex and found that both synaptic and intrinsic properties of Pv+ interneurons changed dramatically during the critical period of OD plasticity. Using a simple integrate-and-fire neuron model driven by inputs modifiable by STDP, we determined that a sufficient amplitude of synaptic inhibition along with an increase in gain of GABAergic neuron spike output was required to ensure that temporal coherence, rather than initial synaptic strength, controlled postsynaptic spike timing. Inhibition exerted this effect by preferentially reducing the synaptic efficacy of the less coherent inputs. The modeling results predict that the developmental increase in inhibition should decrease synaptic efficacy during the critical period. Indeed, using acute cortical slices of visual cortex, we found that stimulus-evoked synaptic inhibition potently reduced synaptic efficacy at the peak of but not prior to the onset of the critical period. Precocious OD plasticity can be triggered by enhancing GABAergic transmission within V1 [36], [39], [40], suggesting that the machinery for OD plasticity is operational before its natural onset, but lies dormant until local GABAergic inhibitory circuits mature. To examine whether the basic mechanisms for synaptic plasticity are present at glutamatergic connections prior to the onset of OD plasticity, we compared the ability to induce STDP at layer 2/3 synapses in acute slices of V1 prior to the onset (postnatal day 16–18) and at the peak (P26–30) of OD plasticity. Long-term synaptic depression (LTD) or potentiation (LTP) was induced using a STDP protocol. Postsynaptic action potentials were evoked by current injection from the recording electrode, bypassing the need for inputs to summate to bring the cell to spike threshold. Whole-cell current-clamp recordings were made from layer 2/3 pyramids (Figure 1A), in the presence of 10 µM picrotoxin. EPSPs were continuously evoked at a frequency of 0. 2 Hz throughout the experiment from a field electrode placed in layer 4. EPSPs were monitored for a baseline period of 5 minutes, then paired 100 times with an action potential (AP), and further monitored for 20–45 minutes. To induce LTD, the AP was timed to precede the EPSP by 9+/−2 ms. To induce LTP, the AP was timed to follow the EPSP by 9+/−2 ms. An example of LTD induction from a P17 slice is shown in Figure 1C. The average initial slope of the EPSP during the baseline period was 0. 44 mV/ms. Following the AP-EPSP pairing protocol, the average initial slope of the EPSP decreased to 0. 27 mV/ms. We calculated the EPSP slope ratio (EPSP slope post-pairing/EPSP slope pre-pairing) to compare plasticity across slices and ages. In both young and mature slices there was a significant reduction of the mean EPSP slope ratio following the LTD protocol. The mean EPSP slope ratio was 0. 72+/−0. 09 in young slices (p<0. 05, paired t-test, n = 12); and 0. 72+/−0. 13 in mature slices (p<0. 05, paired t-test, n = 12). There was no significant difference between the two ages (Figure 1B), determined using either a t-test (p = 0. 97) or a Kolmogorov-Smirnov (KS) test (p = 0. 99), which is sensitive to differences in data distribution as well as the mean. An example of LTP is shown in Figure 1d. The baseline EPSP slope was 0. 28 mV/ms. Following the AP-EPSP pairing protocol, the EPSP slope increased to 0. 42 mV/ms. Similar to the LTD protocol, the LTP protocol induced significant plasticity at both ages. The mean EPSP slope ratio was 1. 32+/−0. 10 in young slices (p<0. 05, paired t-test, n = 11), and was 1. 36+/−0. 09 in mature slices (p<0. 05, paired t-test, n = 11). There was no significant difference between the two ages (t-test, p = 0. 76; KS, p = 0. 81). We also compared the ability to induce STDP at local recurrent connections within layer 2/3 in the two age groups (Figure 1B). To stimulate local recurrent connections, the field electrode was placed laterally within 50 microns of the recorded cell. Similar to layer4→layer2/3 connections, we found that the activated synapses were modifiable by STDP in both young and mature slices. In response to the LTD protocol the mean EPSP slope ratio was 0. 79+/−0. 09 in young slices (p<0. 05, Wilcoxon signed rank, n = 10), and was 0. 78+/−0. 09 in mature slices (p<0. 05, Wilcoxon signed rank, n = 9). There was no significant difference between the two ages (KS, p = 0. 25). In response to the LTP protocol the mean EPSP slope ratio was 1. 47+/−0. 16 in young slices (p<0. 05, Wilcoxon signed rank, n = 8), and 1. 29+/−0. 10 in mature slices (p<0. 05, Wilcoxon signed rank, n = 9). There was no significant difference between the two ages (KS, p = 0. 90). Therefore, by bypassing the requirement for input summation, we demonstrated that STDP was similarly induced at layer 4→ layer 2/3 connections as well as local recurrent connections in mouse primary visual cortex both prior to the onset and at the peak of the critical period of OD plasticity. Our results, along with others [41], raise the possibility that the ability to induce plasticity at glutamatergic synapses may not be a primary factor in determining the onset of OD plasticity. To examine the changes of GABAergic inhibition onto V1 pyramidal neurons during the critical period, we assayed the maximal inhibitory input onto layer 2/3 pyramids at two developmental ages, just prior to the onset (young) and during (mature) the critical period. Postsynaptic responses in layer 2/3 pyramidal neurons were recorded in response to stimulation of layer 4, which evoked a mixed excitatory-inhibitory response (Figure 2A). Similar to previous reports [42], we found that inhibitory drive increased with age relative to excitatory drive (Figure 2A), and that the maximal inhibitory charge significantly increased with age, while the maximal excitatory charge was stable (Table 1). Parvalbumin-containing (Pv+) basket cells make up ∼50% of GABAergic interneurons in rodent V1, and it has been shown that there is a ∼2-fold increase in the number of Pv+ basket presynaptic terminals surrounding pyramidal somata during the critical period [34]. To determine if there was a corresponding increase in synaptic function, we recorded from synaptically connected Pv+ interneuron to pyramidal neuron pairs in layer 2/3. Pv+ interneurons were recorded using either BAC transgenic mice in which the Pv promoter drives GFP [43] or Pv-cre mice [44] injected with a recombinant adeno-associated virus that expresses GFP specifically in Pv+ basket cells [45]. We found that peak inhibitory synaptic conductance increased by 1. 8-fold during the critical period compared to prior to the onset of critical period, while there was a 25% decrease in the synaptic decay time-constant (Figure 2B, Table 1). In contrast to inhibitory connections, paired recording of pyramidal neurons revealed that the peak excitatory synaptic conductance was similar between young (n = 10) and mature (n = 10) layer 2/3 connections (0. 17+/−0. 09 and 0. 23+/−0. 14 nS, respectively). In addition to synaptic properties, we characterized the intrinsic properties of Pv+ interneurons (Table 2) and found that the current input/spike out curve shifted during the critical period: for the same stimulus input, the number of output spikes was greater during the critical period compared to that prior to the onset of the critical period. Thus, the gain of Pv+ interneurons increased during the critical period. In addition, there was a corresponding decrease in spike half-width (Figure 2C). These electrophysiological results are summarized in Tables 1 and 2. Relative to excitatory connections, synaptic inhibition significantly matured with age. These results do not exclude the possibility that there are subtle developmental changes in synaptic excitation. In contrast to synaptic properties, the intrinsic properties of pyramidal cells, including input resistant, changed with age, as previously reported [46]. Modeling studies have demonstrated that in response to a change in the temporal pattern of presynaptic spike times, STDP implemented in its most basic form, re-organizes the population of synapses converging onto a postsynaptic neuron such that the most coherent inputs are strengthened, while the remaining synapses are weakened [22]. The outcome of STDP driven re-organization is that the net excitatory drive across a population of synapses converging onto a single postsynaptic neuron is stabilized and a subset of presynaptic inputs controls postsynaptic spike timing. Here we extended the Song-Miller-Abbott (2000) model to include two distinct convergent input pathways, and tested the effects of altering presynaptic spike times within a pathway on the ability of the pathway to control postsynaptic spike timing, across a range of inhibitory levels. It has been observed that strong synapses can in some situations undergo less potentiation than weak synapses [47], therefore we also ran the simulation in a weight-dependent mode in which the amount of potentiation was inversely related to synaptic size. We used an integrate-and-fire model neuron driven by 2 presynaptic pathways that each contained 40 synaptic inputs. As in Song et. al. (2000), a function F (Δt) determined the amount of excitatory synaptic modification arising from a single pair of pre- and postsynaptic spikes separated by a time Δt (Methods). Excitatory synaptic conductance was not allowed to exceed a maximum value gmaxex. If the modification function pushed the synaptic weight past the gmaxex value, the weight was reset to the appropriate limiting value. The maximum amount of modification for a single pre- and postsynaptic spike pair corresponded to a 0. 5% change of gmaxex. This function provides a reasonable approximation of the dependence of synaptic modification on spike timing observed experimentally, and makes no assumptions regarding the mechanism (s) of STDP. The model neuron also received inhibitory conductance: each excitatory conductance was followed by an inhibitory conductance of fixed amplitude with a delay randomly varying between 4 to 10 ms. For the simulations in Figure 3, the ratio of the amplitude of inhibitory conductance over gmaxex (gI/gmaxex) was 0. 264, and the average initial synaptic strength (ISS), was the same for both pathways, set to 25% of gmaxex. This amount of inhibition was sufficient to maintain the postsynaptic neuron in a balanced mode, defined by an excitatory-inhibitory ratio of 1. 1–1. 2 at the threshold for action potential generation [14]. A minimal number of parameters were used to generate presynaptic spike trains (see Methods). We then altered two of these parameters to modify the temporal relationships among inputs within and between the two pathways. First, we altered the temporal correlation between pathway 1 (P1) and pathway 2 (P2), defined as whether or not the two pathways share coincident presynaptic spike times. Second, we altered the temporal coherence (1/σ) among inputs within a single pathway, which refers to the degree of temporal clustering of spike times; the value σ represents the temporal jitter (ms) of presynaptic spike times (Figure 3). In our baseline condition, Input Regime I (Figure 3A–C), presynaptic spike times were correlated between P1 and P2, and presynaptic spike times within the two pathways had the same degree of high temporal coherence (σP1 = 3, σP2 = 3). This input regime represents features of a normal binocular neuron, which receives converging and correlated inputs from the two eye pathways, and activity within each pathway displays high temporal coherence driven by the same visual stimulus. Ten independent trials of the simulation were run. Cross-correlation analysis of individual presynaptic spike trains versus the postsynaptic spike train demonstrated that each pathway was capable of driving postsynaptic events during the initial 50 seconds of simulated time, and also in the final phase of the simulation (Figure 3B). Cross-correlation results are schematized in Figure 3a. As expected for this input regime, the mean synaptic weight for each pathway was unchanged during the course of the simulation (Figure 3C), both pathways maintained the ability to control postsynaptic spike timing throughout the simulation, as indicated by the dashed lines in Figure 3A. In Input Regime II (Figure 3D–F), presynaptic spike times between the two pathways were de-correlated, while the high degree of temporal coherence within each pathway (σP1 = 3, σP2 = 3) was preserved. This input regime represents features of a binocular neuron during strabismus, in which inputs from the two eyes are de-correlated and spike times within each eye-specific pathway display high temporal coherence. Ten independent trials of the simulation were run. The mean synaptic weight of one pathway strengthened, at the expense of the opposing pathway, such that the total excitatory driving force was maintained and the firing rate of the postsynaptic neuron was stable. The outcome of which pathway, P1 or P2, dominated occurred at chance level. For Input Regime III (Figure 3G–J), in addition to temporal de-correlation between the pathways, temporal coherence was reduced in Pathway2 (σP1 = 3, σP2 = 6, a σ ratio of 1∶2). This input regime represents features of a binocular neuron during monocular deprivation, in which the two eye-specific pathways are uncorrelated and the temporal structure of activity differs between the two pathways. The open eye views high-contrast patterns, therefore the open-eye pathway likely has a relatively higher degree of temporal coherence compared to the closed-eye pathway [26]–[28]. The temporal structure of the presynaptic spike trains used in the simulation is shown in Figure 3J. Ten independent trials were run. Pathway1, with higher temporal coherence, always attained the higher mean synaptic weight and emerged to drive spike output. In addition, the spike times of the postsynaptic neuron became controlled by P1 in all trials. Our model thus demonstrates that the stabilizing and competitive properties of STDP first described by Song et. al. (2000), also apply to the condition of two independent convergent pathways, and that the pathway with relatively higher temporal coherence will dominate in driving postsynaptic spike times when the initial synaptic strength is equal between pathways. In addition to the temporal structure of presynaptic inputs, initial synaptic strength plays a major role in driving postsynaptic spiking and is also likely to contribute to the outcome of STDP-driven re-organization of inputs. Inputs with higher synaptic strength have an advantage because fewer active synapses are required to evoke a postsynaptic action potential. Indeed, at the retinotectal projection in tadpoles, it has been demonstrated that the extent to which a given pathway potentiates in response to asynchronous stimulation of convergent inputs is dependent on initial synaptic strength [9]. Using Input Regime III (σ ratio of 1∶2), we challenged the ability of P1, the temporally more coherent pathway, to control postsynaptic spike timing by increasing the initial synaptic strength (ISS) of P2. The ISS ratio (P2/P1) was varied from 1. 0 to 2. 0. Using the same level of inhibition as in Figure 3 (gI/gmaxex = 0. 264), we found that the fraction of simulation trials in which P1 controlled postsynaptic spike timing decreased with increasing strength of P2 (Figure 4A, dashed line). When P2 was initially 50% stronger (ISSP2/P1 ratio = 1. 5), it dominated in driving postsynaptic spike activity in only half of the trials. Thus when the ISSP2/P1 ratio was ≥1. 5, P1, the pathway with higher temporal coherence, failed to direct postsynaptic spike timing above chance level. This result, however, formally contradicts with the results of OD plasticity in V1. The majority of binocular neurons in V1 are not equally driven by the two eye-specific pathways. In rodents, approximately 70% of binocular neurons [25] are characterized as class 2/3 cells, preferentially driven by the contralateral eye inputs. Despite this contralateral bias, contralateral eye closure during the critical period shifts the response properties of class 2/3 cells such that they become dominated by the initially weak, but temporally more coherent ipsilateral eye inputs. Our result in Figure 4A, in which the pathway with higher temporal coherence fails to direct postsynaptic spikes, is thus inconsistent with the OD shift of class2/3 neurons induced by monocular deprivation. In the following section we tested the hypothesis that synaptic inhibition can constrain STDP and promote the selective strengthening of temporally coherent inputs at convergent pathways, even when challenged with inputs of higher initial synaptic strength. The stabilizing influence of STDP on net excitatory drive is highly dependent on the non-linearity of the spike generation process [22]. Given that synaptic inhibition has been shown to potently influence input summation by restricting the temporal window over which inputs are able to effectively cooperate [29]–[31], we tested if increasing the amplitude of synaptic inhibition in Input Regime III could bias the weaker but temporally more coherent pathway (P1) to control postsynaptic spike timing. The simulation was run at 6–12 different levels of inhibition and the strength of P2 was increased 2 to 3-fold relative to P1. The simulation was run for 30–50 independent trials for each parameter pair, and the same presynaptic spike train was used for each level of inhibition for a given ISSP2/P1 ratio (Figure 4A). We found that the range of ISSP2/P1 ratios in which P1 out-competed P2 was extended when inhibition was high, and that for a given ISSP2/P1 ratio, higher levels of inhibition increasingly biased the outcome of STDP-driven competition to favor P1 over P2 (Figure 4A). For example, in the case that P2 was set to be 50% stronger than P1 (ISSP2/P1 = 1. 5) and the amplitude of inhibition was set to ≥0. 792 gI/gmaxex, P1 dominated in 100% of the trials, while at lower levels of inhibition (gI/gmaxex = 0 to 0. 264), P2 out-competed P1 in roughly 50% of the trials. We examined a range of relative temporal coherence values and found that increasing inhibition had a similar effect as above in cases that the ISS ratio was sufficiently high to give an advantage to P2 (Figure 4B). As previously reported [46], we found that the input resistance of layer 2/3 pyramidal cells decreased with age (Table 1). A change in input resistance (Rin) could potentially influence summation and therefore impact the effect of inhibition in our simulation. However, there was a parallel change in the membrane time-constant (τmem). Given the relationship, τmem = Rin * whole-cell capacitance, whole-cell capacitance remained stable. In acute slice experiments, values of Rin are typically about 30% higher than in vivo studies [48]. We examined a physiologically realistic range of whole-cell capacitance values in our simulation (from 0. 125 nF to 0. 375 nF) and found that the effect of increasing inhibition was independent of whole-cell capacitance (Figure S3). In the above simulations we implemented contralateral bias as an increase in initial synaptic strength. However, contralateral bias in vivo could be the result of an increased number of inputs rather than increased synaptic strength (or a combination of the two). Therefore we examined the effect of increasing inhibition across an increasing number of P2 inputs and found that as with synaptic strength, increased inhibition helped to ensure that the temporally coherent pathway out-competed the pathway with an initially stronger synaptic drive (Figure 4C). A likely mechanism by which stronger inhibition biased STDP to favor P1 is that inhibition narrowed the window of input cooperation, thereby preferentially restricting the less coherent inputs in P2 from contributing to postsynaptic spike generation. If this is the case, then the relative ability of P1 synapses to drive a postsynaptic spike event (synaptic efficacy) should increase with inhibition, independent of the STDP learning rule. To examine this possibility, we compared the efficacy of P1 and P2 synapses in a simulation implemented as in Figure 3A, except without applying the STDP learning rule. As expected, increasing the level of inhibition (from gI/gmaxex = 0. 264 to 0. 729) decreased the relative efficacy of P2 synapses. At a high level of inhibition, P1 synapses had a slight advantage in driving postsynaptic events 2–8 ms preceding the postsynaptic spike (Figure 5, compare A&B), a temporal window for which STDP-mediated potentiation is the strongest. This slight advantage of P1 was robustly magnified when the STDP learning rule was applied in the simulation. Within the first 100 spikes of the simulation, the relative efficacy of P1 synapses increased compared to the no-STDP condition (Figure 5, compare B&D). P1 synapses completely dominated by the end of the simulation (Figure 5, compare D&F). Analysis of synaptic efficacy was also performed on the results from the initial simulations shown in Figure 3 (see Figure S4). Similarly, we found that the pathway having a slight advantage after the first 100 spikes of the STDP simulation would ultimately dominate. We previously showed that at a low level of inhibition (gI/gmaxex = 0. 264), P2 out-competed P1 in 50% of the STDP simulation trials (Figure 4A, ISSP2/P1 ratio = 1. 5). It was surprising therefore to find that in the absence of STDP, P2 dominated in driving postsynaptic spikes for the full 20 ms time window (Figure 5A, upper left). Given this advantage, P2 would be expected to dominate in 100% of the STDP simulation trials rather than only 50%. The reason that P2 did not dominate in 100% of the trials was that the number of presynaptic spikes occurring after each postsynaptic spike was greater for P2 than P1 (Figure 5G), thus leading to more LTD in P2 than in P1 when the STDP learning rule was applied. Importantly, the difference in the number of presynaptic spikes following postsynaptic spike events between the two pathways was similar for both low and high levels of inhibition (Figure 5G, H). Therefore, the preferential restriction of P2 inputs (Figure 5 A, B) was due to a specific decrease in their synaptic efficacy rather than a increase of their LTD at higher levels of inhibition. It has been shown that under some conditions, the amount of synaptic potentiation is dependent on synaptic strength [13], [49], and that this can impact the outcome of STDP [47]. Therefore, we ran the same simulation shown in Figure 3A with an additional weight-dependent rule in which the amount of LTP was inversely related to synaptic size (Figure 4D). The ability of stronger inhibition to confer a competitive advantage to P1 was maintained. Our whole-cell recordings from pyramidal neurons in cortical slices revealed that while the amplitude of unitary IPSCs increased at the onset of the critical period, there was a concomitant decrease in the synaptic decay time-constant, thus the developmental change in unitary synaptic charge does not scale equally with amplitude. When the 25% decrease in synaptic decay time-constant that we experimentally observed was implemented in the simulation, we found that the effectiveness of increased inhibition in ensuring that P1 out-competed P2 was diminished (Figure 6). We hypothesized that the developmental increase in gain of spike output in GABAergic interneurons that we experimentally observed could compensate for the decrease in the synaptic decay time-constant. We tested this hypothesis by first demonstrating that an increase in gain of spike output increased the probability of P1 out-competing P2 across a range of initial synaptic strength values (Figure 6). In this case the synaptic inhibitory decay time-constant was set to the original value of 5. 75 ms, and increased gain was implemented as the probability of a given inhibitory input spiking twice. For example, a gain of 1. 0 corresponded to a given presynaptic input spiking only once, a gain of 1. 5 corresponded to a 50% probability that a given presynaptic would generate a second spike, and a gain of 2 signified that a presynaptic input would spike twice. Pv+ interneuron-to-pyramidal synapses display short-term depression in which the amplitude of the second IPSC is 10–30% reduced compared to the first IPSC (Figure S5). To account for short-term depression in our simulation, the amplitude of the conductance on the second spike was reduced by 20%. We found that increasing spike output gain increased the probability of P1 out-competing P2. Next, the gain was increased at the same time the synaptic decay time-constant was reduced to 75% of the original value. We found that an increase in gain could indeed compensate for the decreased synaptic decay time-constant. In summary, our modeling results show that both temporal coherence and initial synaptic strength of synaptic inputs can confer a competitive advantage at convergent pathways modifiable by STDP, but synaptic inhibition can constrain STDP to favor temporally coherent inputs at the expense of inputs with stronger initial synaptic strength. These results have implications for ocular dominance plasticity, particularly for class 2/3 neurons. Given that OD plasticity involves correlation-based Hebbian mechanisms [25], a prerequisite for a binocular neuron to shift its ocular dominance towards the open eye is that its spike times must correlate with the open eye pathway. For a class 2/3 neuron during contralateral eye closure, this means that the weaker but more coherent inputs of the open eye pathway must increase the correlation of their spike times with the spike times of the postsynaptic neuron. Studies using single unit recordings generally classify cells as 2/3 if the contralateral drive is 1. 5 fold or greater than the ipsilateral drive [50], thus the ISSP2/P1 ratios we used are well within the range of experimental observations, and the effects that we observe on STDP may also apply to cells that are borderline class 4 cells. Our modeling results demonstrate that the maturation of GABAergic inhibition can constrain STDP so that the spike output of class 2/3 neurons, which are dominated by the contralateral eye at the time of its closure, become increasingly correlated with the ipsilateral, open-eye input. Our simulation further shows that inhibition mediates such an effect by reducing the synaptic efficacy of the less coherent, even though stronger, inputs. Therefore, a prediction from our model is that maturation of GABAergic inhibition must be sufficiently strong to more potently decrease synaptic efficacy at the peak versus prior to the onset of the critical period of OD plasticity. We used a visual cortical slice preparation to examine whether the maturation of GABAergic inhibition reduces synaptic efficacy, and whether this effect correlates with the onset of OD plasticity. Postsynaptic responses in layer 2/3 pyramidal neurons were recorded following stimulation of layer 4, which evoked a mixed excitatory-inhibitory response (Figure 2A). The time course of evoked IPSCs outlasted the EPSCs by roughly 7-fold. This was due to the slower kinetics of the GABAA receptors compared to that of the AMPA receptors and to the presence of polysynaptic IPSCs. Previous studies have shown that stimulus-evoked inhibition can reduce the synaptic efficacy of asynchronous EPSPs for up to 30–50 ms [30], [51], a time course that matches the duration of the evoked GABAA current measured here. We used a two-pathway stimulation paradigm to examine the effect of synaptic inhibition on synaptic efficacy at layer 4 to layer 2/3 connections (Figure 7A). The stimulation intensity of both pathways was normalized to spike threshold in layer 2/3 pyramidal neurons to facilitate comparison across slices and animals. A test pathway (Ptest) was stimulated at threshold intensity such that an action potential in a layer 2/3 pyramidal neuron was generated with a probability of approximately 0. 5 (see methods). The ability of Ptest to trigger a postsynatpic action potential was then challenged by stimulating a leading pathway (Plead) 40 ms earlier. The stimulation intensity of the Plead was set such that an action potential in the postsynaptic pyramidal neuron was triggered with >0. 9 probability. Ptest and Plead were verified to be independent pathways to avoid short-term plasticity such as synaptic depression. The use of the 40 ms interval between the two stimuli ensured that postsynaptic spikes evoked by Plead stimulation did not over lap with the postsynaptic responses evoked by Ptest stimulation. Trials in which only Ptest (Test Only) was stimulated were interleaved with those in which Plead and Ptest were sequentially stimulated (Lead-Test). The first experiment was done in cell-attached mode, with an intact intracellular chloride gradient. We found that at the peak of the critical period (P26–30), Plead stimulation reduced the spike probability (ρ) of layer 2/3 pyramidal neurons in response to Ptest stimulation by 37+/−0. 09% (lead-test: ρ = 0. 32+/−0. 08, test alone: ρ = 0. 69+/−0. 08, n = 9, Wilcoxon signed rank, p<0. 02, Figure 7B, C). Prior to the onset of the critical period (P16–18), however, there was little if any effect of Plead stimulation on spike probability of layer 2/3 pyramidal neurons triggered by Ptest stimulation (lead-test: ρ = 0. 66+/−0. 09, test alone: ρ = 0. 62+/−0. 06, mean difference: Δ+0. 04+/−0. 08, n = 9, Wilcoxon signed rank, p = 0. 25, Figure 7E, F). We then repeated the above stimulation protocol using whole-cell recordings of layer 2/3 pyramidal neurons. Similar to cell-attached recording, spike probability in layer 2/3 pyramidal neurons in response to Ptest stimulation was significantly reduced by Plead stimulation at the peak of the critical period (lead-test: ρ = 0. 25+/−0. 07, test alone: ρ = 0. 58+/−0. 07 mean difference: Δ−0. 034+/−0. 08, n = 12, Wilcoxon signed rank, p<0. 003, Figure 7D), while no such effect of Plead stimulation was found prior to the onset of the critical period (lead-test: ρ = 0. 56+/−0. 08, test alone: ρ = 0. 55+/−0. 06, mean difference: Δ+0. 01+/−0. 11, n = 12, Wilcoxon signed rank, p = 0. 58, Figure 7G). To examine whether the reduction in synaptic efficacy of Ptest was mediated by synaptic inhibition, we repeated the whole-cell experiment in a bathing solution containing 3 mM divalent cations and 3 µM bicuculline methiodine (BMI). This resulted in an 80% block of synaptically evoked GABAA current (data not shown) without inducing epileptic activity in cortical slice. In the presence of raised cation concentration but in the absence of BMI, spike probability in layer 2/3 pyramidal neurons in response to Ptest stimulation was significantly reduced by Plead stimulation at the peak of the critical period (lead-test: ρ = 0. 23±0. 10, test only: ρ = 0. 56±0. 07, mean difference: Δ−33+/−0. 10, n = 12, Figure 8A, B), similar to the above results of Figure 7. The effect of Plead stimulation was blocked in the presence of BMI (lead-test: ρ = 0. 75±0. 05, test only: ρ = 0. 56±0. 05, n = 11, mean difference: Δ+0. 18±0. 08, Figure 8E). In contrast, prior to the onset of the critical period, BMI had little impact on the ability of Plead to reduce spike probability in response to Ptest stimulation (no BMI, lead-test: ρ = 0. 45±0. 07, test alone: ρ = 0. 53±0. 02, mean difference: Δ−0. 08±0. 07, n = 12, Figure 8C; BMI, lead-test: ρ = 0. 60±0. 50, test alone: ρ = 0. 49±0. 02, mean difference: Δ+0. 11±0. 06, n = 10, Figure 8F). A 2-way ANOVA was performed to determine whether the age-dependent effect of GABAA blockade was significant. The change in spike probability (lead-test – test only) for layer 2/3 pyramidal cells was compared across treatment groups (Figure 8G). The interaction between age and BMI treatment was significant (p<0. 05), and the effect of BMI on spike probability was significant (p<0. 001). Subsequent pairwise comparisons using the Holm-Sidak method revealed a significant difference between BMI treated and control cells in the mature age group, but not in the young age group. We conclude that there is a developmental increase in the ability of synaptic inhibition to decrease synaptic efficacy during the critical period in mouse visual cortex. We noted that reduced GABAA conductance revealed the presence of a slight summation among inputs. This effect was also seen by Mittman et. al. (2005); given the rapid kinetics of AMPA receptors, this effect was unlikely due to synaptic conductance evoked by stimulation of Plead. The effect could not be explained by a change in input resistance (young: control, 262. 38±3. 89, BMI, 261. 62±4. 96; mature: control, 132. 50±3. 38, BMI, 135. 57±2. 34 MΩ), suggesting that the slight summation observed in the condition of 80% GABAA block may be due to a voltage-dependent persistent sodium conductance induced by stimulation of Plead [52]. Further evidence for the role of chloride conductance in mediating the reduction in spike probability at the peak of the critical period was obtained by using DIDS-fluoride in the recording pipette to block anionic conductances, including GABAA [53]. Because the drug was in the pipette, the blockade was specific to the recorded cell. Spike probability in response to Ptest stimulation was significantly reduced by Plead stimulation (lead-test: ρ = 0. 61±0. 07, test alone: ρ = 0. 32±0. 06, mean difference: Δ+0. 28+/−0. 01, n = 8, Wilcoxon signed rank, p<0. 03, Figure 8G). In this case, there was a 2. 8 fold change in input resistance that likely contributed to summation (break in: 154±37, stable: 348±50 MΩ). The change in spike probability due to stimulation of Plead for all treatments is summarized in Figure 8G. In summary, synaptic inhibition evoked by the Plead was effective in reducing synaptic efficacy of Ptest at the peak of but not prior to the onset of the critical period of OD plasticity. The initial synaptic strength among converging inputs in neural circuits may be determined by genetic mechanisms and/or prior activity-dependent modifications. As an effective plasticity mechanism, STDP must maintain the capacity to modify synaptic strength, including the weakening of strong synapses, according to the on-going patterns of input activity. This is crucial in order for circuits to refine connectivity based on experience. By using a simple integrate-and-fire model neuron driven by two input pathways, we compared the effectiveness of temporal coherence versus initial synaptic strength in shaping the outcome of STDP. In addition, we compared the effectiveness of temporal coherence versus number of inputs within a given pathway. Here we show that the competitive advantage that temporally coherent inputs have over initially stronger inputs does not result from the intrinsic properties of STDP, but rather requires constraints by synaptic inhibition. Synaptic inhibition potently influences input summation required for spike generation by restricting the temporal window over which inputs are able to effectively cooperate [29]–[31]. In vivo recordings in primary sensory cortex demonstrate that inhibition exerts this effect by increasing the requirement for temporal coherence among inputs to evoke spiking beyond what is set by the membrane time-constant of the postsynaptic neuron [31], [54]. The precise manner by which synaptic inhibition is recruited during sensory experience is likely influenced by many factors [34], [55], [56]. In our model therefore, we implemented synaptic inhibition using the least number of assumptions: every EPSP was followed by an IPSP with a delay ranging between 4 to 10 ms; and for a given simulation, the amplitude of inhibition was fixed. We then systematically varied the amplitude of inhibition across simulation trials. We found that, independent of STDP, an increase of inhibition reduced the synaptic efficacy of both P1 and P2 pathways, but the reduction was more profound for the less coherent, even though stronger, P2 pathway (Figure 5; compare A&B). The small difference in synaptic efficacy between the two pathways brought about by synaptic inhibition had a major impact on STDP-mediated re-organization of synaptic weights (Figure 5; compare C, E, &D, F). Our results thus suggest that, at low levels of inhibition, strong but less coherent inputs effectively competed with weaker but more coherent inputs; the stabilizing property of STDP favors the maintenance of the existing synaptic weight distribution over updating the distribution according to novel temporal patterns of input. At higher levels of inhibition, on the other hand, the efficacy of the less coherent inputs is preferentially reduced, biasing STDP to increasingly favor the more coherent inputs. Therefore, a sufficient level of inhibition is crucial to regulate STDP such that inputs are modified according to their correlation structure, a parameter that is often controlled by peripheral sensory events [27]. Monocular deprivation during a critical period induces a shift in the ocular dominance (OD) of binocular neurons in V1 such that the open eye pathway dominates in driving spiking activity. The OD shift involves both a reduced drive from closed eye inputs [57], [58] and an increased drive from open eye inputs [59]. The recent finding that monocular blurring rapidly shifts OD indicates that it is the quality rather than the quantity of retinal illumination that is the key factor for OD plasticity [26]. These results suggest that it is the temporal pattern rather than the overall rate of activity that drives receptive field plasticity [27]. Accordingly, in our model we altered temporal correlation of inputs but held the overall rate of the two pathways fixed at 20 Hz. Multiple forms of cellular plasticity, operating on different time-scales, likely mediate OD plasticity in vivo at excitatory synapses. MD has been shown to induce classic homosynaptic LTD [58], [60], modeling studies suggests that STDP may also account for some forms of OD plasticity [23]. Homeostatic mechanisms also likely contribute on a slower time-scale [61], [62]. Similarly, it appears that STDP alone cannot account for the MD-induced ocular dominance plasticity that is observed in fast-spiking interneurons [63]. In addition to STDP, simulations of interneuron plasticity must add additional rules of synaptic elimination to recapitulate experimental results. Furthermore, it was recently demonstrated that in acute slices, the polarity and magnitude of associative plasticity can be regulated by neuromodulators applied generally to the bathing perfusion, raising the possibility that under some neuromodulatory states in vivo, timing rules of STDP are not rigorously bi-directional [64]. Here we find that both layer 4→ layer 2/3 inputs and local recurrent inputs are modifiable by STDP in V1 prior to and during the critical period in mice. Our results are consistent with and extend previous work that examined local recurrent connections in rat V1 [16]. It has been demonstrated that cortical synapses can be modified by STDP rules in vivo in response to visual stimulation [15], [17], [18], indicating that the temporal precision required for cortical STDP is propagated to upper cortical layers during sensory experience. The current available evidence supports the view that in response to MD, the synaptic weight distribution of synapses converging onto binocular neurons is re-organized and stabilized in part by STDP. Regardless of the precise contribution of these cellular mechanisms to the OD shift itself, a prerequisite for plasticity to proceed according to Hebb' s rule is that the temporally coherent open eye inputs must correlate their spike times well with those of their postsynaptic binocular neuron. Our results suggest that GABAergic inhibition is required for this prerequisite to be met. Maturation of GABAergic inhibitory circuits has been implicated in the regulation of critical period plasticity in visual cortex [37], [38]. Particularly compelling is the finding that direct enhancement of GABAergic transmission induces precocious OD plasticity [36], [39], [40]. This result indicates that the machinery for OD plasticity is operational even before its natural onset but lies dormant, and can be triggered by the maturation of GABAergic transmission. However, the cellular and synaptic mechanisms by which GABAergic inhibition regulates OD plasticity remain elusive. It is also unclear how GABAergic inhibition is related to correlation-based plasticity mechanisms. Our modeling study now demonstrates that a sufficient level of synaptic inhibition is crucial to constrain STDP so that synaptic strengths are modified according to their correlation structures rather than their initial synaptic drive. We further provide experimental evidence that maturation of inhibition is sufficient to potently reduce synaptic efficacy during the critical period. These results are consistent with the notion that inhibition preferentially reduces the synaptic efficacy of the less coherent inputs among convergent pathways. Due to the dominance of crossed contralateral retinal projections, most binocular neurons in rodent V1 are characterized as class 2/3 neurons, driven strongly by contralateral input, and weakly by ispilateral eye input. Closure of the contralateral eye likely reduces the temporal coherence of inputs in this pathway compared to that of the ipsilateral, open eye pathway. Prior to the onset of OD plasticity, GABAergic inhibition is immature (Figure 9, upper row) and is ineffective in reducing the synaptic efficacy of inputs (regardless of their temporal coherence), and the temporal window of input summation is wide. The stronger contralateral inputs, though less coherent, continue to drive the postsynaptic neuron and correlate with postsynaptic spiking. Thus, at low level of inhibition, temporal coherence is not sufficient to control postsynaptic spike timing. As a consequence, correlation-based plasticity mechanisms cannot be engaged and the shift in OD fails to occur. STDP in fact disrupts Hebbian processes from selecting temporally coherent inputs. At the peak of critical period, inhibition is mature (Figure 9, lower row), which restricts the temporal window of input summation and thus preferentially reduces the efficacy of the less coherent contralateral inputs. As a consequence, the spike times of the ipsilateral pathway are better correlated with postsynaptic spike times compared to those of the contralateral pathway (Figure 9, lower row). In the presence of the STDP rule, the ipsilateral pathway consistently dominates the control of spike timing of postsynaptic neurons. Therefore, correlation-based mechanisms are able to strengthen ipsilateral inputs, weaken the contralateral inputs, and OD plasticity proceeds. By modeling specific features of temporal input structure that are disrupted during MD, we were able to characterize the effect of increasing inhibition on STDP. Other Hebbian and non-Hebbian mechanisms operating at various time scales are likely involved in mediating the loss of responsiveness of the contralateral pathway (Figure 9, far right) [55], [58], [65]. In summary, the maturation state of GABAergic inhibition likely has a major impact on whether the synaptic weight distribution can be updated to reflect to novel patterns of sensory input, including monocular deprivation at the onset of the critical period for OD plasticity. A complete understanding of the impact that maturation of inhibition has on OD plasticity will require closer examination of interneuron binocular plasticity and initial contralateral bias during development. A recent study examining layer 2/3 interneuron calcium activity reported an initial contralateral bias in interneurons similar to that of pyramidal cells, and a delayed shift in ocular dominance relative to pyramidal cells in response to monocular deprivation. Modeling results indicated that a delayed shift of interneurons potently increased the rate of pyramidal ocular dominance plasticity [66]. If it is the case that the inhibition/excitation ratio for a given postsynaptic pyramidal cell is higher in response to contralateral stimulation compared to ipsilateral stimulation during the pre-critical period, the need for maturation of inhibition to regulate STDP as we describe may not be as strong. However, two other recent studies observe a different pattern of interneuron recruitment in which there is little or no initial contralateral bias of interneurons [63], [67]. Under conditions in which the contralateral pathway does not preferentially drive inhibition the requirement for inhibition to regulate STDP is maintained. It has been proposed that maturation of inhibition may promote STDP to efficiently induce LTD of deprived inputs during OD plasticity [10], [68]. In this scenario, loss of responsiveness of deprived inputs is directly mediated via STDP. Furthermore, there is precedent for STDP to actively participate in the selective weakening of deprived inputs in barrel cortex [19]. Interestingly, our modeling results demonstrate that the mechanism by which increased inhibition promotes STDP-mediated LTD does not necessarily involve more LTD of the less coherent inputs due to increasing the probability that these inputs fall in the LTD portion of the STDP rule (Figure 4G, H), which is extended compared to the LTP portion [10]. Instead, we describe a set of conditions in which increased inhibition increases the relative number of coherent inputs that fall into the LTP window (Figure 4A–F). Thus, it appears that inhibition can exert a potent effect on the outcome of STDP-mediated competition by regulating synaptic efficacy. Synaptic inhibition is mediated by diverse types of GABAergic interneurons [69]. Our point-conductance model implies that perisomatic inhibition is a candidate source of synaptic inhibition. Consistent with this notion, parvalbumin-positive (Pv+) perisomatic GABAergic synapses structurally mature during the critical period of OD plasticity [34], and there is evidence that α1-containing GABAA receptors, which are enriched at the perisomatic region, contribute to OD plasticity [40]. Here we demonstrated that on average, there was a 2-fold increase in IPSC amplitude at unitary Pv+ interneuron-to-pyramidal connections during the critical period compared to that prior to the critical period. We also found that the stimulus input/spike output curve of Pv+ interneurons matured during the critical period, raising the possibility that an increase in the gain of Pv+ interneurons may contribute to the onset of the critical period. Indeed, increasing gain or IPSC amplitude had a similar effect on STDP-mediated redistribution of synaptic weights in our simulation. Furthermore, the increase in gain was sufficient to compensate for the developmental decrease in IPSC decay time-constant. It is important to note that in addition to spike generation, inhibition may constrain STDP by regulating the propagation of dendritic action potentials [70]. This effect is not simulated in our point conductance model. Although dendtritic-targeting interneurons also mature during postnatal development [71], whether their maturation correlates with the critical period remains to be investigated. By modeling primary features of MD-induced alterations in temporal input structure we demonstrated that regardless of the extent to which STDP mediates the shift in ocular dominance, the potent stabilizing property of STDP can in fact disrupt ocular dominance plasticity from proceeding unless constrained by inhibition. As predicted by the model, we found that maturation of inhibition decreases synaptic efficacy at the peak of the critical period. Our results highlight the need for circuits to regulate powerful stabilizing mechanisms such as STDP in order for experience-dependent plasticity to proceed. All procedures were approved by CSHL IACUC. Acute cortical slices of visual cortex were prepared from C57B6 mice, age postnatal day (P) 16-18 (young), or P26–30 (mature), unless otherwise noted in the text. Brain slices (300 microns thick) were cut in the coronal plane with a vibroslicer (Vibratome, St. Louis, MO) in ice-cold dissection ACSF (in mM): 212. 7 sucrose, 2. 5 KCl, 1. 25 NaH2PO4,3 MgSO4,1 CaCl2,10 D (-) -glucose, and 26 NaCHO3, continuously bubbled with 95%O2/5%CO2 and allowed to recover for >30 minutes in normal ACSF (in mM): 126 NaCl, 2. 5 KCl, 1. 25 NaH2PO4,1 MgSO4,2 CaCl2,10 D (-) -glucose, and 25 NaCHO3 continuously bubbled with 95%O2/5%CO2 and then transferred to the recording chamber. Slices were viewed with infrared differential interference contrast optics on an upright microscope (Axioskop, Zeiss, Thornwood, NY). Slices were submerged in normal ACSF containing 50 µM APV (Tocris, Ellisville, MO) and 1 µM CGP55845 (Tocris, Ellisville, MO), except as noted, and perfused at a rate of 2–3 ml/min (33+/−1°C). Recording were made using a Multiclamp 700A amplifier (Molecular Devices, Sunnyvale, CA). For current-clamp recordings using the STDP protocol, the intracellular solution contained (mM): 110 K-gluconate, 20 KCl, 10 HEPES; 4 MgATP, 10 phosphocreatine (Na), and 0. 3 NaGTP, pH 7. 3,300 mOsm. To avoid confounding effects of inhibition when assaying plasticity at glutamatergic synapses, 10 µM picrotoxin was included in the bath perfusion to block GABAA receptors. Picrotoxin is a preferred blocker over bicuculine methiodine (BMI) in synaptic plasticity assays because BMI used at concentrations sufficient to completely block GABAA receptors has been shown to block SK potassium channels [72], which could potentially alter local dendritic excitability and thereby impact the induction of plasticity. For voltage-clamp recordings used in the EPSC-IPSC maximal charge assay, the intracellular solution contained (mM): 130 Cs-gluconate, 8 KCl, 10 HEPES, 10 EGTA, 10 QX-314 (Alomone, Jerusalem, Israel). For current-clamp recordings in the synaptic efficacy assay, the intracellular solution contained (mM): 135 K-gluconate, 4. 3 KCl, 2 NaCl, 10 HEPES, 0. 5 EGTA, 4 MgATP, 20 phosphocreatine (Na), and 0. 3 NaGTP, pH 7. 3,300 mOsm. Methods currently available to block GABAA receptors in cortical slices during protocols that require synaptic stimulation intensities strong enough to bring postsynaptic cells to spike threshold are limited because full blockade of GABAA receptors, such as achieved with 10 µM picrotoxin in the bathing medium, will cause epileptic-like activity in response to strong synaptic stimulation. Therefore, we employed two different methods to reduce GABAA receptor conductance, both have non-overlapping drawbacks. Low-concentration BMI (3 µM) in combination with raised cation concentration was previously shown to significantly reduce GABAA receptor conductance without inducing epilepsy in cortical slices [16], this method was employed in Figure 8. Anion channels and pumps can be blocked intracellularly with a fluoride-based internal solution in combination with 4,4' -diisothiocyanatostilbene-2,2' -disulfonic acid (DIDS) [73], [74]. The DIDS internal solution contained (in mM): 120 KF, 8 KCL, 10 HEPES, 10 EGTA, 1 DIDS. Whole-cell recordings pipettes had a tip resistance of 3–4 MΩ. Data were digitized at 10 kHz, filtered at 2 kHz, and analyzed with Clampfit 9 (Molecular Devices, Sunnyvale, CA). EPSP/Cs were evoked by focal extracellular stimulation (0. 2 ms, 10–100 µA) with commercial bipolar electrodes (FHC, Bowdoin, ME), except in STDP protocols, a small glass bipolar electrode was used [16]. In STDP protocols, the initial EPSP slope (mV/ms) was set to be the same across ages and cells. Vertical LTP: young, 0. 44+/−0. 11; mature, 0. 46+/−0. 06 (t-test, p = 0. 88). Vertical LTD: young, 0. 45+/−0. 05; mature, 0. 34+/−0. 06 (t-test, p = 0. 16). Horizontal LTP: young, 0. 38+/−0. 08; mature, 0. 42+/−0. 11 (t-test, p = 0. 79). Horizontal LTD: young, 0. 48+/−0. 08; mature, 0. 35+/−0. 04 (t-test, p = 0. 13). Input resistance was monitored with hyperpolarizing current pulses (25 pA, 100 ms); cells were excluded if input resistance changed >30% over the entire experiment. The change in initial EPSP slope (first 2 ms) was calculated as the EPSP slope ratio by dividing the average initial slope after pairing (10–20 minutes following the EPSP and AP pairing) by the average baseline initial slope. In the synaptic efficacy paradigm, the two pathways were considered independent if EPSP slopes summed linearly [75]–[77]. Plead stimulation intensity did not exceed 1. 5x threshold stimulation intensity. Chemicals were purchased from Sigma-Aldrich (St. Louise, MO) except as noted. For statistical analysis, two-tailed parametric tests were used unless the data were not normally distributed. In such cases, Wilcoxon signed rank was used for paired samples, and Mann-Whitney for unpaired samples. Error is reported as ± standard error of the mean, unless noted. We simulated a conductance-based integrate-and-fire model in Matlab (Mathworks, Natick, MA), using the difference equation, where grest = 12. 5 nS and capacitance, C, was set to C = 0. 25 nF to give a membrane time-constant of 20 ms. Ee = 0 mV, and Ei = −70 mV. When the membrane potential Vt reached a threshold value of −54 mV, V was reset to –65 mV with a refractory period of 2 ms (20 iterations or time-steps). Excitatory synaptic inputs were modeled as conductances given by the function, g (Δt) = αe-Δt/τ, where τ = 2 ms, and α was equal to the synaptic weight value. N = 80 excitatory synapses for all simulations. As in Song et. al. (2000), the synaptic weight value α was updated every iteration based on the STDP function F (Δt): Where A+ = 0. 5%* gmaxex, A- = 0. 45%* gmaxex, τA+ = 20 ms, and τA- = 35 ms. gmaxex was set to 15* 150 pS to give an output firing rate of 5–20 Hz for N = 80 synapses. In the weight-dependent mode shown in Figure 4, potentiation was updated as A+ * (1-α). Inhibitory conductances were also given by the function, g (Δt) = αe−Δt/τ, the decay time-constant for inhibitory synapses was initially set to τ = 5. 75 ms and varied as noted in text, the value of a was fixed for the duration of a particular simulation, the amplitude varied between simulations as noted in the text. Excitatory presynaptic spike trains were generated in the following manner: In the case of Input Regime (I), the spike trains activating the synapses of the two sets of inputs were generated from a single Poisson process. In the case of Input Regime (II) and (III), the spike trains activating the synapses of the two sets of inputs were generated from two independent Poisson processes. The correlation coefficient between any two cells within the same pathway for the total number of spikes fired was fixed at r = 0. 5 in all simulations. An additional parameter was used to define the temporal coherence (1/σ) among spike times within a given pathway. Sigma (σ) defined the temporal jitter among spike times within a given pathway, and controlled the width of the cross-correlogram peak between any two trains within the same pathway, we defined temporal coherence as the inverse of σ. Inhibitory presynaptic spike trains were implemented in a feedforward manner, every excitatory presynaptic spike was followed by an inhibitory presynaptic spike with a delay randomly ranging between 4–10 ms. The initial amplitude of inhibition was set via matching the total conductance of one inhibitory synaptic event to the total conductance of one initial excitatory event, this amplitude corresponded to 8. 8% of gmaxex, and is indicated as 0. 09 gI/gMaxex in Figures 4 and 6. Simulations were run for 40–80 minutes of simulated time, except for the simulations shown in Figure 3 which were run for 12 minutes of simulated time. The Matlab M files used to generate the simulation are included in the supplement (Protocol S1).

Evidence suggests that maturation of inhibition is required for the development of plasticity to proceed in the visual cortex. However, the mechanisms by which increased inhibition promotes plasticity are not clear. Here we characterized the maturation of synaptic and intrinsic ionic properties of parvalbumin-positive interneurons, a prominent subtype of inhibitory neuron in the cortex. We used a simple integrate-and-fire model to simulate the influence of maturation of inhibition on associative plasticity rules. We simulated two input pathways that converged onto a single postsynaptic neuron. The temporal pattern of activity was constructed differently for the two pathways: one pathway represented visually-driven activity, while the other pathway represented sensory-deprived activity. In mature circuits it is established that postsynaptic cells can select for sensory-driven inputs over deprived inputs, even in the case that deprived inputs have an initial advantage in synaptic size or number. We demonstrated that maturation of inhibition was required for postsynaptic cells to appropriately select sensory-driven patterns of activity when challenged with an opponent pathway of greater size. These results outline a mechanism by which maturation of inhibition can promote plasticity in the young, a period of development that is characterized by heightened learning.

Abstract Introduction Results Discussion Methods

neuroscience neuroscience/theoretical neuroscience neuroscience/neurodevelopment neuroscience/sensory systems

2010

Maturation of GABAergic Inhibition Promotes Strengthening of Temporally Coherent Inputs among Convergent Pathways

Tissue-encysting coccidia, including Toxoplasma gondii and Sarcocystis neurona, are heterogamous parasites with sexual and asexual life stages in definitive and intermediate hosts, respectively. During its sexual life stage, T. gondii reproduces either by genetic out-crossing or via clonal amplification of a single strain through self-mating. Out-crossing has been experimentally verified as a potent mechanism capable of producing offspring possessing a range of adaptive and virulence potentials. In contrast, selfing and other life history traits, such as asexual expansion of tissue-cysts by oral transmission among intermediate hosts, have been proposed to explain the genetic basis for the clonal population structure of T. gondii. In this study, we investigated the contributing roles self-mating and sexual recombination play in nature to maintain clonal population structures and produce or expand parasite clones capable of causing disease epidemics for two tissue encysting parasites. We applied high-resolution genotyping against strains isolated from a T. gondii waterborne outbreak that caused symptomatic disease in 155 immune-competent people in Brazil and a S. neurona outbreak that resulted in a mass mortality event in Southern sea otters. In both cases, a single, genetically distinct clone was found infecting outbreak-exposed individuals. Furthermore, the T. gondii outbreak clone was one of several apparently recombinant progeny recovered from the local environment. Since oocysts or sporocysts were the infectious form implicated in each outbreak, the expansion of the epidemic clone can be explained by self-mating. The results also show that out-crossing preceded selfing to produce the virulent T. gondii clone. For the tissue encysting coccidia, self-mating exists as a key adaptation potentiating the epidemic expansion and transmission of newly emerged parasite clones that can profoundly shape parasite population genetic structures or cause devastating disease outbreaks. Population genetic studies of pathogenic microbes have been paramount to our understanding of disease resulting from emerging and re-emerging infectious organisms [1]. Studies performed to determine the relative contributions of drift and recombination in the production of genetic diversity have identified that most pathogens have methods to alter, exchange and acquire genetic material that are intimately associated with pathogenicity [1], [2]. For viral pathogens, enhanced levels of drift, genomic reassortment [3], and incorporation of host genes [4] have all been linked to emergence of virulence. Likewise, horizontal gene transfer between bacterial species has facilitated assimilation of pathogenicity islands, plasmids, prophages, and other insertional elements essential for disease and drug resistance phenotypes [5]–[9]. For eukaryotic pathogens, meiotic sex serves an analogous purpose functioning to alter the genetic make-up, and therefore the biologic and virulence potential of strains [10]–[17]. A general paradigm describing disease epidemics for many pathogens is that genetic diversification, complemented by the acquisition of traits that enhance relative fitness and facilitate clonal expansion, leads to the emergence of novel, virulent genotypes. Just as the life history traits for generating genetic diversity vary widely among pathogen types, it is often the case that the mechanistic basis for subsequent clonal expansion of pathogenic strains is unique on a taxonomic level. Determining the mechanisms and contribution of these life history traits to disease is important for focusing prevention and treatment strategies to the most relevant pathogen strains and life cycle stages. For the cyst-forming coccidia, which comprise a diverse group of parasites belonging to the phylum Apicomplexa, complex lifecycles that include both sexual and asexual stages have led to unusual population genetic structures for several species. For the widespread zoonotic pathogen, Toxoplasma gondii, the majority of strains infecting birds and mammals throughout North America and Europe are comprised of just three clonal lineages which exist as successful clones from a genetic out-cross [14], [18]. These three lineages have apparently emerged only recently due to an enhanced fitness that facilitated their ability to effectively outcompete other genotypes [13], [19]–[21]. Likewise, the veterinary pathogen Sarcocystis neurona possesses a surprisingly simple population genetic structure punctuated by the dominance of a few clonal lines in North America [22]–[25]. Similar clonal structures have been reported for other parasitic protozoa that possess sexual cycles [26] but identifying the precise genetic mechanisms that have led to the emergence of distinct clones among the different species in nature remains enigmatic. In combination with population genetic data, the contributions of sexual out-crossing and clonal expansion as factors governing the emergence and eventual dominance of distinct, disease-producing clones have largely been inferred from laboratory studies of T. gondii among the cyst forming coccidia. Prior experiments demonstrated that a sexual cross between mouse-avirulent strains can produce genotypes representing a range of virulence in the mouse model, including some progeny several logs more virulent than the parents [14]. This study identified that natural out-crosses likely produce at least some virulent genotypes, which may subsequently have potential to emerge through clonal amplification to cause extensive disease [20], [21]. Clonal propagation is possible since T. gondii can effectively bypass the sexual stage in felid definitive hosts and cycle, presumably indefinitely, among intermediate hosts. This can occur horizontally via oral transmission through carnivory among intermediate hosts [19], [20] or vertically by transplacental transmission [27]–[30]. Toxoplasma gondii can also functionally bypass genetic diversification during the sexual stage by self-mating in the definitive host. Self-mating (also termed selfing, uni-parental mating, or self-fertilization) occurs when a single parasite clone can give rise to both male and female gametes capable of undergoing fertilization and producing viable offspring [31], [32]. In other words, no predetermined mating types are apparent and the end result is effectively clonal expansion via sex and meiosis. Despite these important laboratory studies, the implications of these life-history traits and their relative effects on population genetic structures, especially in the context of virulence and disease outbreaks, have not been extensively studied in T. gondii or other cyst forming coccidia in a natural setting. Parasite life stages that are most important for causing mass-morbidity and mortality may be revealed through review of past, large-scale T. gondii-associated human outbreaks. For eleven reports of T. gondii-associated disease outbreaks in immune-competent people, eight events, including the four most devastating that caused disease or death in hundreds of individuals, were attributed to the oocyst form of the parasite, which is only produced during the sexual life cycle stage in the definitive feline host [20]. Furthermore, an outbreak of the related veterinary pathogen Sarcocystis neurona that resulted in the death of nearly 1. 5% of the threatened Southern sea otter population over the course of a single month is thought to have resulted from exposure to infectious sporocysts originating in the definitive opossum host [33]. Circumstantial evidence, such as a complete lack [34] or much reduced [35]–[37] prevalence of T. gondii in certain island environments without cats, also gives weight to the importance of the definitive host stage in the parasite life cycle. Similarly, S. neurona has not been identified outside of its definitive host range in the Americas. The apparently profound importance of this stage in the lifecycle of not just T. gondii, but other related parasites, warrants further study to determine the influence it could impart to shaping parasite population genetic structures and which genetic mechanisms inherent to this life stage (i. e. selfing or out-crossing) are more likely to precede a disease outbreak in nature. To determine the genetic basis governing the exposure, evolution, and emergence of virulent genotypes during natural outbreaks linked to sexual stages of these parasitic protozoa, we tested whether epidemic isolates exist as: 1. a diverse array of multiple, novel genotypes that are the products of an out-crossing event in the definitive host, or 2. epidemic clones of a single genotype derived via selfing in the definitive host. To distinguish between these two possibilities, high resolution genetic typing was used to characterize parasite strains associated with a T. gondii outbreak in humans [38] and a S. neurona outbreak in sea otters [33], both of which were associated with unusually high levels of morbidity and mortality. The population level genetic studies presented here argue that selfing in the definitive host plays a central role in the epidemic expansion of newly emerged, recombinant parasite strains, thus potentiating clonal outbreaks caused by tissue cyst-forming coccidia. A microsatellite-based typing scheme using the markers B17, B18, TgMa, TUB2, W35 [39], and M95 [40] was applied to determine the molecular genotypes of T. gondii isolates associated with a human water-borne outbreak in Brazil. This outbreak, which occurred over a short time span in 2001, was linked to oocyst-contamination of a municipal water supply in the town of Santa Isabel do Ivai and resulted in infection and symptomatic disease in hundreds of people [38]. Initial genetic typing analyses performed on two T. gondii strains isolated from the water cistern implicated as the source of the outbreak [38], as well as isolates from chickens [41] and cats [42] from the immediate environment were limited to PCR-RFLP at a single locus, SAG2, leading to the conclusion that the outbreak strain was a canonical Type I strain (see below). Later, more extensive analysis by PCR-RFLP [43] and DNA sequencing on a limited set of markers [44] showed that the outbreak-associated strains from the water cistern were clonal and non-archetypal. The majority of people who seroconverted during the outbreak also possessed a serologic profile consistent with infection by the outbreak clone, and the outbreak genotype appeared to be highly prevalent in the surrounding environment immediately following the outbreak event, infecting 4/11 chickens (TgBrCk98, TgBrCk101–103) and 1 cat (TgCatBr85) [44] (Figure 1). To determine the extent of genetic relatedness among the outbreak-associated strains, high resolution MS typing and DNA sequencing using markers distributed on 11 of the 14 chromosomes was applied. This dataset distinguished the two water cistern, outbreak-associated strains at the genetic level from all others present in the environment, except for one chicken isolate (TgCkBr103) (Figure 1). Unfortunately, insufficient DNA remained from the cat isolate, TgCatBr85, which precluded testing whether it was genetically identical to the cistern isolates. Utilizing the MS typing scheme confirmed the conclusion that the causal agent was a unique, emergent T. gondii strain with a potential for enhanced virulence. The additional typing provided in the current study refined the conclusions of previous studies in two key aspects. First, the much higher level of resolution provided by the markers used and the sequence level analysis imparts a higher level of confidence to the conclusion that the outbreak was in fact clonal. The possibility that the outbreak-associated clones are not genetically identical in lieu of additional typing cannot be excluded, but several facts strongly argue against this: 1. The 18 markers were distributed across all but three of the 14 chromosomes; 2. MS markers are prone to rapid evolution and therefore provide high resolution; 3. Strains from Brazil are genetically divergent from archetypal lines, as evidenced by the segregation of alleles amongst strains in Figure 1, and hence, less prone to linkage disequilibrium effects. Furthermore, only a single, oocyst-derived clonotype was isolated from independent filters collected from two different water-holding tanks providing additional evidence that these isolates resulted from self-mating rather than a genetic out-cross. Second, this study refines previous work on the Santa Isabel outbreak by showing that the outbreak strain was actually rare in the surrounding environment, opposed to the high prevalence reported previously [44]. Moreover, close examination of the environmental isolates reveals that many of them, including those previously identified as the outbreak clone, and the outbreak clone itself, resemble recombinant progeny; only two allelic types are present that segregate independently across the loci examined (see TgCkBr98,99,100,101,102,103, TgCatBr85 and Outbreak 1 and 2 in Figure 1). These data argue that prior to the outbreak, the epidemic clone was produced by a genetic out-cross and was subsequently expanded by self-mating. This confirms that the more extensive resolution provided by the current study was necessary to truly distinguish an epidemic clone in a region known to contain a diverse array of T. gondii genotypes, including many that are apparently siblings of this strain [44]. This result also speaks to the important role selfing in the definitive host can play; allowing a single, emergent genotype of low environmental prevalence to rapidly rise to dominance in the surrounding population by infecting several hundreds of hosts over a short time span. Collectively these data support high-resolution genotyping schemes as important tools for detecting informative genetic signatures in this parasite species. Initial population genetic studies showed that T. gondii strain diversity was comprised of three main clonal groups: Type I, II, and III [18]. As a result of these early studies, many broader population genetic studies have since relied on typing at only one or just a few loci to classify strains as type I, II, or III. However, it is now apparent that strains from diverse geographic locales and host species are more often infected with strains bearing unique alleles or allelic combinations, so relying on a few markers is insufficient for robust conclusions [20]. The first quantitative analysis testing the accuracy of single locus typing found a very low predictive value for the loci analyzed to correctly identify strain genotype [45]. Indeed, results presented in the current study, when compared with results from more limited genetic studies of the same strains conducted previously [38], [41]–[44], provide a clear illustration of the value more extensive genetic typing can have in refining conclusions. This is especially relevant in outbreak investigations where variations in parasite genotype can be highly informative for explaining disease manifestation. High-resolution genetic typing appears to be critical for eliminating preconceived biases in epidemiologic investigations to ensure accurate discernment of disease-associated T. gondii strains and to recognize clonal outbreaks. These results validate the utility of testing for epidemic clones from prospective and retrospective studies of T. gondii disease outbreaks [20]. In support of this, Dumar and colleagues applied a similar typing scheme to a T. gondii outbreak in Suriname and discovered that all five patients from whom they isolated parasites were infected with the same, previously undiscovered genotype [46]. Importantly, the outbreak in Suriname was another waterborne outbreak attributable to human exposure by infectious oocysts, further evidencing selfing in the definitive host as a key mechanism for allowing clonal expansion of virulent genotypes, ultimately resulting in disease epidemics. Since parasite genetic material from past T. gondii outbreaks in humans is in limited supply for the majority of cases, we sought to further assess the role of self-mating in disease outbreaks by examining an epizootic of the related veterinary pathogen, Sarcocystis neurona, infecting the Southern sea otter (Enhydra lutris nereis) of California. As a threatened species, the Southern sea otter population is well monitored and accounted for by conservation groups, creating a unique opportunity to investigate infectious disease in a natural setting. Sea otters are also aberrant hosts for many terrestrial pathogens that can be washed to sea and their high susceptibility to many of these pathogens allows them to serve as a sentinel species for pathogens circulating in the adjacent terrestrial environment [44]. During April, 2004, the highest monthly mortality rate ever recorded in nearly 30 years of data collection occurred among Southern sea otters [33]. Over the course of approximately one month, at least 40 sea otters stranded dead or dying along an 18 kilometer stretch of coast within the 500–600 kilometer Southern sea otter range. Sixteen otters were in sufficient condition to allow for complete post-mortem analysis inclusive of PCR assessment and microscopic examination of tissues. Among these otters, the major cause of death for 15 of the 16 examined animals was S. neurona-associated brain and/or systemic disease [33]. Preliminary genetic analysis using only four polymorphic markers against parasite strains infecting a subset of these otters (n = 7) suggested they were genetically homogenous [25]. However, the limited polymorphism present in the markers used, and lack of information about the population genetic structure of S. neurona in California prevented a confident conclusion that they represented an epidemic clone. The present study developed and applied a battery of higher resolution, polymorphic microsatellite and gene-coding markers to type S. neurona strains. Additional samples were included, encompassing 12 S. neurona strains from otters that died during the outbreak, as well as additional strains from other geographic locations and/or time periods. The high number of sea otter deaths associated with this epizootic provided a unique opportunity to test whether self-mating, as identified in the human T. gondii outbreaks, could explain the genetic origin for the S. neurona strains that caused the outbreak. In addition, genetic data from the current study was combined with S. neurona typing data reported by Rejmanek et al. [23] to determine the population genetic structure of S. neurona in California spanning 15 years of study. Sequence-level analysis of five surface antigen (Ag) genes (SnSAG1,3, 4,5, and 6) [25] and nine microsatellite (MS) markers (Sn2–Sn5, Sn7–Sn11) [23], [25] identified 12 Ag types and 33 MS types among 87 S. neurona-infected samples based on the allele combinations detected at each locus (Table 1; See Table S1 for complete strain and typing information). Seventy-four of the 87 samples were from mammals in California; other states represented include Georgia (n = 2), Illinois (n = 1), Missouri (n = 3), Washington (n = 5), and Wisconsin (n = 2). Combining Ag and MS alleles could distinguish 35 total genotypes, but for this study these typing schemes were analyzed independently because of the likelihood that these parts of the genome are under different selection pressures and subject to differing evolutionary processes [2]. The majority (56/87) of S. neurona strains were classified as either Ag type I or Ag type II (Figure 2A). Certain MS types were also over represented in the sample set, with MS types ‘a’, ‘c’, and ‘g’ accounting for 47/87 samples (Figure 2B). Importantly, 11/12 S. neurona strains from sea otters stranding during the mortality event in 2004 were an exact genetic clone at each marker analyzed (Ag type I, MS type ‘c’). The remaining outbreak sample (Ag type I, MS type ‘d’) differed from the other outbreak strains by only a single stepwise mutation at MS marker Sn4 (Table S1). Since this and all previous studies of S. neurona have found a high level of sequence homology among strains [22]–[25], [47], we chose to analyze strain relatedness with the eBURST algorithm [48], [49]. This program helps eliminate confounding effects that low sequence diversity and moderate levels of recombination can have on other methods of intra-specific sequence analysis, such as clustering, dendrograms, and phylogenetic trees, as demonstrated in [22]–[24], by only focusing on single clones and their most recent descendents [48]–[50]. We adapted the MS data for the nine markers that permit simultaneous comparison of all strains (Sn2–Sn5, Sn7–Sn11) to serve as a multi-locus typing scheme. This typing scheme, which is based on the number of repeats at each locus, was amenable to use with this program. Using the default settings, which group isolates based on the premise that they are single locus variants (SLVs), or share 8 out of 9 alleles, we identified 8 clonal complexes (CC1–8), only 3 of which contained more than two genotypes, and 8 singletons (genotypes differing by 2 or more alleles from all others) (Table 1; Figure 3). Intriguingly, just two clonal complexes, CC1 and CC2, accounted for almost 64% (56/87) of the strains analyzed in this study (Figure 2C). This result held true even when correcting for bias introduced by the outbreak event by removing these samples from the data set, as 44/75 samples (59%) still belonged to CC1 or CC2. All SLVs identified in this study differed by a single stepwise (i. e. a single di-nucleotide repeat) mutation, which supports the assumption that the eBURST groupings represent clonal complexes in which allelic variation is a result of mutation/drift and not recombination (Table S1) [50]. The only exceptions to this were SLVs ‘l’ and ‘o’, members of CC3, that differed by 3 di-nucleotide repeats at MS Sn11. These isolates were from a sea otter in California and a horse from Missouri so the greater number of stepwise mutations detected may be a result of extended geographic isolation, thus allowing time for more drift to occur (Table S1). A single mutation event that resulted in multiple stepwise mutations is also plausible. Since recombination appeared to be rare between clonal complexes based on MS markers, we decided to overlay the results of the Ag typing analysis on the eBURST output (Figure 3). The results were consistent with previous claims of an intermediate population structure for S. neurona [22]–[25], [47] in that both clonal propagation and sexual recombination were supported. All members of CC1 and 29/30 members of CC2 possessed an identical Ag type (Ag types I and II, respectively). In contrast, all MS types in CC3 and CC8 possessed a distinct Ag type. There were also two cases (MS types ‘x’ and ‘bb’) where the same MS type was identified with two distinct Ag types (Ag types VII and VIII) and the reverse scenario also occurred where the same Ag type (VI) characterized three clonal complexes based on MS types (CC4, CC5, CC6), all of which could potentially indicate recombination events (Figure 3). Overall, these data support a population structure that is highly clonal, though evidence for recombination is present as well. This intermediate population structure is similar to that described for T. gondii, though definitive conclusions will require a sample set less biased towards diseased animals [2]. It is worth noting here that the population structure of the organisms described in this study is, like all population genetic structures, only as resolved as the markers allow. For example, finer resolution can be achieved by applying the marker SnD2 from Rejmanek et al. [23] to SO4711, SO4786 and O7 to show that they are different strains. What this does not change, though, is that these strains are members of the same clonal complex and that resolution at this level is sufficient to identify an outbreak clone and to document geographic partitioning of strains along the California coastline (see below). This level of resolution is more robust to the possibility of strand slippage and evolution of new alleles during PCR that could make identical clones appear distinct with finer levels of resolution. An example of this may have occurred with SO4387, identified in this study as MS type ‘g, ’ but by Rejmanek et al. [23] as MS type ‘i. ’ These types differ by a single repeat at MS Sn9 (Table S1). It is also possible that this otter was co-infected with two closely related strains. Consistent identification of SLVs in many samples increases the confidence that they represent truly different strains. The outstanding potential these microsatellite markers have for more robust strain resolution, if interpreted cautiously, can facilitate addressing more specific questions, such as the identity and point source of an epidemic clone. The majority of strains (72/87; 83%) evaluated in this study were collected from two distinct 200 km stretches along the California coast or the adjacent terrestrial environment (Table S1; Figure 4). As such, we utilized this subset of the data to examine the temporal stability of strains and their geographic and host distribution in central California. The total time period covered by the strains analyzed in this study is 15 years (1994–2009). Sample sizes were not evenly distributed across each year and some years (1996–1998) had no representative samples, so it is likely that genotype life spans are underestimated. Despite this, at least one clonal complex, CC2, appears to be very stable in nature over time, exhibiting a lifespan encompassing the entire length of this study. CC2 was sampled during 12 of the 13 years for which a sample was collected (Table S2). Within this complex, Ag type II, MS type ‘g’ had a lifespan of the full time period examined (15 years) and was the longest lived of any Ag or MS type (Figure 5; Table S2). The other clonal complexes present in California, CC2, CC3, CC6–CC8, appeared to be stable as well, with life spans ranging from 5–8 years (Table S2). Collectively these data provide supporting evidence for S. neurona' s ability to propagate clonally. However, it will be important to test whether or not these allelic combinations appear more often than would be expected by chance to confirm clonal propagation as more sequencing data becomes available from strains collected from non-diseased animals and the position of the markers in the genome is identified [51]. Interestingly, the genotype associated with the outbreak, Ag type I, MS type ‘c’, was only found during 2004 (Figure 5). These samples were all associated with otters dying during the epizootic in April, 2004, except for two samples that were obtained from sick otters in the same area four months after the event ended (Table S1). The implications these observations may have for strain virulence are discussed below. On visual inspection, it appeared that the genetic composition of S. neurona strains from the Monterey Bay area was distinct from the southern strains obtained in or near Morro Bay (Figure 4; Table 1). We further tested this hypothesis by conducting χ2 analysis on the proportion of the majority clonal complexes (CC1 and CC2) that comprised each population. There was a highly significant difference between northern and southern strains (Figure 6). Significance remained when analysis was restricted to sea otter samples, in order to eliminate any confounding effects due to host species, because all southern strains were from sea otters (Figure 6). This conclusion is consistent with data reported previously on S. neurona strains from coastal California [24], [25], but contrasts with the conclusions of Rejmanek et al. [23]. We also sought to identify a potential terrestrial source for S. neurona strains present in the marine environment. Experimental evidence for the model organism, T. gondii, supports a route of infection for sea otters through ingestion of S. neurona sporocysts that were washed to the ocean in contaminated fresh water and then concentrated in the otters' filter-feeding invertebrate prey [52]–[54]. Implicating opossums as the ultimate terrestrial source of infection is supported by comparing the prevalence of the majority clonal complexes (CC1 and CC2) in sea otters and opossums in the northern, Monterey Bay area study site (the only locale from which opossum samples were obtained). Strain prevalence differences between these groups were not statistically different, suggesting that monitoring strain types in coastal dwelling opossums will be predictive of genotypes infecting adjacent marine dwelling otters (Figure 6). Observational data from the outbreak noting an abundance of razor clams and evidence of sea otter movement into the area for feeding (i. e. accumulation of broken shells on the shore) just prior to the event, further support this model of land-to-sea parasite transfer [33]. Sea otters very rarely consume known intermediate hosts of S. neurona [55], leaving the ingestion of sporocysts as the most biologically plausible route for sea otter infection regardless of the land-to-sea transport mechanism, and strongly supporting the conclusion that this outbreak originated from a selfing event in the opossum host. Disease is a complex manifestation of the interplay between intrinsic pathogen factors (i. e. pathogen genotype) and numerous external factors, including dose, host immune status, and environmental conditions such as weather that can influence transmission. Delineating the relative contribution of each of these factors to a given disease outbreak is a difficult process, as is illustrated by the outbreaks described in this study. It is plausible that the S. neurona strain associated with the 2004 epizootic is intrinsically more virulent than other strains since it was only identified during the time period surrounding the outbreak and may have been too virulent for continued propagation. Also, the majority of otters infected died within 24–48 hours of stranding and had high IgM titers [33]. The rapid rise and subsequent fall of a virulent strain type is a phenomenon noted in many outbreaks of a diverse array of pathogens from viruses (e. g. Influenza virus [3]) to bacteria (e. g. Leptospira interrogans [56]) to fungi (e. g. Coccidioides immitis [57]). However, this phenomenon may also be attributable to sampling biases [2] or environmental factors [57] making the assumption that the virulent genotype is not adaptive inaccurate. Equally in the case of the sea otter outbreak, numerous external factors, including concurrent infection with other pathogens and domoic acid poisoning, abundant food source with potential for contamination with sporocysts, and a large rainstorm preceding the event that could have increased sporocyst deposition, may have played a contributing role in conferring this S. neurona strain with a virulent phenotype [33]. Similarly, the T. gondii strain implicated in the 2001 Brazil outbreak appeared to rise in prevalence during the outbreak but then decline over time in the local environment [44]. This was also a unique, newly identified genotype that caused symptomatic disease in 155 immune-competent individuals—an unusual phenomenon for this normally asymptomatic parasite. Importantly, though, ∼270 other individuals with access to the same water cistern seroconverted during this time with no overt signs of disease [44], invoking a role for environmental and host factors in this outbreak. A striking character of both these outbreak events is the key role self-mating in the definitive host served as a catalyst allowing virulent pathogen genotypes to rapidly reach high levels under the right conditions to precipitate a disease epidemic. Epidemic clonality associated with sporocyst or oocyst ingestion strongly suggests that self-mating in the definitive host was the key event leading to these outbreaks. Selfing in the definitive host has been confirmed experimentally for T. gondii [31], [32] but only indirectly assumed for S. neurona [58]. Prior to this study, rigorous genetic characterization of selfing events in nature were lacking and the question as to whether a productive sexual out-cross or a selfing event precedes an outbreak linked to oocysts or sporocysts had not previously been tested. Early population genetic studies using limited, poorly resolved markers identified a paucity of mixed strain T. gondii or S. neurona infections in nature and these data have previously been interpreted to suggest that most definitive host infections would be by a single strain and therefore out-crossing would be rare in nature [59]. However, more recent studies using unbiased, multi-locus typing schemes have consistently identified mixed strain infections among natural intermediate hosts suggesting that prey species of definitive hosts are more frequently harboring mixed strain infections than previously envisaged [60]–[72]. Hence, the lack of mixed strain infections identified in earlier studies may simply reflect the techniques used, such as bioassay or limited genetic typing, that were biased toward certain strains and likely missed multiple infections and the true diversity of genotypes present. As more high resolution, multilocus genetic markers are being applied against previously characterized strains of T. gondii, an increasing number are being re-classified as recombinants, defined as products of sexual out-crossing events, including strains previously linked to outbreaks [20]. Given the virulent nature of the two outbreaks examined here, and the evidence that out-crossing between two avirulent, haploid parents can produce progeny with enhanced virulence [14], we originally hypothesized that out-crossing might explain the genetic origin and expansion of the outbreak strains, rather than self-mating. Intriguingly, close examination of the environmental isolates surrounding the T. gondii outbreak supported this hypothesis because the epidemic clone was one of many progeny produced by a local genetic out-cross. However, the available evidence indicated that, while out-crossing certainly preceded the outbreak, it was the subsequent selfing event that was responsible for the epidemic expansion and transmission of the virulent clone that caused the outbreak. Certainly this dataset argues that sex and self-mating combined to produce the T. gondii clonal outbreak. Further typing of additional outbreaks is warranted to examine whether or not an out-cross is independently sufficient to cause an epidemic attributable to multiple, recombinant progeny. This two-step process of local epidemic expansion via a sexual out-cross followed by clonal propagation of a few progeny with enhanced adaptations or virulence is reminiscent of the process envisioned on a larger scale for the pandemic rise of the archetypal T. gondii clones (Types I, II, and III), also found to be the progeny of an out-cross [13], [14]. Documenting this process in real time at a local level has provided key insight into mechanisms that account for clonal propagation in nature. It was previously proposed based on laboratory studies that clonal dominance of archetypal T. gondii strains was attributable to an enhanced ability for oral transmission through carnivory, a hypothesis which certainly warrants further investigation in natural settings [19]. However, recent studies have since shown that this trait does not operate as originally proposed [73], [74]. These findings raised the possibility that other life history traits may likewise be important in perpetuating clones. In this light, it is worth noting that all aspects of the parasite lifecycle that promote clonal propagation, namely selfing, oral transmission through carnivory, and transplacental transmission, contribute in part to clonality in the population structure. However, when considering their relative roles, the advantage in fecundity the sexual stage can impart during a selfing event to a single parasite genotype, as documented in this study, provides strong evidence this mechanism is likely the major contributor to localized or regional clonal dominance of certain strains. The basic reproductive number (R0), or number of secondary infections a single infected individual will cause, is many orders of magnitude greater in the definitive host (which releases millions of environmentally stable, infectious propagules capable of waterborne or aerosolized transmission [75]) compared to an intermediate host (in which the infectious units produced can only be passed to those directly feeding on tissues). Oocysts or sporocysts can also successfully infect intermediate hosts at much lower doses (even a single oocyst) than tissue cysts [76], [77]. Oocyst deposition therefore exists as a potent mechanism for causing widespread epidemics and establishes a plausible rationale for explaining how selective sweeps can occur among these heterogamous pathogens. Determining what factors govern whether these sweeps occur on a local, and presumably more frequent, epidemic level or reach pandemic proportions are important subjects for future research. Our results also confirm that fecal contamination of food and water sources represents a major threat to human and animal health, hence targeting the definitive host or the oocyst stage of these parasites is an excellent first-step strategy to disrupt transmission. This conclusion is further supported by studies showing the importance of the definitive host stage for maintaining continued transmission of this parasite in island communities [34]–[37] and how local vaccination of definitive feline hosts can significantly reduce T. gondii infection rates [78]. The scope of explanatory power for this selfing model can also be extended to other highly clonal, cyst forming parasites, including the clonal outbreak linked to S. neurona and likely other pathogenic Sarcocystis spp. and Neospora spp. This finding is significant since many aspects of the T. gondii life cycle have previously been proposed to be unique to this species among the tissue encysting coccidia, including its broad host range inclusive of nearly all warm-blooded vertebrates and its ability to be transmitted through carnivory among intermediate hosts [19]–[21, ] (but also see: [79], [80]–[83]). Notably, selfing has also been demonstrated in more distantly related Apicomplexan parasites, including Eimeria spp. and Plasmodium spp. [31]. In addition, the processes of homothalism and same-sex mating identified in fungi serve the analogous purpose of clonal propagation via a mechanism more generally thought to serve in genetic recombination and out-crossing [84]. This suggests that selfing, as a genetic mechanism of clonal propagation, has potential to play a pivotal and previously under-recognized role for a diverse array of eukaryotic pathogens in the expansion of genotypes that cause disease epidemics and/or emerge as highly successful clonotypes to rapidly alter population genetic structures. Work in California was conducted under United States Fish and Wildlife Service (USFWS) permit MA 491 672724-9 issued to United States Geological Survey Biological Resource Discipline (USGS492 BRD). Harbor seal carcasses were gathered and samples processed as part of Northwest Marine Mammal Stranding Network activities authorized under Marine Mammal Protection Act (MMPA) Stranding Agreements (SA), and Section 109 (h) (16 U. S. C. 1379 (h) ). Additional specimens were acquired under MMPA Section 120, and the National Marine Fisheries Service (NMFS) MMPA Research Permit 782–1702. Parasite DNA was obtained either from infected host tissues or parasite isolates maintained in tissue culture as described previously [25]. Samples were analyzed using a typing scheme that included the surface antigen markers: SnSAG1, SnSAG3, SnSAG4, SnSAG5, SnSAG6 [25] and 9 microsatellite markers Sn2–Sn5 and Sn7–Sn11 originally described by Asmundsson and Rosenthal [85] but applied as modified in Wendte et al. [25] and Rejmanek et al. [23]. Three additional microsatellite markers were designed by the following method: Publically available Sarcocystis neurona expressed sequence tags (ESTs) were downloaded from the NCBI dbEST database (http: //www. ncbi. nlm. nih. gov/dbEST) and the S. neurona Gene Index (maintained by the Computational Biology and Functional Genomics Laboratory at the Dana Farber Cancer Institute, http: //compbio. dfci. harvard. edu/tgi/) databases. The downloaded ESTs were assembled into contigs using the SeqMan (Lasergene) application. Contig sequences were then processed with the MISA microsatellite identification program (http: //pgrc. ipk-gatersleben. de/misa/) with the following repeat parameters: definition (unit size-minimum repeats): 2-12,3-7,4-5,5-4,6-3,7-3,8-2,9-2,10-2,11-2,12-2,13-2,14-2,15-2; interruptions (maximum difference between 2 simple sequence repeats): 25. Approximately 50 microsatellites of sufficient length and/or complexity were identified. Three (Sn1520, Sn1863 and Sn515) of these markers were not previously published and possessed sufficient non-redundant flanking sequence to allow for nested primer design and produced robust size-polymorphic PCR amplification products. Primers were validated as described [25] and found to be specific and sensitive for S. neurona DNA in tissues (data not shown). The primers designed are as follows: Sn1520 Fext- GGGGCAGAACCATCGTAGTA, Rext- GTGAAGCATTTCCCCTACGA, Fint- GGCGGTAGTCACTTGCTGA, Rint- GTGGGAGAAGACGGTCGTTA; Sn1863 Fext- CATGGCGTGCGTTAACTAAA, Rext- CGTACAAACACACGCTCCAC, Fint- CCATTCATCGACAGCGACTA, Rint- TGAGACAGCCGTCAAACACT; Sn515 Fext- CTTCTAGCGGCTGTTTCTCC, Rext- TCTGTGTGGGTGTGGAAGTC, Fint- GACCCCCTCTCTGCTTCTCT, Rint- ACGCAAATGCGAACATATCA. Representative sequences for each allele at each locus were placed in GenBank under the following accession numbers: Sn1520: HM851251, HM851252, HM851253, HM851254, HM851255; Sn1863: HM851256, HM851257, HM851258, HM851259; Sn515: HM851249, HM851250. PCR, DNA sequencing and analysis were conducted as described previously, except, to control for bias in scoring results, random sample IDs were assigned to samples before sequencing so that sequence analysis for some loci was blinded [25]. For this study, S. neurona DNA from 15 sea otters and 4 harbor seals was analyzed. Additionally, samples from 21 sea otters, 2 harbor seals, 3 horses, and 2 raccoons previously described by Wendte et al. [25] at the SnSAG antigen loci and MS Sn9, were further typed in this study at the remaining 10 MS loci. Finally, S. neurona DNA from 21 sea otters, 1 porpoise, 4 horses, 13 opossums, and 1 cat that was previously typed by Rejmanek et al. [23] at SnSAG3, SnSAG4, and MS markers Sn2–Sn5 and Sn7–Sn11 were combined with the data in this study for a total sample set that included 87 samples from 57 sea otters, 6 harbor seals, 2 raccoons, 13 opossums, 7 horses, 1 porpoise, and 1 cat. In all, 75 of the 87 samples were from California. Other states represented include Georgia (n = 2 samples), Illinois (n = 1), Missouri (n = 3), Washington (n = 4), and Wisconsin (n = 2). Some overlap existed between the samples typed in this study and those reported by Rejmanek et al. : samples SO4387, SO4413, H1, H2, and H3 in this study are reported as SO1, SO2, Horse 1, Horse 2, and Horse 3 in Rejmanek et al. [23], respectively. Complete information about the sample origins is found in Table S1. Toxoplasma gondii isolates from a water cistern (n = 2), chickens (n = 11), and one cat associated with a human waterborne toxoplasmosis outbreak [44], as well as laboratory strain CEP were typed at microsatellite loci B17, B18, TgMA, TUB2, W35 [39] and M95 [40]. Markers were PCR amplified and sequenced to assign alleles as for S. neurona markers [25]. Representatives of each microsatellite allele at each locus were placed in Genbank under accession numbers: B17: HM851260–67; TgMA: HM851268–73; W35: HM851274–77; M95: HM851278–81. Because different parts of the genome are likely under different selective pressures, all S. neurona samples were categorized by an antigen (Ag) type designated by roman numerals and a microsatellite (MS) type indicated by a lowercase letter designation. Ag types were defined by the presence/absence of mutually exclusive antigen genes (SnSAG1, SnSAG5, or SnSAG6) and the inheritance pattern of alleles at SnSAG3 and SnSAG4 [23], [25]. MS types were assigned on the basis of allele combinations defined by the number of di- or tri- nucleotide repeats at each locus (Sn2–Sn5 and Sn7–Sn11, Sn1520, Sn1863). Sn515 was a complex repeat in which each isolate possessed one of two alleles. Samples from the study by Rejmanek et al. [23] were not typed at the SnSAG1-5-6 loci, but were placed into Ag groups based on the allelic profile at SnSAG3 and SnSAG4 and by the Ag group their MS type was associated with in samples typed at all markers. For example, based on the alleles at SnSAG3 and SnSAG4, sample SO4 (Table S1) could be placed either in Ag type II or V, but its MS type was only found associated with Ag type II in samples where all markers were typed, making this the most likely, though not definitive, Ag type designation. The S. neurona strains assessed by Rejmanek et al. [23] were also not typed at MS markers Sn1520, Sn1863, and Sn515. Presumptively classifying these samples into MS types based on alleles at Sn2–Sn5 and Sn7–Sn11 is likely accurate, though, since these three markers did not provide additional resolution to MS types for the 46 additional S. neurona strains described in this study. The alleles present at MS markers Sn2–Sn5 and Sn7–Sn11 were used for creation of a multi-locus sequence typing scheme by which all isolates could be compared. The numerical designation of alleles allowed the detection of which MS types formed clonal complexes using the eBURST program [48]. Default settings were used which grouped MS types on the basis of sharing alleles at 8 of the 9 markers analyzed. To assess T. gondii isolates for clonality, MS alleles were combined with previously published DNA sequence analysis at three genetic loci, PCR-RFLP or DNA sequencing at 10 loci, and serologic analysis as described by Vaudaux et al. [44]. Statistical analyses were performed using GraphPad Prism 5 and χ2 values were considered significant at P = 0. 05.

The parasites Toxoplasma gondii and Sarcocystis neurona have lifecycles that include a sexual stage in a definitive host and an asexual stage in intermediate hosts. For T. gondii, laboratory studies have demonstrated that the sexual stage can serve the dual purpose of producing new, virulent genotypes through recombination and promoting expansion of single clones via self-mating. Self-mating and other life history traits of T. gondii, including transmission of asexual stages among intermediate hosts, are assumed to account for the clonal population genetic structure of this organism. However, the relative contributions of sexual recombination and self-mating verses other life history traits in causing disease outbreaks or in shaping Toxoplasma' s population genetic structure have not been verified in nature, nor have these traits been extensively examined in related parasites. To address this knowledge gap, we conducted population genetic analyses on T. gondii and S. neurona strains isolated from naturally occurring outbreaks affecting humans and sea otters, respectively. Our results identify self-mating as a key trait potentiating disease outbreaks through the rapid amplification of a single clone into millions of infectious units. Selfing is likely a key adaptation for enhancing transmission of recently emerged, recombinant clones and reshaping population genetic structures among the tissue-cyst coccidia.

Abstract Introduction Results/Discussion Materials and Methods

infectious diseases/protozoal infections infectious diseases/epidemiology and control of infectious diseases genetics and genomics/population genetics

2010

Self-Mating in the Definitive Host Potentiates Clonal Outbreaks of the Apicomplexan Parasites Sarcocystis neurona and Toxoplasma gondii

Human fungal pathogens like Candida albicans respond to host immune surveillance by rapidly adapting their transcriptional programs. Chromatin assembly factors are involved in the regulation of stress genes by modulating the histone density at these loci. Here, we report a novel role for the chromatin assembly-associated histone acetyltransferase complex NuB4 in regulating oxidative stress resistance, antifungal drug tolerance and virulence in C. albicans. Strikingly, depletion of the NuB4 catalytic subunit, the histone acetyltransferase Hat1, markedly increases resistance to oxidative stress and tolerance to azole antifungals. Hydrogen peroxide resistance in cells lacking Hat1 results from higher induction rates of oxidative stress gene expression, accompanied by reduced histone density as well as subsequent increased RNA polymerase recruitment. Furthermore, hat1Δ/Δ cells, despite showing growth defects in vitro, display reduced susceptibility to reactive oxygen-mediated killing by innate immune cells. Thus, clearance from infected mice is delayed although cells lacking Hat1 are severely compromised in killing the host. Interestingly, increased oxidative stress resistance and azole tolerance are phenocopied by the loss of histone chaperone complexes CAF-1 and HIR, respectively, suggesting a central role for NuB4 in the delivery of histones destined for chromatin assembly via distinct pathways. Remarkably, the oxidative stress phenotype of hat1Δ/Δ cells is a species-specific trait only found in C. albicans and members of the CTG clade. The reduced azole susceptibility appears to be conserved in a wider range of fungi. Thus, our work demonstrates how highly conserved chromatin assembly pathways can acquire new functions in pathogenic fungi during coevolution with the host. Eukaryotic chromatin is densely packed with the nucleosome being its basic repeating unit [1]. This structure represents a barrier for enzymes reading or modifying genomic DNA. Thus, disassembly and reassembly of histones, the core components of nucleosomes, is essential for various biological processes, including transcription, replication and DNA repair [2–7]. Two key players in the deposition of newly synthesized histones into chromatin are type B histone acetyltransferases (HATs) and histone chaperones. Type B HATs specifically acetylate free histones immediately after synthesis and show at least partial cytoplasmic localization. Hat1 was the first type B HAT identified and was found conserved throughout the eukaryotic kingdom [8,9]. Together with the Hat2 subunit (RbAp46/48 in higher eukaryotes), Hat1 acetylates histone H4 at lysine 5 and 12 in Saccharomyces cerevisiae and Candida albicans [9,10]. After binding of an additional subunit in the nucleus, the so-called NuB4 complex is formed, which is responsible for histone deposition at sites of DNA damage [11,12]. Rtt109 is another fungal-specific Type B HAT involved in DNA damage repair and associated histone deposition by acetylating free histone H3 at lysine 56 [2,13]. Interestingly, telomeric silencing is also defective in S. cerevisiae and Schizosaccharomyces pombe hat1Δ mutants [14,15]. Thus, Hat1 appears to be also involved in the generation or maintenance of repressive chromatin structures. Hat1 operates with various histone chaperones [16–18]. This class of proteins is able to bind histones thereby avoiding unspecific interactions with DNA and facilitating correct incorporation into nucleosomes [19]. Two main chromatin assembly pathways include the hallmark histone chaperone complexes CAF-1 and HIR. While CAF-1 is well known to function in replication-coupled chromatin assembly, recent reports indicate also a role in transcription regulation [20–22]. In contrast, HIR is involved in replication-independent chromatin assembly and acts as a repressor of histone genes outside of S-phase [23]. Of note, Hat1 was found in complexes together with CAF-1 or HIR, thus suggesting a central role in the delivery of newly synthesized histones for incorporation via distinct pathways [16,18]. The opportunistic human fungal pathogen Candida albicans is the most frequent cause of Candida-derived invasive infections [24]. Candida spp. can cause diseases ranging from chronic mucocutaneous to life-threatening systemic infections in immunocompromised patients. C. albicans belongs to the CTG clade, a group of closely related species which translate the CUG codon as serine instead of leucine [25]. During the infection process, C. albicans encounters various environmental stress conditions, including reactive oxygen species (ROS) produced by innate immune cells such as macrophages, dendritic cells or neutrophils dedicated to kill invading pathogens [26]. Furthermore, treatment with antifungal agents such as azoles, which inhibit fungal ergosterol biosynthesis and are commonly used to treat Candida infections, also represents severe stress to the pathogen [27]. At least two major mechanisms are therefore essential for C. albicans to be able to survive under these conditions: the fungus has to be able to repair damaged cellular components efficiently, and it has to respond rapidly by adapting transcriptional programs to counteract immune defense [28–32]. Changes in chromatin structure are involved in both mechanisms. Disassembly and reassembly of histones is required for efficient repair of DNA damage [2,7]. Furthermore, transcriptional modulation is intimately linked to the histone density at the corresponding loci [6,33–35]. Interestingly, several studies reported functions of chromatin-remodeling factors in the regulation of stress-responsive genes [33,36–38]. For instance, incorporation of histones into chromatin is particularly important for efficient gene repression, as the increase of histone density impairs binding of activators and inhibits RNA polymerase progression [6,39,40]. We have recently shown that C. albicans NuB4 complex is required for efficient repair of DNA damage resulting from endogenous or exogenous impact [10]. Here, we report a novel function for NuB4 in the negative regulation of oxidative stress resistance and azole tolerance. The Hat1 component of NuB4 acts in concert with distinct chromatin assembly pathways, which is independent of its conserved role in DNA damage repair. We show that the loss of the NuB4 complex markedly increases oxidative stress resistance through an accelerated induction of oxidative stress genes. Furthermore, this renders the pathogen resistant to killing by innate immune cells and promotes persistence of a hat1Δ/Δ mutant in a murine infection model. Interestingly, loss of Cac2, a subunit of the CAF-1 histone chaperone complex, mimics the oxidative stress resistance phenotype of hat1Δ/Δ cells. Furthermore, transcriptional profiling by RNA-Seq confirms overlapping functions of NuB4 and CAF-1 in C. albicans. Moreover, we also discover a novel role for NuB4 requiring the HIR complex for the negative regulation of azole tolerance in C. albicans. Interestingly, whereas the oxidative stress phenotype of hat1Δ/Δ cells has exclusively evolved in the Candida CTG clade, the reduced azole susceptibility seems to be conserved in a wider range of fungal species. Thus, our work demonstrates how highly conserved chromatin assembly pathways can acquire new functions, as for example in pathogenic fungi during coevolution with the host. Thus, host-specific immune defense mechanisms can act as drivers of evolutionary adaptation, enabling pathogens to cope with specific stress conditions. In a previous study, we identified the C. albicans NuB4 complex being essential for efficient repair of both exogenous and endogenous DNA damage [10]. C. albicans DNA damage repair mutants show increased susceptibility to ROS produced by immune cells [30]. Therefore, we asked if inactivation of the NuB4 complex would render this pathogen hypersusceptible to H2O2. Surprisingly, however, deletion of HAT1, HAT2 or both genes increased the resistance to H2O2 as determined by spot dilution assays (Fig 1A). Importantly, reintegration of HAT1 or HAT2 at its endogenous locus fully restored the wild-type phenotype (Fig 1A). Due to this unexpected resistance phenotype, we subjected the hat1Δ/Δ mutant to phenotypic analysis by applying a set of different stress conditions including cell wall stress (Calcofluor White, Congo Red), osmotic stress (NaCl), oxidative stress (tert-Butyl hydroperoxide (tBOOH), diamide), heavy metal stress (CdCl2), as well as antifungal drugs (Voriconazole, Itraconazole, Amphotericin B). For most conditions, we did not observe any difference between the wild-type and the hat1Δ/Δ strain (S1A Fig). However, lack of Hat1 markedly increased resistance to the oxidizing agents tBOOH and diamide, indicating a specific role for Hat1 in the regulation of oxidative stress resistance (Fig 1B and S1B Fig). Interestingly, we also observed that deletion of HAT1 increased tolerance to different azole drugs (Fig 1C). Azoles represent a widely used class of antifungals targeting the lanosterol 14-α-demethylase, thereby blocking fungal ergosterol biosynthesis. Furthermore, deletion of HAT2 or HAT1 and HAT2 mimicked deletion of HAT1 and reintegration of both genes at their endogenous loci fully restored the wild-type phenotype (Fig 1C). To confirm that the observed resistance phenotypes are independent of the NuB4 function in DNA damage repair, we determined sensitivities of different DNA damage repair mutants to H2O2 and voriconazole. Importantly, neither deletion of the gene encoding the histone H3 specific acetyltransferase Rtt109 nor the absence of the repair protein Rad52 yielded in comparable oxidative stress or voriconazole phenotypes (Fig 1B, 1D and S1C Fig). Hence, not all defects in DNA repair lead to oxidative stress resistance. These data suggest that Hat1 has an additional and novel function in C. albicans in the regulation of oxidative stress resistance and antifungal drug tolerance. Hat1 is involved in the deposition of histone H3–H4 dimers at sites of DNA damage and in heterochromatic regions in different species [14,41–43]. Different histone chaperones are responsible for the incorporation of histones into nucleosomes via distinct chromatin assembly pathways [20,44–47]. We hypothesized that a defect in chromatin assembly is causing the observed resistance phenotypes. Therefore, we created deletion mutants of a set of genes encoding histone chaperones or histone chaperone complex components known to interact or copurify with Hat1 or to regulate gene expression in other species (Table 1). Genes were deleted in wild-type cells as well as in a hat1Δ/Δ background and sensitivities to H2O2 and voriconazole were tested. Strikingly, lack of CAC2, a subunit of the CAF-1 histone chaperone complex, also strongly increased resistance to H2O2 and tBOOH (Fig 2A and S1D Fig). Furthermore, a quantification of H2O2 resistance by determination of the survival rate in liquid culture confirmed this result (Fig 2B). Interestingly however, spot dilution assays suggested a minor effect on azole susceptibility of cac2Δ/Δ cells (Fig 2C). By contrast, deletion of HIR1, a component of the HIR histone chaperone complex, dramatically increased tolerance to voriconazole, but did not alter H2O2 susceptibility (Fig 2A and 2C). Importantly, HAT1 and HIR1 are epistatic, since a double deletion strain failed to show increased azole resistance when compared to the corresponding single deletions based on spot dilution assays and growth inhibition in liquid culture (Fig 2C and 2D). Deletion of RTT106 did not alter susceptibility to H2O2 and voriconazole (Fig 2A and 2C). Unfortunately, we and others were unable to construct homozygous deletion mutants for ASF1 and SPT16, indicating that they might be essential [51,52]. We also included the C. albicans mutant lacking the histone chaperone Spt6 [53]. Similar to HAT1, deletion of SPT6 increased H2O2 resistance (Fig 2E and 2F). However, a reliable and accurate determination of azole sensitivities of the homozygous spt6Δ/Δ mutant was not possible due to its pronounced slow-growth phenotype. Of note, deletion of one SPT6 allele also decreased susceptibility to voriconazole (Fig 2F). In a previous study, we showed that reducing histone H4 gene dosage mimics the genetic deletion of HAT1 with respect to sensitivity to genotoxic stress [10]. Furthermore, a common set of genes is differentially regulated upon depletion of histone H4 and deletion of CAC2 in S. cerevisiae [37]. Thus, we determined the effect of histone H4 reduction on oxidative stress and azole susceptibility. Interestingly, strains harboring a single copy of the histone H4 gene remaining also displayed increased H2O2 resistance and slightly elevated tolerance to voriconazole (Fig 2G). In summary, these data indicate that defects in distinct C. albicans chromatin assembly pathways can alter susceptibilities to H2O2 and voriconazole, thus regulating oxidative stress response and tolerance to azole antifungals, respectively. Next, we wanted to determine if the role of Hat1 in the regulation of oxidative stress resistance and azole tolerance is conserved in other fungal species. Therefore, we analyzed the effect of HAT1 deletion on H2O2 and voriconazole susceptibility in the distantly related fungi Saccharomyces cerevisiae, Candida glabrata and Schizosaccharomyces pombe. However, loss of Hat1 did not lead to increased resistance to H2O2 in any of these species (Fig 3A). Furthermore, lack of Hat1 failed to lower voriconazole sensitivity in S. cerevisiae and C. glabrata (Fig 3B), but increased azole tolerance in S. pombe (Fig 3Bc). Based on these results, we speculated that Hat1 might have acquired a role in the regulation of oxidative stress resistance only later in evolution. To prove this, we constructed homozygous HAT1 deletion mutants in the more closely related CTG clade members Candida parapsilosis and Candida tropicalis and determined their H2O2 sensitivity. Strikingly, lack of Hat1 decreased the susceptibility to H2O2 (Fig 3C) and increased tolerance to voriconazole in both species (Fig 3D). Thus, the role of Hat1 in regulating oxidative stress resistance appears specific for C. albicans and related species within the CTG clade, while the function in azole tolerance is present in a wider range of fungal species. To investigate the effect of HAT1 deletion during normal growth, and to elucidate the mechanism of oxidative stress resistance in the hat1Δ/Δ mutant, we determined transcriptional profiles of cells during the logarithmic growth phase and upon H2O2 treatment. Therefore, we performed RNA sequencing (RNA-Seq) analysis on cells before and after exposure to 1. 6 mM H2O2 for 30 minutes. Importantly, no loss of viability was observed even after 1 hour treatment at this peroxide concentration (S2A Fig). Furthermore, transcriptional profiles were also determined for the cac2Δ/Δ strain, since this mutant also showed an oxidative stress resistance phenotype. In addition, we included rtt109Δ/Δ cells as a control, as it mimics lack of Hat1 concerning morphology and accumulation of DNA damages [10,30], yet shows wild-type susceptibility to H2O2 (S1C Fig). RNA-Seq analysis of logarithmically growing cells showed that genetic removal of HAT1 primarily upregulates a large number of genes. We found 743 genes at least 2-fold induced and 209 genes 2-fold repressed in the hat1Δ/Δ mutant when compared to the wild-type. Furthermore, the amplitude of gene repression was clearly lower when compared to the upregulation of genes, implying that Hat1 exerts primarily repressing rather than activating functions (Fig 4A). Also for the cac2Δ/Δ and rtt109Δ/Δ mutants, we observed an almost exclusive upregulation of gene expression when compared to the wild-type (Fig 4B and 4C). However, lack of Cac2 or Rtt109 had a less pronounced effect concerning the number of upregulated genes when compared to the loss of Hat1. In the cac2Δ/Δ and the rtt109Δ/Δ strains, 464 and 243 genes were transcriptionally upregulated, respectively. Comparison of regulated genes in the hat1Δ/Δ, the cac2Δ/Δ and the rtt109Δ/Δ mutants revealed large overlaps between all datasets (Fig 4D) with 123 genes upregulated in all three mutants. Interestingly, the majority of upregulated genes in the cac2Δ/Δ mutant were also induced in the hat1Δ/Δ strain. As expected, we also detected a large overlap between the hat1Δ/Δ and the rtt109Δ/Δ strain, since both mutants share functions in DNA damage repair and cell morphology [10,30]. Following H2O2 treatment, downregulated genes in the hat1Δ/Δ mutant increased to 459 when compared to the wild-type, although under these conditions the majority of differentially regulated genes remain up (907). The comparison of upregulated genes in the hat1Δ/Δ, the cac2Δ/Δ and the rtt109Δ/Δ mutants again revealed large overlaps between the three datasets (Fig 4E). Similar to the untreated condition, most genes with higher expression levels in the cac2Δ/Δ strain were also upregulated in the hat1Δ/Δ mutant, suggesting redundant functions of Hat1 and Cac2 in the regulation of gene expression. Furthermore, we again detected a large overlap between the hat1Δ/Δ and the rtt109Δ/Δ strains, and a common set of genes upregulated in all three mutants (Fig 4E). In addition, a large fraction of upregulated genes (449) in the hat1Δ/Δ mutant were repressed in the wild-type upon treatment with H2O2, indicating a defect in repression due to the lack of Hat1 (Fig 4F). However, we also detected a smaller set of upregulated genes (104) in the hat1Δ/Δ strain that were induced in the wild-type. Thus, lack of Hat1 primarily leads to upregulation of genes in logarithmically growing cells, as well as upon H2O2 treatment indicating a repressive function of Hat1 in C. albicans. In addition to protein-coding genes, we also analyzed the expression of non-coding elements, including tRNAs and small RNAs. Without treatment, most non-coding RNAs were not differentially expressed in the hat1Δ/Δ mutant (S1 Table). Out of 193 non-coding RNAs, 64 were repressed more than two-fold in wild-type cells upon H2O2 treatment (S2B Fig). Interestingly, in the hat1Δ/Δ mutant, 20 of these non-coding RNAs were upregulated at least 2-fold when compared to the wild-type, again indicating impaired repression upon loss of Hat1. To further characterize genes upregulated upon loss of Hat1, we performed a GO term enrichment analysis. Without H2O2 treatment, at least 2-fold upregulated genes in the hat1Δ/Δ mutant were strongly enriched for genes involved in lipid catabolic processes and oxidation-reduction processes (Fig 5A). Interestingly, the latter group includes genes encoding for proteins with functions in oxidative stress tolerance like the catalase (CAT1), superoxide dismutases (SOD3-6) and a thiol peroxidase (orf19. 87) [26,54–56]. Preliminary proteomics data also identified this group of proteins as being upregulated in the hat1Δ/Δ mutant (S2C Fig). Of note, we failed to observe enrichment for DNA damage response genes most likely due to the fact that Hat1 is involved in different processes in C. albicans. This leads to a high number of differentially expressed genes in the mutant and could hamper detection of enriched GO groups. Therefore, we analyzed subsets of differentially expressed genes based on their expression in the three mutants. As expected, due to their DNA damage phenotype, genes significantly upregulated only in the hat1Δ/Δ and the rtt109Δ/Δ mutants were strongly enriched for DNA damage repair genes (Fig 5B). The specific overlap of regulated genes between hat1Δ/Δ and cac2Δ/Δ was still enriched for genes involved in oxidation-reduction processes and arginine metabolism, implying that Hat1 and Cac2 might be involved in the same processes (Fig 5C). Finally, we identified genes involved in mitochondrial degradation and microautophagy which were significantly upregulated only in the hat1Δ/Δ mutant (Fig 5D). Next, we analyzed genes with differential expression levels in the mutants relative to the wild-type upon H2O2 treatment. The whole group of genes significantly upregulated in the hat1Δ/Δ mutant was enriched for genes involved in rRNA processing and transport reactions (Fig 5E). We failed to detect any gene sets enriched specifically in the overlaps between the hat1Δ/Δ and the rtt109Δ/Δ or the hat1Δ/Δ and cac2Δ/Δ strains. However, we observed a strong enrichment of genes involved in non-coding RNA (ncRNA) /rRNA processing and ribosome biogenesis as upregulated specifically in the hat1Δ/Δ mutant (Fig 5F). In summary, loss of Hat1 mainly leads to increased expression of genes belonging to different functional groups, which is in accordance with Hat1 being involved in different processes in C. albicans. Furthermore, distinct functional groups affected by the loss of Hat1 are also upregulated in either the rtt109Δ/Δ or the cac2Δ/Δ strain indicating that Hat1 might function in specific processes together with Rtt109 or Cac2. Transcriptional profiling of hat1Δ/Δ and cac2Δ/Δ cells revealed the upregulation of various gene sets encoding for proteins involved in the response to oxidative stress. Both mutants displayed elevated transcriptional levels of CAT1 encoding catalase, which is responsible for the decay of H2O2 [54]. Of note, RT-qPCR analysis confirmed derepression of CAT1 in hat1Δ/Δ and cac2Δ/Δ cells in the absence of H2O2. However, as expected from the RNA-Seq data, CAT1 derepression was also observed for the rtt109Δ/Δ strain, which is not resistant to H2O2 (S2D Fig). Thus, this moderate derepression of CAT1 is most likely due to a general stress response in these deletion mutants. Therefore, we investigated CAT1 induction upon treatment with H2O2. Interestingly, lack of Hat1 as well as Cac2 caused increased CAT1 expression, indicating that these two proteins negatively regulate the induction kinetics of CAT1 (S2E Fig). However, induction levels in the rtt109Δ/Δ control strain were similar to the wild-type (S2E Fig). Since lack of Hat1 affected induction levels of CAT1, we investigated the induction kinetics of this gene in detail. Thus, we determined transcript levels upon treatment with H2O2 over time. Interestingly, lack of Hat1 resulted in a significantly faster induction of the CAT1 gene when compared to the wild-type (Fig 6A). Since Hat1 is involved in the deposition of histones into chromatin, we determined the effect of HAT1 deletion on the histone density at the CAT1 locus by chromatin immunoprecipitation (ChIP) using antibodies against histone H3. Without treatment, the hat1Δ/Δ mutant showed a reduced histone density at the CAT1 promoter (Fig 6Ba). There was no difference in the occupancy in the CAT1 coding sequence (CDS) between the mutant and the wild-type (Fig 6Bb). Interestingly, however, treatment with H2O2 decreased histone density at the promoter as well as in the CDS significantly faster in hat1Δ/Δ cells when compared to the wild-type (Fig 6B). This explains the increased induction rate in the mutant, since nucleosomes represent a physical barrier for RNA polymerase II (RNAPII) and reduced nucleosome density can facilitate transcription [6,35,57]. Thus, lower histone density in the hat1Δ/Δ mutant could promote higher RNAPII processivity leading to increased mRNA production. Alternatively, elevated mRNA levels could also be due to increased RNAPII recruitment, which could be facilitated by the reduction in nucleosome density at the promoter. To clarify whether elevated mRNA levels are the consequence of increased RNAPII recruitment, we determined RNAPII levels at the CAT1 gene using ChIP. Interestingly, we detected increased recruitment of RNAPII to the CAT1 CDS, thus explaining the elevated mRNA levels (Fig 6C). Notably, the oxidative stress response set upregulated in the hat1Δ/Δ mutant also comprised several glutathione-utilizing enzymes. Thus, we determined the induction kinetics for two of these genes upon treatment with H2O2. Similar to CAT1, we detected a faster induction of GPX1 (orf19. 86) encoding a glutathione peroxidase (GPx) as well as of GST1 (orf19. 3121), encoding a glutathione S-transferase (Fig 6D). In addition, we also detected increased RNAPII levels at both genes upon treatment with H2O2 (Fig 6E). Notably, we observed similar hyperinduction of GPX1 and GST1 upon H2O2 treatment for the cac2Δ/Δ mutant as well (Fig 6F and 6G). Furthermore, we tested whether the increase in transcription of oxidative stress genes in the hat1Δ/Δ mutant is paralleled by elevated activity of the corresponding enzymes. Therefore, we prepared whole cell extracts from cells before and after 60 min exposure to H2O2 and determined catalase activity as well as GPx activity spectrophotometrically. We detected elevated catalase activity in the hat1Δ/Δ mutant already under non-inducing conditions, albeit as expected at a low level (Fig 7A). However, H2O2 treatment of hat1Δ/Δ cells dramatically increased catalase activity when compared to the wild-type, indicating that the mutant is more efficient in degrading hydrogen peroxide (Fig 7A). Likewise, the GPx activity assay detected some increase already under non-inducing conditions (Fig 7B), but H2O2 treatment of the hat1Δ/Δ mutant significantly increased GPx activity (Fig 7B). These data demonstrate that the increased induction rate of oxidative stress genes in the hat1Δ/Δ mutant increases the activities of the enzymes encoded by the target genes. Furthermore, to determine whether catalase hyperinduction is the main reason for the increased resistance of the hat1Δ/Δ mutant, we deleted the CAT1 gene in wild-type and hat1Δ/Δ cells and quantified their H2O2 susceptibilities. As expected, cat1Δ/Δ cells were highly sensitive to H2O2 showing only 6% survival after treatment with a low H2O2 concentration of 1 mM (Fig 7C). However, the double deletion strain maintained increased H2O2 resistance when compared to the cat1Δ/Δ single mutant indicating that increased catalase expression is not the main reason for the hydrogen peroxide resistance of the hat1Δ/Δ strain (Fig 7C). To determine if increased induction of glutathione-utilizing enzymes is causing the increased H2O2 resistance, we genetically depleted cells for glutathione by removing the GSH2 and GCS1 glutathione biosynthesis genes in wild-type and hat1Δ/Δ backgrounds. The first step in the synthesis of glutathione is catalyzed by the gamma-glutamylcysteine synthetase (Gcs1), followed by the second step carried out by the glutathione synthase (Gsh2). Deletion of one of these genes disrupts glutathione biosynthesis in C. albicans leading to glutathione auxotrophy and to increase oxidative stress sensitivity [58,59]. To determine the sensitivity to H2O2 of wild-type and hat1Δ/Δ cells upon deletion of GSH2 or GCS1, the strains were grown in the absence of glutathione and treated with H2O2 for 2h. As expected, lack of Gsh2 as well as Gcs1 strongly increased the sensitivity to H2O2 (Fig 7D). Strikingly, however, the absence of Hat1 did not significantly change H2O2 resistance in the absence of Gsh2 or Gcs1 (Fig 7D). Thus, the data strongly suggest that H2O2 resistance caused by deletion of Hat1 is mainly mediated via the glutathione system. The absence of Hat1 and Cac2 leads to upregulation of oxidative stress genes and resistance to hydrogen peroxide. ROS are produced by immune cells to kill C. albicans during the infection process. The fungus counteracts this attack by upregulating ROS detoxifying enzymes of which the superoxide dismutases Sod5 and to some extend Sod4 were shown be essential for survival of C. albicans upon phagocytosis [26]. Interestingly, our RNA-Seq data not only revealed an upregulation of the catalase and glutathione-utilizing enzymes, but also markedly increased expression of genes encoding the superoxide dismutases Sod4 and Sod5 in cells lacking Hat1. The significant induction of SOD4 and SOD5 in the hat1Δ/Δ strain was also confirmed by RT-qPCR analysis (Fig 8A). Therefore, we asked if the deletion of HAT1 influences the ability to detoxify and survive ROS released by immune cells. ROS production during interaction of C. albicans with bone marrow-derived murine macrophages and bone marrow neutrophils was determined by measuring luminol dependent chemiluminescence [26]. Interestingly, deletion of HAT1 strongly reduced total ROS levels during interaction with macrophages (Fig 8B and 8C), as well as neutrophils (S3A Fig), when compared to the wild-type. Phagocytosis is required for ROS production by the NADPH oxidase [26]. To exclude that differences in phagocytosis of the pseudohyphal hat1Δ/Δ cells contribute to reduced ROS levels, we performed a phagocytosis assay. However, phagocytosis was similar for the hat1Δ/Δ strain or wild-type (S3B Fig). Furthermore, we also determined the NADPH oxidase activation in macrophages interacting with wild-type and hat1Δ/Δ cells by immunoblotting. However, we observed no difference in the phosphorylation levels of the NADPH oxidase subunit p40phox upon interaction with hat1Δ/Δ cells or wild-type cells (S3C Fig). Therefore, reduced ROS accumulation upon interaction with hat1Δ/Δ cells most likely results from increased ROS detoxification by the mutant. In addition, interaction of macrophages with rtt109Δ/Δ cells did not lead to significant changes in the ROS levels when compared to the wild-type (Fig 8B). Finally, we determined the resistance of hat1Δ/Δ cells to killing by bone marrow neutrophils. Strikingly, lack of Hat1 clearly increased the survival rate upon interaction with the immune cells. Furthermore, reintegration of HAT1 fully restored the wild-type sensitivity (Fig 8D). These data strongly suggest that genetic removal of HAT1 strongly promotes increased survival to neutrophil attack due to the impaired ROS accumulation. Deletion of HAT1 causes reduced growth rate in vitro with morphological defects even in complete media (Fig 9A) [10], which has been shown to reduce virulence of several C. albicans mutants [28,30,60–63]. However, cells lacking Hat1 are also more resistant to killing by immune cells (Fig 8D). Therefore, we wanted to test how these seemingly opposing phenotypes caused by the deletion of HAT1 affect virulence of C. albicans. We used a mouse model of systemic candidiasis. Infection was performed via the tail vein and fungal burdens in kidneys were followed at day 1,3 and 7. Interestingly, after 24 hours mice infected with the hat1Δ/Δ mutant showed significantly reduced CFUs in the kidneys when compared to the wild-type or the restored strain (Fig 9B). However, the fungal burden of the mutant increased until day 7 after infection reaching the levels of the wild-type and the reintegrant (Fig 9B). Thus, cells lacking Hat1 are not efficiently cleared from infected mice as they are able to compensate the in vitro growth defect in vivo. To further investigate the virulence properties of the hat1Δ/Δ strain, we determined the survival rate of infected mice. Interestingly, 15 days post infection the majority of wild-type infected mice had died, whereas all of the mice infected with the hat1Δ/Δ strain were still alive (Fig 9C). Even after 32 days, only one mouse infected with the hat1Δ/Δ mutant had died. Of note, mice infected with the reintegrant showed an intermediate survival rate, which was however not significant when compared to the wild-type strain (Fig 9C). Although the majority of mice survived the infection with cells lacking Hat1, mutant cells were not cleared from the kidneys in 4 out of 5 individuals (Fig 9D). Instead, the fungal burden stayed high until the end of the experiment. Furthermore, two mice survived infection with the restored strain and for both Candida was not cleared (Fig 9D). Thus, again the revertant strain showed an intermediate phenotype most likely due to haploinsufficiency. We also determined the effect of deleting CAC2 in vivo using the same murine model of systemic infection as for the hat1Δ/Δ mutant. Interestingly enough, cac2Δ/Δ cells showed reduced fungal burdens at day 1 and 3 after infection. However, similar to the hat1Δ/Δ mutant, the strain lacking Cac2 was also protected from clearance by the immune system, since CFUs in kidneys increased during the course of the experiment (Fig 9E). In addition, body weight measurements of infected mice revealed a striking difference between the wild-type and hat1Δ/Δ as well as the cac2Δ/Δ cells. As expected, mice infected with the wild-type rapidly lost body weight after infection. However, upon infection with the hat1Δ/Δ or the cac2Δ/Δ strain, weight loss was clearly reduced, suggesting a reduced burden to the immune system or increased tolerance to the infecting fungal strain (S4A and S4B Fig). Nonetheless, although showing increased persistence in the host, deletion of HAT1 strongly attenuates virulence in a mouse model of systemic candidiasis. Even though fungal burdens are comparable between the hat1Δ/Δ mutant and the wild-type 7 days post infection, the mutant is strongly attenuated in killing the host. Since kidney failure is the primary cause of death in this particular mouse infection model [64,65] we determined the degree of kidney injury or function in infected mice by measuring serum urea levels. As expected due to the lower fungal burden, mice infected with the hat1Δ/Δ mutant showed reduced urea levels at day 3. The reintegrant again yielded in an intermediate phenotype (Fig 9F). However, at day 7 lack of Hat1 still yielded in lower urea levels when compared to the wild-type (Fig 9F). Thus, lack of Hat1 seems to cause reduced kidney injury even at comparable fungal burdens. In summary, our data suggest that the hat1Δ/Δ strain is able to compensate its in vitro growth defect in vivo and persist in the host. However, cells lacking Hat1 are severely compromised in killing the host. In this work, we provide compelling evidence for a novel function of the eukaryotic histone acetyltransferase Hat1 in the pathogenic fungus C. albicans. Hat1 is required for regulating oxidative stress response, antifungal drug tolerance as well as virulence. We show that Hat1 functions in distinct chromatin assembly pathways acting in concert with well-known histone chaperones CAF-1 and HIR. Hence, in addition to an evolutionary conserved role of Hat1 in DNA damage repair [10,41,66,67] our data demonstrate a novel function for Hat1 in regulating the response to oxidative stress and azole treatment. Furthermore, certain resistance phenotypes as well as gene sets misregulated upon loss of Hat1 are also affected by distinct histone chaperones operating in independent chromatin assembly pathways. Thus, our data suggest that the Hat1 histone acetyltransferase has a central function in mediating the flux of histones necessary for nucleosome remodeling in C. albicans thereby affecting various cellular processes, including fungal virulence and persistence in the host. Interestingly, the function in the regulation of the ROS response seems restricted to C. albicans and related fungal CTG clade members. This has not been reported in other organisms so far. This function of Hat1 might have developed only recently, providing an explanation why this is only observed in a small number of fungal species. Of note, Rtt109 from non-pathogenic S. cerevisiae negatively regulates catalase expression in response to H2O2 together with Asf1 via histone deposition, whereby lack of Asf1 or Rtt109 increases RNAPII recruitment [33]. Likewise, we detect increased RNAPII levels at several oxidative stress genes immediately after induction in the hat1Δ/Δ mutant (Fig 6). While the S. cerevisiae rtt109Δ mutant displays slightly increased H2O2 resistance (S5 Fig), the C. albicans RTT109 deletion fails to effect catalase induction or peroxide resistance (S1C and S2E Figs). Thus, it is very tempting to speculate that Hat1 and Rtt109 are involved in the regulation of distinct gene sets in different species. Noteworthy, Asf1 interacts with Hat1 in S. cerevisiae [68]. Whether Hat1 regulates oxidative stress response in C. albicans also via Asf1 remains unclear, since Asf1 is an essential gene in C. albicans. Asf1 is also essential in S. pombe and Drosophila melanogaster [69,70], implying increased functional redundancy in histone chaperone functions in S. cerevisiae when compared to C. albicans and other species. Of particular interest is that Spt6 negatively regulates catalase expression in S. cerevisiae [33]. The observed H2O2 resistance phenotype of the C. albicans spt6Δ/Δ mutant might indicate conservation of this function in both species. Although interesting, further experiments are needed to determine the molecular mechanism (s) of oxidative stress regulation in cells lacking Spt6. We show here that genetic removal HAT1 accelerates the induction kinetics of oxidative stress genes, including catalase as well as other genes encoding glutathione-utilizing enzymes. However, the lack of catalase in a hat1Δ/Δ background does not lower the H2O2 sensitivity to the level of the cat1Δ/Δ single knock-out (Fig 7C). On the other hand, the absence of Hat1 combined with defects in glutathione biosynthesis does not significantly increase H2O2 tolerance (Fig 7D). Therefore, the H2O2 resistance phenotype of the hat1Δ/Δ deletion strain is likely to be due to the upregulation of glutathione-utilizing enzymes, although a contribution of Cat1 cannot be fully excluded at this point. Interestingly, novel functions for the CAF-1 and HIR histone chaperone complexes in the regulation of white-opaque switching were reported for C. albicans [71]. The CAF-1 and HIR chaperone complexes are essential for histone deposition via distinct pathways. While the HIR complex functions in replication-independent chromatin assembly, CAF-1 is thought to mediate replication-coupled histone deposition [21,44,72]. To the best of our knowledge, no reports exist suggesting a role for Hat1 acting through the HIR complex to modulate antifungal drug resistance in any other eukaryote. Of note, this Hat1 function may have been conserved in some fungal species, since elevated azole tolerance is at least observed in S. pombe lacking the Hat1 orthologue. Of note, HAT1 is indeed repressed in C. albicans cells treated with itraconazole, implying a function for Hat1 in the regulation of azole susceptibility [73]. Furthermore, our RNA-Seq data demonstrate a downregulation of ergosterol biosynthesis genes in the hat1Δ/Δ strain, including ERG3 whose lack confers pronounced azole resistance [74]. Thus, a similar mechanism may explain the reduced susceptibility of hat1Δ/Δ and hir1Δ/Δ strains. Another mechanism mediating the increased azole tolerance could be the upregulation of multidrug transporters [75,76]. The two major ATP-binding cassette transporters responsible for azole resistance in C. albicans are Cdr1 and Cdr2 [77–80]. Whereas only CDR2 was slightly upregulated in hat1Δ/Δ cells, the major facilitator superfamily transporter Mdr1, which is also induced upon oxidative stress and confers azole resistance [77,81], is significantly upregulated in the hat1Δ/Δ mutant upon treatment with H2O2 (S1 Table). However, the exact mechanism by which Hat1 and Hir1 regulate azole tolerance remains to be determined in future studies. Although, the CAF-1 complex can assemble histones in a replication-coupled manner, it can also influence the rate of replication-independent histone incorporation and transcription [22,82,83]. Therefore, NuB4 and CAF-1 might act in concert to regulate specific target genes such as oxidative stress response sets in C. albicans. We propose that they do so by facilitating histone incorporation concomitant with transcription, thereby increasing the histone density. Indeed, we observe a large overlap in differentially expressed gene sets in hat1Δ/Δ and the cac2Δ/Δ mutants, the latter lacking a subunit of CAF-1 (Fig 4D and 4E). Unfortunately, we were not able to obtain a hat1Δ/Δ cac2Δ/Δ double deletion strain for epistasis analysis. Thus, we cannot completely rule out the possibility that Hat1 and Cac2 regulate H2O2 resistance independently. One explanation for a possible synthetic lethality of hat1Δ/Δ cac2Δ/Δ double mutants may be that both are associated with essential DNA replication [21,84,85]. The genome-wide transcriptional RNA-Seq profiling of logarithmically growing hat1Δ/Δ cells mainly revealed upregulation of genes, suggesting repressive functions of Hat1. Notably, distinct GO terms are enriched in the hat1Δ/Δ mutant alone or in overlaps with other mutants. These data suggest specific functions for Hat1 in different processes, including DNA damage repair, arginine biosynthesis and mitochondrial degradation (Fig 5). Interestingly, Rtt109 inhibits arginine biosynthesis genes in S. cerevisiae together with Asf1 under repressing conditions [86]. By contrast, this gene set is derepressed in C. albicans hat1Δ/Δ and cac2Δ/Δ, but not in rtt109Δ/Δ cells (Fig 5B and 5C). Thus, we believe that Hat1/Cac2 have taken over this function from Rtt109 in C. albicans. Interestingly enough, mitochondrial dysfunction has been linked to azole and oxidative stress sensitivity in C. albicans, as well as in other fungal species [87–89]. However, depletion of Fzo1, which is involved in mitochondrial fusion, and Goa1, a protein localizing to mitochondria upon oxidant treatment [87,88], decreases azole and oxidative stress tolerance. Thus, although genes involved in mitochondrion degradation are upregulated in hat1Δ/Δ cells, the loss of mitochondrial function is unlikely to explain the increase in stress resistance. Interestingly, upon H2O2 treatment we observed increased expression of genes involved in ncRNA/rRNA processing specifically in the hat1Δ/Δ mutant (Fig 5E and 5F). However, this group of genes is repressed in the wild-type upon H2O2 stress. Only 4 out of a total of 119 hat1Δ/Δ upregulated genes in this group were less than 2-fold repressed in the wild-type. Thus, lack of Hat1 seems to cause a defect in repression of genes involved in ncRNA/rRNA processing and ribosome biogenesis. Although a faster induction kinetics of genes connected to H2O2 resistance occurs upon loss of Hat1 (Fig 6A and 6D), many oxidative stress-regulated genes appear not significantly modulated in the RNA-Seq dataset making GO term enrichment for this group futile. A possible explanation is that many stress conditions in fungi, including oxidative stress or osmostress, follow a stress-specific and often transient regulation [90], which may result in difficulties choosing the correct timing for collecting data points in hat1Δ/Δ and wild-type cells. Nevertheless, we speculate that the Hat1-mediated oxidative stress regulation may relate to host infection conditions, where severe oxidative stress is a major immune defense during phagocytosis of fungal pathogens [26,91]. Although infection of macrophages or neutrophils by the hat1Δ/Δ mutant strongly reduces total ROS accumulation, we observed faster accumulation of ROS in the beginning of the interaction (Fig 8B and S3A Fig). Differences in ROS production cannot be simply explained by the pseudohyphal morphology of cells lacking Hat1, since the rtt109Δ/Δ strain has the same morphological defect [30] yet does not show any differences in ROS accumulation (Fig 8B). Thus, it may specifically result from Hat1-mediated gene regulation and from upregulated factors triggering ROS production. Interestingly enough, a mouse model of virulence yields compelling data showing that hat1Δ/Δ cells display strongly attenuated virulence. Of note, infected mice show significantly reduced organ burdens of hat1Δ/Δ cells immediately after infection, but reach comparable levels of fungal burdens between day 3 and day 7 post infection (Fig 9B). This implies a faster growth in vivo, which may be explained by the reduced susceptibility to neutrophil-mediated killing, essentially providing a “fitness” advantage of hat1Δ/Δ in vivo. Moreover, cells lacking Hat1 cause less damage to host organs and therefore enable better mouse survival. Reduced urea levels in sera of infected mice support this notion (Fig 9F). In line with this hypothesis is also the fact that hat1Δ/Δ cells degrade ROS more efficiently, protecting the hat1Δ/Δ mutant from clearance by the immune surveillance. This may also result in reduced tissue damage, since reactive oxygen species originating from innate immune response are harming both the pathogen and the host due to increasing inflammation [92,93]. Indeed, a downmodulation of the inflammatory immune response can be beneficial for the host during invasive Candida albicans infections [65,94]. Finally, ROS is also known as a signaling molecule during infections and reduced levels upon challenge with the hat1Δ/Δ mutant might impact the immune response [95]. Furthermore, compelling evidence exists suggesting that the hat1Δ/Δ virulence phenotype is not just due to its role in DNA damage repair. Deletion of RAD52 or RTT109 also causes pseudohyphal cell morphology and strong virulence defects in mouse models of systemic candidiasis [28,30,96]. However, the rtt109Δ/Δ mutant is efficiently cleared from kidneys of infected mice 3 days post infection [30]. Likewise, kidney fungal burdens of rad52Δ/Δ cells decline over 3 days post infection even with an inoculum concentration ten times higher than in our setup [28]. By sharp contrast, the hat1Δ/Δ fungal burdens in kidneys increase, reaching wild-type levels 7 days post infection. Nonetheless, despite increased growth in vivo, removal of Hat1 renders cells unable to kill the host efficiently over 32 days, consistent with a persistent but avirulent invasive infection (Fig 9C). Our data unequivocally show that loss of Hat1, Cac2 and Hir1 can be beneficial for C. albicans in our experimental setup. However, the inability to efficiently repress genes under stress conditions might also be detrimental for the organism. Furthermore, loss of Hat1 and Cac2 also impairs DNA damage repair and thus promotes genome instability [10,71]. Thus, the resulting fitness loss due to the absence of these factors might be disadvantageous in the long run, despite a gain in oxidative resistance. Further experiments are required to identify additional factors functioning together with NuB4/CAF-1 and NuB4/HIR in the regulation of oxidative stress resistance or azole tolerance. This might also lead to the discovery of potential antifungal targets, which could be used to render cells more sensitive to ROS or azoles. Importantly, the data from the mouse model imply that inhibitors of HATs could have beneficial effects in clinical therapeutic settings but pose the risk of promoting persistent or latent infections. Interestingly, we don’t observe comparable resistance phenotypes upon genetic removal of Hat1 in S. cerevisiae, C. glabrata and at least for the oxidative stress resistance also in S. pombe (Fig 3). Thus, in C. albicans, the NuB4 and CAF-1 complexes might have gained functions during coevolution with the human host, which appears restricted to the CTG clade. Hence, deletion of HAT1 in C. glabrata, the second-most prevalent human fungal pathogen [24], does not result in similar phenotypes. Nonetheless, our results clearly demonstrate substantial differences in the functions of even highly conserved chromatin modification mechanisms between C. albicans and other fungi. Therefore, results obtained from classical model systems such as baker’s yeast often cannot simply be transferred to pathogenic fungi. The regulation of virulence-associated traits by chromatin modification has been reported for several pathogens and it is now generally accepted that the chromatin status affects virulence [97,98]. Moreover, several reports suggest that chromatin modifiers are involved in the regulation of fungal virulence factors including antifungal drug tolerance [97,99–102]. Hence, modulation of chromatin function or chromatin-modifying pathways including nucleosome remodeling may be a common strategy during coevolution of microbial pathogens with the host to promote immune evasion. All animal experiments were evaluated by the ethics committee of the Medical University of Vienna and approved by the Federal Ministry for Science and Research, Vienna, Austria (GZ: BMWF- 68. 20n5/231-II/3b/2011) adhering to European legislation for animal experimentation. Rich medium (YPD) and synthetic complete medium (SC) were prepared as previously described [103]. Minimal medium contained 8. 38 g/l yeast nitrogen base (BD Biosciences), 6. 25 g/l ammonium sulfate (Sigma Aldrich) and 2 g/l glucose (Sigma Aldrich). Fungal strains were routinely grown on YPD plates at 30°C. Hydrogen peroxide, tert-butyl hydroperoxide, Calcofluor White, Congo Red, cadmium chloride, sodium chloride, luminol and HRP Type VI were obtained from Sigma Aldrich. Voriconazole, Itraconazole and Amphotericin B were purchased from Discovery Fine Chemicals Ltd. DMEM and RPMI media were purchased from PAA. A list of fungal strains, plasmids and primers used in this study is shown in Tables A, B and C in S1 Text, respectively. All C. albicans strains constructed in this work were derived from the clinical isolate SC5314 [104]. For deletion of HAT1 in C. parapsilosis the clinical isolate GA1 was used [105]. Disruption of HAT1 in C. tropicalis was done in the clinical isolate AKH2249. Deletion of CAC2 and HIR1 was done using a modified version of the SAT1 flipper method [106]. Briefly, the marker cassette was amplified using the pSFS3b plasmid and primers containing some 80bp homologous region to replace the whole coding sequence of the corresponding gene [10]. For deletion of RTT106 two primer pairs were used to add the homologous regions in two sequential PCR steps. For deletion of CAT1, YEp352-SAT1-CAT1urdr was constructed by in vivo cloning in S. cerevisiae exactly as described previously [107]. Plasmids for deletion of GCS1, GSH2, C. tropicalis HAT1 and C. parapsilosis HAT1 were constructed by fusing ~500bp fragments upstream and downstream of the corresponding gene with a FRT-FLP-NAT1-FRT cassette derived from pSFS3b [10] and a fragment containing an ampicillin resistance cassette and an E. coli replication origin derived from YEp352 [108] via in vivo recombination in E. coli EL350 as described in [109]. Due to low transformation efficiency for C. parapsilosis and C. tropicalis a second set of deletion vectors was constructed in the same way to obtain the homozygous knock-outs. These vectors contained 300–350bp of the beginning and the end of the corresponding CDS for homologous recombination to avoid integration at the deleted allele. For construction of the CAC2 reintegration plasmid the coding sequence plus ~500 up- and downstream was amplified and cloned into pSFS3b via KpnI and ApaI. Transformation of C. albicans was done via electroporation [106]. Correct integration of the deletion cassette and loss of the corresponding gene were confirmed by colony PCR. Colony PCR assays were used to verify correct genomic integration of deletion constructs. Briefly, a colony was resuspended in 25 μl H2O in a PCR tube and incubated at 95°C for 10 minutes. Cell debris was spun down briefly and 5 μl of the supernatant was used as template for the PCR, which was performed using the DreamTaq Green DNA Polymerase (Thermo Scientific) according to the manufacturer’s instructions. Spot dilution assays were performed as described previously [10]. For determination of H2O2 survival logarithmically growing cells in YPD were treated with the indicated concentrations of hydrogen peroxide for 2 hours at 30°C. Before and after treatment cells were diluted and plated on YPD plates. For the H2O2 survival shown in Fig 7D cells were grown overnight in minimal medium, diluted to OD600 of 0. 2 and further incubated at 30°C for 5 hours. Cells were harvested by centrifugation at 1500 g for 3 minutes, resuspended at an OD600 of 0. 5 and treated with 0. 5 mM H2O2 for 2 hours. Before and after treatment cells were diluted and plated on YPD plates containing 1 mM glutathione. Colonies were counted after 3 days incubation at 30°C and viability was determined relative to the samples plated before H2O2 addition. To quantify growth inhibition by azole treatment cells were grown to logarithmic phase in SC medium at 30°C. Cultures were diluted to an OD600 of 0. 01 in SC medium with or without voriconazole at the indicated concentrations. OD600 was determined after growth at 30°C for 18–24 hours. For C. parapsilosis cultures were incubated for 41 hours due to the slow growth rate of the hat1Δ/Δ mutant. Growth inhibition was calculated relative to untreated controls. Cells were grown in YPD overnight to an OD600 of 1 at 30°C. For hydrogen peroxide treatment 1. 6 mM H2O2 was added to the culture for the indicated period of time. RNA isolation and qPCR analysis was done as described previously [10]. For RT-qPCRs shown in Fig 8A, RIP1 was used as reference gene [99]. All other RT-qPCRs were normalized to PAT1 [110]. After RNA isolation, 10 μg total RNA were treated with DNase I (Thermo Scientific) and purified using the RNeasy MinElute Cleanup Kit (Qiagen). 5 μg DNase treated RNA were used for rRNA depletion with the RiboMinus Eukaryote System v2 (Life Technologies, Carlsbad, CA). rRNA depleted samples were fragmented using the NEBNext Magnesium RNA Fragmentation Module (New England Biolabs) and purified with the RNeasy MinElute Cleanup Kit (Qiagen). SuperScript III reverse transcriptase (Life Technologies, Carlsbad, CA) was used for first strand synthesis. Priming was done with 3 μg random hexamers (Life Technologies). Samples were purified using Mini Quick Spin Columns (Roche) and second strand synthesis was done with the NEBNext mRNA Second Strand Synthesis Module (New England Biolabs). Final purification of double stranded cDNA was done with the MinElute PCR Purification Kit (Qiagen). Samples were further processed and sequenced on a HiSeq 2000 instrument (Illumina) at the Next Generation Sequencing Facility (CSF NGS unit, http: //www. csf. ac. at) of the Campus Vienna Biocenter. For both conditions five biological replicates for the wild-type as well as the hat1Δ/Δ strain and three biological replicates for the cac2Δ/Δ as well as the rtt109Δ/Δ mutants were sequenced. Reads were mapped onto the Assembly 21 of the C. albicans genome (http: //www. candidagenome. org) using NextGenMap [111]. Read counts were determined with HTSeq using the union mode [112] and a reference annotation (C_albicans_SC5314_version_A21-s02-m07-r10; http: //www. candidagenome. org). The annotation of the coding sequence assembly was used as transcript coordinates. For short non-coding RNAs (tRNAs, snRNAs, snoRNAs and ncRNAs) 20bp up- and downstream of their chromosomal coordinates were added before mapping [113]. Differential expression analysis was done with edgeR [114]. Benjamini-Hochberg adjusted p-values were used to determine differentially regulated genes [115]. Venn diagrams were created using Venny 1. 0 [116]. Gene ontology (GO) term enrichment was determined using the GO Term finder (http: //www. candidagenome. org). Overlapping GO terms were merged manually. Chromatin immunoprecipitation was performed as described previously [113]. One mg whole cell extract was used per ChIP. For determination of histone density an antibody against the C-terminus of histone H3 was used (ab1791, Abcam). Detection of RNAPII was done with an antibody against the C-terminal domain (05–592, clone 8WG16, Millipore). To analyze the CAT1 promoter region primers amplifying a fragment ranging from -315 to -163 with respect to the start codon were used. To determine enrichment within the CAT1, GPX1 and GST1 genes primers within these coding regions were used. Signals were normalized to an intergenic region on chromosome R. To quantify catalase activity, cells were grown overnight to an OD600 of 1 at 30°C. For hydrogen peroxide treatment 1. 6 mM H2O2 was added to the culture for one hour. Before and after treatment 20 ml culture were harvested at 1500 g for 3 min at 4°C and washed once with 20 ml cold H2O. Pellets were resuspended in 250 μl lysis buffer [50 mM Tris-HCl pH 7. 5; 10% glycerol, complete protease inhibitor cocktail (Roche) ] and an equal volume of glass beads (425–600mm, Sigma Aldrich) was added. Cells were lysed by shaking 5 times at 6 m s-1 for 30 s on a FastPrep instrument (MP Biomedicals). Extracts were cleared by centrifugation at 14000 g for 5 min at 4°C. Protein concentration in the extracts was determined by measuring absorption at 280 nm. For catalase activity measurement 5–40 μl whole cell extract were added to 3 ml of catalase assay buffer [384 mM Na3PO4; 0. 015 mM Triton X-100 (Sigma Aldrich); 11. 4 mM H2O2] and degradation of H2O2 was determined by measuring absorbance at 240 nm for up to 2 min. Catalase activity was calculated in μM H2O2 per minute per mg of whole cell extract as described previously [117] (ε = 43. 6). Glutathione peroxidase assays were performed as previously described [118] using some modifications. The lysis buffer used contained 50 mM Tris-HCl, pH 7. 5,150 mM NaCl, 0. 5 g/100 ml Nonidet P-40 and complete protease inhibitor cocktail (Roche, Basel, Switzerland). Cell lysis was performed as described for the catalase assay. 10–50 μl cleared whole cell extract was used for the assay exactly as described in [118]. Sample preparation and western blot analysis were essentially carried out as described previously [119]. A MOI of 5: 1 (fungi to macrophages) was used and samples were harvested after 30 min of interaction. Activated NADPH oxidase was detected using an antibody against the phosphorylated p40phox subunit (Cell Signaling 4311). A panERK antibody (BD 610123) was used as loading control. Whole cell extracts for mass spectrometric analysis were prepared by lysing logarithmically growing cells in MS lysis buffer [10 mM Tris-HCl pH 7. 5; complete protease inhibitor cocktail (Roche) ] with a French press. Extracts were lyophilized, resuspended in 2 M urea and used for trypsin digestion followed by liquid chromatography—tandem mass spectrometry analysis on a LTQ Orbitrap Velos system (Thermo Scientific). For all experiments, 7–10 week old C57BL/6 wild-type mice were used. Isolation and cultivation of primary bone marrow-derived macrophages was done as described previously [26]. Isolation of bone marrow neutrophils and subsequent C. albicans survival assays were performed as described earlier [65]. A MOI of 1: 10 (fungi to neutrophils) was used, and cells were harvested after a 1-hour interaction. Mouse infections were carried out through lateral tail vein injections as described previously with some minor modifications [65]. Briefly, C. albicans strains were grown overnight to an OD600 of around 1, washed twice and finally resuspended in PBS. For infection, 1 x 105 Candida cells per 21 g mouse body weight were injected via the lateral tail vein. For survival experiments, mice were monitored for 32 days. Analysis of fungal burdens in the kidneys at day 1,3 and 7 post infection, as well as determination of serum urea levels was done exactly as described previously [65]. Statistical analysis was carried out using the Prism software (Graphpad Software Inc.). ROS assays were done exactly as described previously [26]. The multiplicity of infection (MOI) for all ROS assays was 5: 1 (fungi to immune cells). Phagocytosis assays were performed essentially as described with some modifications [119]. C. albicans cells were grown overnight to an OD600 of around 1, washed twice in PBS and stained with 10 mg ml-1 Alexa Fluor 488 (Life Technologies) in 100 mM HEPES buffer (pH 7. 5) for 60 min at 30°C shaking in the dark. After staining cells were washed 3 times, resuspended in HEPES buffer and used for interaction with BMDMs. Stained Candida cells were added to macrophages and incubated for 45 min at 37°C and 5% CO2. Control reactions were kept on ice during the whole procedure. A MOI of 2: 1 (fungi to macrophages) was used. Phagocytosis was terminated by chilling on ice. Plates remained on ice during subsequent detaching and fixation in 1% formaldehyde. Extracellular fluorescent C. albicans cells were quenched by addition of 0. 4% trypan blue. Samples were subject to flow cytometry analysis with FL1-H on a FACSCalibur instrument (BD Biosciences).

Candida albicans is the most prevalent fungal pathogen infecting humans, causing life-threatening infections in immunocompromised individuals. Host immune surveillance imposes stress conditions upon C. albicans, to which it has to adapt quickly to escape host killing. This can involve regulation of specific genes requiring disassembly and reassembly of histone proteins, around which DNA is wrapped to form the basic repeat unit of eukaryotic chromatin—the nucleosome. Here, we discover a novel function for the chromatin assembly-associated histone acetyltransferase complex NuB4 in oxidative stress response, antifungal drug tolerance as well as in fungal virulence. The NuB4 complex modulates the induction kinetics of hydrogen peroxide-induced genes. Furthermore, NuB4 negatively regulates susceptibility to killing by immune cells and thereby slowing the clearing from infected mice in vivo. Remarkably, the oxidative stress resistance seems restricted to C. albicans and closely related species, which might have acquired this function during coevolution with the host.

Abstract Introduction Results Discussion Materials and Methods

2015

The Candida albicans Histone Acetyltransferase Hat1 Regulates Stress Resistance and Virulence via Distinct Chromatin Assembly Pathways

Small RNAs regulate diverse biological processes by directing effector proteins called Argonautes to silence complementary mRNAs. Maturation of some classes of small RNAs involves terminal 2′-O-methylation to prevent degradation. This modification is catalyzed by members of the conserved HEN1 RNA methyltransferase family. In animals, Piwi-interacting RNAs (piRNAs) and some endogenous and exogenous small interfering RNAs (siRNAs) are methylated, whereas microRNAs are not. However, the mechanisms that determine animal HEN1 substrate specificity have yet to be fully resolved. In Caenorhabditis elegans, a HEN1 ortholog has not been studied, but there is evidence for methylation of piRNAs and some endogenous siRNAs. Here, we report that the worm HEN1 ortholog, HENN-1 (HEN of Nematode), is required for methylation of C. elegans small RNAs. Our results indicate that piRNAs are universally methylated by HENN-1. In contrast, 26G RNAs, a class of primary endogenous siRNAs, are methylated in female germline and embryo, but not in male germline. Intriguingly, the methylation pattern of 26G RNAs correlates with the expression of distinct male and female germline Argonautes. Moreover, loss of the female germline Argonaute results in loss of 26G RNA methylation altogether. These findings support a model wherein methylation status of a metazoan small RNA is dictated by the Argonaute to which it binds. Loss of henn-1 results in phenotypes that reflect destabilization of substrate small RNAs: dysregulation of target mRNAs, impaired fertility, and enhanced somatic RNAi. Additionally, the henn-1 mutant shows a weakened response to RNAi knockdown of germline genes, suggesting that HENN-1 may also function in canonical RNAi. Together, our results indicate a broad role for HENN-1 in both endogenous and exogenous gene silencing pathways and provide further insight into the mechanisms of HEN1 substrate discrimination and the diversity within the Argonaute family. Argonautes are an evolutionarily conserved family of proteins implicated in diverse cellular processes. They function as effector proteins in the RNA-induced silencing complex (RISC), a gene regulatory complex that binds small, non-coding RNAs to target its silencing effects. Small RNAs are broadly segregated into groups that differ in their mechanisms of biogenesis and silencing, as well as in the subsets of Argonaute effectors that bind them. The microRNAs (miRNAs) are highly conserved small RNAs processed from endogenous hairpin precursors that regulate networks of mRNAs primarily through post-transcriptional repression [1], [2]. The piRNAs, so named for the Piwi Argonautes that bind them, function predominantly in maintenance of germline integrity, often through repression of repetitive transposable elements. The small interfering RNAs comprise a more heterogeneous group that includes small RNAs derived from cleavage of exogenous double-stranded RNA (exo-siRNAs) or generated endogenously (endo-siRNAs). Chemical modification has emerged as an important theme in regulation of small RNA function (for a review, see Kim et al. , 2010 [3]). Internal editing has been found to occur in select miRNA precursors through the action of ADAR (adenosine deaminase acting on RNA) enzymes, with consequences for miRNA processing efficiency, stability, and targeting [4]–[8]. Some siRNAs generated in fly and mouse also show evidence of editing by ADARs [9], [10], but the significance of such internal editing among siRNAs is not yet known. In contrast, terminal editing through 2′-O-methylation, addition of untemplated nucleotides, or exonucleolytic trimming plays a more general role in small RNA metabolism. These terminal modifications are not unrelated. Evidence in plants and animals suggests that methylation of the 3′ terminal nucleotide protects small RNAs from polyuridylation and polyadenylation, signals that direct exonucleolytic degradation [11]–[16]. Thus, terminal methylation plays an important role in regulating small RNA turnover. Formation of the 2′-O-methyl group is catalyzed by HEN1, a methyltransferase discovered in Arabidopsis thaliana that is conserved across metazoa, fungi, viridiplantae, and bacteria [17]. Although plant and animal HEN1 orthologs exhibit 40–50% amino acid similarity in the conserved methyltransferase domain [18], the proteins differ in their substrate specificity. Plant HEN1 acts on small RNAs in duplex and methylates both siRNAs and miRNAs [19]–[21]. In contrast, animal HEN1 orthologs modify only single-stranded small RNAs [22]–[24], enabling methylation of small RNAs such as piRNAs, which are not derived from double-stranded RNA intermediates [25]–[29]. While animal piRNAs appear to be universally methylated [24], [26], [27], [30]–[32], animal miRNAs are generally not methylated [19], [26], [31], and the mechanisms by which animal HEN1 orthologs discriminate between substrates are not entirely clear. HEN1 orthologs that catalyze terminal methylation of small RNAs have been characterized in mouse, fish, and fly, among other organisms [15], [22]–[24], [33], yet the orthologous methyltransferase in worm [18] has yet to be investigated. With its expanded Argonaute family and diverse small RNA classes, Caenorhabditis elegans provides an advantage for studying HEN1 substrate specificity. Since the discovery of the founding members of the microRNA family in C. elegans [1], [2], [34], many additional classes of small RNAs have been characterized. A large-scale small RNA sequencing effort revealed a class of terminally methylated 21-nucleotide RNAs with 5′ uridines [27]. These 21U RNAs were subsequently determined to represent the piRNAs of C. elegans based on their germline-specific expression, association with worm Piwi Argonautes PRG-1 and PRG-2, and function in transposon silencing and maintenance of temperature-dependent fertility [35]–[38]. Also found through small RNA cloning and deep sequencing were populations of 26- and 22-nucleotide RNAs with a 5′ preference for guanosine (the 26G RNAs and 22G RNAs, respectively) that constitute the endo-siRNAs of C. elegans [27], [39]. The 26G RNAs are primary endo-siRNAs generated in the germline to regulate spermatogenic and zygotic gene expression. They are divided into two non-overlapping subclasses named for the Argonautes that bind them: the ERGO-1 class 26G RNAs, which are generated in the maternal germline and distributed into the embryo, and the ALG-3/ALG-4 class 26G RNAs, which are specific to the male germline and required for sperm function [40]–[42]. The 22G RNAs are composed of many small RNA classes, all of which are bound by worm-specific Argonautes (Wagos). A large population of 22G RNAs are secondary endo-siRNAs whose production by RNA-dependent RNA polymerases is triggered by the activity of 21U RNAs and 26G RNAs [36], [41]–[43]; however, many other 22G RNAs are independent of these primary small RNAs [44], [45]. Secondary siRNAs serve to amplify the signal of primary small RNAs to effect robust silencing. Production of 22G secondary siRNAs is also triggered by exogenously introduced dsRNAs [43], [45]–[47], suggesting convergence of endogenous and exogenous RNAi pathways at the level of the secondary siRNA response. Among C. elegans small RNAs, only 21U RNAs and 26G RNAs are known to be methylated [27], [42]; 22G RNAs triggered by either primary endo- or exo-siRNAs appear to be unmethylated [45], [46]. Although the significance of worm small RNA methylation is unknown, loss of terminal methylation has been shown to decrease stability of piRNAs in many animal models [15], [22], [24] and both endo- and exo-siRNAs in fly [22], [48]. Methylation may therefore represent an essential step in stabilization of some classes of worm small RNAs. In this study, we characterize the C. elegans hen1 ortholog, which has been named henn-1 (hen of nematode), as the name hen-1 has already been assigned to an unrelated C. elegans gene. We demonstrate that HENN-1 methylates small RNAs bound by Piwi clade Argonautes: the 21U RNAs and the ERGO-1 class 26G RNAs. However, we show that 26G RNAs bound by Ago clade Argonautes ALG-3 and ALG-4 are not methylated and are therefore henn-1-independent. Differential methylation of 26G RNAs provides evidence for an existing model [13], [22], [23], [49], [50] wherein evolutionarily divergent Argonautes either direct or prohibit HEN1-mediated methylation of associated small RNAs. In further support of this Argonaute-dictated methylation model, we find that small RNAs are likely methylated after associating with an Argonaute: the Argonaute ERGO-1 is required for 26G RNA methylation, but methylation is not required for ERGO-1 to bind a 26G RNA. In the henn-1 mutant, levels of both 21U RNAs and ERGO-1 class 26G RNAs drop precipitously after their deposition into embryo, suggesting that HENN-1-mediated methylation is essential for perdurance of the maternal small RNA load during filial development. Accordingly, the henn-1 mutant shows enhanced somatic sensitivity to exogenous RNAi, a phenotype associated with loss of ERGO-1 class 26G RNAs. Surprisingly, however, the henn-1 mutant germline exhibits an attenuated response to RNAi, suggesting that HENN-1 may also function in the exogenous RNAi pathway. Altogether, our study supports a role for HENN-1 in diverse small RNA pathways in C. elegans and offers further insight into the mechanisms governing substrate discrimination for animal HEN1 orthologs. To examine small RNA methylation in C. elegans, we began by characterizing C02F5. 6, the gene previously predicted to encode the HEN1 ortholog in worm [18]. This gene, subsequently named henn-1, encodes a protein that exhibits significant amino acid similarity across the conserved HEN1 methyltransferase domain relative to established members of the HEN1 family (Figure S1). Although two henn-1 gene models with differing 3′ ends have been proposed, 3′RACE and protein studies using a rabbit polyclonal antibody generated against a common N-terminal HENN-1 epitope detected only the longer isoform (Figure S2A, S2B). To facilitate our studies of the function of HENN-1, we isolated and characterized the henn-1 (tm4477) allele. This allele carries a deletion that encompasses henn-1 exon four, which encodes 65% of the conserved methyltransferase domain as annotated by Kamminga et al. [15]. Sequencing of the henn-1 (tm4477) mRNA indicates that loss of exon four activates a cryptic splice donor site in the third intron, resulting in an extended third exon that encodes a premature termination codon (Figure S2B). The henn-1 (tm4477) mRNA is readily detected by RT-PCR but does not produce a detectable protein product (Figure S2A) or exhibit methyltransferase activity (see below), suggesting that henn-1 (tm4477) (hereafter, henn-1) represents a functional null allele. Like piRNAs in fly [22], [23], [32], mouse [30], [31], and zebrafish [26], the C. elegans 21U RNAs are terminally methylated [27], but the factor catalyzing this modification has not yet been identified. To determine if 21U RNA methylation depends on henn-1, we assessed methylation status using the β-elimination assay [51]. A small RNA molecule whose terminal nucleotide has been 2′-O-methylated is resistant to this treatment, whereas the cis-diols of an unmodified 3′ terminal nucleotide are oxidized by sodium periodate, rendering the nucleotide susceptible to β-elimination under basic conditions. The resulting size difference can be resolved on a polyacrylamide gel to determine methylation status. All 21U RNAs examined were found to be terminally methylated in a henn-1-dependent manner (Figure 1A, Figure S3A), whereas a control miRNA was not methylated in either wild-type or henn-1 mutant animals (Figure 1B). Although 21U RNAs are still detectable in the henn-1 mutant, the abundance of the full-length species is visibly decreased for some 21U RNAs; this correlates with the appearance of putative degradation products of unmethylated, unprotected 21U RNAs. To demonstrate that loss of 21U RNA methylation in the henn-1 mutant is specifically due to the absence of henn-1, we used the Mos1-mediated single copy insertion technique [52] to introduce a henn-1: : gfp transgene driven by the promoter of the polycistronic mRNA that encodes henn-1 (xkSi1) or by the germline-specific pie-1 promoter (xkSi2) into the henn-1 mutant (Figure S2C). Both endogenous and germline-specific expression of henn-1: : gfp restore 21U RNA methylation in the henn-1 mutant (Figure 1A). To investigate the relationship between terminal methylation and piRNA accumulation, we used Taqman RT-qPCR to assess 21U RNA levels in wild-type and henn-1 mutant animals across development at 25°C. Importantly, the Taqman stem-loop RT primer is capable of distinguishing between full-length and terminally degraded small RNAs [53]. For example, the let-7e miRNA differs from let-7a only in the absence of the final nucleotide and U>G substitution at the ninth nucleotide, a position likely not represented in the stem-loop Taqman primer. Absence of this final nucleotide decreases detection of let-7e by the let-7a Taqman assay by more than a thousandfold [53]. henn-1 mutant embryo and early larva show dramatically reduced detection of female germline-enriched piRNA 21UR-1848 (Figure 2A), consistent with decreased embryonic detection for some 21U RNAs observed by northern blot (Figure 1A, Figure S3A). 21U RNA levels recover to wild-type in late larval stages, coincident with the onset of germline proliferation and de novo 21U RNA biosynthesis; however, in gravid animals at 56 hours, 21UR-1848 levels in the henn-1 mutant have declined to less than 50% of those observed in wild-type (P = 0. 0005; two-tailed t-test). Eight additional 21U RNAs examined show a similar pattern (Figure S4). These data suggest that henn-1 is dispensable for piRNA biogenesis but essential for robust inheritance of piRNAs. Parallel analysis of miR-1 and several additional miRNAs across development shows that effects of loss of henn-1 are specific to its substrates and not due to generalized small RNA dysregulation in the henn-1 mutant (Figure 2B, Figure S5). We next sought to determine the extent to which decreased abundance of piRNAs in the henn-1 mutant compromises activity of the piRNA pathway. Unlike in fly, where many selfish genetic elements are desilenced in the absence of piRNAs [32], C. elegans at present has only a single established molecular readout for piRNA pathway function: increased expression of transposase mRNA from Tc3, a Tc1/mariner family transposon [35], [36]. Two 21U RNAs have been found to map to Tc3, but both map in the sense direction and thus are unlikely to act directly in Tc3 repression via canonical RNAi mechanisms [35], [36]. Rather, 21U RNAs likely mediate their repressive effects through triggering production of secondary siRNAs, 22G RNAs, that engage worm-specific Argonautes (Wagos) to effect Tc3 gene silencing [36], [45]. We therefore identified a 22G RNA that shows complete antisense complementarity to Tc3 and can be classified as a Wago-dependent, 21U RNA-dependent secondary siRNA based on its total depletion both in the MAGO12 mutant, which lacks all Wagos, and in the prg-1 (n4357); prg-2 (n4358) double mutant, which lacks piRNAs [36], [45]. Levels of this 22G RNA in the henn-1 mutant are reduced by 44% in embryo but not significantly altered in hatched L1 larva (Figure S6A). This suggests that the low embryonic and early larval levels of 21U RNAs in the henn-1 mutant are still sufficient to trigger production of secondary siRNAs, although to a lesser degree than in wild-type. Consistent with the modest effect of loss of henn-1 on accumulation of piRNA-triggered secondary siRNAs, henn-1 mutant animals exhibit only a small increase (35% in starved L1 larva, 25% in L1 larva fed for 4 hours at 25°C) in Tc3 transposase mRNA levels relative to wild-type (Figure 2C). This is not unexpected due to the poor coincidence of the time intervals corresponding to piRNA dysregulation in the henn-1 mutant and Tc3 sensitivity to 21U RNAs; the henn-1 mutant shows the greatest disparity in piRNA levels in early larval development, whereas Tc3 levels are most sensitive to piRNAs in germline and embryo (Figure 2A, 2C). These findings suggest that HENN-1 is not strictly required for piRNA target repression, but contributes to robust silencing of Tc3. In addition to Tc3 dysregulation, loss of prg-1 also results in a temperature-sensitive sterility phenotype [38], [43]. To determine if the henn-1 mutant also exhibits a fertility defect, we assessed fertility at 20°C and 25°C. At 20°C, brood size of the henn-1 mutant does not differ significantly from that of wild-type. In contrast, henn-1 mutant animals maintained at 25°C exhibit a 25% decrease in brood size relative to wild-type (P = 0. 0059; two-tailed t-test) that can be rescued by germline expression of henn-1: : gfp from the xkSi2 transgene (Figure S7). The impaired fertility of the henn-1 mutant is consistent with abnormal fertility phenotypes associated with loss of HEN1 methyltransferase activity in other animals. Loss of HEN1 in Tetrahymena thermophila depletes Piwi-interacting RNAs called scan RNAs, impairing DNA elimination and, consequently, the viability of progeny [24]. The zebrafish hen1 mutant fails to maintain a female germline, resulting in an exclusively male population [15]. Nevertheless, we cannot conclude that the temperature-sensitive fertility defect of the henn-1 mutant is due exclusively to compromise of the 21U RNA pathway. 26G RNAs were reported to be methylated in the first C. elegans small RNA deep sequencing study [27]. Subsequent studies concluded that the species assessed was an ERGO-1 class 26G RNA [40]. Consistent with these data, we found that ERGO-1 class 26G RNAs, found in female germline and embryo, are methylated. As was the case for piRNAs, this methylation occurs in a henn-1-dependent manner (Figure 3A, Figure S3B). Surprisingly, however, ALG-3/ALG-4 class 26G RNAs, specific to the male germline, showed no evidence of methylation even in wild-type animals (Figure 3B, Figure S3C). One potential explanation for this observation would be that female germline small RNAs are universally methylated, whereas male germline small RNAs are not. To explore this possibility, we assessed 21U RNAs in male and female germlines. Both were methylated (Figure 3C), indicating that differential 26G RNA methylation cannot be explained simply by a lack of methyltransferase functionality in the male germline. Because the two classes of 26G RNAs bind unique Argonautes in male and female germlines, we hypothesized that the Argonaute ERGO-1 might direct methylation of 26G RNAs. To address this question, we sought to assess methylation of an ERGO-1 class 26G RNA in the absence of ERGO-1. As 26G RNAs are dramatically depleted in the absence of their respective Argonautes [40], we queried published wild-type and ergo-1 (tm1860) gravid adult deep sequencing libraries [42] to identify an ERGO-1 class 26G RNA that still accumulates to levels sufficient for visualization by northern blotting in the ergo-1 (tm1860) mutant. 26G-O1, an extremely abundant ERGO-1 class 26G RNA, is present at roughly 0. 5% wild-type levels in the ergo-1 (tm1860) mutant, but still abundant enough to detect by northern blotting. Consistent with our hypothesis that ERGO-1 is required for 26G RNA methylation, we found that 26G-O1 is unmethylated in the ergo-1 (tm1860) mutant embryo (Figure 4A). We next asked the converse question: Is 26G RNA methylation required for association with ERGO-1? We immunopurified ERGO-1 complexes from wild-type and henn-1 mutant embryo lysates (Figure 4B) and extracted RNA. In both wild-type and henn-1 mutant samples, ERGO-1 class 26G RNAs are readily detected (Figure 4C), indicating that ERGO-1 effectively binds both methylated and unmethylated 26G RNAs. Taken together, these data suggest that 26G RNAs bind ERGO-1 and are subsequently methylated by HENN-1. To test whether HENN-1-mediated methylation is required to maintain levels of all substrate small RNAs, we assessed ERGO-1 class 26G RNAs for defects in accumulation in the henn-1 mutant. Loss of henn-1 has more severe consequences for this class of small RNAs than are observed for 21U RNAs: ERGO-1 class 26G RNA 26G-O3 fails to accumulate to wild-type levels at any stage of development, although the disparity is less pronounced in adulthood, during peak 26G RNA biogenesis (Figure 5A). For comparison, we assayed levels of ALG-3/ALG-4 class 26G RNA 26G-S5 across the developmental window during which it is readily detected by Taqman RT-qPCR. Levels of 26G-S5 are similar in the henn-1 mutant relative to wild-type (Figure 5B), consistent with the idea that HENN-1 is required for accumulation of ERGO-1 class 26G RNAs but dispensable for that of ALG-3/ALG-4 class 26G RNAs. Analysis of seven additional ERGO-1 class 26G RNAs and two additional ALG-3/ALG-4 class 26G RNAs corroborated these observations (Figures S8, S9). To determine the effect of loss of henn-1 on the silencing of ERGO-1 class 26G RNA targets, we assayed levels of a panel of mRNAs targeted by ERGO-1 class 26G RNAs for desilencing in henn-1 mutant animals. During time points at which ERGO-1 class 26G RNAs are abundant, only modest upregulation of some, but not all, targets was detected; furthermore, no single target shows consistent desilencing in the henn-1 mutant (Figure 5C, Figure S10A). This is not unexpected, however, as the targets themselves vary in both expression and sensitivity to small RNA-mediated silencing across development [40]. To determine the specificity of this effect, two non-targets were examined in parallel. The maximal upregulation for either non-target does not exceed the maximal upregulation observed for any target, suggesting that the upregulation of ERGO-1 class 26G RNA targets in the henn-1 mutant may be a consequence of 26G RNA depletion (Figure 5C, Figure S10B). This connection is supported by our observation that a Wago-dependent and ERGO-1 class 26G RNA-dependent secondary siRNA that presumably enhances target silencing also shows defects in accumulation in embryo (Figure S6B). The effect is modest, indicating that, as observed for the piRNA pathway, the depleted pool of ERGO-1 class 26G RNAs in the henn-1 mutant is still sufficient for triggering fairly robust production of secondary siRNAs. Nevertheless, in an accompanying manuscript, Montgomery et al. observe that HENN-1 is required for silencing activity of a similar secondary siRNA upon a sensor transgene [54], suggesting that this pathway may indeed be compromised by loss of henn-1. ALG-3/ALG-4 class 26G RNAs are restricted to the male germline, and their mRNA targets are enriched for genes involved in spermatogenesis [40]. Accordingly, loss of ALG-3/ALG-4 class 26G RNAs results in male-associated sterility at non-permissive temperatures due to defects in sperm activation that are thought to arise from target dysregulation [41]. ERGO-1 class 26G RNAs, in contrast, are dispensable for fertility and target mostly poorly conserved and incompletely annotated genes, many of which reside in duplicated regions of the genome [42]. It is therefore not unexpected that the ergo-1 (tm1860) mutant, which lacks ERGO-1 class 26G RNAs, exhibits no overt phenotypes that can be traced to target dysregulation. Rather, the ergo-1 (tm1860) mutant exhibits an enhanced RNAi sensitivity (Eri) phenotype that is attributed to effects of loss of the ERGO-1-dependent small RNAs themselves; presumably, depletion of ERGO-1 class 26G RNAs and dependent secondary siRNAs liberates limiting RNAi factors shared between the endogenous and exogenous RNAi pathways [43], [55], [56]. To determine whether loss of henn-1 depletes ERGO-1 class 26G RNAs sufficiently to produce an Eri phenotype, as observed in the ergo-1 mutant, we subjected L1 larvae from a panel of strains to feeding RNAi targeting various genes in the soma or germline. In order to expose subtle differences in RNAi sensitivity, we modulated the degree of knockdown, attenuating the dose of dsRNA trigger by diluting the bacterial RNAi clone with a bacterial clone harboring empty vector. RNAi of the somatic gene lir-1 causes larval arrest and lethality in wild-type animals at full strength, but dilution 1∶1 with empty vector largely eliminates the effect. In contrast, the eri-1 (mg366) mutant, which lacks 26G RNAs, is affected severely by even dilute lir-1 RNAi. The henn-1 mutant also shows dramatically increased sensitivity to lir-1 feeding RNAi relative to wild-type (Figure 6A, 6B). A henn-1; eri-1 double mutant, however, shows RNAi sensitivity that is virtually identical to that of the single eri-1 mutant, suggesting that the Eri phenotype of each allele likely stems from the same defect, namely, loss of ERGO-1 class 26G RNAs. While the somatic Eri phenotype of the henn-1 mutant shows partial rescue by the germline-specific henn-1: : gfp transgene xkSi2, henn-1: : gfp expression under the native promoter from transgene xkSi1 rescues wild type RNAi sensitivity completely in the henn-1 mutant (Figure 6B). These findings suggest that loss of henn-1 in both germline and soma contributes to the Eri phenotype of the henn-1 mutant. The henn-1 mutant exhibits a similar somatic Eri response to RNAi of dpy-13 and lin-29 (Figure S11). While the somatic Eri phenotype of the henn-1 mutant was expected, knockdown of genes required for germline development or embryogenesis revealed that, incongruously, the henn-1 mutant maternal germline exhibits an RNAi defective (Rde) phenotype. Animals subjected to pos-1 RNAi lay dead embryos because maternally loaded pos-1 mRNA is required for specifying cell fate of many tissues during embryonic development [57]. On pos-1 RNAi diluted 1∶2 with empty vector (1/3 strength), knockdown in wild-type animals is still sufficiently robust to reduce average brood size to fewer than five offspring per animal. henn-1 mutant animals at this dilution, however, produce an average brood greater than tenfold that of wild-type, suggesting that loss of henn-1 confers resistance to RNAi-mediated knockdown of this maternally deposited mRNA (Figure 6C). A lesser but statistically significant effect was observed for RNAi of the germline-expressed transcripts par-1, par-2, pie-1, and glp-1 (Figure S12). Sensitivity to pos-1 RNAi is effectively rescued by either endogenous or germline-specific expression of henn-1: : gfp, likely due to the fact that both transgenes are expressed in germline. HEN1 orthologs appear to be restricted to the germline in vertebrates [15], [33]; however, we observe phenotypes in both the germline and soma of the henn-1 mutant that suggest broader activity. To investigate expression of HENN-1 in C. elegans, we assessed henn-1 mRNA and protein levels throughout development. henn-1 mRNA levels are lowest in young larva and increase as the germline proliferates, peaking in gravid adult (Figure 7A, Figure S13A). Germline-deficient glp-4 (bn2) adult hermaphrodites show approximately a 50% reduction in henn-1 mRNA levels relative to wild-type (Figure S13B), indicating that henn-1 mRNA is expressed in both germline and soma. Embryonic levels of henn-1 are high but decrease rapidly; this pattern suggests that, unlike in zebrafish [15], henn-1 mRNA may be maternally deposited into the embryo. HENN-1 protein is detectable throughout development and in both hermaphrodite and male adults (Figure 7B). We next assessed the distribution of HENN-1: : GFP fusion protein expressed from xkSi1, the rescuing henn-1: : gfp transgene driven by the endogenous promoter, in the henn-1 mutant background. Although single copy transgene expression levels are too low for direct visualization by fluorescence microscopy, HENN-1: : GFP is readily detected using a mouse monoclonal anti-GFP antibody. Whole-mount immunostaining of transgenic L4 larvae reveals that HENN-1: : GFP is expressed broadly in diverse somatic tissues and germline (Figure S13C). Non-transgenic larvae show no signal, indicating that detection of HENN-1: : GFP is specific. In extruded gonads of xkSi1; henn-1 hermaphrodites, HENN-1: : GFP is detected throughout the germline. Notably, the proximal oocytes show cytoplasmic and intense nucleoplasmic HENN-1: : GFP expression (Figure 7C). Although nucleoplasmic enrichment is lost following fertilization, HENN-1: : GFP is also abundant in embryo, with ubiquitous expression prior to gastrulation (Figure S13D). HENN-1: : GFP is also expressed throughout the germline of xkSi1; henn-1 males (Figure 7D). During sperm maturation, we detect enrichment of HENN-1: : GFP in residual bodies, but we cannot definitively conclude that it is excluded from sperm (Figure 7D, inset). In wild-type animals, studies of endogenous HENN-1 using the rabbit polyclonal antibody generated against an N-terminal HENN-1 epitope corroborate the above findings, although the signal is more difficult to detect (Figure 7E). Staining in the henn-1 mutant yields no signal for anti-GFP and anti-HENN-1 antibodies (Figure 7F); this demonstrates that detection of transgenic and endogenous HENN-1 proteins is specific. Together, these data define an expression pattern consistent with a role for HENN-1 in modifying small RNAs in both male and female germlines as well as in soma. The 21U RNAs and 26G RNAs appear to be significantly stable only in the presence of their respective Argonaute proteins [35], [36], [40]; accordingly, the localization patterns of the Argonaute proteins reflect the distribution of the different classes of small RNAs. We therefore wanted to compare the expression patterns of HENN-1 and the 26G RNA-binding Argonautes to determine whether the small RNA substrate specificity of HENN-1 could be explained by differential access to Argonaute-bound small RNAs. ERGO-1, which binds methylated 26G RNAs, is abundant in embryo [42], and its transcript is enriched during oogenesis [58], but its localization has not yet been reported. We assessed the staining pattern of ERGO-1 in hermaphrodite gonad and embryo using a polyclonal antibody generated against a C-terminal ERGO-1 epitope. ERGO-1 expression in the hermaphrodite germline begins at pachytene exit and persists in embryo (Figure S13D, S13E). ERGO-1 shows cytoplasmic enrichment both in germline and embryo, suggesting that the cytoplasmic pool of HENN-1 may act in methylating 26G RNAs bound by ERGO-1. This interaction may, however, be transient, as we were unable to identify HENN-1 by mass spectrometry of immunopurified ERGO-1 complexes, nor could we detect ERGO-1 in immunopurified HENN-1: : GFP complexes by western blot (data not shown). Notably, both HENN-1 and ERGO-1 remain abundant in early embryo (Figure S13D). This is consistent with the proposed existence of a somatic endo-siRNA pathway that promotes continued biosynthesis of ERGO-1 class 26G RNAs after fertilization [59]. We next assessed co-localization of HENN-1 and ALG-3. ALG-3 and its close paralog, ALG-4, bind unmethylated 26G RNAs, and their transcripts are enriched during spermatogenesis [58]. In the male gonad, a rescuing gfp: : alg-3 transgene was reported to express in the proximal male germline, with localization to P granules beginning at late pachytene [41]. During sperm maturation, GFP: : ALG-3 is relegated to residual bodies. Dual immunostaining of GFP: : ALG-3 and endogenous HENN-1 demonstrates a large region of overlap (Figure S13F), but HENN-1 does not appear to localize to P granules. This does not explain why ALG-3/ALG-4 class 26G RNAs are not methylated, because it is likely that HENN-1 can access P granules transiently: PRG-1 localizes predominantly to P granules [35], [37], and the PRG-1-bound piRNAs are methylated. This is in contrast to zebrafish Hen1, which carries a poorly conserved C-terminal domain (Figure S1) that directs localization of Hen1 to nuage, perinuclear granules similar to C. elegans P granules, to methylate piRNAs [15]. We have shown that HENN-1 is essential for methylating select classes of C. elegans small RNAs, namely, 21U RNAs and ERGO-1 class 26G RNAs. As is the case in other animals, small RNAs in C. elegans that associate with Piwi clade Argonautes require HENN-1 for maintenance of wild-type levels. Ago clade Argonaute-associated microRNAs and ALG-3/ALG-4 class 26G RNAs, in contrast, are HENN-1-independent (Figure S14A). It has been proposed that spatial and temporal regulation of HEN1 ortholog expression may contribute to small RNA substrate specificity in metazoans [24]. However, our immunostaining studies indicate that HENN-1 is coexpressed in the same tissues and subcellular compartments as Argonautes ERGO-1, PRG-1, and ALG-3 and their respective small RNAs (Figure 7, Figure S13). Therefore, differences in gross sub-cellular localization cannot explain the failure of ALG-3/ALG-4 class 26G RNAs to be methylated. Furthermore, although the two subclasses of 26G RNAs are generated in different germlines from non-overlapping targets, their sequences exhibit no obvious distinguishing characteristics that might account for their non-uniform methylation status. One model of small RNA methylation posits that animal HEN1 orthologs only methylate small RNAs bound by Argonautes [15], [22]–[24], [49]. In support of this, work in fly shows that siRNA methylation requires assembly of DmAgo2 RISC [22], [50], and in vitro studies using lysate from a silkworm ovary-derived cell line show that methylation of synthetic RNA only occurs after the longer substrate is bound by a Piwi protein and trimmed to piRNA size [60]. This model predicts that all 26G RNAs are bound as unmethylated species by either ERGO-1 in the female germline or ALG-3/ALG-4 in the male germline and subsequently methylated or not, respectively. This is consistent with our findings in vivo that ERGO-1 is required for methylation of 26G RNAs (Figure 4A) and associates with 26G RNAs of either methylation status (Figure 4C). It has been further proposed that the identity of the Argonaute determines whether bound small RNAs are methylated [22], [23], [49], [50]. An elegant illustration of this is provided by fly miR-277, which associates with both Ago1, the canonical fly miRNA Argonaute, and Ago2, which binds methylated siRNAs [61]. The miR-277 pool contains both methylated and unmethylated species. Depletion of Ago2 in cell culture results in loss of methylated miR-277, whereas Ago1 depletion results in a completely methylated miR-277 population [22]. Similarly, fly hairpin derived hp-esiRNAs sort into Ago1 and Ago2, but accumulate mainly in Ago2 because only hp-esiRNAs bound by Ago2 are methylated and therefore protected against degradation triggered by their extensive target complementarity [50]. In C. elegans, the model of Argonaute-dictated methylation can be invoked to explain the disparate methylation of the 26G RNAs: in the male germline, only ALG-3/ALG-4 are expressed, resulting in an unmethylated male 26G RNA population, whereas exclusive expression of ERGO-1 in the female germline and embryo directs methylation of female and zygotic 26G RNAs. This raises the intriguing possibility that selective expression of Argonautes that permit or prevent methylation could represent a new mechanism for differentially regulating small RNA turnover. It is important to note that our results do not definitively exclude an alternative model wherein 26G RNAs are methylated prior to association with Argonautes and subsequently bound by ALG-3/ALG-4 only if unmethylated or by ERGO-1 only if methylated. In this model, HEN1 would methylate 26G RNAs in both germlines, but degradation of labile unbound siRNAs would result in a purely unmethylated or methylated population of 26G RNAs in male and female germlines, respectively. Because 26G RNAs assessed in embryo are fully methylated (Figure 3A, Figure S3B), such a mechanism would require that ERGO-1 exhibit very unfavorable kinetics for association with unmethylated small RNAs. We do not find this to be the case, as ERGO-1 binds some 26G RNAs with similar efficiency when methylated and unmethylated (Figure 4C). Our data therefore provide stronger evidence for a model of Argonaute-dictated methylation of small RNAs. Differential germline expression of Argonautes could have evolved in C. elegans because of advantages conferred by selective stabilization of female germline 26G RNAs. Unlike ALG-3/ALG-4 class 26G RNAs, which appear to function exclusively during sperm development [40], [41], ERGO-1 class 26G RNAs exert much of their influence during embryonic and larval development, well beyond initiation of their biogenesis in the hermaphrodite germline [40]. Accordingly, their targets are depleted of germline-enriched genes [40], [59]. The oocyte contributes the vast majority of the initial zygotic cellular contents; therefore, methylation of 26G RNAs originating in the female germline may ensure robust inheritance and perdurance of primary small RNAs. Methylation of 26G RNAs in the male germline would likely not significantly increase their representation in sperm or zygote, as ALG-3/ALG-4 are relegated to residual bodies during spermatogenesis and exert effects in mature sperm only indirectly through dependent secondary 22G RNAs [41]. Nonetheless, it would be interesting to express ERGO-1 ectopically in sperm and determine whether ALG-3/ALG-4 class small RNAs are methylated. Such a strategy may reveal unexpected consequences related to inappropriate methylation and stabilization of ALG-3/ALG-4 class 26G RNAs. In the absence of henn-1, we show that response to RNAi-mediated knockdown is enhanced for somatic genes (Figure 6A and 6B, Figure S11). This is likely due to destabilization of ERGO-1 class 26G RNAs in the henn-1 mutant, which reduces competition with primary exo-siRNAs for stimulating secondary siRNA activity mediated by somatic Argonautes such as SAGO-1 and SAGO-2 [43], [55]. While germline-specific expression of henn-1: : gfp only partially rescues this somatic Eri phenotype, henn-1 mutant animals rescued with an endogenous henn-1: : gfp transgene, which drives both somatic and germline expression, show wild-type RNAi sensitivity. Under the model of competing endo- and exo-RNAi pathways, this suggests that HENN-1-mediated methylation of ERGO-1 class 26G RNAs in the germline alone cannot maintain small RNA levels sufficient to sequester an appropriate proportion of the limiting RNAi factors. It is possible that ERGO-1 class 26G RNA biogenesis continues in embryo and larva, as previously suggested [59], and that high concentrations of HENN-1 are necessary for continued stabilization of these small RNAs. Such a model would be consistent with our characterization of the distributions of HENN-1 and ERGO-1, both of which are still detected in abundance in developing embryo (Figure S13D, S13E). While the majority of the phenotypes observed in the henn-1 mutant can be attributed to destabilization of endogenous small RNA substrates, the germline Rde phenotype suggests a role for HENN-1 in exogenous RNAi. It is unclear why HENN-1 is dispensable for robust exogenous RNAi in the soma but required in the germline. While this may be an indirect effect, as suggested in concurrent work by Kamminga et al. [62], one possible explanation is that HENN-1 stabilizes primary exo-siRNAs or dependent 22G secondary siRNAs. There is support in fly for methylation of exo-siRNAs and transgenic hairpin-derived siRNAs [22], [63], but this has not yet been demonstrated in C. elegans. 22G RNAs triggered by primary exo-siRNAs appear not to be methylated [47], consistent with our and others' observations that Wago-dependent 22G RNAs from diverse endogenous sources are unmethylated (Figure 3A, Figure S3B, and [45]). The methylation status of worm primary exo-siRNAs has not been definitively established, although a 22-nucleotide siRNA generated from a transgene encoding a perfect hairpin was not found to be methylated [46]. All Argonautes contain two signature domains, PAZ and Piwi [64]. The Piwi domain, unique to Argonautes, adopts an RNase H-like configuration and serves as the catalytic core of RISC [65], [66]. The PAZ domain recognizes and anchors the 3′ end of the small RNA [67], [68]. Comparison of Piwi and Ago clade Argonautes reveals that Piwi proteins contain a small insertion in their PAZ domains in a loop connecting two β strands [69]. Crystal structures of a human Piwi Argonaute PAZ domain suggest that this insertion results in the formation of a more spacious binding pocket capable of accommodating the 2′-O-methyl group of a piRNA. Interactions between the methyl group and hydrophobic residues lining the pocket confer a threefold to sixfold higher binding affinity for 2′-O-methyl than 2′-OH [69]. In C. elegans, only PRG-1/PRG-2 and ERGO-1 show evidence of a PAZ domain insertion (Figure S14B), consistent with their designation as Piwi clade Argonautes and association with methylated small RNAs. In spite of their shared classification, ERGO-1 exhibits far less homology than PRG-1/PRG-2 to mammalian and insect Piwi proteins (Figure S14A) [43]. Similarly, among worm, fly, and human Argonautes, DmAgo2 and C. elegans Argonaute RDE-1 are among the most divergent members of their clades [43]. In fact, so divergent is RDE-1 that its cladistics are ambiguous, with our and other published alignments variably assigning it to each of the three clades (Figure S14A and [43], [70]). Both DmAgo2 and RDE-1 bind exo-siRNAs, although only the former has been shown to permit methylation [22]. Interestingly, both lack the insertion found in Piwi Argonaute PAZ domains (Figure S14B). The absence of this insertion in DmAgo2 suggests that it is not required for association with methylated small RNAs, raising the possibility that RDE-1 too may permit methylation of associated small RNAs. If HENN-1 does not methylate RDE-1-bound small RNAs, it is unclear what specific role HENN-1 plays in exo-RNAi in the germline. Nevertheless, its dual functions in endogenous and exogenous RNAi place HENN-1 in the company of DCR-1 and the Wago proteins at the intersection between these two RNAi pathways. C. elegans were maintained according to standard procedures. The Bristol strain N2 was used as the standard wild-type strain. The alleles used in this study, listed by chromosome, are: unmapped: neIs23[unc-119 (+) GFP: : ALG-3]; LGI: glp-4 (bn2), prg-1 (tm872); LGII: xkSi1[PC30A5. 3: : henn-1: : gfp: : henn-1 3′UTR cb-unc-119 (+) ] II, xkSi2[Ppie-1: : henn-1: : gfp: : tbb-2 3′UTR cb-unc-119 (+) ] II; LGIII: rde-4 (ne301), henn-1 (tm4477); LGIV: eri-1 (mg366), fem-1 (hc17), him-8 (e1489); LGV: ergo-1 (tm1860). The neIs23[unc-119 (+) GFP: : ALG-3] strain was generously provided by Craig Mello (University of Massachusetts, Worcester, MA). For embryo samples, L1 larvae were grown at 20°C until gravid. Embryos were isolated using sodium hypochlorite solution; an aliquot of embryos was allowed to hatch overnight at room temperature to determine viability. For male samples, synchronized him-8 (e1489) L1 larvae were grown at 20°C for 72–75 hours. Males were isolated by filtering through 35 µm mesh [71]. For female samples, synchronized fem-1 (hc17) L1 larvae were plated and grown at 25°C for 52 hours. For time course samples, synchronized wild-type (N2) and henn-1 (tm4477) L1 larvae were grown at 25°C until gravid; embryos were extracted and harvested for RNA or hatched overnight at room temperature and then grown at 25°C for the specified number of hours before harvest. The prg-1 (tm872) time course samples were prepared in the same way, except that animals were grown for the first generation at 20°C to evade temperature-sensitive sterility. Samples were processed by either three rounds of freeze/thaw lysis or two rounds of homogenization for 15 sec using the Tissue Master-125 Watt Lab Homogenizer (Omni International) and the RNA was extracted in TriReagent (Ambion) following the vendor' s protocol, with the following alterations: RNA was precipitated in isopropanol for one hour at −80°C; RNA was pelleted by centrifugation at 4°C for 30 min at 20,000× g; the pellet was washed three times in 75% ethanol; the pellet was resuspended in water. For detection of small RNAs, 10 or 40 µg of total RNA were β-eliminated as described [51]; control samples were processed in parallel without sodium periodate. Northern blot analysis was performed as described [72]. In brief, 5 or 10 µg of β-eliminated total RNA were resolved on 17. 5% or 20% denaturing Urea-PAGE gels (SequaGel, National Diagnostics) and transferred to Hybond-NX membrane (Amersham). 21 and 26 nt synthetic RNAs were run as size markers and visualized in tandem with rRNA by ethidium bromide staining. Pre-hybridization/hybridization and washes were performed at 48°C or 50°C. Oligonucleotides corresponding to the antisense sequences of the small RNAs (Table S1) were synthesized and end-labeled with [α-32P]-dATP using the miRNA StarFire kit (Integrated DNA Technologies). To test the response to exogenous RNAi, bacterial clones from the Ahringer RNAi library [73] were diluted with bacteria harboring the empty vector L4440 to achieve a level of RNAi sensitivity that allowed us to differentiate the RNAi responses in the strains examined. To determine lir-1 RNAi sensitivity, the lir-1 RNAi bacterial clone diluted with L4440 bacterial clone at a 1∶1 or 1∶2 ratio (1/2 or 1/3 strength) was used; >50 L1 larvae were plated per plate and the number of total animals assayed per plate was determined at day two after plating; the percent of animals exhibiting the larval arrest phenotype was determined at 70 hours at 20°C. Sensitivity to RNAi of dpy-13 and lin-29 was also assessed using this method, where animals subjected to dpy-13 RNAi were imaged at 70 hours and those subjected to lin-29 RNAi were evaluated for the absence of protruding vulva or bursting phenotype. For pos-1 RNAi, synchronized L1 larvae were singled onto plates with pos-1 RNAi diluted with empty vector at a 1∶2 ratio (1/3 strength) that had been induced overnight at 25°C. Animals were grown at 20°C for six days and progeny were counted. Sensitivity to RNAi of pie-1, par-1, and par-2 was assessed similarly at the indicated dilutions with 4 plates of 4 P0 animals per strain. Sensitivity to glp-1 RNAi was determined at the indicated dilutions by plating 4 plates of >50 L1 larvae per strain per gene and scoring for the absence of oocytes and embryos in both arms of the germline at 70 hours at 20°C. For all RNAi sensitivity assays, data are representative of at least two independent experiments. To determine brood size, synchronized L1 larvae from gravid adults grown at 20°C or shifted to 25°C for two generations were singled onto plates with OP50 and grown to adulthood at their respective temperatures. Once egg-laying began, animals (N≥13 per strain) were transferred to fresh plates daily until the supply of fertilized eggs was exhausted. Progeny of the singled parents were counted as late larvae/adults. Results are representative of two independent experiments. Taqman small RNA probes were synthesized by Applied Biosystems (Table S2) [74]. For each reaction, 50 ng of total RNA were converted into cDNA using Multiscribe Reverse Transcriptase (Applied Biosystems). The resulting cDNAs were analyzed by a Realplex thermocycler (Eppendorf) with TaqMan Universal PCR Master Mix, No AmpErase UNG (Applied Biosystems). We could not identify a small RNA whose levels were consistent across development for use in normalization. Therefore, to preserve the developmental profile of each of the small RNA assessed, back transformation was used to calculate relative small RNA levels from qRT-PCR cycle numbers. As a control for RNA quality, miR-1 Taqman assays were run in parallel for all samples excluding the ERGO-1 RNA immunoprecipitation samples, in which miRNAs are absent. For quantification of mRNA levels, 100 ng of total RNA were converted into cDNA with Multiscribe Reverse Transcriptase (Applied Biosystems) following the vendor' s protocol with the following changes: 25 units of RT and 7. 6 units of RNAse OUT (Invitrogen) were used per reaction. cDNAs were analyzed using Power Sybr Green PCR Master Mix (Applied Biosystems) (primers, Table S3). Relative mRNA levels were calculated based on the ΔΔ2Ct method [75] using eft-2 for normalization. For all qPCR, 40 cycles of amplification were performed; reactions whose signals were not detected were therefore assigned a cycle number of 40. All results presented are the average values of independent calculations from biological triplicates unless indicated. To determine average upregulation of ERGO-1 26G RNA targets in henn-1 relative to wild-type (Figure 5C), the mean was calculated for all of the ratios generated by dividing each henn-1 biological replicate by each wild-type biological replicate. 3′ RACE was performed using the 3′ RACE System for Rapid Amplification of cDNA ends (Invitrogen) according to the manufacturer' s protocol. henn-1 gene-specific primer (5′ GCAGTATGTCGCCTCCAAGTAGAT 3′) was used to amplify henn-1 3′ ends from cDNA generated from embryo. Product corresponding to only the seven-exon gene model of henn-1 was observed, consistent with detection of a single protein isoform corresponding to this model on western blot analysis. The endogenous henn-1: : gfp reporter construct (xkSi1) was generated by introducing the following fragments into pCFJ151: endogenous promoter of the henn-1-containing operon CEOP3488 [76] (3. 9 kb PCR fragment immediately upstream of the C30A5. 3 start codon), henn-1 genomic coding region (1. 8 kb PCR fragment with mutated termination codon), gfp coding region (0. 9 kb fragment with multiple synthetic introns and termination codon), and henn-1 endogenous 3′UTR (1. 1 kb PCR fragment immediately downstream of henn-1 termination codon). The germline-only henn-1: : gfp reporter construct (xkSi2) was generated as above with the following substitutions: CEOP3488 operon promoter was replaced with the pie-1 promoter (2. 4 kb PCR fragment immediately upstream of pie-1 start codon) and henn-1 endogenous 3′UTR was replaced with the C36E8. 4 3′UTR (0. 3 kb PCR fragment downstream of C36E8. 4). Constructs were cloned into the pCFJ151 vector, confirmed by sequencing, and used to generate single-copy integrated transgenes via the MosSCI technique [52]. Gene fusion products of the expected size were specifically detected by western blot with both anti-HENN-1 and anti-GFP antibodies. Synthetic antigenic peptides were conjugated to KLH and each was used to immunize two rabbits (Proteintech). Antisera were subsequently affinity purified using Affi-Gel 15 gel (Bio-Rad). Antigenic peptide sequences are as follows: N-terminal HENN-1 peptide with N-terminal added cysteine (CTYVEAYEQLEIALLEPLDR), C-terminal ERGO-1 peptide (CEVNKDMNVNEKLEGMTFV). Proteins immobilized on Immobilon-FL transfer membrane (Millipore) were probed with anti-HENN-1 rabbit polyclonal antibody (1∶2000), anti-γ-tubulin rabbit polyclonal antibody (LL-17) (Sigma) (1∶2000), or anti-ERGO-1 rabbit polyclonal antibody (1∶1000). Peroxidase-AffiniPure goat anti-rabbit IgG secondary antibody was used at 1∶10000 (Jackson ImmunoResearch Laboratories) for detection using Pierce ECL Western Blotting Substrate (Thermo Scientific). Wild-type, henn-1, or eri-1 (mg366) embryos isolated from gravid adults grown at 20°C were frozen in liquid nitrogen and homogenized with a Mixer Mill MM 400 ball mill homogenizer (Retsch) Homogenates were suspended in lysis buffer (50 mM HEPES (pH 7. 4), 1 mM EGTA, 1 mM MgCl2,100 mM KCl, 10% glycerol, 0. 05% NP-40 treated with a Complete, Mini, EDTA-free Protease Inhibitor Cocktail tablet (Roche Applied Sciences) ) and clarified by centrifugation at 12,000× g for 12 minutes at 4°C. Aliquots of homogenate were reserved as crude lysate for western blot to confirm that immunoprecipitations were performed in lysates of equivalent protein concentration (2 mg/mL). For immunoprecipitations, embryo homogenates were incubated at 4°C for one hour with 75 µg anti-ERGO-1 rabbit polyclonal antibody conjugated to Dynabeads Protein A (Invitrogen), after which the beads were washed (500 mM Tris-HCl (pH 7. 5), 200 mM KCl, 0. 05% NP-40) and associated proteins were eluted with 200 µL glycine. Three quarters of each eluate were precipitated overnight at 4°C in trichloroacetic acid, pelleted, washed with acetone, and resuspended for western blot analysis. The remaining eluate was treated with 2 mg/ml Proteinase K (Roche) and incubated at 37°C for 30 minutes. RNA was isolated from the eluate by incubation with TriReagent and processed as described above. RNA pellets were resuspended in 10 µL water and 5 µL were used for each Taqman RT reaction. Primary antibodies were applied according to the following specifications: anti-GFP mouse monoclonal antibody 3E6 (Invitrogen) was diluted 1∶1500 to detect HENN-1: : GFP and 1∶200 to detect ALG-3: : GFP; anti-ERGO-1 rabbit polyclonal was diluted 1∶200; and anti-HENN-1 rabbit polyclonal antibody was preabsorbed as described [77] with henn-1 (tm4477) mutant extract and diluted 1∶200. Alexa Fluor 555 goat anti-rabbit IgG and Alexa Fluor 488 goat anti-mouse IgG (Molecular Probes) secondary antibodies were diluted 1∶500. All antibodies were diluted in 0. 5% bovine serum albumin (Sigma). For immunostaining of gonads and embryos, synchronized gravid hermaphrodites or adult males grown at 20°C were dissected on Superfrost Plus positively charged slides (Fisherbrand) with 27 G×1/2 inch BD PrecisionGlide needles (Becton, Dickinson and Company) as described by Chan and Meyer in WormBook [78] Protocol 21 with 1. 5% paraformaldehyde (Sigma). Slides were incubated with primary antibodies overnight at 4°C and with secondary antibodies for three hours at room temperature. Slides were mounted with VECTASHIELD Mounting Medium with DAPI (Vector Laboratories). For whole-worm immunostaining, synchronized late L4 larvae grown at 20°C were transferred to subbed slides [77] in M9, fixed for six minutes in 1. 5% paraformaldehyde, freeze-cracked, and incubated for 15 minutes in ice cold methanol. After fixation, slides were processed as above. Images were captured on an Olympus BX61 epifluorescence compound microscope with a Hamamatsu ORCA ER camera using Slidebook 4. 0. 1 digital microscopy software (Intelligent Imaging Innovations) and processed using ImageJ.

Small RNAs serve as sentinels of the genome, policing activity of selfish genetic elements, modulating chromatin dynamics, and fine-tuning gene expression. Nowhere is this more important than in the germline, where endogenous small interfering RNAs (endo-siRNAs) and Piwi-interacting RNAs (piRNAs) promote formation of functional gametes and ensure viable, fertile progeny. Small RNAs act primarily by associating with effector proteins called Argonautes to direct repression of complementary mRNAs. HEN1 methyltransferases, which methylate small RNAs, play a critical role in accumulation of these silencing signals. In this study, we report that the 26G RNAs, a class of C. elegans endo-siRNAs, are differentially methylated in male and female germlines. 26G RNAs derived from the two germlines are virtually indistinguishable, except that they associate with evolutionarily divergent Argonautes. Our data support a model wherein the methylation status and, consequently, stability of a small RNA are determined by the associated Argonaute. Therefore, selective expression of Argonautes that permit or prohibit methylation may represent a new mechanism for regulating small RNA turnover. As we observe this phenomenon in the germline, it may be particularly pertinent for directing inheritance of small RNAs, which can carry information not encoded in progeny DNA that is essential for continued transgenerational genome surveillance.

Abstract Introduction Results Discussion Materials and Methods

rna interference gene regulation rna stability animal models caenorhabditis elegans model organisms epigenetics molecular genetics gene expression biology molecular biology rna rna processing nucleic acids gene identification and analysis genetics molecular cell biology genetics and genomics

2012

The Caenorhabditis elegans HEN1 Ortholog, HENN-1, Methylates and Stabilizes Select Subclasses of Germline Small RNAs

Hemoglobin is the prototypic allosteric protein. Still, its molecular allosteric mechanism is not fully understood. To elucidate the mechanism of cooperativity on an atomistic level, we developed a novel computational technique to analyse the coupling of tertiary and quaternary motions. From Molecular Dynamics simulations showing spontaneous quaternary transitions, we separated the transition trajectories into two orthogonal sets of motions: one consisting of intra-chain motions only (referred to as tertiary-only) and one consisting of global inter-chain motions only (referred to as quaternary-only). The two underlying subspaces are orthogonal by construction and their direct sum is the space of full motions. Using Functional Mode Analysis, we were able to identify a collective coordinate within the tertiary-only subspace that is correlated to the most dominant motion within the quaternary-only motions, hence providing direct insight into the allosteric coupling mechanism between tertiary and quaternary conformation changes. This coupling-motion is substantially different from tertiary structure changes between the crystallographic structures of the T- and R-state. We found that hemoglobin' s allosteric mechanism of communication between subunits is equally based on hydrogen bonds and steric interactions. In addition, we were able to affect the T-to-R transition rates by choosing different histidine protonation states, thereby providing a possible atomistic explanation for the Bohr effect. Hemoglobin (Hb) consists of two and two protein chains that bind oxygen to a heme group for the transport through the blood. These binding sites act cooperatively, i. e. the binding affinity for in one site is increased after binding in one of the other sites. The remarkable efficiency of Hb' s cooperative oxygen binding gave rise to extensive experimental and theoretical work [1]–[9]. Today, many high-resolution structures are available, including oxy states, deoxy states, carbon monoxide bound structures and structures of a large number of point mutations [10]. Dioxygen dissociation curves are recorded for many of the mutants, making it possible to see the effect of individual residues on the cooperativity. Dynamical information is obtained from e. g. spectroscopic studies, observing transition states in the oxy to deoxy transition, and analysing specific bonds during CO dissociation [11]. In addition to experiments, Molecular Dynamics (MD) simulations have provided insight for shorter timescales from down to ps while maintaining the full atomistic picture of Hb. The work of Shadrina et al. and of Lepeshkevich et al. focused on diffusion in Hb studied with MD [3], [4]. Ramadas and Rifkind simulated conformational changes due to perturbations of the heme pocket for methemoglobin dimers [5]. In the work by Mouawad et al. , the Hb T-to-R transition was enforced by restraining the Hb coordinates with decreasing structural distance to the R-state structure [6]. The study from Yusuff et al. focused on 100 ns simulations from different crystallographic structure models [9]. Recently, Hub and co-workers observed for the first time spontaneous reproducible transitions from the T- to the R-state during MD simulations [7]. They described a tendency for the -chains to couple more strongly to the quaternary motion than the -chains. In the present study we investigated how the local intra-chain motions couple to the global inter-chain motions on a molecular level. To this end, we enhanced the statistical basis to 21 transition trajectories, and developed a method which allows to characterize the coupling between global and local motions. For this purpose, we first separated global from local motions and then identified the coupling mechanism between them. We analysed the resulting coupling collective coordinate on the level of molecular contacts, shedding light on the molecular allosteric mechanism of hemoglobin. In addition, by using a different set of histidine protonations to mimick simulations at different pH, we were able to show a reduction in the number of transition trajectories. This constitutes a possible explanation for the Bohr effect in Hb [12], [13]. From the Hb simulations carried out by Hub and co-workers [7], the ones starting from the T-state with doubly protonated and thus positively charged (all other histidines neutral) showed transitions to the R-state in all three runs. In our study, we extended these simulations to improve the statistical basis of the T-to-R transitions. From a total of 50 simulations (200 ns long each; total simulation time) 22 showed a spontaneous transition. For the further steps, only transition trajectories were taken into account. Starting from our T-to-R transition trajectories, to analyse the interplay of local and global motions, we separated local from global motions as the first step. Here, the MD transition trajectories were decomposed into two trajectories: quaternary-only (Q) and tertiary-only (T). The first consists of inter-chain motions with the Hb chains translating and rotating as rigid bodies and the second contains intra-chain motions, omitting the global movements. The combination of the Q and T trajectories yields the full MD trajectories. For a visual explanation of the basic idea of the decomposition see Figure 1, for a detailed description see Materials and Methods. The two corresponding subspaces for Q and T are orthogonal by construction, but the actual motions inside them may still be correlated, thus reflecting the underlying allosteric mechanism in Hb. We therefore investigated if there was a coupling between local (T) and global (Q) motions. In other words: can we construct a linear combination of the T coordinates that is correlated to the Q motion? We simplified the Q trajectory by only considering the first eigenvector of a Principal Component Analysis (PCA) as the most dominant motion (referred to as cQ). For obtaining a collective coordinate within T that maximally correlates to cQ, we applied Functional Mode Analysis ( (FMA), [14]) based on Partial Least Squares [15]. In our case the projection on cQ was the functional property and T the coordinates that were used as a basis to cQ from. We assessed the risk of overfitting, arising from the high dimensionality of the T space by cross-validation. The resulting tertiary FMA model correlated to the cQ motion with a Pearson correlation coefficient of (see Figure 2). This means that despite the orthogonal nature of the two underlying subspaces, T and Q, we found a collective coordinate (named cT) within T that is strongly coupled to Q. This allows us to predict the quaternary state from the internal subunit coordinates alone, thereby providing insight into the allosteric mechanism of subunit communication through quaternary conformation changes. The cT coordinate is not similar to the tertiary difference vector of the T- and R-state – the scalar product of the two motions is 0. 01 –, and thus yields novel information based on the dynamics during quaternary transitions. Within the FMA framework it can be desirable to reweigh the individual latent vectors with their contribution to the overall variance (see [14]): While the FMA mode cT is the maximally correlated motion, which may actually be restricted, the ensemble-weighted mode cTew is the most probable motion that correlates with the functional property. For further analysis, we used this ensemble-weighted motion (cTew). To elucidate the mechanism of coupling between T and Q, we focused on the hyperplane spanned by T and Q. In this plane we chose a specific path starting from the T-state and moving first parallel to cQ and then parallel to cTew. The path is marked in white in Figure 3 and will be referred to as cQ-cTew. It artificially separates motions that are occurring simultaneously in the simulations. This allows us to classify the contacts according to their decomposition into global and local motions. In order for the information of a local conformation to flow from one protein chain to another, it has to cross the corresponding interface. It is therefore of interest to investigate interactions at the subunit interfaces. We focused on inter-chain van der Waals overlaps and general distance based contacts to investigate the interactions underlying the allosteric coupling mechanism. To estimate the influence of interatomic repulsion due to steric interactions as a driving force for the allosteric coupling, we calculated the overlap of atomic van der Waals (vdW) spheres. We did this for structures in the plane spanned by cQ and cTew, and took into account only inter-chain overlaps. (Note that these structure are projected onto this plane and hence differ from the actual MD structures.) The higher the overlap in a specific structure, the more energetically unfavourable it is. As can be seen in Figure 3, the overlaps are minimal along the main diagonal while increasing when moving orthogonally. Projecting the structures from the MD simulations (white dots) onto this plane shows that they coincide with the low vdW overlap region. For a broader view including also attractive interactions, we monitoredinter-chain atom pairs showing a distance smaller than 3 nm. This analysis was carried out along cQ-cTew which allows us to classify the contacts according to when the contact is or is not formed along this specific path. For specific interaction types following contact patterns are expected. The two residues in contact are either: Figure 4 depicts the contact analysis and classification at the example of the contact cluster around. In Table 1 all observed contacts are listed that fall in the first category. These contacts only stay intact if the system moves along cQ and cTew together, but break if moving in one or the other direction independently. This is the expected behaviour for contacts which must remain intact for the allosteric mechanism to function. Exemplarily, this was observed for and. The hydrogen bond between the carboxylic oxygen of Phe and the side chain of Arg breaks while moving from the T-state towards the off-diagonal intermediate artificial states (see Figure 3), and forms again when approaching the R-state. Contacts of the second category, which appear only while moving along cQ and cTew individually, are listed in Table 1. One scenario how these contacts could be leading to the allosteric mechanism is the residues getting too close when moving only along cQ or cTew. This could be the case for vdW overlaps or repulsive coulomb interactions. A clear example for this is the interaction of and we observed: Close to the T-state both side chains are pointing into the solvent. While moving along cQ, the two chains approach each other and bring both positively charged side chains unfavourably close. The motion along cTew relaxes this repulsive interaction by bending the N-terminal ends of the F helices (the helix notation goes back to Watson, Kendrew and Perutz [16]). Experimental studies introduced cross-links between the two lysines [17], [18]. The derived structure was described to be an intermediate between T- and R-state with characteristics of both states but no cooperativity. This is in accord with our analysis, from which we saw that a linker between the lysines would make the F helix bending impossible. If during the transition one residue switches an interaction partner, we expect to see the first contact disappearing and a contact with the new residue appearing. This was observed e. g. for the C-terminal. Its side chain interacts with the carboxyl group of, and switches along cQ-cTew so that a salt-bridge is formed between the Arg terminus and the side chain of. This event also has been seen in the symmetry-related counterpart independently. Further contacts of this type are listed in Table 1. We aimed to analyse the effect of the histidine protonation states on the transition probabilities to elucidate the possible pH-dependence underlying the Bohr effect. For this purpose, a second set of protonation states was simulated. We chose to use the protonation states as reported by Kovalevsky et al. , who used neutron protein crystallography to measure the protonation of histidine residues in the T-state [19]. Both histidine protonation states are listed in Table 2 with the Kovalevsky protonation states corresponding to a lower pH. In the case of the protonations used by Hub et al. , 13 out of 20 simulations showed a transition, while in the case of the protonation state described by Kovalevsky et al. , only 4 out of 20 simulations did (see Table 3). This suggests a clear dependence of the transitions on the histidine protonation state. We present a novel allosteric mechanism coupling quaternary and tertiary transitions in Hb. One fundamental component of the applied approach is a strict separation of local/tertiary and global/quaternary degrees of freedom. This guarantees that any observed coupling between both subspaces is not due to linear dependence of the respective basis vectors, but represents a true feature of the allosteric mechanism. The suggested separation algorithm is not limited to hemoglobin and can be applied to other systems with multiple chains. Also, the algorithm can be used for any definition of domains in the broader sense to separate the motions within the domains and between the domains. The second component of the presented approach, Functional Mode Analysis, allowed us to find a linear coupling coordinate between tertiary and quaternary motions and thereby identify the allosteric coupling. Applied to hemoglobin, FMA revealed a remarkable correlation between collective coordinates from quaternary-only and tertiary-only motions. The tertiary-only coupling mode is markedly different from the tertiary structure differences between the known crystallographic R- and T-states. Thus, this mode could not have been derived solely based on the X-ray structures, but yields novel information directly based on transition trajectories between the T- and R-state. For a direct comparison of the collective coordinates cQ, cT, cTew and the crystallographic T-R difference vector (and its quaternary-only and tertiary-only component), scalar products between the coordinates are summarized in Table 4. The correlation of for the model training (and for the cross-validation) between the quaternary mode and the detected tertiary mode is high, and allows us to predict Hb' s quaternary conformation for a given tertiary conformation with high accuracy. We were able to detect this coupling despite the fact that we did not take the full Q motions into account, but rather reduced the motions to the first PCA eigenvector cQ. In a future study, a more complete interaction picture may be derived by coupling the tertiary motions to the full 18 dimensional quaternary subspace. The high correlation between cQ and cT despite the near-zero overlap between the modes (scalar product of 0. 01) clearly indicates towards the strong coupling between tertiary and quaternary motions as basis for the cooperativity in Hb. The model that we used to describe the coupling of quaternary and tertiary motions is of linear nature. On the one hand, since there is no necessity for a linear coupling, non-linear models could be more suitable. On the other hand, the fact that we found a linear model with an substantial correlation in the cross-validation makes us confident that already a major part of the coupling can be described linearly. We investigated only instantaneous cQ to T coupling, although in reality there might be a lag-time due to the time required for the signal to pass. Our results, however, show that the coupling can already be identified from an analysis of instantaneous correlations. Future work may include non-linear models as well as delayed responses for the coupling, especially for systems in which the order of events is known. An analysis of the identified coupling mode focused on the protein chain interfaces revealed key interaction residues. We suggest that the allosteric coupling between local and global motions in Hb consists of an interplay of repulsive and attractive interactions at the subunit interfaces on a similar scale. The interaction picture we derived from our vdW overlap analysis suggests that vdW overlaps are a global driving force for the allosteric coupling. The fact that MD trajectories projected onto the plane spanned by cQ and cTew coincide with the region around the diagonal in Figure 3 that corresponds to low vdW overlap strongly points at the underlying coupling mechanism: With cTew derived to optimize the coupling between local and global motions, it also shows a strong coupling of steric repulsions between local and global motions. Nevertheless, efforts to break down the global interactions to individual repulsive contact pairs did not yield a conclusive picture. This suggests that the vdW overlap at Hb' s inter-chain interfaces does not act on a residue level, but on a broader, collective scale. The contact analysis along cQ-cTew allowed us to classify contacts according to their behaviour along this path. By picking this path in the plane spanned by cQ and cTew, we ensure that the observed contacts are important for the coupling. If any contact pair did not play a role in the allosteric coupling, it would not have been part of the coupling of global and local motions. For a number of contacts we observed also the symmetry-related residue-pairs (if not a contact of the same type, at least the contact itself), which further indicates the significance of these contact pairs. The similar number of pulling and pushing interface interactions suggests that both types contribute to the allosteric coupling on a similar scale. The fact that steric repulsions and hydrogen bonds could not be unambiguously traced down to a residue level individually, but could in the generalized contact analysis, points at an interplay of repulsive and attractive interactions. In our analysis the two residues and stay in contact along the full cQ-cTew path, that is close to R and T as well as in the off-diagonal intermediate artificial states. Hence, since the contact is not changing in the coupling space spanned by cQ and cTew, this salt-bridge does not seem to play a role in the coupling of local and global motions. Nevertheless, it was shown in Hb Montefiore that the mutation of to a Tyr breaks down the cooperativity [20]. Also, while the contact is present in the crystallographic T-state structure, it is broken by an outward flip of the Aspartates in the R-state. Further studies are needed to investigate why this contact pair did not show up in our analysis. The applied dimensionality reduction from Q to cQ may have caused this. Simulations of the Tyr mutant may reveal the underlying mechanism. Further, has been shown to play an important role during the quaternary transition [6], [21]. In our analysis this residue is in contact with, but behaved similarly to the contact and stayed intact along the cQ-cTew path. The neighbouring switch region residues and were, however, identified as ‘switching’ and ‘pushing’ contacts, respectively, confirming the crucial role of the switch region in the transition. In this study, we assigned different roles in the coupling mechanism (like forming hydrogen bonds or repulsion due to van der Waals overlap) to individual amino acids. Mutagenic studies of these amino acids provide a direct means to validate the predicted role of individual amino acids in the allosteric mechanism. The observed contacts which are caused by van der Waals overlap may be reduced by mutating to residues with smaller side chain sizes. In the case of charged side chains introducing an additional charge of the same sign may increase the repulsion. Contacts including hydrogen bonds can be suppressed by using unfavourable mutations. We suggest mutations affecting the “hinge” and “switch” contacts [11] in Table 1. Our observations extended this region by interaction partners in the R-state allowing to choose mutations affecting either the T- or R-state. By introducing a hydrogen bond donor the mutation may stabilize the R-state. The central role of the in this area makes it an interesting mutation site since it might separate the two timescales associated with the “hinge” and “switch” contacts as described by Balakrishnan et al. [11]. To explore the mutant we performed five simulations at 200 ns with this mutant and observed two quaternary transitions. Even though we originally expected this mutation to affect the transition rates, the fraction of transition trajectories does not significantly differ from our original simulations. A closer look at these preliminary simulations indicates that an additional hydrogen bond is indeed formed but not inter-subunit but rather with within the same chain and may thereby only contribute a minor effect to the transition rates. The repulsive interaction of the two can be explored by either a mutation to Arg or even by switching both charges by a mutation to Glu, keeping the repulsive Coulomb interaction. Further, we suggest a mutation of the to analyse the details of this salt-bridge by this conservative mutation, leaving the charges untouched and only changing the side chain length. In our simulations, we were able to lower the T-to-R transition probability by mimicking a pH shift, thereby providing a molecular picture of the Bohr effect. Future mutational studies may verify our predicted interactions and consolidate the molecular interaction mechanism of hemoglobin' s allosteric coupling. The measured histidine protonations in the T-state by Kovalevsky et al. [19] showed a large number of doubly protonated and thus positively charged side chains. By applying these protonations to our MD simulations we were able to observe significantly lowered transition probabilities from the T- to R-state as compared to the original setup with singly protonated histidines. This is consitent with the Bohr effect: a preference for the T-state at lower pH. Even though our simulations are not long enough to have reached equilibrium, our observations can be taken as a proof of concept of the electrostatic interactions stabilizing the T-state and thereby underlying the pH-driven Bohr effect. During the T-to-R transition residues at the and the interfaces get closer. Positively charged histidine residues at these interfaces add a repulsive Coulomb force, rendering the transition energetically more unfavourable. Interestingly, most of the histidines that are changed from neutral to positive when applying the T-state protonations (see Table 2), are located on the outside of Hb and not at chain interfaces. This might point at long-range interactions complementing the short-range interface interactions. A careful introduction of additional histidines by mutation – as pH sensitive switches – may increase the Bohr effect. Also, calculation of free energy differences between the T- and R-state upon protonating histidine residues may yield direct, quantative insight into the contribution of individual histidines to the total Bohr effect. To estimate if our choice of force-field influences the transition process, we carried out further simulations using the CHARMM27 force-field [22]. Three of those are using the TIP3P water model and the other three the CHARMM water model TIPS3P. Within 60 ns, we see that all three simulations with TIP3P and two out of three TIPS3P simulations showed a quaternary T-to-R transition. In addition to that, in their recent work Yusuff et al. also see quaternary transitions using the AMBER-99SB force field [23] when starting from the T-state in two simulations at 100 ns [9]. Both results show that the spontaneous transitions are reproducible across force-fields and render us confident that our choice of force-field yields an adequate description of the transition process. The starting structure for all simulations was the Hb T-state X-ray structure (Fermi et al. [24], PDB id: 2HHB). Each simulation was run for 200 ns with independent starting velocities. All simulations were carried out using the GROMACS 4. 5. 3 software [25], [26] with the GROMOS 43a2 force field [27] in explicit solvent and parameters as described before [7] with the exception of the Lennard-Jones cut-off (rvdw). Ten simulations used the same rvdw = 1. 0 nm as Hub et al. Twenty additional simulations were run using a larger rvdw = 1. 4 nm as consistent with the GROMOS force field parameterization. We did not observe differences in transition probabilities due to rvdw. Finally, 20 simulations using a second set of histidine protonations were performed (see Table 3). In total of Hb simulations were carried out. To judge whether a T-R transition occurred in the individual simulations, the following criterion was applied: Each simulation was projected onto the difference vector between the T-state and the R-state X-ray structure (Park et al. [28], PDB id: 1IRD; with applied symmetry for the full tetrameric state). If a projection at any time covered 80% of the T-to-R distance, the whole simulation was considered a transition simulation. All transition trajectories covered a similar range on the T-R vector whereas one simulation exceeded that strongly. We excluded this outlier, because it would have dominated the global motions in terms of covariance despite its low statistical weight. This left us with 21 transition trajectories and simulation time in total. The following method was applied to hemoglobin, but it can be used to analyse other systems with multiple domains as well. Here, we were interested in the coupling of inter- and intra-subunit motions and hence, we defined the protein chains to each be a local domain and focused on the chain-chain interfaces. Note that also other choices would have been possible, e. g. grouping Hb into two dimers instead of four monomers. When analysing the motions of a two domain protein consisting of only one chain, it may be of interest to consider each domain as one local entity. To obtain tertiary-only (T) coordinates, we superimposed the coordinates of each chain individually onto the respective chain in the T-state X-ray structure. This yielded an artificial trajectory with all chains having the same center-of-mass and orientation as in the T-state, not moving relative to each other and displaying only subunit-internal fluctuations. All four subunits were superimposed individually, reducing the degrees of freedom (DOF) of each by six (three translational and three rotational DOF) and resulting in 3N-24 DOF for the T trajectory (with N = 4556 being the number of non-hydrogen atoms). The complementary quaternary-only (Q) coordinates were obtained by doing the opposite and superimposing the individual chains of the rigid T-state structure onto the respective chains of the MD structures, thereby leaving out the internal motions and only keeping the rigid body motions of the chains relative to each other. Thus each chain was represented by a rigid body (six DOF) and lost all information on internal coordinate changes. Since the six global DOF for the whole protein were removed, the Q motions sample 18 DOF. For the coupling analysis, we simplified the Q motions further using PCA. The eigenvalue spectrum shows a steep decrease among the 18 non-zero eigenvalues with the first eigenvector describing 22% of the total variance of the Q motions. This principal mode (referred to as cQ) was used for FMA. Functional Mode Analysis (FMA) identifies collective motions in a coordinate space (in our case T) maximally correlated to a specific functional property (in our case cQ). To avoid overfitting the high dimensionality of the T space needs to be reduced. The original implementation of FMA [14] accomplished a reduction of dimensionality through PCA on the underlying coordinates. Recently, a new version of FMA based on Partial Least Squares (PLS) was developed [15] where a linear model is constructed based on a given number of vectors in the coordinate space (so-called latent vectors) that are subsequently optimized. Thereby the number of latent vectors forming the subspace for the linear combination needs to be chosen correctly to avoid overfitting. We constructed the FMA model on the first half of the concatenated transition trajectories using the second half for cross-validation. 20 components were found optimal, with decreasing predictive power for a higher number of components (indicative of overfitting). We constructed the final model with 20 latent vectors on the full data set. The constructed model and the cross-validation are shown in Figure 2. The calculated Pearson correlation coefficients were 0. 98 for fitting part and 0. 83 for the cross-validation part indicating a strong coupling between motions along cQ and cT. For the further analyses of the interactions, the individual latent vectors underlying cT were reweighted with their contribution to the overall variance as described in [14] to yield the ensemble weighted coordinate cTew. For both analyses, we reassembled the coupled tertiary and quaternary motions in the following way: The MD trajectories were projected onto the plane spanned by cQ and cTew. The space between the minimal and maximal projections was divided equidistantly into 20 parts, yielding grid, covering all MD structures. Starting from the T-state we moved stepwise in this plane until each gridpoint was visited and mapped projections back onto the full structural space of the protein. This provided a subspace of backprojected structures showing the coupling of local and global motions as derived from our simulations. Note that one would expect a vdW overlap of zero for any full-dimensional MD structure. The fact that we did observe non-zero overlap values arises from the reduced dimensionality of the coupling plane: The vdW overlaps have been calculated for structures in the cQ-cTew hyperplane, which is a two-dimensional representation of the coupling process. Any such projection cannot fully cover the full-dimensional motion (and overlap) of atoms and therefore only shows the global trends. The fact that the maximal correlated coordinate from T (cTew) is reducing the vdW overlap when moving along cQ at the same time, points at (the prevention of) vdW overlaps as a global driving force for the coupling mechanism. For each of these 400 structure the inter-chain vdW overlap was computed using a modified version of the dist program from the CONCOORD software [29]: For each atom we calculated how much its vdW sphere penetrates vdW spheres of atoms from other chains. The sum of vdW overlap values for all atoms gives a length in nm, representing a measure for the energetically unfavourable vdW overlap for each structure. For the contact analysis, we picked a specific pathway (referred to as cQ-cTew) in this hyperplane. Starting from the T-state, the first part of the path is along cQ, and the second along cTew For each of the 17 structures on this path, atoms closer to each other than 0. 3 nm were monitored and defined a residue contacts. The calculation was carried out using g_contacts by Blau and Grubmüller [30].

Hemoglobin transports oxygen from our lungs to other tissues. Its effectiveness in binding and unbinding of oxygen is based on a type of regulation called allostery. Thereby, already bound oxygen molecules in either of the four binding sites of hemoglobin increase the likelihood of oxygen binding in the other sites; the sites act cooperatively. It is known that the four protein chains of hemoglobin – each containing one oxygen binding site – need to rearrange globally or the cooperativity is lost. But how do the individual chains communicate with each other in this rearrangement? This is still not resolved on the molecular level. By applying the computational technique of molecular dynamics simulations we were able to simulate the motions of the individual atoms of hemoglobin during such a global rearrangement. We present a novel method that allows to directly analyse the coupling of motions within and between the four protein chains and thereby reveal the allosteric mechanism. That allows us to classify which amino acids are important for the cooperativity and in what way: by either pushing other amino acids away or attracting them. Also, we have observed in simulations how pH differences could affect the cooperativity.

Abstract Introduction Results Discussion Materials and Methods

2013

Collective Dynamics Underlying Allosteric Transitions in Hemoglobin

Visceral leishmaniasis (VL) is hypoendemic in the Mediterranean region, where it is caused by the protozoan Leishmania infantum. An effective vaccine for humans is not yet available and the severe side-effects of the drugs in clinical use, linked to the parenteral administration route of most of them, are significant concerns of the current leishmanicidal medicines. New drugs are desperately needed to treat VL and phenotype-based High Throughput Screenings (HTS) appear to be suitable to achieve this goal in the coming years. We generated two infrared fluorescent L. infantum strains, which stably overexpress the IFP 1. 4 and iRFP reporter genes and performed comparative studies of their biophotonic properties at both promastigote and amastigote stages. To improve the fluorescence emission of the selected reporter in intracellular amastigotes, we engineered distinct constructs by introducing regulatory sequences of differentially-expressed genes (A2, AMASTIN and HSP70 II). The final strain that carries the iRFP gene under the control of the L. infantum HSP70 II downstream region (DSR), was employed to perform a phenotypic screening of a collection of small molecules by using ex vivo splenocytes from infrared-infected BALB/c mice. In order to further investigate the usefulness of this infrared strain, we monitored an in vivo infection by imaging BALB/c mice in a time-course study of 20 weeks. The near-infrared fluorescent L. infantum strain represents an important step forward in bioimaging research of VL, providing a robust model of phenotypic screening suitable for HTS of small molecule collections in the mammalian parasite stage. Additionally, HSP70 II+L. infantum strain permitted for the first time to monitor an in vivo infection of VL. This finding accelerates the possibility of testing new drugs in preclinical in vivo studies, thus supporting the urgent and challenging drug discovery program against this parasitic disease. Neglected Tropical Diseases (NTDs) are severe scourges that affect the less protected layer of the poorest population of low-income countries [1,2]. Their neglected consideration implies a poor attention by most of the actors involved in their eradication including Pharma Industries, which are more concerned about those diseases affecting people from more developed countries [3]. The fight against these diseases is mainly based on preventive measures, but when the latter fails and a sudden outbreak emerges, no effective vaccines exist and treatment is based on the administration of drugs [4]. However, due to the low investments in R & D to develop new compounds most of them are outdated, toxic in many cases, full of undesirable side effects and their route of administration requires hospitalization [5,6]. All these are major drawbacks for the frequently overwhelmed health systems of these countries. Recent drug discovery programs sponsored by public or private initiatives pursue a crushing defeat of major NTDs during this decade. Visceral leishmaniasis (VL) is one of the diseases that accomplish all the conditions to be a NTD. Unfortunately, no effective vaccine candidates, either prophylactic or preventive, are under clinical trials [7], the treatment is still mainly based on old-fashioned antimony derivatives and the administration route of these drugs is parenteral. To overcome these gaps, several Big Pharma companies have made available to academic researchers and supranational institutions myriads of small molecules to be tested on Leishmania on recently developed High Throughput target-based and target-free Screenings platforms (HTS) [8,9]. High Content Screening (HCS) image-based readouts using confocal microscopy, or genetically modified parasites expressing easily detectable reporters are in the pipeline of target-free (phenotypic) in vitro screenings [10,11]. The use of transgenic light-emitting parasites has an additional advantage since they permit a further scaling up to in vivo preclinical trials using rodent models of both visceral and cutaneous infections for monitoring parasite loads by means of bioimaging devices [12]. So far, luminescent transgenic parasites that express genes encoding the firefly or Renilla sp. luciferases are the only systems that permit a rapid readout in vitro under HTS conditions and the assessment of parasitic burdens in internal organs of living mice [13–15]. The major pitfall of luminescence is the need of adding a light-emitting substrate–luciferin or coelenterazine–that is time-consuming and significantly increases the cost of the analysis. For this reason, standard fluorescent proteins are more suitable for in vitro assays [16]. There are many drug screening systems using GFP, RFP or mCherry-transformed parasites that have been useful to evaluate libraries of compounds on Leishmania promastigotes and amastigotes [17–19]. Moreover, the appraisal of in vivo infections by using these genetically-modified strains is currently limited to cutaneous leishmaniasis (CL) models. However, fluorescence emission in the visible spectrum has low tissue penetration to be recorded by the standard optical imaging platforms [20]. This drawback has been overcome by the arising of a new developed protein (IFP 1. 4) from the bacteriophytochrome of Deinococcus radiodurans, whose emission in the near-infrared region prevents any background interference derived from organs or tissues [21]. This protein has been successfully used to image the in vivo infection of the adenovirus serotype 5 (Ad5) that specifically infects the mouse liver in vivo. More recently, a new-engineered infrared fluorescent protein (iRFP) from the photosynthetic bacterium Rhodopseudomonas palustris showed a clear superiority over IFP 1. 4 to evidence the viral load of Ad5 in mouse liver [22]. In this report, we have established for the first time a platform based on the comparative analysis of IFP 1. 4- and iRFP- transfected L. infantum BCN 150 amastigote-infecting splenocytes ex vivo to perform phenotypic screenings with several drug collections of small molecules. In addition, the latter strain was used to monitor the development of an in vivo infection of VL, showing the higher tissue penetration of iRFP over IFP 1. 4, which represents a useful tool to assess the parasite load by non-intrusive bioimaging techniques. Female BALB/c mice (6–8 weeks old) were obtained from Harlan Interfauna Iberica SA (Barcelona, Spain) and housed in specific-pathogen-free facilities for this study. L. infantum (strain MCAN/ES/96/BCN 150) promastigotes were obtained from J. M. Requena (Centro de Biología Molecular Severo Ochoa, Madrid, Spain). Parasites were routinely cultured at 26ºC in M199 medium supplemented with 25 mM HEPES pH 6. 9,10 mM glutamine, 7. 6 mM hemin, 0. 1 mM adenosine, 0. 01 mM folic acid, 1x RPMI 1640 vitamin mix (Sigma), 10% (v/v) heat-inactivated foetal calf serum (FCS) and antibiotic cocktail (50 U/ml penicillin, 50 μg/ml streptomycin). The 987-bp IFP1. 4 coding region was amplified by PCR from pENTR1A vector, a kindly gift from Dr. Roger Y. Tsien, Departments of Pharmacology and Chemistry & Biochemistry, UCSD (USA). The oligonucleotides used as primers (RBF696 and RBF687 in Table 1) introduced NcoI-NotI as restriction sites. The 948-bp iRFP coding region was digested with BglII and NotI from pShuttle CMV-iRFP vector obtained from Dr. Vladishlav V. Verkhusha −Department of Anatomy and Structural Biology and Gruss-Lipper Biophotonics Center, Albert Einstein College of Medicine, Bronx, New York (USA). The successful cloning of these ORFs in pLEXSY-hyg2 (Jena Bioscience) yielded pLEXSY-IFP1. 4-HYG and pLEXSY-iRFP-HYG vectors, respectively. Parasites expressing IFP1. 4 and iRFP reporters were obtained after electroporation of L. infantum BCN150 promastigotes with the linear SwaI-targeting fragment obtained from the above described vectors. Subsequent plating on semisolid media containing 200 μg/ml hygromycin B, allowed the isolation of individual clones that were subcultured in liquid media with antibiotic pressure. Correct integration of each fragment into the 18S rRNA locus of the resulting clones (IFP 1. 4+L. infantum and iRFP+L. infantum) was confirmed by PCR amplification analysis, using appropriate primers (Table 1). To achieve a higher stability of the reporter mRNAs 3’-untranslated regions (3’-UTRs) and intergenic regions (IR) were also included. For this purpose, three fragments derived from L. donovani A2 (LinJ. 22. 0670) (3400 bp) [23], L. infantum AMASTIN (LinJ. 34. 1010) (1200 bp) [24] and HSP70-II (LinJ. 28. 3060) (1600 bp) [25], were amplified using forward primers, which introduced a 5’-NotI site, and reverse primers that introduced a 3’-XbaI site (Table 1). Each amplicon was NotI/XbaI-digested and ligated into the parental construct pLEXSY-iRFP-HYG previously cut with the same enzymes to remove the urt2, thus yielding pLEXSY-iRFP-A2-HYG, pLEXSY-iRFP-AMASTIN-HYG and pLEXSY-iRFP-HSP70-HYG, respectively. Electroporation and semisolid selection, as well as correct integration, were done as described above. Several clones from each electroporation were grown in liquid medium in the presence of antibiotic selection and those clones with higher infrared fluorescent emission were selected after flow cytometry analysis (Cyan ADP, Dako). In addition, each L. infantum modified strain was used to infect mice in order to recover the lost infectivity after cloning and plating. We will hereafter refer to these strains as A2+L. infantum, AMASTIN+L. infantum and HSP70+L. infantum. The human acute leukemia monocyte cell line (THP-1, ATCC TIB-202), was cultivated in RPMI medium supplemented with 10% heat-inactivated FCS and 1% streptomycin/penicillin at 37ºC and 5% CO2. The cultures were diluted every 3 or 4 days to maintain the cell density between 105 cells/ml and 8 x 105 cells/ml. THP-1 cells at 5 x 105 cells/ml were differentiated with 50 ng/ml of phorbol 12-myristate 13-acetate (PMA) for 48 h at 37ºC and 5% CO2. Stationary promastigotes of each infrared strain (5–6 day cultures) freshly transformed from lesion amastigotes were added to differentiated-macrophages at 10: 1 ratio for 24 h at 37ºC. Then, the parasites that have not been internalized were removed by washing with phosphate-buffered saline (PBS). Intracellular infrared signal was analyzed after 72 h post-infection by flow cytometry (Cyan ADP, Dako). Female BALB/c mice (5 animals per group) were injected intraperitoneally with either 107 infective-stage metacyclic promastigotes of L. infantum BCN 150 wild type, IFP 1. 4+, iRFP+, A2+, AMASTIN+ and HSP70+L. infantum strains. Metacyclic parasites were purified from stationary cells freshly transformed from lesion amastigotes by negative selection with peanut agglutinin [26]. At different times post-infection, five animals were euthanized and their spleens and livers were aseptically recovered, washed in cold PBS and placed in Petri dishes. Spleens were used for evaluating the proliferative response of spleen cells after concanavalin A (Con A) addition. Briefly, small pieces were obtaining by using a scalpel. In order to obtain a single cell suspension, the tissue was incubated with 2 ml of collagenase D (Roche) at 2 mg/ml of buffer (10 mM HEPES pH 7. 4,150 mMNaCl, 5 MmKCl, 1 mM MgCl2,1. 8 mM CaCl2) for 20 min at 37ºC. The cell suspension and remaining tissue fragments were gently passed through a 100 μm-cell strainer to remove the tissue fragments. After erythrocyte lysis, splenocytes were washed 3x with PBS by centrifugation (500 x g for 7 min at 4ºC). Splenocytes were seeded at a density of 2 x 105 cells/well in 96-wells plates in RPMI medium, 20% FCS, 1 mM sodium pyruvate, 1x RPMI vitamins, 10 mM HEPES and 100 U/ml penicillin and 100 μg/ml streptomycin supplemented with 50 μM 2-mercaptoethanol. Cells were cultured in medium alone (control) or stimulated with Con A (5 mg/ml) for 3 days. One μCi of [3H]-thymidine was added for the final 16 to 18 h of the culture. Subsequently, cells were washed, lysed and [3H]-thymidine incorporation was measured onto glass fiber filters for scintillation counting. The splenocyte stimulation index was determined by dividing the cpm of Con A-stimulated and non-stimulated splenocytes [27]. The total number of living parasites in the organs was calculated from single-cell suspensions that were obtained by homogenization of the tissue through a wire mesh. Briefly, liver and spleen homogenates (100 mg/ml) were serially diluted in complete Schneider’s medium and distributed in 96-well plates at 26ºC. After 10 days, each well was examined and categorized as positive or negative according to the presence of viable promastigotes. The number of parasites was calculated as follows: Limit Dilution Assay Units (LDAU) = (geometric mean of titer from quadruplicate cultures) x (reciprocal fraction of the homogenized organ added to the first well). The titer was the reciprocal of the last dilution in which parasites were observed [28]. Intracellular amastigotes were purified from infected THP-1 macrophages and from spleen tissues in order to evaluate the infrared emission of intracellular parasites and for developing a standard curve up to the fifth week post-infection (5 wpi) in mice. On the one hand, 72 h post-infection THP-1 macrophages were washed twice with cold PBS, and scraped off using a rubber policeman. On the other hand, the spleens at different time points were collected and a splenocyte suspension was obtained by passing the spleen through a wire mesh. Both cellular suspensions from in vitro infections and from splenocytes were disrupted by passing sequentially through 27G1/2 and 30G1/2 needles, and polycarbonate membrane filters having pore sizes of 8 μm, 5 μm and 3 μm (Isopore, Millipore) [29]. Released amastigotes (free of host cells), were washed twice with PBS (3000 x g for 10 min at 4ºC) and counted by direct microscopy. The standard curves were developed using microscopic quantification of amastigotes and the corresponding infrared signal at 2-fold serial dilutions. Several parameters were included in the development of the explant culture. i) Parasite infection rate is not equal in all animals from the same batch, thus in order to normalize the ex vivo culture, only spleens with weights ≥ 0. 7 g were selected to prepare the ex vivo explants. ii) The starting cell density was adjusted in terms of fluorescence emission in a wide range of arbitrary units (0. 25,0. 5,1. 0, and 2. 0 x 105 AU). Since we intend to evaluate the drug inhibition in ex vivo cultures for at least 4 days, the initial number of AU that matched this condition was chosen. iii) Discrimination between drug-treated and untreated control cells was assessed by adding amphotericin B (AmpB) to the explants cultures. iv) To prevent the edge effect caused by evaporation from outer wells in assays with multi-day incubations, the top and bottom rows as well as the first and last columns were filled with ddH2O. In addition, the positive and negative controls were symmetrically distributed through the columns [30]. A total of 298 compounds belonged to three distinct collections including i) indenoisoquinolines obtained from Dr. Mark S. Cushman, Department of Medicinal Chemistry and Molecular Pharmacology (Purdue University, Indiana, USA.); ii) podophylotoxin and quinones were a kindly gift from Dr. Arturo San Feliciano, Deparment of Química Farmacéutica (Universidad de Salamanca, Spain) and iii) carbolins obtained from Dr. Sankaran Murugeshan, Department of Pharmacy, (Birla Institute of Technology & Science, Pilani, India). Stock solutions consisted of test compound at a concentration of 200 μM in 20% DMSO. The 200 μM master plates were further diluted by transferring aliquots of 5 μl into black 384-well plates with clear bottom containing 45 μl of culture medium. The final concentration of DMSO in both assay and control wells not exceeded 1% (v/v). Splenocytes obtained from infected mice (50 μl), were added to Testing Plates (containing the compounds to be test at single 10 μM concentration) and Control Plates (containing positive and negative controls). The plates were incubated at 37ºC and 5% CO2 and at 0,24,48 and 72 h the infrared signal was recorded using the Odyssey infrared system (LI-COR, USA). The anti-leishmanial reference drugs were amphotericin B, miltefosine, and paromomycin sulfate. The antileishmanial effect of the compounds was calculated as the percentage inhibition in relation to AmpB and DMSO as follows: (mean of 1%DMSO–Experimental Value) / (mean of 1%DMSO–mean of AmpBEC100) x 100. The positive control (10 μM AmpB), and the negative control (1% DMSO), placed in columns 2 and 23, were included in all Testing Plates. To normalize compound activity in relation to plate-to-plate variations in the assay signal, and to calculate Z’-factors, Control Plates containing also 10 μM AmpB and the vehicle 1% DMSO (in alternating fashion) were also included [30]. The quality of the assay was given by Z’ factor that was calculated as 1 –[ (3SD positive controls + 3SD negative controls) / (mean of the positive controls–mean of the negative controls) ]. Data from plates were used only if Z’ factors were > 0. 5 [31]. Primary hits were defined as compounds displaying 60% inhibition of signal readout. To test the toxicity of the identified hit compounds we established splenic ex vivo explant cultures of uninfected BALB/c mice. Briefly, the animals were euthanized and the spleens were aseptically removed and homogenized as described above. The cell suspension culture was conducted in RPMI medium, 10% FCS, 1 mM sodium pyruvate, 1x RPMI vitamins, 10 mM HEPES and 100 U/ml penicillin and 100 μg/ml streptomycin. Cells were microscopically counted and distributed into clear 96-well plates; 50 μl of uninfected splenocytes were added to each well (5 x 105 cells), containing 50 μl of serial 2-fold dilutions of the test compounds (200–0. 1 μM) or the DMSO control. After 72 h of incubation at 37ºC and 5% CO2, the viability of the cells was assessed using the Alamar Blue assay according to manufacturer’s recommendations (Invitrogen). Mice infected with wild type and HSP70+L. infantum parasites were imaged using an intensified charged coupled device camera of the In Vivo Imaging System (IVIS 100, Xenogen). To minimize background when imaging in the near infrared, the normal diet was replaced by a purified diet during 7 days (AIN-93M, LabDiet, UK), since chlorophyll fluoresces naturally, emitting between 675 and 685 nm, and is detected in the 700 nm channel. Nevertheless, animals were fasted overnight, although water was allowed freely before acquiring the images in order to assure that faeces were removed from intestines. The animals were shaved to reduce background signal due to the fur. The animals were then lightly anesthetized with 2. 5–3. 5% isoflurane (then reduced to 1. 5–2. 0%), placed in the camera chamber, and the fluorescence signal was acquired for 1 s. Fluorescence determinations, recorded by the IVIS 100 system, were expressed as a pseudocolour on a grey background, with yellow colour denoting the highest intensity and dark red the lowest one. Aimed to create stably-integrated infrared L. infantum BCN 150 strains we electroporated wild-type promastigotes with the lineal 6156 and 6110 bp SwaI-SwaI fragments containing the ORFs encoding IFP 1. 4 and iRFP, respectively, as well as the selection marker of the pLEXSY-hyg2 plasmid. We will refer to them as IFP 1. 4+L. infantum and iRFP+L. infantum. After selection on semisolid plates containing 200 μg/ml hygromycin B, individual colonies were seeded in M199 liquid medium supplemented with 10% FCS and hygromycin B. Genomic DNA isolated from these cultures was used to confirm the correct integration of the target sequence into the 18S rRNA locus of L. infantum genome (data not shown). Fig. 1A shows that the genomic manipulation of the engineered strains did not affect the growth rate of the parasites compared to wild-type promastigotes. Further comparative studies aimed to characterize the two infrared strains were performed. In order to measure the fluorescence emission of both strains, we employed logarithmic promastigotes of IFP 1. 4+L. infantum and iRFP+L. infantum (Fig. 1B), as well as intracellular amastigotes isolated from infected THP-1 cells in vitro (Fig. 1C), and lesion-derived amastigotes obtained from infected spleens of BALB/c mice at 5 wpi (Fig. 1D). A clear positive correlation was observed between the fluorescence signal and the number of logarithmic promastigotes placed in 96-well plates (Fig. 1B). Surprisingly, intracellular parasites expressing both infrared reporters (1C and 1D) displayed higher differences than promastigote cultures. All results clearly showed the superior fluorescence signal of the iRFP protein over IFP1. 4, particularly at the really relevant stage of the parasite life cycle, the amastigotes, both isolated from THP-1 macrophages and from lesions. Moreover, the stability of iRFP expression was monitored over a period of 6 months after transfection and no change was observed in fluorescence intensity during this period, even in the absence of hygromycin B (Fig. 1B). Once the iRFP reporter was chosen, we optimized the expression by replacing the original utr2 from pLEXSY-hyg2 vector with three sequences containing both 3’UTRs and IR (we will refer to as downstream regions, DSR). Two of these DSR belong to stage-regulated genes in the amastigote stage, the A2 and AMASTIN genes [23,24]. The third DSR belongs to the HSP70 II gene of L. infantum, a sequence involved in mRNA stability at 37ºC [25]. Fig. 2A shows the schematic view of the different constructs containing the iRFP reporter gene. In a similar way to IFP 1. 4 and iRFP-expressing parasites, we proceeded to compare the fluorescent signal of iRFP+, A2+, AMASTIN+ and HSP70+L. infantum strains. A clear correlation between the fluorescence emission and the number of parasites was also detected by using logarithmic promastigotes (Fig. 2B). Interestingly, replacement of the utr2 from pLEXSY vector with the corresponding DSR sequences from the A2 and AMASTIN genes resulted in a reduction of the fluorescent levels in the promastigotes. However, using the same strains but isolating the intracellular amastigote stage, from both infected THP-1 cultures (Fig. 2C) and lesions (Fig. 2D), we observed a 2–2. 5-fold increase in fluorescence emission. Nevertheless, the most significant increase (p<0. 001) was obtained with amastigotes carrying the plasmid including HSP70 II as DSR. Infection of laboratory mice with either L. infantum or L. donovani vary markedly between different organs. In the liver, the infection can resolve, unlike in the spleen, where Leishmania parasites may persist [32]. Therefore, we established a model of visceral L. infantum infection that allowed us to collect large numbers of naturally-infected cells from spleen to use as ex vivo explant. Infected mice were euthanized at different times post-infection. The spleens were weighed, processed to obtain a cellular suspension culture and the burden parasite and the proliferative response of mouse spleen cells were evaluated. There was a dramatic increase in spleen size, up to 1. 0 g (Fig. 3A), splenocyte culture, (up to 12 x 108 cell/spleen, Fig. 3B), and parasite burden (up to 25 x 106/100 mg tissue, Fig. 3C), whereas the proliferative response dropped to basal levels at five wpi (Fig. 3D). Consequently, for drug screening we established five wpi as the time point for the ex vivo splenic explant culture. Finally, to be sure that fluorescence signal was stable in infected animals throughout all this time, parasites from infected spleens were also isolated and amastigote standard curves were prepared at 3,4 and 5 wpi (Fig. 3E). As previously described by other authors, the loss of fluorescent signal during the evaluated period of time was not significant [18,33]. In addition, the amastigote standard curves provided us the number of amastigotes that were used during the screening assays. Firstly, we checked the fluorescence dynamic range in which we could assure the effective and quantifiable replication of intracellular parasites inside macrophages in the absence of active drug over the course of the ex vivo culture. However, since there were individual-to-individual differences in parasite load, we first needed to establish the appropriate starting cell-density well that allowed the detection of parasite proliferation. For this purpose, we fixed the fluorescence level per well, instead of adjusting the number of cells. We followed the replication of the amastigotes within the splenic culture each 24 h during 5 days, using four different starting densities ranging from 0. 25 x 105 to 2. 0 x 105 arbitrary units (AU). The signal was measured in black 384-well plates with clear bottom using the infrared imager Odyssey (LI-COR). After 5 days of culture, only those wells containing 2. 0 x 105 AU at the beginning of the assay supported the active replication of the parasite in the absence of active drug over the whole experiment, indicating ongoing multiplication of the amastigote cells (Fig. 4A). The next step was to assure that the selected starting cell density was accurate to guarantee the correct discrimination between drug-treated and untreated control wells. For doing this, different concentrations of AmpB (20,6. 6,2. 2,0. 74 μM) were added to the explant and the reduction in parasite load was measured by quantifying the loss of infrared fluorescence for 72 h (Fig. 4B). To confirm these results, and also to establish the AmpBEC100, as well as the appropriate period of incubation, ten serial 2-fold AmpB dilution steps covering a range from 10 to 0. 02 μM were assayed against three starting cell densities (from 1. 0,2. 0 and 4. 0 x105 AU). Readouts were harvested at 72 and 96 h. Surprisingly, IC50 values at 72 h were very similar in all the starting cell densities, ranging between 0. 27–0. 38 μM, (Fig. 4C-E) which corresponded with data collected by other authors [10,34]. These results suggest that all these conditions allow the screening of compounds. To confirm the quality of the assay in discriminating active from inactive compounds, we calculated the Z prime (Z’) factor in 3 different screening experiments using 6 different plates. All Z’ were higher than 5. 0, which demonstrated the correct reproducibility and the suitable quality of the assay [31]. Fig. 4F shows the fluorescence image of a 384-well plate containing the ex vivo culture treated with different compounds, where the red colour shows different intensities, which reflects the viability of the amastigotes in each well. The course of the ex vivo infection was followed over a period of 72 h by measuring the absolute fluorescence of the infection and the percentage of infected macrophages by classical Giemsa staining to determine parasite load (data not shown, Fig. 4G). No differences between both methods were accounted, thus pointing to the suitability of fluorescence analyses to assess the infectivity of HSP70+L. infantum strain on mouse macrophages. Moreover, intracellular forms were analyzed by confocal microscopy at 633 nm, after 5 wpi. A strong red fluorescence emission from round-shaped HSP70+L. infantum-emitting amastigotes was observed inside parasitophorous vacuoles in the cytoplasm of the infected macrophages (Fig. 5), confirming that the fluorescent signal was not lost during the ongoing proliferation of intracellular amastigotes. A library of 295 compounds comprising three collections of chemically distinct structures was screened at a concentration of 10 μM. The threshold for selecting hits was set as inhibitory activity of ≥ 60% after 3 independent assays. Twenty-four primary hits were obtained out of the 295 compounds (8%). Our next goal was to select, amongst the 24 confirmed hits, the best candidates for further investigation. Firstly, compounds were tested in dose-response curves using 7–8 concentrations 2-fold diluted from 50 μM, 100 mM and 2. 5 mM respectively for combrestatins, carbolins and indenoisoquinolins. Then, the cytotoxicity of these compounds was tested in different concentrations with a 3-day assay on non-infected ex vivo explant cultures (CC50) using Alamar Blue (data not shown). Five out of the 24 hits were selected based on a threshold of Selective Index (SI = CC50/IC50), score of ≥30. They included combrestatins (8%; 2/24), carbolins (8%; 2/24) as well as indenoisoquinolines (4%; 1/24) (Table 2). Finally, the applicability of the HSP70+L. infantum for imaging in BALB/C mouse was tested. Animals (n = 10) were inoculated intraperitoneally with 107 stationary-phase HSP70+L. infantum promastigotes. The mice were imaged at 0,10 and 20 wpi. A bright and intense signal was detected after parasite injection allowing even to trace the needle entry site. The transitory hepatic episode (14 days post-infection) was out of the scope of this experiment. On the contrary, spleen colonization was evidenced after 10 wpi and until the end of the experiment (Fig. 6A). Five animals were euthanized and after removal of the overlying skin (Fig. 6B top) and peritoneum (Fig. 6B down), increasing fluorescence signal was detected, confirming the above-mentioned results. The dissected organs were entirely fluorescent (Fig. 6C). Because of a lower parasite burden in the hepatic organ, the high level of fluorescence emitted by amastigotes in the spleen might have blocked the detection of parasites in the liver when imaging both organs at the same time. In order to tackle a future drug trial in vivo, the appraisal of infection was monitored for up to 2 wpi, showing that the fluorescent signal was still visible in the remaining group of mice (Fig. 6A). These results support future long-treatment schedules to tackle in vivo drug trials against VL. In this work we present a potent tool based on infrared fluorescent proteins to achieve a double goal; on the one hand a robust platform for HTS screening of compounds and on the other hand an in vivo imaging of visceral Leishmania infections. Firstly, we chose between two different sources of infrared fluorescent proteins (Deinococcus radiodurans and Rhodopseudomonas palustris) [21,22], and then we optimised the expression level by introducing different DSR of well-known virulence factors and heat shock proteins. Initially, we compared the suitability of two fluorescent infrared stably-modified strains, in terms of brightness and fluorescence-emission properties in promastigotes and amastigotes of L. infantum BCN 150. iRFP+L. infantum reached significantly higher emission than IFP1. 4+L. infantum throughout the different stages of L. infatnum life cycle. Fluorescent levels varied 2–3 times between both strains, and allowed the detection of as little as 100 intracellular amastigotes. In general, the intracellular stage has a lower emission due to their low metabolism, a fact that has been previously pointed out by other authors [35,36]. Unlike other fluorescent reporters whose emission wavelengths are in the visible spectrum, near infrared fluorescent proteins have improved optical properties since the absorbance of haemoglobin, protein and lipids, as well as light-scattering, is minimal [21]. Thereafter, optimization of the vector was achieved by including different sequences; two of them previously described as virulence factors [23,24], and the third one having a protection role against heat and oxidative stress in VL [25]. The highest levels of expression were associated with HSP70+L. infantum, which contains the downstream sequences from HSP70-II gene, with influence on transcript stability at conditions that facilitate the survival of the parasites within mammalian hosts [37]. Furthermore, the fluorescence emission of this reporter has been enhanced via stable integration into 18S rRNA locus, which represents an efficient and effective strategy that enables expression of proteins, avoiding the need of drug selection for both in vitro screenings and in vivo infections as previously described [17,18,35,38,39,40,41,42]. Once the tool was selected, its applicability for drug discovery in VL in terms of searching new hits was assessed. Leishmania represents an enormous challenge as an intracellular parasite, since the drug must kill the parasite without affecting the host cells. Recently, HCS using image-based microscopy have been proposed as a robust platform for evaluating intracellular infections [10,36]. In this system, axenically parasite cultures are used to infect macrophage’s cell lines. However, this approximation has several drawbacks: i) it is highly complex, ii) axenic parasites have notable differences with amastigotes [43], and when they have been used in screening assays a high false-positive rate was associated [36], and iii) the use of macrophage lines as host cell has also been questioned [34]. A novel approach based on the physiological and immunological environment of the infection site has been proposed [29]. Ex vivo explant is a primary culture that contains the full variety of cells that allow the parasite progression in the infected organ, and not only inside macrophages. Moreover, the infection is not recently established, but installed several weeks before the addition of the drug and the evaluation of its effect, and thus it greatly resembles human infections. From the lab point of view, the manipulation is minimal and a large number of test wells are easily prepared from a single infected animal. In this work, the ex vivo splenic explant used for drug screening was developed in BALB/c mice infected with L. infantum, showing the typical features of VL with enlargement of spleens, high parasite burden, and loss of proliferation in splenic cells. The system congregates the basic rules previously defined for drug-screening systems for trypanosomatids [44]. i) it uses dividing populations of the mammalian stages of the parasite, ii) the drug activity is readily quantified by changes in infrared fluorescence signal, iii) a standard drug for VL (AmpB) showed activity at concentrations near to the one reached in serum [45,46]. A crucial point was to establish the signal discrimination referred to the number of amastigotes in the infected cells per well. The system was assessed with a selected set of compounds coming from three different collections and well-known active drugs were also included. The system adaptation presented here, allows for rapid identification of active compounds without the requirement of substrate reagents to get the readouts, and hence, the cost is reduced compared to other systems. The model uses a stable-modified Leishmania with no significant loss of reporter signal after more than 20 wpi. The fluorescence signal of the infected splenic explant has lineal correlation with the number of amastigotes, which makes the system easily adaptable to quantitative assays. Furthermore, there is no need of expensive infrastructure, such as the HCS, which are not affordable for some labs. Since the assay could be performed in 384-well plates, a single infected animal could be enough to screen more than 500 compounds. However, host factors involved in the pharmacokinetics and pharmacodynamics of drugs are out of the scope of in vitro drug screening, consequently in vivo drug assays are still necessary. Using the same tool, parasite dissemination in an in vivo chronic model of VL has been traced for the first time based on infrared fluorescence signal. At 5 wpi, the animals were imaged, but not clear fluorescence signal was recorded. After 10 wpi an undisputed infrared signal was achieved, showing that the splenomegaly had been already established time before. Many studies initiate drug administration trials after one or two weeks post-infection assessing the effect against the liver infection time with a peak in the parasite burden [47]. However, our goal was to use this model for longer times in order to perform future drug administrations against spleen infection, so the animals were pictured after 20 wpi and the fluorescence signal was still strongly detected. Therefore, this in vivo imaging model could be used without significant loss of reporter emission, allowing the appraisal of chronic infections after drug treatments. In conclusion, the near-infrared fluorescent L. infantum strain used in the current work represents an important step forward in bioimaging research of VL, providing a robust model of phenotypic screening suitable for HTS of collections or libraries of small molecules in the amastigote form of mouse splenic explants. In addition, this novel L. infantum strain could be used in preclinical in vivo studies of potential lead compounds against visceralizing Leishmania species.

Visceral leishmaniasis (VL), caused by Leishmania infantum or L. donovani, is still one of the most threatening diseases affecting poor people in developing countries, with a fatality rate as high as 100% in two years in infected and untreated people. With no vaccine available and ineffective and toxic chemotherapy, the search for new potential drugs that accelerate the urgent drug discovery process are highly needed. A novel technology that addresses this important issue has been developed, by performing High Throughput Screening (HTS) assays in 384-well plates format in combination with an infrared L. infantum-expressing strain. The system was further validated to identify active compounds against VL in an ex vivo splenic culture. In addition, in vivo non-invasive imaging of the visceral infection in BALB/c mice was achieved for the first time by using transgenic fluorescent parasites. These findings open up the possibility of testing vast amounts of potential compounds and allow in vivo screening of drug candidates against this severe parasitic disease in an attempt to speed up the vital drug discovery program.

Abstract Introduction Methods Results Discussion

2015

Infrared Fluorescent Imaging as a Potent Tool for In Vitro, Ex Vivo and In Vivo Models of Visceral Leishmaniasis

Selection acting on genomic functional elements can be detected by its indirect effects on population diversity at linked neutral sites. To illuminate the selective forces that shaped hominid evolution, we analyzed the genomic distributions of human polymorphisms and sequence differences among five primate species relative to the locations of conserved sequence features. Neutral sequence diversity in human and ancestral hominid populations is substantially reduced near such features, resulting in a surprisingly large genome average diversity reduction due to selection of 19–26% on the autosomes and 12–40% on the X chromosome. The overall trends are broadly consistent with “background selection” or hitchhiking in ancestral populations acting to remove deleterious variants. Average selection is much stronger on exonic (both protein-coding and untranslated) conserved features than non-exonic features. Long term selection, rather than complex speciation scenarios, explains the large intragenomic variation in human/chimpanzee divergence. Our analyses reveal a dominant role for selection in shaping genomic diversity and divergence patterns, clarify hominid evolution, and provide a baseline for investigating specific selective events. The action of natural selection on genome sequences is most directly revealed by a deficit or excess of substitutions relative to the neutral rate, but detecting this requires sequences that have been diverging long enough to experience a high density of mutations [1]. An alternative approach, applicable over shorter evolutionary time periods, is to look for indirect effects of selection on neutral sequence variation [2], [3]. Directional selection reduces population diversity at linked neutral sites by eliminating chromosomes bearing a less fit variant from the population, an effect known as ‘hitchhiking’ in the case of positive selection [3] and ‘background selection’ in the case of negative or purifying selection [2], [4], [5]. The magnitude of the diversity reduction depends upon the density of selected sites, the amount of time during which selected variants segregate in the population prior to fixation or loss, and the rate at which recombination decouples neutral sites from selected variants [2], [4], [5]. In Drosophila a positive correlation between recombination rate and nucleotide diversity is well established and there is strong evidence for background selection or hitchhiking [2], [4], [6]–[10]. In hominid evolution, the roles of background selection and hitchhiking are less certain. Human diversity is positively correlated with recombination on a large scale [11]–[13] and negatively correlated with coding sequence density [14], consistent with a role for selection in recent human evolution. However, whole genome scans have identified relatively few regions with convincing evidence of positive selection [15], [16], an important role for background selection has generally been discounted [5], [17], [18], and it has been suggested that the association with recombination may reflect a mutagenic effect rather than selection [11], [17]. Consequently a clear picture of the importance and nature of selection in human evolution is still lacking. Here we conduct a broader and more systematic search for signatures of selection. We look more widely in hominid evolution, augmenting human polymorphism data [19], [20] with orthologous sequences for five primate species ([21] and our laboratory) (Figure 1). The latter sequences carry information about ancient population diversity, because some sequence differences between any two species represents polymorphic variation that existed in their common ancestral population [22]. We used mammalian sequence conservation to identify two classes of genomic segments: “conserved” segments, which appear to be under long-term purifying selection, and “neutral” segments which are putatively free of selective constraint. Specifically, we employed a phylogenetic Hidden Markov Model (HMM) [23], which we extended to improve sensitivity by incorporating information from alignment gaps. We ran the HMM on a multiple alignment of placental mammals [24], but intentionally excluded data from the great apes (including human) and rhesus macaque to avoid biasing our subsequent analysis of sequence divergence in these species. Less than one-fourth of conserved bases identified by this approach are protein-coding, with the remainder largely of unknown function [23]; moreover, conserved segments are much more uniformly distributed in the genome than coding sequences, with most genomic bases surprisingly close to a conserved site (Figure 2). Thus it is desirable to take into account the detailed genomic distribution of all conserved sequences, and not just coding sequences, in investigating the effects of selection on diversity. Using sequence conservation rather than existing gene annotations has the advantage that it is unbiased by assumptions about which annotated features are functional. We next compared levels of variation at putative neutral sites in the 10% of the genome nearest to conserved segments, to that in the 50% of the genome farthest from such segments, hypothesizing that selection should have a reduced effect on more distant regions. Human diversity and human/chimpanzee (H/C) divergence are indeed both substantially reduced near conserved segments, and using genetic instead of physical distance magnifies this effect (Figure 3). An even stronger reduction in neutral divergence and diversity is observed if distances are calculated with respect to annotated exons rather than conserved segments, suggesting that selection acting on exonic sequences has a greater effect on nearby diversity than selection on non-exonic conserved sequences. The effect is not limited to sites which are closest to exons; across the genome, H/C divergence exhibits a strong dependency on distance from conserved exonic segments (Figure 4, Table S1). Somewhat surprisingly, a fine-scale recombination map that incorporates ‘hotspot’ patterns [25] provides significantly better discrimination than a coarse pedigree-based map [26], even though many hotspots have moved in recent evolution [27], [28]. This suggests the finescale map may be more accurate than the pedigree map at smaller scales despite the hotspot movement. The trends described above are consistent with selection at conserved segments acting to reduce diversity in both the human and human-chimpanzee ancestral populations. As a more sensitive indicator for the latter population, we also examined neutral sites where human and gorilla, or chimpanzee and gorilla, share one nucleotide and the other 3 primates share a different nucleotide (‘HG’ and ‘CG’ sites). At such positions, the human-chimpanzee coalescent predates the gorilla split [29] (see Figure S1) and so is very old. Since directional selection reduces time to coalescence at linked neutral sites, the density of HG and CG sites should be depleted near elements under selection, and this is indeed the case around conserved segments (Figure 3). To control for the possibility that the lower diversity and divergence near conserved segments are due to the presence of unidentified sites under negative selection, or to a lower neutral mutation rate, we calculated human/macaque (H/M) and human/dog (H/D) divergence in the same bins. Only a small portion of divergence between distantly related species should reflect ancestral population diversity, so background selection or hitchhiking should have a minor effect on H/M divergence and a negligible effect on H/D divergence. There is a small reduction in both H/M and H/D divergence near conserved segments, suggesting that some of the trend is attributable to mutation rate variation or direct selection. However, normalizing by H/M divergence to cancel such effects does not change the overall trends (Figure 3) suggesting they are mainly due to indirect effects of selection. (Since some fraction of H/M divergence itself reflects ancestral diversity, normalizing in this way is an overcorrection, which is presumably why it reverses the trend for H/D divergence). We also confirmed that the same trends are seen separately for introns and for intergenic sequences upstream and downstream of transcripts (Figure S2). Normalizing by H/M divergence would not correct for lineage-specific mutation rate variation. For example, if recombination is itself mutagenic [11], [13] and recombination rates have changed in primate evolution, normalizing by H/M divergence may fail to cancel recombination-induced mutation rate variation among hominids. However, we are unable to envision a plausible scenario along these lines that could explain the trends in Figure 3. In particular, changes in recombination would not explain the dependence on physical distance from exons. We next examined the evolutionary rates within conserved sequences and putatively neutral sequences near conserved sequences, calculating divergence relative to the genome average at all putatively neutral sites (Figure 5). Relative divergence is much lower in exonic than non-exonic conserved segments, suggesting that selection is weaker on the non-exonic sites. The relative divergence in conserved segments decreases with evolutionary distance (e. g. relative divergence is lowest for the H/D comparison) consistent with weaker selection in the hominid lineage [30]–[32]. The opposite trend is observed for fourfold degenerate (4D) sites, and neutral sites near exonic conserved segments. In these cases relative divergence increases with evolutionary distance, which is consistent with background selection or hitchhiking, rather than direct selection. Divergence in 4D sites is substantially lower than the overall neutral rate even for the human-dog comparison, possibly because a subset of these sites are under direct selection. H/C and H/M divergence are only slightly lower in neutral sites near non-exonic conserved segments suggesting that background selection or hitchhiking in these regions is very weak. The preceding analysis indicates a role for selection in shaping population diversity, but does not allow quantitative conclusions about selection strength. We therefore undertook a more detailed analysis, applying a theoretical model [5] of background selection to compute the expected reduction in nucleotide diversity at a neutral site due to purifying selection at other sites, as a function of recombination rates, selected site locations, deleterious mutation rate, and the distribution of selection strengths. We use a model of background selection rather than hitchhiking because it should provide a reasonable baseline estimate for the effects of selection, given that purifying selection is thought to be widespread (affecting most functional elements), while the relative importance of positive selection is still controversial. Because strength of selection in hominids may depend on the type of functional element [31] we distinguish exonic (protein-coding and UTR) from non-exonic selected sites, allowing them to have different mean selection strengths and deleterious mutation rates. From these calculations we obtain a background selection (B) value for each position in the genome. B indicates the expected fraction of neutral diversity that is present at a site, with values close to 0 representing near complete removal of diversity as a result of selection and values near 1 indicating little effect. We then represented the probability of the observed primate sequence alignment data as a function of species divergence times, mutation rates, ancestral effective population sizes, and B, and estimated all parameters by maximum likelihood (Table 1). Additionally, our model corrects for intragenomic mutation rate variation by allowing the mutation rate to vary with local H/D divergence. The model provides a good fit to the alignment data (Figure S3), indicating a strong dependence of divergence on predicted background selection in each ancestral population. Our speciation time and effective population size estimates are broadly consistent with previous analyses [29], [33], [34] (Table 1). The mean selection strength (t) estimate for autosomal exonic conserved segments is 0. 0025, within the range of those from recent studies of human coding sequence polymorphisms [35], [36]. For non-exonic conserved sites, t is very low (0. 00001); moreover fitting a reduced model that allows only for selection on conserved exonic segments gives essentially the same likelihood (Table S2) and parameter estimates (Table S3). This suggests that many non-exonic conserved segments are false-positives or are no longer under selection in hominids. The latter possibility accords with promoter region analyses that suggest weaker selection on regulatory elements in hominids than rodents, possibly because hominid effective population sizes are smaller [31]. If selection is weaker on non-exonic conserved elements in hominids then they should evolve more quickly in the human and chimpanzee lineages. A comparison of H/C, H/M and H/D divergence in these elements confirms that this is indeed the case (Figure 5). Our estimate of the deleterious mutation rate at exonic selected sites (Table 1) substantially exceeds the per base mutation rate estimates from other studies [37], [38]. In part this excess may reflect background selection on deleterious mutations occurring outside our designated conserved segments, including mutations in other coding or exonic sites (only 63% of annotated coding bases meet our conservation threshold), and intronic mutations (including transposable element insertions) that affect splicing or polyadenylation. Widespread positive selection [39], fluctuating selection (which tends to amplify hitchhiking effects [40]), or biased gene conversion that increases the frequency of deleterious alleles [41], [42] may also contribute to the diversity reduction. We cannot at present distinguish among these possibilities, and consequently our B estimates should be interpreted as perhaps only partly reflecting background selection. A recent examination of human segregating sites by Hellman et al. found that both hitchhiking and background selection explain the relationship between diversity and recombination rate better than neutral models [43]. In their analysis the hitchhiking model gave a slightly better fit, but their results are not conclusive because their models are greatly simplified and in particular do not consider the distribution of conserved segments in the genome. We attempted to discriminate between background selection and hitchhiking models by examining allele frequency distributions in regions near or far from conserved segments (as in [8]). However, we were not able to find conclusive evidence that favored one model over the other (see Text S1, Table S4, and Figure S4). Both hitchhiking and background selection are likely to contribute to patterns of genomic diversity and future work would ideally take both forces into account [44]. The mean autosomal B value predicted by our model is 0. 74–0. 81 (bootstrap 90% CI), indicating selection has reduced autosomal diversity by 19–26% on average during hominid evolution. Genome-wide H/C divergence shows a strong dependence on B (Figures 6A and 7), as does human diversity (Figure 6B, C) even after stratifying by local GC content or recombination rate (Figure S5). This genome-wide dependence is striking given that the model parameters were estimated using only a small set of genomic data (about 8. 5 million filtered alignment columns for which 5-species data was available). To further quantify how well regional variation in neutral H/C divergence and human diversity can be explained by selection, we calculated correlations with divergence and diversity in non-overlapping genomic windows (Figures 7C and S6). Both B values and H/M divergence are well correlated with H/C divergence and human diversity. The correlation with H/M divergence is consistent with the action of selection because at least some variation in H/M divergence is attributable to selection in the ancestral population. H/D divergence exhibits a much weaker, but still substantial, correlation with H/C divergence. Since very little variation in neutral H/D divergence is likely to reflect selection in the ancestral population, this correlation is probably attributable to variation in the neutral mutation rate. H/C divergence is also well correlated with the density of protein coding sequences but not with the density of conserved segments (the majority of which are non-exonic). Thus, although selection on coding sequences appears to exert a strong influence on levels of neutral diversity, selection on non-exonic conserved segments may be too weak to have much effect in hominids. We can also now interpret several puzzling observations in the literature. H/C divergence was observed to be elevated both in high-recombination and in A+T rich regions [45], which was attributed to the action of two different mutagenic forces. Both trends are at least partly explained by the association of divergence with B, since the effects of selection are weakest in regions where recombination is high or gene density is low, and A+T-rich regions tend to be gene-poor [46]. In comparison to B, factors previously proposed to influence local mutation rates such as recombination rate and GC content [17] are only weakly correlated with diversity and H/C divergence (Figure 7C and S6). This again suggests that selection, rather than mutation rate variation, is the principal reason for these associations. Patterson et al. [21] proposed that the large variation in H/C divergence within the genome reflects relatively recent hybridization events following a much earlier split (a similar proposal was made earlier by Osada and Wu [47]). In contrast, Innan and Watanabe found no evidence supporting a model of gene flow following an initial speciation event [48] and Barton argued that much of the variation in divergence could instead be explained by a simple speciation model and a large ancestral effective population size [49]. Although a large ancestral population would give rise to genomic segments that differ widely in their H/C divergence and HG+CG site density [34], under a neutral model these segments would be scattered randomly throughout the genome. In contrast, we found that H/C divergence and HG+CG site density are preferentially depleted in the vicinity of conserved exonic sequences. This also contradicts the predictions of a complex speciation model that divergence should be lowest in intergenic regions [47]. Our results argue that much of the variation is instead attributable to the action of natural selection in a fairly large ancestral population (Figure 7). An additional anomaly identified by Patterson et al. is the unexpectedly low divergence of the X chromosome relative to the autosomes. We analyzed the X chromosome using our likelihood model (Table 1) and found that, as with the autosomal analysis, the model provides a good fit to the data and reveals a strong dependence of ancestral population diversity on B (Figure 6). The estimated average diversity reduction for chromosome X is 12–41% (bootstrap 90% CI). At neutral sites not influenced by selection the estimated effective population size for the X is only 24% that of the autosomes (Table 1), however the large confidence intervals imply that this is not significantly different from the 75% expectation of random mating models. Because of the uncertainty in our chromosome X parameter estimates we cannot determine whether the low H/C divergence across the chromosome can be explained by selection. The future availability of complete genome sequences from gorilla and orangutan should enable a more precise comparison of chromosome X and the autosomes. In a recent study of human diversity (published while this manuscript was under review) Cai et al. estimated hitchhiking or background selection has reduced neutral diversity by 6% genome-wide (11% in gene-rich regions) [50]. Their estimate is substantially lower than our own (19–26% for autosomes), but the discrepancy can potentially be explained by several aspects of their analysis. They exclude all sites near genes (within 5 kb of transcript start and ends and within 1 kb of any exon); since about 11% of the genome is within 1 kb of an exon this omits a large fraction of the sites that are the most influenced by selection. In addition, their analysis uses very large windows (400 kb) which will tend to dilute some of the effects of selection. Finally, they normalize human diversity by H/C divergence as a correction for mutation rate variaton. This normalization is overly conservative because as we have shown here, a substantial fraction of H/C divergence is itself affected by selection. In summary, our analyses reveal a dominant role for selection in shaping genomic patterns of diversity and divergence, and appear to resolve several controversies regarding hominid evolution. Our results have several implications for studies that involve human diversity or H/C divergence. Findings of reduced H/C divergence in some regions may reflect the indirect effects of selection at nearby sites, rather than direct selection or reduced mutation rates. For example, the lower H/C divergence in short introns [51] might reflect selection on nearby exons. In addition, estimates of the effective population size or neutral mutation rate should be based on regions that are distant from selected sites. The B values computed by our model should provide a useful baseline for future studies, allowing regions to be stratified by their predicted levels of neutral diversity or divergence. Loci that depart significantly from our diversity predictions warrant more detailed investigation because they may have undergone unusually strong selective or mutagenic events. Genome sequences for the human [46] (version hg18), chimpanzee [45] (version panTro2), and rhesus macaque [52] (version rheMac2) genomes, and human genome annotation files were obtained from the University of California at Santa Cruz Genomic Informatics (UCSC) web site [53]. Human protein-coding sequences and exons were identified using UCSC ‘known gene’ files [54] (downloaded Sept. 2007). Repetitive regions were identified using the UCSC lower-case markup (which is based on RepeatMasker [55] and Tandem Repeats Finder [56] analysis). Simple repeats identified by Tandem Repeats Finder were also downloaded from the UCSC simpleRepeats track so that they could be used independently. Files indicating map distance per nucleotide for deCODE [26] and Myers et al. [25] recombination maps were downloaded Feb 2007 (we used snpRecombRateHapmap files for the Myers et al. map), and transferred from hg17 using the UCSC liftOver tool. X chromosome values were multiplied by 2/3 to correct for non-recombination in males. Chromosome regions missing from the recombination maps were ignored for most analyses; however for use in calculating background selection values, we assigned each base in missing regions a recombination rate equal to that of the nearest defined base (for terminal regions of chromosomes) or the mean of the nearest defined bases from each side (for internal regions). We downloaded ‘chained and netted’ pairwise whole genome alignments from UCSC [57]–[59] for human (hg18), chimp (panTro2) and macaque (rheMac2). We converted these ‘best’ alignments to be best-reciprocal by splitting alignment blocks to omit portions that were non-reciprocal between forward (e. g. hg18 vs. panTro2) and reverse (e. g. panTro2 vs. hg18) alignments. Next, blocks aligning parts of non-orthologous chromosomes or unassigned to a chromosome region were discarded. Human and macaque chromosome regions were considered orthologous if their pairing was consistent with the synteny map of Rogers et al. [60] and for the sex chromosomes, only X to X and Y to Y alignment blocks were kept, in accordance with the synteny map of Murphy et al. [61]. We filtered out putative copy number variants and segmental duplications since these are likely to be enriched for non-orthologous alignments. Alignment blocks were omitted if more than 50% of a block overlapped regions identified as having an excessive depth of shotgun sequence reads (WSSD regions). WSSD features generated from Celera, Venter, and Watson human genome sequences as well as chimpanzee and orangutan sequencing projects were combined in order to create the set used for filtering [62], [63]. Additionally, human-chimp alignment blocks were excluded if the chimp sequence overlapped WSSD features identified by aligning chimp reads to the chimp genome, and human-macaque alignment blocks were excluded if they overlapped WSSD features identified by aligning macaque reads to the macaque genome [63]. We then grouped remaining alignment blocks into ‘chains’. Blocks were chained when their chromosomal ordering was consistent for both species. We eliminated chains with fewer than 250 kb in the human-chimp alignment, or 50 kb in the human-macaque alignment. We further excluded blocks with lengths less than 2 kb from both alignments. The remaining pairwise human-chimp and human-macaque alignments were then used to define a three-species alignment. We applied a set of site filters to individual alignment columns. We used only sequence with high-confidence base calls, requiring that each site was flanked by five sites with minimum quality scores 25 (in both chimp and macaque), and that the site itself had a quality score of at least 40. We ignored columns in the alignment that included gap characters, were adjacent to mismatches, gaps or undefined bases, or that overlapped a CpG dinucleotide in any of the three species. We also imposed a ‘symmetry’ filter to eliminate potential non-orthologous alignments by using macaque as an outgroup to assign (where possible) human-chimpanzee sequence differences to either the human or chimp branch, and eliminating regions in which more than 16 out of 20 successive substitutions were on the same branch. We downloaded pairwise human (hg18) and dog (canFam2) [64] alignments, converted these to best-reciprocal alignments as described above, and discarded blocks of length<100 bp and blocks that were unassigned to a chromosome region. Chains of blocks that had a combined length of <5000 bp were then discarded in a subsequent pass. Regional human/dog divergence was estimated by counting alignment columns in putative neutral sites (defined below) in 1 Mb sliding windows that were advanced by 1 bp at a time. Only transversion substitutions were used in divergence calculations because they gave better correlations to human/macaque divergence than transition substitutions or Kimura-corrected divergence (presumably because many sites have multiple transition substitutions). Sites were excluded from the analysis if they were in a potential CpG context (following a C or preceding a G) or if they were adjacent to a gap or N. Windows were discarded if they contained fewer than 200,000 sites post-filtering. For comparison of H/C, H/M and H/D divergence (Figures 3,7C, S2 and S6) we constructed the implied human-dog-macaque alignment and applied the same filtering criteria used to calculate regional H/D divergence, again using only transversion substitutions. The same alignment was used to calculate levels of divergence in several site classes relative to the neutral rate (Figure 5). In this case, less stringent filtering was applied (sites in a potential CpG context were still excluded), and all four species were required to have an aligned base at each site. We downloaded HCGOM alignments utilized in [21] from http: //genepath. med. harvard. edu/~reich, and extracted columns identified as used in that study (these all have at least one nucleotide difference among species). We eliminated columns that overlapped a CpG dinucleotide in any species or that were not flanked by invariant columns (i. e. columns for which all species shared the same nucleotide). In addition, we identified all invariant columns, and selected at random a subset of these equivalent to the number reported in [21]. We augmented these data with a smaller dataset generated in our own laboratory as follows. We chose primer pairs to amplify 591 human genome loci of size 1 to 2 kb spaced roughly every 5 Mb on the autosomes and 2. 5 Mb on the X. Primers were chosen to avoid repetitive regions, positions where human, chimp, or macaque differed, and the dinucleotide CpG, and trinucleotides ACA, or TGT which we have found to have higher than average mutability in primate sequences (data not shown). DNA samples from a male chimpanzee (Pan troglodytes), a male bonobo (Pan paniscus), a female gorilla (Gorilla gorilla), a female Sumatran orangutan (Pongo pygmaeus) and a female rhesus monkey (Macaca mulatta), purchased from the Coriell Institute (Camden, NJ), were PCR amplified and sequenced in both directions, using standard protocols and an ABI Prism 3100 Genetic Analyzer. After basecalling with phred [65], [66], we searched each read against the human genome and eliminated reads for which the best match was to a non-target location. The remaining 4759 reads were aligned to the human sequence using a banded Smith-Waterman algorithm (cross_match; www. phrap. org). Analyzed data were required to pass quality and alignment filters similar to those described [21]. Traces have been deposited in the NCBI trace archive (http: //www. ncbi. nlm. nih. gov/Traces). A total of 8. 5 million alignment columns passing these filters in the combined 5-species dataset were analyzed. Human single nucleotide polymorphism (SNP) data was obtained from Perlegen Sciences [19], HapMap phase II (non-redundant October 2008 update) [67], the SeattleSNPs NHLBI Program for Genomic Applications (PGA) [68] and the NIEHS Environmental Genome Project (EGP) [20], [69] (downloaded July 2008). To estimate nucleotide diversity we averaged combined-population heterozygosity for all di-allelic polymorphisms in scanned regions (assuming heterozygosity of 0 for monomorphic sites). For the HapMap data the CEU panel was used instead of combined-population heterozygosity. As with divergence calculations, we required the presence of an aligned macaque base (necessary for normalization) and excluded sites in a CpG context, sites with poor quality scores, and sites adjacent to a gap or N in the human/macaque alignment. For both HapMap and Perlegen datasets we omitted sites that fell within annotated repeats. Diversity for a given class of genomic sites (e. g. putative neutral sites having a specified background selection value) was estimated by summing the estimated heterozygosities (computed from observed allele frequencies in the samples) of SNPs in that class and dividing by the total number of scanned sites in the class. To avoid ascertainment bias in the Perlegen data we only used class A SNPs, which had been identified using array-based resequencing [19]. We converted the NCBI34/hg16 coordinates of the Perlegen SNPs to NCBI36/hg18 coordinates using the UCSC-annotated positions of the associated dbSNP identifiers. The following sites were considered unscanned in order to correct for biases in Perlegen array-based detection: sites within 25 bases of an annotated repetitive region; sites with <6 or >13 G or C nucleotides among the 24 bases (12 to each side) flanking the site (our unpublished analyses indicate that class A SNPs are strongly depleted at such positions); regions >100 kb that completely lacked class A SNPs; and regions present in the NCBI36 assembly but not the NCBI34 assembly (as identified by mapping non-overlapping 1 kb segments from NCBI36 to NCBI34 using liftOver). To address the possibility that Perlegen or HapMap SNP ascertainment strategies could bias our estimates of human diversity [70], we employed an ascertainment correction that takes into account the size of the discovery sample [71]. The discovery sample size of the Perlegen data is 20–50 chromosomes (see supplemental data for [19]). We were unable to obtain per-SNP discovery sample sizes so we calculated corrected nucleotide diversity values assuming uniform discovery sample sizes of either 20 or 50. Note that this ascertainment correction does not account for failure of the array technology to identify SNPs during the discovery process, but it should not bias our Figure 3 analyses (which compare regions near and far from conserved segments) provided discovery sample size and technology failures are not themselves biased with respect to distance from conserved segments. As expected, our ascertainment corrected nucleotide diversity estimates are higher than our uncorrected estimates, but our diversity ratios from regions near-to and far-from conserved sites are essentially unchanged (Table S5). Moreover, consistent with this expectation, we obtained similar results from HapMap phase II data, which used different ascertainment methodologies, and from SeattleSNPs EGP/PGA data derived from complete resequencing. We implemented a program, gcons, to identify evolutionarily conserved segments from aligned genomic sequences. Gcons extends the two-state phylogenetic Hidden Markov Model (phylo-HMM) approach used by phastCons [23] by incorporating alignment gap information. We define separate substitution models for nucleotide and gap evolution, and estimate substitution probability matrices on each branch of the phylogenetic tree without assuming a common rate matrix (phastCons uses a single rate matrix). Our probability matrices are constrained to be strand-symmetric (e. g. A→G substitutions must occur at the same rate as complementary T→C substitutions) but may be non-reversible. Our gap substitution model is a simple site-independent deletion model with three symbols representing defined bases (b), sites in short gaps of length≤10 bp (-), and sites in long gaps or unaligned regions (D). Because we consider only ‘ancient’ sites present in the root and assume that orthologous nucleotides are aligned, it is unnecessary to model insertions. Thus the only non-zero substitution rates are b→-, b→D, and -→D. In high coverage genomes, the absence of long gaps is indicative of functional constraint [72], but in low-coverage genomes, long gaps may simply represent coverage gaps and are therefore less informative [24]. Because our model uses separate long and short gap symbols and allows rates to vary on different branches, it can be applied effectively to a mixture of high and low sequence coverage genomes. From a set of alignment columns we obtain maximum likelihood estimates of the substitution probability matrices on each branch of a phylogenetic tree using an EM algorithm [73]. For our purposes, it is sufficient to estimate substitution probabilities directly, rather than the underlying substitution rate matrices and branch lengths. We downloaded a multiple alignment of 28 vertebrate genomes from UCSC in August 2007 [24] and extracted from this the alignment of placental mammal species. To avoid biasing our primate sequence analyses we excluded chimpanzee and rhesus macaque sequences from the alignments, leaving a total of 15 sequences plus human and we treated the human sequence as missing data for the likelihood calculations described below. For these sequences we assume the following fixed phylogenetic tree topology obtained from UCSC (http: //hgdownload. cse. ucsc. edu/goldenPath/hg18/multiz28way/28way. nh): ( ( ( (hg18, otoGar1), tupBel1), ( ( (rn4, mm8), cavPor2), oryCun1) ), ( (sorAra1, eriEur1), ( ( (canFam2, felCat3), equCab1), bosTau3) ) ). We restrict our analysis to ‘ancient’ sites defined as those present in at least one species on either side of the internal node ( ( ( (hg18, otoGar1), tupBel1), ( ( (rn4, mm8), cavPor2), oryCun1) ), ( (sorAra1, eriEur1), ( ( (canFam2, felCat3), equCab1), bosTau3) ) ). This node is used instead of the root because the three species on one side of the root (armadillo, elephant, tenrec) have low-coverage (2×) assemblies with many gaps. We estimate neutral substitution probabilities using multiple alignment columns from ancient repeats. Repeats were identified using the lower-case markup in the UCSC human hg18 sequence; to allow for repeat alignment ambiguity we excluded the 5 bp at each end of the repeat. Ancient sites that fulfilled these criteria were considered to be ancient repeats. Similarly, we used first and second codon positions in annotated coding sequences to estimate the conserved region substitution probabilities. Alignments from odd-numbered autosomes were used as a ‘training set’ input for the EM algorithm, and data from even-numbered autosomes were used as a ‘test’ set. Substitution probabilities for the X chromosome were estimated using the full set of alignment data (i. e. no test set was held out). To approximate flanking nucleotide context effects on substitution rates [74] we categorized sites by their inferred ancestral context and trained separate models for each category. Specifically, for each ‘ancient’ alignment column we designated an ancestral nucleotide by choosing the nucleotide with the highest posterior root probability, as calculated using our initial (context-free) neutral evolutionary model. We then grouped alignment columns into categories based upon their ancestral purine and pyrimidine contexts because these contexts have previously been shown to capture a substantial proportion of mutation rate variation [74]. The four possible context categories are RRR, RRY, YRR, YRY, where the center symbol is the ancestral state at the site of interest and R and Y denote purine and pyrimidine, respectively (note that reverse-complement pairs of contexts, e. g. RRY and YYR, are equivalent by virtue of the strand symmetry condition on our substitution matrices). After grouping columns by their ancestral contexts we trained separate conserved and neutral evolutionary models for each possible context as described above, retaining the initial context-free model for sites where one of the flanking ancestral states is unknown. We then computed a conserved/neutral log-likelihood ratio (LLR) for each ancient site in the human genome using these models. The LLR for non-ancient and unaligned sites was taken as the log of the rate of occurrence of such sites in conserved regions (first and second codon positions) divided by the rate of such sites in neutral regions (ancient repeats). To avoid biasing our primate sequence analyses, human sites were treated as missing data in LLR calculations. The sum of the nucleotide substitution and gap LLRs at each site may be interpreted as the log of the ratio of the emission probabilities of the corresponding alignment column by a Hidden Markov Model having two states, ‘conserved’ and ‘neutral’. We assigned state transition probabilities of 1/7 (conserved→neutral) and 0. 0075 (neutral→conserved), implying an expected conserved segment length of 7 bp and an expected conserved portion of the genome of 5%, and computed a score, S, for each site which is related to the posterior probability, PP, of being in the conserved state by PP = es/ (es+1). To identify potentially incorrect portions of the multiple alignment we used a similar procedure, defining an HMM with a neutral and a ‘high substitution’ state. Emission probabilities for the neutral state used the context-free substitution probability matrices from ancient regions, whereas those matrices raised to the 5th power defined emission probabilities for the high substitution state. State transition probabilities were chosen such that high substitution segments were expected to be of length 25 bp and span 10% of the genome. We then computed scores as above, and defined contiguous regions with scores greater than 0. 0 (posterior probability 0. 5) as high substitution segments; these comprise 8% of aligned ancient repeats, 2% of aligned intergenic bases and 0. 2% of aligned first and second codon positions. These segments likely reflect misalignments, and we excluded them before re-extracting alignment columns and re-performing the training of the conserved and neutral models described above (i. e. the columns were omitted for training, but retained in other analyses). We defined conserved segments to be contiguous sets of bases in the human genome having gcons score≥10; these are the bases with the strongest evidence for being under purifying selection. Note however that because the gcons model is designed to detect segments of a given minimal length rather than individual conserved bases, some bases within a conserved segment may be under little or no selection pressure (e. g. synonymous bases within coding exons), and short evolutionarily constrained segments may have low gcons scores. Approximately 39% of annotated exonic bases and 4. 3% of non-exonic bases meet our gcons score threshold. We classified conserved segments as ‘exonic’ if they contain any annotated exonic base, and as ‘non-exonic’ otherwise. Exonic and non-exonic conserved segments comprise 1. 1% and 4. 2% of all genomic bases respectively. Except where indicated, all analyses use putative neutral sites, which are required to be ≥10 bases away from any annotated exon, have gcons score<−10, and to pass the additional filters indicated above. Gcons score thresholds of −10 and 10 were chosen to designate neutral and conserved sites because we found that they provide good separation between putative functional sites (e. g. known protein coding sequences) and putative non-functional sites (e. g. ancient repeats) (data not shown). Following filtering, our primary datasets consisted of the following numbers of neutral sites: 1. 2 billion autosomal and 48 million X chromosome human/chimp/macaque alignment columns; 550 million autosomal human/macaque/dog autosomal alignment columns; 5. 3 million 5-species autosomal alignment columns; 6. 5 million autosomal and 0. 23 million X chromosome SeattleSNPs PGA and EGP scanned sites. Additional filtering steps were performed for the maximum likelihood analysis described below (regional H/D divergence was required to be defined, and some column types e. g. HGO, were not used). In total, 4. 7 million alignment columns were retained for the 5-species autosomal dataset and 27 million alignment columns were retained for the chromosome X human/chimp dataset. We applied the model of Nordborg et al. [5] and Hudson and Kaplan [4], which estimates the expected reduction in nucleotide diversity at a neutral site due to purifying selection at other sites as a function of deleterious mutation rate, selection strength, and recombination rate, under the assumptions that selection acts multiplicatively over loci and is strong enough that allele frequencies of deleterious mutations remain low (so that homozygotes for the deleterious allele may be ignored). Specifically, the background selection coefficient B = B (ν) at a neutral site ν is given bywhere We distinguish two different classes of selected sites x, exonic and non-exonic, and allow them to have different u and f. Accordingly B may be expressed as a product B = Bex Bnex where Using the above formulae we computed Bex and Bnex values for each site ν in the human genome, for various selection densities f, and ux fixed (initially) to a value of 1. 2×10−8. The selected sites x are taken to be the bases in exonic or non-exonic conserved segments as defined above, and rx, ν is taken to be the recombination map distance between ν and x (this slightly overestimates the actual recombination rate). For f we use truncated exponential distributions: f (t) = 0 for t>1 and t<10−5, and f (t) = C e−ct for 10−5≤t≤1 where c and C are constants. We considered f having mean values of the form a10b, (or (4/3) a10b in the case of the X chromosome) where a = 5. 0,2. 5, or 1. 0, and b = −2, −3, −4, or −5. As alternative possibilities for f we also considered point distributions, and truncated gamma distributions with shape parameters 0. 25,0. 75 and 2. 0 (using the same grid of mean values). The gamma distribution with shape parameter 0. 75 gave a slightly higher likelihood for the 5-species autosomal dataset, but not significantly so given the additional degree of freedom (Table S2). For the human/chimp chromosome X dataset a point distribution gave a slightly better likelihood (Table S2), but for consistency with the autosomal analysis we use the exponential distribution results. Classifying conserved segments as coding or non-coding rather than exonic or non-exonic, or using the deCODE [26] instead of a finescale recombination map [25], gave somewhat lower likelihoods in our preliminary analyses (data not shown). To accelerate calculations of Bex and Bnex we employed several approximations. We constructed a lookup table giving, for a range of values of r and the length of the conserved segment, values of the integral (evaluated numerically) over f. Integrals were then estimated by performing bilinear interpolation between the nearest values stored in the table. Summations over x were done segment-by-segment, approximating the sum over the segment by a continuous integral. To make this approximation more accurate, segments were broken at points where the recombination map rate per nucleotide changed. The summations over segments were then performed by starting with segments nearest to ν and moving progressively farther away on the chromosome, calculating at each step the maximum possible remainder of the summation for the entire chromosome, and stopping the summation when this maximum remainder fell below a target value (0. 001). Values for the first and second derivatives of the B' s (with respect to the position of ν) were computed by summing the term-by-term derivatives. Finally, we carried out summations only for a subset of ν' s on the chromosome, with B values for other sites estimated by quadratic interpolation using the derivatives. Our B value estimates are available for download from http: //www. phrap. org. We model the probability of the observed 5-species alignment data as a function of species divergence times, ancestral effective population sizes, and background selection on exonic and non-exonic conserved segments, in order to estimate these parameters by maximum likelihood. Our model allows for the fact that the gene tree varies along the sequence, such that at a given site any two of human, chimp, or gorilla may share the most recent common ancestor (Figure S1). Following [21] we ignore alignment columns having more than two distinct nucleotides (implying two or more mutation events at the same position), and we label those with exactly two distinct nucleotides by indicating which species share the same nucleotide; thus an HG (or equivalently COM) column, or ‘site’, is one such that human (H) and gorilla (G) share one nucleotide, while chimp (C), orang (O) and macaque (M) share a different nucleotide. We ignore most site types such as HGO which represent obligate double mutation events, however we use HO and CO counts to help estimate rates of double mutation (described below). We assume each site involves a mutational change along at most two branches of the gene tree at that position; because all branches are short, multiple events are rare. The probabilities that the sequences at the beginning and end of branch i differ by a transition (I) or transversion (V) substitution are given by Kimura' s formulae [75]: where μI and μV are the per-generation per-nucleotide transition and transversion mutation rates (so that the combined mutation rate is μ = μI+2μv), and βi is the branch length (in generations). The probability of an observed column of type k is thenif the column has two distinct nucleotides differing by a transition; if the column has two distinct nucleotides differing by a transversion; andif the column is invariant. Here S and D denote sets of branches that can give rise to the observed column type via a single or double substitution, respectively. Distinct alignment columns are treated as independent observations. The parameters λI and λV are used to scale the rates of double substitution events, which are higher than predicted by the site-independent Kimura substitution model because of mutational hotspots and flanking nucleotide contexts. Patterson et al. [21] observed that it is particularly important to take recurrent mutation into account for HG and CG columns, a significant fraction of which are the result of substitutions on multiple branches. We calculated the expected number of sites that are due to recurrent substitutions under our fitted model and compared the results to those from Patterson et al. Our estimates are in close agreement for the column types that are most frequently due to double substitution (HC, HG, and CG) (Table S6). We estimate lower rates of double substitution for some of the other column types, but since only a small fraction of these are due to double substitutions, differences in these rates should not affect our overall results. To illustrate these issues consider the alignment column GAGAA, where human and gorilla both have a G nucleotide and the other three species have an A. This column could be the result of a single A→G transition substitution on the HG branch (assuming a gene tree that differs from the species tree) but could also be due to A→G transitions on both H and G branches, or an A→G transition on the HCG branch and a back substitution (G→A) on the C branch. In this case S is (HG), and D is either (H, G) or (HCG, C). Expected branch lengths β [21], [29], [76] for each site type are given by: where is the probability that the human-chimpanzee coalescent predates the gorilla speciation, the' represent inter-speciation intervals (measured in generations), the represent ancestral effective population sizes (corresponding to in the formula for B in the Background selection section above, and as depicted in Figure 1), and B = BexBnex is the background selection value. The factor 1. 4 in the βM formula corrects for the estimated mutation rate excess in old world monkeys relative to hominids [77]. Note that in contrast to the other parameters, B depends on the sequence position. B also depends on the choice of recombination map, and on uex, unex, fex, and fnex. We assume the human-estimated B values apply to the orthologous bases in the other species, which is only approximately true because local recombination rates vary over time [27], [28]. Selection strengths may also vary, and even if they do not, differences in effective population size imply that deleterious mutations eliminated by selection in some populations may become fixed in others. For the 3-species (human/chimpanzee/macaque whole genome alignment) analyses we developed a similar model, but ignoring the macaque branch and using. To accelerate the probability calculations, we binned sites by their B values and column types. The log-probability of the data is thenwhere nB, k is the number of filtered columns of type k in bin B, and πk is the probability associated to column type k as given above. For each maximum likelihood analysis, the distribution functions fex and fnex are held fixed to compute B across the genome for a particular ux, and estimates for the remaining parameters are obtained by searching the likelihood surface with the GNU Scientific Library' s [78] implementation of the Broyden-Fletcher-Goldfarb-Shanno (BFGS) quasi-Newton method [79] (with slight modifications to prevent stalling at ridges), using analytically computed first partial derivatives. We varied uex and unex by rescaling B values (computed initially with a fixed u) as follows (where i denotes ex or nex, and and denote the updated values): Because μ is confounded with the and parameters, a calibration is required to infer individual parameter values; we fix to 240,000 generations (assuming a species divergence time of 6 MYA and a 25 year generation time), and adjust the other and values proportionately. Note also that μ is distinct from ux (deleterious mutation rate per selected site, for calculating B): in particular μ reflects alignment filtering whereas ux does not, and the estimate of ux is influenced by background selection arising from deleterious mutations at sites outside the identified conserved segments. Regional variation in neutral substitution rates [80] has the potential to bias our parameter estimates. In particular, a higher average neutral substitution rate in regions which are distant from conserved segments (potentially due to a mutational effect associated with recombination [11], [17] or insertions and deletions [81]), could be misinterpreted as evidence for selection in the ancestral population. To incorporate regional substitution rate variation into our model, we allowed mutation rates to depend upon regional human/dog divergence. Alignment column counts used for maximum likelihood estimation were binned by the regional human/dog divergence D in addition to Bex and Bnex. Rather than estimating the transition and transversion mutation rate parameters (μI and μV) directly we instead estimate parameters μA and μB and define the transition and transversion rates in each bin as μI = μAD and μV = μBD. This correction may not fully accommodate substitution rate variation if the effect is very local or has changed substantially over time. Confidence intervals in Figures 3,4, 5,6, S2 and S5 were calculated using 1000 bootstrap iterations. Correlation confidence intervals (Figures 7 and S6) were calculated by resampling windows; intervals for the other analyses were calculated by resampling counts of sites in bins, which were assumed to be binomially distributed. Confidence intervals for maximum likelihood parameter estimates were also calculated by a bootstrap procedure. In each bootstrap iteration, alignment columns were resampled with replacement. As before, columns were binned by their associated exonic and non-exonic B values (which differ for each pair of selection coefficients tried), and the local human/dog divergence. Maximum likelihood parameter estimation was done using the binned column counts and a new set of parameter estimates was obtained for each iteration. Confidence intervals for each parameter correspond to the central 90% of the ordered set of estimated values. Confidence intervals for mean autosomal and chromosome X B values were calculated using parameter estimates from the same bootstrap iterations. We performed 100 bootstrap iterations, which required approximately six days for the 5-species analysis using a 96-node computer cluster.

Comparisons of the human and chimpanzee genomes have revealed that the frequency of sequence differences between these species varies dramatically across the genome. Previously proposed explanations for this variation include a large ancestral population, variable mutation rates, or a complex speciation scenario in which humans and chimpanzees initially separated but then rehybridyzed several million years later. We consider, here, an alternate possibility; the action of selection to remove less-fit functional variants from a population has significantly reduced the frequency of “neutral” sequence differences at nearby sites. We identified sequences that are likely to be subject to natural selection because they are highly conserved across placental mammals and showed that neutral differences among five primate species are greatly depleted near such sequences. Applying a theoretical evolutionary model, we found that selection has played a greater role in shaping hominid genome evolution than has been appreciated and provides a better explanation for patterns of sequence differences than other hypotheses.

Abstract Introduction Results/Discussion Methods

computational biology/population genetics genetics and genomics/comparative genomics evolutionary biology/evolutionary and comparative genetics evolutionary biology/human evolution computational biology/comparative sequence analysis computational biology/molecular genetics molecular biology/molecular evolution evolutionary biology/genomics computational biology/evolutionary modeling molecular biology/bioinformatics computational biology/genomics computational biology evolutionary biology/bioinformatics genetics and genomics genetics and genomics/bioinformatics

2009

Widespread Genomic Signatures of Natural Selection in Hominid Evolution

ATP-dependent nucleosome-remodeling enzymes and covalent modifiers of chromatin set the functional state of chromatin. However, how these enzymatic activities are coordinated in the nucleus is largely unknown. We found that the evolutionary conserved nucleosome-remodeling ATPase ISWI and the poly-ADP-ribose polymerase PARP genetically interact. We present evidence showing that ISWI is target of poly-ADP-ribosylation. Poly-ADP-ribosylation counteracts ISWI function in vitro and in vivo. Our work suggests that ISWI is a physiological target of PARP and that poly-ADP-ribosylation can be a new, important post-translational modification regulating the activity of ATP-dependent nucleosome remodelers. Eukaryotic chromatin is packaged in a highly organized hierarchy of structural building blocks, all composed of the basic repeating unit of the nucleosome. ATP-dependent nucleosome-remodeling activities as well as covalent modifications of chromatin components underlie the dynamic nature of chromatin structure and function [1,2]. Although it is expected that a crosstalk should exist between ATP-dependent remodelers and covalent modifiers of chromatin, very little is known about how these activities are integrated and coordinated with each other. ISWI is the catalytic subunit of several ATP-dependent nucleosome remodeling complexes. ISWI is highly conserved during evolution and is essential for cell viability [3]. ISWI-containing complexes are thought to play central roles in DNA replication, gene expression, and chromosome organization [4]. ISWI uses the energy of ATP hydrolysis to catalyze nucleosome spacing and sliding reactions [3]. In Drosophila, loss of ISWI function causes global transcription defects and leads to dramatic alterations in higher-order chromatin structure, including the apparent decondensation of both mitotic and interphase chromosomes [5,6]. Recent findings indicate that ISWI controls chromosome compaction in vivo, in part through its ability to promote chromatin association with the linker histone H1 [5]. In vitro and in vivo studies carried out in several model organisms have also shown the involvement of ISWI complexes in a variety of nuclear functions including telomere silencing, stem cell self-renewal, neural morphogenesis, and the epigenetic reprogramming that occurs during nuclear transfer in animal cloning [4,7, 8]. Remarkably, inactivation of ISWI interferes with the Ras pathway [9], and loss of ISWI function seems to be associated with a subset of melanotic tumors and the human multi-systemic disease Williams-Beuren syndrome [10]. The variety of functions associated with ISWI activity are probably connected to the ability of other cellular factors to regulate its ATP-dependent chromatin remodeling activity. Indeed, nucleosome spacing reactions catalyzed by ISWI can be regulated by its associated subunits [11]. However, evidences in vitro and in vivo indicate that ISWI activity can also be directly regulated by acetylation [12] and site-specific acetylation of histones [13]. We recently found that ISWI function could be modulated in vivo by a variety of cellular factors that escaped previous biochemical analyses. Indeed, in an unbiased genetic screen for factors modifying phenotypes caused by loss of ISWI function, we identified new potential regulators of ISWI in the higher eukaryote Drosophila melanogaster [14]. One class of mutants isolated in the screen is made up of chromatin components and nuclear enzymatic activities that could regulate ISWI function by covalently modifying histones or ISWI itself. In this class we found mutants in the gene encoding for the poly-ADP-ribose polymerase—Parp—and the gene encoding for poly-ADP-ribose glycohydrolase—Parg—[14], two conserved nuclear enzymes that catalyze the transfer and the removal, respectively, of ADP-ribose units to a wide variety of target proteins, using NAD+ as a substrate, to regulate chromatin accessibility in Drosophila [15]. The activities of PARP and PARG have been implicated in modulating chromatin structure, gene expression and the response to DNA damage [16–18]. The genetic interactions identified between ISWI, Parp, and Parg suggest that poly-ADP-ribosylation reactions could be coordinated and integrated within the activity of the ATP-dependent chromatin-remodeling factor ISWI. Here we present data showing that the nucleosome remodeling factor ISWI is poly-ADP-ribosylated in vitro and in vivo. The poly-ADP-ribosylation of ISWI inhibits its ATPase activity by reducing the affinity of ISWI with its nucleosomal substrate. We found that ISWI and PARP bind different chromatin domains and that the in vivo induction of chromatin poly-ADP-ribosylation results in loss of ISWI chromatin binding, suggesting that poly-ADP-ribosylation of ISWI might favor its dissociation from chromatin. One of the central questions in the study of PARP biology is the functional role played by the poly-ADP-ribose as a covalent epigenetic mark [22]. Our data suggest a molecular mechanism to explain the coordinated functions played by ISWI and PARP in the regulation of chromatin organization in vivo, providing the first example, to our knowledge, of post-translational regulation of an ATP-dependent remodeler function by poly-ADP-ribosylation. Drosophila has a single PARP gene that spans more than 150 kb of transposon-rich centromeric heterochromatin, which is highly related to the mammalian PARP-1 gene [15]. The NAD+-dependent activity of PARP is reversed by PARG, a poly-ADP-ribose glycohydrolase [15]. PARP activity in flies has been associated with the loosening of chromatin structure that precedes gene expression at heat shock puffs [19]. On the contrary, ISWI preferentially associates with transcriptionally silent chromatin [5,6, 13,20,21], indicating that PARP and ISWI may play opposite roles in regulating chromosome structure and gene expression. Given that ISWI and PARP are both chromatin components with enzymatic activity, one possible molecular explanation for the observed genetic interaction (Figure S1) [14] could be that PARP regulates ISWI function by covalently modifying chromatin components to which ISWI binds. Double immunostaining on wild-type polytene chromosomes with antibodies directed against ISWI and PARP shows that localization of the two proteins is largely nonoverlapping (Figure 1B and 1D). We also examined PARP binding in salivary gland polytene chromosomes prepared from transheterozygous ISWI1/ISWI2 null mutant male third-instar larvae. In ISWI mutant chromosomes, showing chromosome condensation defects (Figure 1F) [5,6], PARP binding levels do not change significantly as compared to wild-type chromosomes (compare Figure 1A and 1B with Figure 1E and 1F). The observation that PARP- and ISWI-bound chromatin domains appear to be nonoverlapping does not exclude the possibility that ISWI binds poly-ADP-ribosylated (PARylated) chromatin domains from which PARP has dissociated. Therefore we conducted double immunostaining with antibodies directed against ISWI and poly-ADP-ribose (PAR) to examine how the pattern of ISWI binding to chromatin domains is correlated with that of PARylated chromatin proteins. We found that like PARP itself, PARylated chromatin proteins show mainly nonoverlapping binding patterns with ISWI on wild-type polytene chromosomes (Figure 2B and 2F). As with PARP binding, PARylated protein binding levels in ISWI mutant chromosomes do not change significantly when compared to wild-type chromosomes (compare Figure 2A and 2B with Figure 2C and 2D; see also Figure S2A). Thus, global chromosome distribution of both PARP and its enzymatic product PAR appears inversely correlated with the distribution of ISWI. To better understand the functional relationship existing between the nuclear enzymatic activities of ISWI and PARP, we examined whether PARylation levels and distributions are altered by increasing the levels of chromatin-bound ISWI. Double-immunostaining with antibodies directed against PAR and ISWI shows that ISWI is overloaded on polytene chromosomes misexpressing wild-type ISWI, as compared to control chromosomes misexpressing green fluorescent protein (GFP) (Figure 3A). The increase in chromatin-bound ISWI is accompanied by a massive increase in PARylation of chromosomes. This hyper-PARylation is not dependent on ISWI ATPase activity, because misexpression of the enzymatically inactive ISWIK159R protein has a similar effect (Figure 3A). Thus, over-PARylation of chromosomes probably occurs as a consequence of the physical binding of a nonphysiological amount of ISWI on chromatin rather than as an indirect response to an increase in ISWI activity. One interesting possibility is that increased PARylation of chromatin-bound proteins, which might include ISWI itself, could be a homeostatic response designed to counteract excessive ISWI chromosome binding. Therefore, we conducted immunoprecipitations on protein extracts derived from salivary glands misexpressing ISWI. Comparison of the inputs of native protein extracts derived from salivary glands misexpressing GFP, and HA-tagged wild-type ISWI or ISWIK159R (Figure 3B, lanes 1,4, and 7), shows that the overall level and pattern of protein PARylation are broadly similar in each. However, two bands, which run above the 125-kDa marker, are particularly abundant in the extracts from glands misexpressing either wild-type or ISWIK159R and are absent in the control expressing GFP (Figure 3B, double arrow). Immunoprecipitation of these extracts with aHA antibody detects a PARylated protein that co-migrates with one of the bands specific to the wild-type and ISWIK159R inputs (Figure 3B, compare lanes 4 and 7 with lanes 6 and 9). Remarkably, this protein migrates at the same molecular weight as the HA-tagged versions of ISWI, as revealed by blotting the same filter with aISWI antibody (Figure 3B, double arrow, compare lanes 6 and 9 with lanes 15 and 18). Our immunofluorescence and immunoprecipitation data indicate that the misexpression of ISWI causes PARylation of specific chromatin components and suggest that a main target of PARylation in ISWI misexpressing extracts could be ISWI itself. Since our data indicate that a fraction of overexpressed ISWI can be target of PARylation in vivo, we next investigated whether the same might be true when physiological levels of ISWI are present in the cell. Nickel chelate affinity chromatography conducted on larval nuclear extracts that were derived from a line expressing HA-tagged ISWI fused to 6-histidines at the C-terminal end of the ISWI coding region showed that a fraction of the eluted proteins contains a PARylated band, which migrates at the same size as ISWI (Figure 4A, double arrow on lanes 3 and 6). Consistently, tandem affinity purification (TAP) from larval nuclear extracts expressing a TAP-epitope tagged ISWI [14] reveals a co-eluting PARylated band migrating at the same molecular weight as ISWI (Figure 4D, lane 6, arrowhead). Our data suggest that ISWI could be a physiological target for PARylation in vivo. If a fraction of ISWI is PARylated in vivo, we might expect this level to increase when the activity of PARP is not counterbalanced by the action of PARG, the poly-ADP-ribose glycohydrolase that reverses the enzymatic activity of PARP [17,18]. We therefore compared the level of ISWI PARylation in wild type and Parg mutants. After immunoprecipitation of either wild type or Parg mutant salivary gland extracts with antibodies directed against PAR, sequential Western blot with aPAR and aISWI antibodies reveals a PARylated protein migrating at the same molecular weight as ISWI (Figure 4E, lanes 1 and 3). However, both the amount of PARylation and of protein detected by the aISWI antibody are higher in the Parg mutant extracts than in those of wild-type origin (Figure 4E, bracket and double arrow on lanes 2 and 4). The immunoprecipitation data we present show that a fraction of ISWI is PARyated under physiological conditions and that ISWI PARylation is increased in Parg mutants. We next explored whether the ISWI PARylation observed in vivo could be recapitulated in vitro. Classic in vitro PARylation assays show that purified PARP is enzymatically active, showing specificity for proteins that are known to be bona fide in vivo and in vitro substrates. Since the number of PAR moieties added to a substrate by PARP is variable and generates branched polymers of ADP-ribose, PARylation in vitro is detected as a signal smearing upward from the molecular weight of the modified target protein. As evidenced by the characteristic smears in an aPAR Western blot of proteins incubated with or without PARP, and the AuroDye staining of the blotted membrane showing the protein migration in the blot, the purified PARP enzyme is able to PARylate histones (Figure 4B, compare lanes 1 and 2) [23] and the recombinant p53 tumor suppressor (Figure 4B, compare lanes 3 and 4) [24] However, PARP shows no activity toward the recombinant p50 subunit of the NFκB complex which is not a target of PARP (Figure 4B, compare lanes 7 and 8) [25], while a low-background signal, derived from self-PARylation, is apparent when the same amount of PARP was incubated alone (Figure 4B, lane 9). Remarkably, when recombinant ISWI and PARP are incubated together, ISWI PARylation is visible as a strong smear that extends upward from the position of unmodified recombinant ISWI (Figure 4B, compare lanes 5 and 6). Thus our PARylation assay clearly indicates that ISWI is a specific substrate of PARP in vitro. To identify the ISWI protein domain that is the target of PARylation, we set up PARylation assays on recombinant truncated forms of ISWI. Drosophila ISWI can be divided in two parts: the N-terminal portion containing the ATPase domain and the C-terminal portion containing the SANT and SLIDE domains (Figure S2B and Figure 4G) [41]. We found that the N-terminal portion of ISWI is specifically PARylated in vitro, whereas the C-terminal portion is not (Figure 4C). Since, both the N- and C-terminal of ISWI contribute to the interaction with nucleosomal DNA [41], our data suggest that PARylation of the N-terminal fragment could target the nucleosomal DNA recognition element of ISWI present in the ATPase domain. If one of the enzymatic targets of PARP is ISWI, we should be able to monitor a physical interaction between the two proteins in vitro or in vivo. Immunoprecipitations conducted on wild-type larval nuclear extracts using antibodies directed against ISWI, or on extract derived from HA-epitope tagged ISWI using the aHA antibody, failed to detect a direct physical association between ISWI and PARP (unpublished data). Since PARP has been shown to be dynamically associated with its target sites on chromatin [26], we may not be able to detect a direct physical interaction with ISWI from crude nuclear extracts. Therefore, we decided to monitor in vitro the association of PARP with ISWI, by pull-down experiments using purified PARP and FLAG-epitope tagged recombinant ISWI. In the absence of DNA, the aFLAG-coupled resin can pull down FLAG-tagged ISWI together with a discrete amount of PARP (Figure 4F, lanes 3 and 4). When DNA is included in the binding reaction, the enzymatic activity of PARP is stimulated, as show by the self PARylation of PARP and the PARylation of ISWI detected with aPARP and aISWI antibodies, respectively (Figure 4F, compare lanes 1 and 5). Remarkably, in the presence of DNA, the fraction of PARP that is found associated with ISWI is dramatically decreased (Figure 4F, compare lanes 3 and 4 with lanes 7 and 8). Interestingly, the residual fraction of PARP that associates with ISWI in the presence of DNA does not appear strongly PARylated, suggesting that actively PARylating PARP does not preferentially associates with ISWI, probably explaining PARP dynamic association with its chromatin targets observed in vivo as well as our failure to detect PARP–ISWI interaction by immunoprecipitation from crude protein extracts (Figure 4F, compare lanes 5 with 7 and 8). To investigate whether PARylation influences ISWI ability to remodel nucleosomes, we first tested whether ISWI ATPase activity is modified in the presence of PARP activity. As PARP and ISWI enzymatic activities are differentially stimulated by linear DNA or nucleosome arrays [20], we conducted a dual assay to test ATPase and PARylation activities in the presence of recombinant ISWI together with purified PARP (with a molar ratio ISWI/PARP ∼2. 5), using both linear DNA and in vitro assembled chromatin as stimulators of these enzymatic activities (Figure 5A and 5B and Figure S3A and S3B). Both the ATPase of ISWI and the PARylation activity of PARP were assayed from the same reaction mixture for each condition tested. Interestingly, ISWI activity in the presence of PARP is reduced by ∼60% for DNA-dependent (Figure 5A, compare lanes 2 and 4), and ∼40% for nucleosome-stimulated ATPase activities (Figure 5B, compare lanes 2 and 4). The observed inhibition correlates with PARylation of ISWI, as indicated by aPAR Western blot analysis and the AuroDye staining of the blotted membrane of the same samples. We were unable to reverse the PARP-dependent ATPase inhibition of ISWI in the presence of 3-aminobenzammide (3-AB), a competitive inhibitor of PARP, in the reaction. However, this is because the amount of 3-AB that is sufficient to inhibit PARP activity (Figure 5A and 5B, lane 5), also strongly inhibits ISWI ATPase activity (Figure 5A and 5B, lane 6). Therefore, to verify that the ISWI ATPase inhibition was caused specifically by PARP activity, we lowered ∼50× the amount of PARP in the ATPase/PARylation assay (molar ratio ISWI/PARP ∼133). At these levels, PARP can be inhibited by 3-AB concentrations that are not inhibitory for ISWI ATPase activity (Figure 5C and 5D, open square graph). Even this very low amount of PARP exerts an inhibitory effect on both ISWI DNA-dependent and chromatin-stimulated ATPase activity over a 120-min time course (Figure 5C and 5D, compare filled diamond with filled triangle graphs). Furthermore, under these conditions, 3-AB reverses the effect of PARP, indicating that ISWI ATPase inhibition is specifically caused by PARP activity (Figure 5C and 5D, open triangle graph). Although chromatin stimulates the ATPase activity of ISWI ∼10 fold more then DNA, we observed a lower inhibition of ISWI by PARP in the presence of chromatin. However, a lower inhibition with the substrate that is able to stimulate more the ATPase activity of ISWI is expected. In fact, during the ATPase/PARylation reaction, the fraction of ISWI that is not yet PARylated should have a higher ATPase activity in the presence of chromatin than with DNA, thus explaining the lower inhibition observed in the presence of chromatin in our assay. During the ATPase/PARylation reaction, PARP can also generate free soluble PAR [17,18]. The PAR that is not covalently attached to ISWI could in theory account for the observed PARP-dependent ISWI ATPase inhibition. We show that under the conditions in which we conducted the ATPase/PARylation assay, free PAR cannot account for the observed ISWI ATPase inhibition (Figure S4). Thus, we conclude that it is indeed the specific PARylation of ISWI that inhibits its DNA and nucleosome-stimulated ATPase activity. The inhibition of ISWI ATPase activity by PARP could indicate either that the PARylation of ISWI directly blocks its ATPase activity after DNA or nucleosome recognition, or that this posttranslational modification counteracts ISWI ability to productively interact with its substrates. To distinguish between such alternatives and gain deeper insight into the molecular mechanism underlying the inhibition of ISWI ATPase upon PARylation, we conducted nucleosome shift assays in which a mononucleosome-enriched polynucleosome fraction purified by sucrose gradient was used as binding substrate with increasing amounts of ISWI, in the presence or absence of PARP (Figure S5A). Under the reaction conditions used, in the absence of PARP, a molar excess of ∼8 fold ISWI/nucleosomes is sufficient to shift the migration of the bulk of the nucleosome population (Figure 6A). In contrast, when PARP is included in the bandshift reaction, the mass excess of ISWI necessary to shift all the nucleosomes is at least 2-fold greater (Figure 6A). These data indicate that the PARylation of ISWI inhibits its ability to interact with arrays of nucleosomes. To exclude the idea that posttranslational modifications of the histones—present in the purified polynucleosome fraction used as substrate—contribute to this effect, we repeated the bandshift experiments using a mononucleosome that was assembled in vitro by salt dialysis (Figure 6B). As with the polynucleosomes, a molar excess of less than 8-fold ISWI/nucleosome is sufficient to up-shift the recombinant mononucleosome if PARP is absent (Figure 6B). However, when PARP is added to the bandshift reaction, there is no sign of change in migration of the DNA–protein complexes even at ISWI/nucleosome mass excess of ∼16-fold and a molar ratio ISWI/nucleosome of ∼32-fold is necessary to start to appreciate a nucleosome shift (Figure 6B and unpublished data). Both the inhibition of PARP on ISWI binding to the purified poly-nucleosomes, and the more pronounced effect on recombinant mononucleosomes, are completely reversed by the addition of 3-AB (Figure 6A and 6B). Core histones present in purified polynucleosomes, and in vitro assembled mononucleosomes can be themselves a target of PARylation and contribute to the reduction in nucleosome affinity observed with ISWI and PARP. However, ISWI DNA bandshift assays in the presence of PARP show a reduction in DNA affinity similar to the one observed with nucleosomal DNA (Figure S3D), indirectly suggesting that histone PARylation does not contribute to the reduction in nucleosome affinity observed with PARylated ISWI. To exclude directly the idea that histone PARylation might contribute to the observed loss of affinity of ISWI for nucleosomes in the presence of PARP, we first PARylated ISWI and then conducted the nucleosome binding step in the presence of 3-AB, to prevent PARylation of histones (Figure S5B). The presence of DNA, to first stimulate PARP activity in the absence of nucleosomes, causes an apparent loss of affinity of ISWI for purified polynucleosomes or in vitro–assembled mononucleosomes due to binding competition. Indeed, a 32-fold ISWI/nucleosome is necessary to shift poly- and mononucleosomes (Figure 6C and 6D). As previously observed, when PARP is added to the bandshift reaction under conditions in which both histones and ISWI could be target of PARylation, there is no sign of poly- and mononucleosome shift even with a 32-fold ISWI/nucleosome ratio (Figure 6C and 6D). Remarkably, when ISWI is PARylated and 3-AB is included to prevent PARylation of histones, we observe a loss of affinity of ISWI with poly- and mononucleosomes that is indistinguishable from conditions when both histones and ISWI could be subjected to PARylation (Figure 6C and 6D). Our data strongly suggest that PARylation of ISWI is sufficient to reduce its affinity with nucleosomal DNA, which in turn may be responsible for the observed reduction in ATPase activity. If PARP inhibits ISWI ATPase activity, by reducing nucleosome affinity through ISWI PARylation as suggested by our in vitro data, we would predict that the loss of PARP function in vivo might result in an increase of chromatin-bound ISWI. We carried out double immunostaining with antibodies directed against PAR and ISWI and compared the chromatin binding patterns obtained from Parp mutants with those of wild-type polytene chromosomes. Remarkably, the levels of ISWI are significantly higher in Parp mutant chromosomes than in wild type (Figure 7A). Thus, loss of PARP activity appears to cause a global increase of chromatin bound ISWI on polytene chromosomes. Given that PARylation of ISWI causes a reduced affinity for nucleosomes in vitro, and that loss of PARP function is correlated with an increase in ISWI chromatin binding, we might also expect that an increase in PARP activity at specific chromosome domains should result in loss of ISWI binding from the same domains. The dual detection by immuno-FISH of the hsp70 genes at cytological location 87A and 87C and ISWI has allowed us to establish that in the absence of heat shock conditions, ISWI very reproducibly binds the hsp70 locus at 87A (Figure 7B, arrowhead). However, upon heat shock, the hsp70 loci acquire elevated levels of PARylated chromatin components and PARP is required in the process of chromatin decondensation that manifests itself as localized heat shocked puffs (see DAPI in Figure 7C and Figure S6A and S6B) [19]. Remarkably, after PARylation of chromatin is induced by heat shock, among other loci we observe that the specific binding of ISWI at 87A is lost (Figure 7C, arrowhead). If, upon heat-shock, PARP activity removes ISWI chromatin binding at the 87A hsp70 locus, we would predict that in the Parp mutant, ISWI should remain bound to chromatin after heat shock. Indeed, in Parp mutant polytene chromosomes ISWI remains bound to the hsp70 locus 87A after heat shock (Figure S6C and S6D). In line with these findings, in heat-shocked salivary gland protein extracts, the amount of PARylated ISWI is significantly higher than in extracts obtained under non–heat-shocked conditions (Figure 7D). Our data indicate that binding of ISWI to the hsp70 gene can be counteracted by PARylation of chromatin at the 87A locus and that the loss of ISWI binding upon heat shock is a physiological response that is directly dependent upon the activity of PARP. Since ISWI is a target of PARylation both in vitro and in vivo, we propose that upon heat shock, ISWI bound at the hsp70 locus could be among the chromatin components that get PARylated and this in turn promotes its release from chromatin. PARP is an abundant nuclear protein that plays important roles in multiple DNA repair pathways [17]. Interestingly, high PARP enzymatic activity has also been observed in chromosomal sites where high transcriptional activity is occurring [17]. One main goal in the study of the diverse physiological roles of PARP is the identification of molecular determinants that can stimulate PARP, besides DNA damage. Recently, it has been shown that PARP activity, in the absence of DNA damage, can be stimulated by transcription-coupled, TopoIIb-dependent transient double-stranded DNA breaks [27]. Interestingly, nucleosomal histone H4 amino terminal tails have been shown to also activate PARP activity in vitro [26]. Remarkably, the amino termini of histone H4 also stimulates ISWI activity [13,28], whereas PARP' s NAD+-dependent activity is known to be inhibited by ATP, a substrate of ISWI [29]. Moreover, ISWI and PARP have opposing effects on the binding of the linker histone H1 to chromatin; human PARP-1 competes with H1 for binding to the nucleosome linkers [30,31], while ISWI promotes chromosome condensation through the loading of histone H1 [5]. PARP and H1 exhibit a reciprocal pattern of chromatin binding at many RNA polymerase II–transcribed promoters [31]. Therefore, one intriguing possibility is that PARP activity can counteract ISWI function. Indeed, ISWI and PARP appear to compete for common chromatin target sites, as supported by their nonoverlapping chromatin binding patterns. The antagonistic dominant action of PARP over ISWI may help to promote transcription at specific chromatin sites by opening chromatin and blocking higher order chromatin structure formation by ISWI, although PARP has also been shown to directly promote the formation of compact, transcriptionally repressed chromatin [30]. We found that ISWI is PARylated in vivo and in vitro, revealing the molecular basis of the genetic interaction we found between ISWI and Parp [14]. We show that PARylation of ISWI inhibits both its ATPase activity and its chromatin binding, in vitro and in vivo. Although, the PARylation of ISWI reduces its chromatin-stimulated ATPase activity by ∼40%, PARylated ISWI reduces its affinity for nucleosomes by nearly 10-fold. These differences can be explained by the different nature of the ATPase/PARylation and band shift assays. In contrast with the band shift assay, during the enzymatic ATPase/PARylation assay, the fraction of ISWI that is not yet PARylated can be strongly and rapidly stimulated by nucleosomes, thus masking the strong inhibition exerted by PARP on the PARylated ISWI fraction. PARP is in a dynamic equilibrium between its chromatin-bound and free nucleoplasm form [26]. Indeed, our in vitro data indicate that the interaction between active PARP and ISWI is likely to be transient, highlighting the dynamic nature of this functional interaction. PARP can promote or inhibit chromatin binding of a variety of nuclear factors [32,33]. The finding that PARylation of ISWI lowers its affinity for chromatin suggests a molecular explanation for the mutually exclusive patterns of ISWI and PARP on wild-type polytene chromosomes, although other mechanisms probably regulate chromatin binding of the two proteins in vivo. Cross-linking studies have shown that ISWI contacts the nucleosome at two locations: (i) on the linker DNA near the nucleosome entry/exit site, and (ii) at an internal site about two turns from the nucleosomal dyad [34]. Because ISWI needs some overhanging piece of DNA to remodel nucleosome, the inhibition of both ISWI-specific DNA and nucleosome stimulated ATPase upon PARylation probably reflects the reduced affinity of PARylated ISWI for free DNA and nucleosomal DNA that we observed in vitro. Indeed, under the conditions we conducted the ISWIK159R screen, we found that the EP3570 insertion in the Parp gene can increase the level of chromatin PARylation, thus reducing ISWI binding on polytene chromosome. On the other hand, the observation that over-expression of ISWI leads to an increase in chromatin-bound PARylated ISWI does not contradict this model; PARylation of ISWI only lowers its affinity for chromatin, but does not entirely prevent it from binding in vitro. It is generally accepted that PARP plays roles in both local chromatin remodeling and the recruitment/modulation of the activity of various factors involved in DNA replication, repair, transcription, and recombination [17,18]. Our work presents the first example of a nucleosome remodeling activity being regulated by PARylation and the first insight into the mechanism of regulation of a chromatin remodeler by PARP activity. Although our data suggest that the modulation of ISWI activity by poly-ADP-ribosylation could be a key regulatory step acting in the context of the heat-shock induction of the hsp loci in flies, further studies will be necessary to fully understand the evolutionary conservation of this mechanism and other physiological context in which ISWI regulation by PARP occurs. Flies were raised at 25 °C on K12 medium [35]. Unless otherwise stated, strains were obtained from Bloomington Stock Center and are described in FlyBase (http: //www. flybase. org). For immunofluorescence staining and protein extract preparations, ISWI1/ISWI2 male larvae were obtained as previously described [14]. Salivary glands misexpressing GFP, wild-type ISWI, and mutant ISWIK159R were obtained as previously described [5]. ParpC03256 is a new loss-of-function allele of Parp that survive until the third instar larval stage. The Parg27. 1 allele was generated as described [36]. For immunostaining of polytene chromosomes with antibody against monoclonal aPAR, slides were washed for 1 min in cold 96% EtOH; then the squashed areas were covered with 10% cold TCA for 10 min. Slides were subsequently washed for 1 min in 70% EtOH, 1 min in 90% EtOH, 1 min in 96% EtOH, and blocked. For immunostaining of polytene chromosomes with antibodies against ISWI and PAR, the slides were blocked in PBS, 3 % BSA, 0. 1 % Triton X-100, whereas for PARP staining and ISWI/PARP double staining, slides were blocked in PBS, 5% nonfat milk, and processed as described previously [21,30]. Immuno-FISH was conducted after immunostaining with ISWI and PAR antibodies, as described previously [37], except that denaturation was done for 8 min; the hsp-70 probe was labeled using Biotin-Nick translation Mix (Roche). Salivary gland protein extracts misexpressing GFP, wild-type ISWI, ISWIK159R, and from wild-type or ISWI1/ISWI2 mutants were obtained as previously described [14]. Larval nuclear extracts were prepared as described [38]. After SDS-PAGE, proteins were transferred to nitrocellulose membrane (Whatman Schleicher & Schuell) and stained with AuroDye Forte (GE Healthcare). Proteins were detected by Western blot using SuperSignal West Femto substrate (Pierce). Chemiluminescent signals were acquired with the ChemiDoc XRS imager (BioRad). 1. 5 μg of anti-HA (Roche) or anti-PAR (10H) antibodies were pre-bound to 30 μl of Protein A-Sepharose 4B Fast Flow resin (Sigma). The resin was incubated with 250 μg of salivary gland protein extracts from wild-type third instar larvae under non–heat shock or heat shock conditions, or third instar larvae misexpressing GFP, wild-type ISWI, or ISWIK159R, or with total extracts derived from 30 wild-type or 30 Parg27. 1/Y mutant adult males. The affinity purification of His-tagged and TAP-tagged ISWI were conducted as described in [14]. In pull-down experiments, 8 nmol of FLAG-ISWI [20] were incubated with 8 nmol of purified PARP-1 (Trevigen) in 1X PARP cocktail, 8ml 100mM Tris-HCl pH8,1mM MgCl2,1 mM DTT with or without 1mg of activated DNA (Trevigen) for 15 min at 25°C. For each condition tested, FLAG-ISWI was pulled-down with 50 ml of “Anti-FLAG M2 Affinity Gel Freezer-Safe” (Sigma). Unless otherwise indicated, the standard reaction contained 4 nmol of one of the following proteins: ISWI, p53 (Santa Cruz Biotech), p50 NFκB (Promega), and histones in 1X PARP Buffer, 1 X PARP Cocktail, 1μg activated DNA with or without 1nmol of PARP-1 in a final volume of 12. 5 μl (Trevigen). Samples were incubated for 1 h at 25 °C, then reactions were stopped by the addition of 3X SDS loading buffer and analyzed on 8% SDS-PAGE. The standard reaction (14 μl) contained 4 nmol of ISWI, 6. 6 mM HEPES (pH 7. 6), 0. 66 mM EDTA, 0. 66 mM 2-mercaptoethanol, 0. 033 % NP-40,1. 1 mM MgCl2,33 μM ATP, 5 μCi [γ-33P]ATP-3000 mmol-1 (GE Healthcare). Either 1 μg of activated DNA (Trevigen) or 100 ng of in vitro assembled chromatin [39] was used as substrate. In ATPase assays containing PARP, the mix was added with 1X PARP cocktail and 8 μl 100mM Tris-HCl pH 8,1mM MgCl2,1mM DTT. The chromatin used as substrate in these assays was treated with UV (290–320 nm) for 20 min on ice. In ATPase time course assays, 0. 03 nmol of PARP and 0. 14 mM 3-AB were used. In all other ATPase assays, the samples contained 1. 5 nmol PARP and 14 mM 3-AB. Unreacted ATP and free γ-phosphate were separated by thin layer chromatography as previously described [20]. ATP hydrolysis quantification was done with the Personal Molecular Imager FX System (BioRad). Poly-nucleosomes were prepared from chicken erythrocytes by sucrose gradient [40]. Mononucleosomes were assembled by salt gradient dialysis on 146-bp labeled DNA fragments as previously reported [41]. Increasing amount of ISWI were incubated in the presence or absence of PARP, 3-AB, poly-nucleosomes, or in vitro–assembled mononucleosomes in 5 μl of 100mM Tris-HCl pH 8,1mM MgCl2,1mM DTT, 1X PARP cocktail for 15 min at 25 °C in a final volume of 10 μl. Then 5 μl of 50 mM Tris-HCl (pH 8), 50 mM NaCl, 1 mM MgCl2,100 μg/ml chicken albumin, 0. 05 % NP-40,10 % glycerol, were added to each sample and the reaction was incubated for 10 additional min at 25 °C. To poly-ADP-ribosylate ISWI in absence of nucleosomes, ISWI was first incubated with 50 ng of activated DNA and PARP in 5 μl of 100 mM Tris-HCl pH 8,1 mM MgCl2,1 mM DTT, 1X PARP cocktail for 15 min at 25 °C. Then 3-AB was added to the mix, together with 125 ng poly-nucleosome or 0. 5 nmol of mononucleosomes. Poly-nucleosomes were resolved on 1. 4% agarose gel in 0. 3 X TBE at 4 °C for 50 min as previously reported [41] and detected by EtBr staining. Gels containing labeled mononucleosomes were dried and detected by Personal Molecular Imager FX System (BioRad).

The ISWI protein is a highly conserved nucleosome remodeler that plays essential roles in regulating chromosome structure, DNA replication, and gene expression. The variety of functions associated with ISWI activity are probably connected to the ability of other cellular factors to regulate its ATP-dependent nucleosome-remodeling activity. We identified one factor—the poly-ADP-ribose polymerase, PARP—that can counteract ISWI function. PARP is an abundant nuclear protein that catalyzes the transfer of ADP-ribose units to specific proteins involved in DNA repair, transcription, and chromatin structure. Our work suggests that the activity of an ATP-dependent remodeler can be modulated by poly-ADP-ribosylation in order to regulate chromatin function in vivo.

Abstract Introduction Results Discussion Materials and Methods

biochemistry cell biology genetics and genomics

2008

The Nucleosome-Remodeling ATPase ISWI Is Regulated by Poly-ADP-Ribosylation

Both for understanding mechanisms of disease and for the design of transgenes, it is important to understand the determinants of ribosome velocity, as changes in the rate of translation are important for protein folding, error attenuation, and localization. While there is great variation in ribosomal occupancy along even a single transcript, what determines a ribosome' s occupancy is unclear. We examine this issue using data from a ribosomal footprinting assay in yeast. While codon usage is classically considered a major determinant, we find no evidence for this. By contrast, we find that positively charged amino acids greatly retard ribosomes downstream from where they are encoded, consistent with the suggestion that positively charged residues interact with the negatively charged ribosomal exit tunnel. Such slowing is independent of and greater than the average effect owing to mRNA folding. The effect of charged amino acids is additive, with ribosomal occupancy well-predicted by a linear fit to the density of positively charged residues. We thus expect that a translated poly-A tail, encoding for positively charged lysines regardless of the reading frame, would act as a sandtrap for the ribosome, consistent with experimental data. While it is known that there is great variation in ribosomal velocity along even a single transcript [1], what determines how fast a transcript (or part thereof) is processed is unresolved. Resolving this issue is important for understanding causes of disease and for the generation of transgenes, as changes in the local translation rate along mRNAs have been implicated in the regulation of protein folding [2], error attenuation processes such as no-go decay in yeast [3], transcription attenuation in bacterial systems [4], and correct protein localization [5], [6]. For some time it has been hypothesized [7]–[10], and commonly assumed (e. g. , [11], [12]), that codons matching rare tRNAs slow ribosomes along transcripts due to differential tRNA availability. The supposition is that codons corresponding to less abundant tRNAs are translated at slower rates as the ribosome must pause while the appropriate tRNA becomes available. This, for example, is held up to explain the usage of codons specified by the most abundant tRNAs in the most highly expressed genes [13], [14]. Although the notion that rare codons must stall ribosomes is commonplace, recent work has started to undermine the supposition that differential usage of synonymous codons will significantly alter the rate of ribosomal translocation within a transcript under normal conditions [15]–[17]. Indeed, much of the evidence cited as support for an effect on translational speed is questionable (see Note S1) and many of the patterns attributed to selection for translational speed are better explained in terms of selection on codon usage for translational accuracy [18]–[21]. Codon usage, however, is not the only potential factor affecting elongation speed. Double-stranded mRNA hairpin or pseudoknot structures are thought to impede progress of the ribosome [22], [23]. The generality of this during elongation, however, is unclear, as other studies [24] suggest that the ribosome can more readily melt moderately stable secondary structures once initiation has taken place. While the above factors consider ribosomal velocity to be modulated by properties of the mRNA, much less attention has been paid to the possibility that the resultant protein might impact translation rates. However, recent experimental work on recombinant peptides has shown that positive charges on the newly synthesized peptide might slow ribosomes [25], [26]. This is conjectured to be owing to an electrostatic interaction between the cation in the emerging polypeptide and the negatively charged exit tunnel of the ribosome [25], [26]. Following on from this, it has been suggested that positive charges, codon usage bias, and transcript folding play a role in ribosomal stalling at 5′ transcript ends [27], [28]. Here we ask not whether certain features can sometimes modulate translation speed along a transcript (e. g. , when grossly overrepresented in transgenes; see Note S1), but if they do as evolved in endogenous genes when expressed at “normal” levels, and to what extent. Ribosomally protected mRNA footprints from an experimental Saccharomyces cerevisiae dataset [29] enable us to profile the location of ribosomes across the S. cerevisiae transcriptome. Under the assumption that ribosomal densities inversely reflect ribosomal velocity [30], [31], we independently examine the effects of codon usage, mRNA folding, and positive charge on ribosomal speed throughout endogenous yeast genes. We show that positive charges in the nascent peptide slow the ribosome along transcripts in an additive manner in vivo, and that this slowing effect cannot be accounted for by mRNA structure, and even far surpasses that (if any) induced by codon usage bias. Within transcripts, those regions with the highest ribosomal occupancy are those most likely to be just downstream of positively charged residues. The cation sandtrap effect has potential relevance for the evolution of the poly-A tail, specifying as it does a series of positively charged amino acids if translated. Ribosomal footprint data [29] allow us to examine changes in the rate of translation given the assumption that the slower a ribosome travels along a given portion of a transcript, the more likely it is to be found there at any point in time [30], [31]. In the case of codon usage, we expect to see any possible ribosomal stalling centered over the rare codon (s) while the ribosome awaits a tRNA to enter its A-site. Hence to examine the effect of a sequence feature such as rare codons on the speed of translation, we calculate the relative change in stringently mapped ribosomal densities that occurs within a single transcript as ribosomes begin to translate regions of transcript enriched for rare codons (see Methods and Figure 1). To this end, within each transcript we compared the ribosomal occupancy at codon positions (rpos) in the vicinity of clusters of rare codons (rpos) to the average ribosomal occupancy of the 30 codons immediately preceding the first rare codon in the cluster (rprec30). We then averaged the relative increase or decrease in ribosomal occupancy across transcript sections aligned by rare codon clusters. A mean rpos/rprec30 after the clusters >1 indicates a denser sampling of ribosomal footprints on average and hence slowing at that codon position, while a mean rpos/rprec30<1 denotes sparser ribosomal coverage, consistent with acceleration. In our main analysis we make use of the tAI (range 0–1) as a measure of codon optimality as this metric uniquely reflects the tRNA pool. The tAI of a sequence is defined as the geometric mean of the relative adaptiveness of its constituent codons to the tRNA pool available in that organism [32]. A higher tAI indicates the codon has a high abundance of decoding isoacceptor tRNAs and, according to the codon usage hypothesis of translational speed, should be translated faster on account of its ready coupling with an aminoacylated tRNA. A lower tAI conversely indicates a codon that is matched by a low number of tRNAs and is therefore putatively slowly translated and nonoptimal. Here we define “rare” codons to be those in the lowest quartile of tAI values (Methods, “The Average Effect of Codon Usage on Ribosomal Densities”) (see also Figures S1, S2, S3 and Table S1 for analysis of rare codons defined according to genomic frequency). Our results show inconsistent trends in ribosomal occupancy after rare codon clusters when all clusters of a given size are aligned and the average increase in ribosomal density after the cluster (here uncontrolled for covariates) is plotted (Figure 2A). This inconsistency is still apparent when we consider rare codons to be not those with a low tAI but those that are genomically infrequent (Figure S1). If there is any slowing due to rare codons, we should expect an increase in the amount of slowing along the mRNA as the number of rare codons increases. However, no such trend is evident (Figure 3A). This lack of influence of rare codon usage on ribosomal speed is not owing to a covariance between rare codon clusters and expression levels (Table S2). Shifting the location of the “preceding 30 codons” we use to normalize footprint values slightly upstream, to accommodate the 5′ portion of the ribosome potentially slowed over a rare codon, still detects no slowing due to codon usage (Figure S4). As it has been postulated that tandem nonoptimal codons may more strongly inhibit progression of the ribosome than scattered rare codons [33], [34], we also investigated whether consecutive rare codons (adjacent codons, each from the lowest quartile of tAI values) may be affecting ribosomal velocity. Examining changes in ribosomal densities after pairs, triplets, and so forth of rare codons, however, also indicates that consecutive rare codons do not systematically slow ribosomes (Figure S5 and Figure 3B). We achieve similar findings when defining rare codons according to their genomic frequency (Figure S2). If the above results are correct, then we should also find that codon usage cannot explain ribosomal slowing when we compare sites within a given mRNA. Upon locating the highest and lowest ribosomal occupancy portions within a given mRNA, we determined whether the denser region was associated with a putative ribosome-slowing feature: lower tAI, or more rare codon pairs or rare 6-mers (two adjacent in-frame codons that, as a pair, come from the lowest 10% of all 6-mers within the genome) (see Methods, “The Relative Contributions of Charge, Folding, and Codon Usage to Extremes of Slowing Within Transcripts”). Considering all transcripts, the most slowly translated region within an mRNA in fact tends to be comprised of more optimal codons or fewer rare pairs, suggesting low codon optimality does not cause slowing (Table 1A, B and Figure S6). These results are not affected if we consider suboptimal codons to be those that are genomically infrequent (Table S1 and Figure S3). Nor do we find that transcript similarity to the yeast Kozak sequence can explain slowing within these regions (Figure S7 and Table S3). Additionally, as the difference in ribosomal occupancy between the two intra-transcript windows increases (and hence the presumed difference in the inferred ribosomal velocities between the two windows grows all the more), the already low proportion of transcripts for which tAI, genomic infrequency, or presence of rare pairs could possibly explain ribosomal pausing in fact decreases (Table 1A, B and Table S1). In other words, in transcripts that have the greatest differences in ribosomal densities along their length (as inferred from the highest and lowest ribosomal occupancy windows), and hence that contain the greatest degree of internal slowing relative to maximum translation speed, the most ribosomally occluded windows are even more likely to be comprised of more optimal codons. This indicates that not only is low codon optimality incapable of explaining ribosomal slowing in general, it is even less capable of explaining the greatest relative slowing within a transcript. We note that the decrease in the ability of codon usage to explain slowing in the upper quantiles (Table 1A) is simply a side effect of differential amino acid usage between the two windows. When we control for differential amino acid content between the two windows, we no longer see the decrease in the ability of codon usage to explain slowing, but codon usage still remains unable to explain the slowing that is observed in any of the quantiles (Table S4). Thus, in addition to the above finding that codon usage becomes less able to explain slowing as the degree of slowing grows (as deduced from observed transcripts), this amino-acid-controlled analysis suggests that even if amino acid sequence had evolved in any other way, codon usage would still not be a factor in the slowing of ribosomes. It is possible that codon usage could have different effects during different times of cell cycle if tRNA levels fluctuate [35]. We do not, however, detect a systematic influence of codon usage on ribosomal speed even under amino acid starvation conditions (Figures S8, S9, S10 and Table S5) when presumably tRNA charging levels are lower, making codon usage potentially more rate-limiting [36], [37]. If neither codon usage nor consecutive rare codons can explain variation in ribosomal speed, then what can? As it has been suggested that transcript structure can impede ribosomes along the length of the transcript [28], we next investigated whether RNA structure might be the major contributor to slowing. We used empirically determined (rather than computationally predicted) RNA structure data (PARS values, see Methods, “The Average Effect of Transcript Structure on Ribosomal Densities”) [38]. S. cerevisiae protein-coding sequences were scanned for stretches whose average PARS value was 0 or negative (and hence tending to be single-stranded), which were immediately followed by a block of codons whose average PARS value was positive (i. e. , with propensity for double-strandedness). The general contribution of folding to slowing was examined by calculating the relative change in ribosomal density (rpos/rprec30) at each position of the identified region of a transcript, where rprec30 is the average ribosomal occupancy in the single-stranded block. We then take the average of this ratio across transcripts aligned by identified blocks of structure. The method is similar to that used above with codons, but with one complication. In the case of codon usage, we have a prior expectation that any ribosomal pausing should occur while the ribosome is positioned over the “slow” codon. It is not immediately clear, however, where along the transcript we should expect any structure-induced pausing to take place. After translating an unstructured span of mRNA, will the ribosomal active site be able to get very close to the first double-stranded ribonucleotide it meets before it is finally slowed, or might pausing take place more 5′ if the ribosome progression is sterically occluded at some distance upstream? We investigated both hypotheses. We cannot immediately distinguish between the possibilities that mRNA folding has an effect on ribosomal progression either upon or upstream of the folded ribonucleotides in question, as some degree of pausing is observed in both cases (Figure 4). But how strong is this slowing effect? Could mRNA folding account for the bulk of the variance in ribosomal speed observed along transcripts? We find, again comparing the slowest and fastest translated regions within a given mRNA, that not only is secondary structure incapable of systematically explaining the slowest regions of translation, but the presence of secondary structure decreases as the difference between the ribosomal density (i. e. , difference in translation speed) of the two intra-transcript windows increases (Table 1C). Hence we conclude something other than mRNA folding must be responsible for the greatest slowing within transcripts. We performed a parallel version of the codon cluster analysis to look for changes in ribosomal density after differently sized clusters of encoded positive charges (see Methods, “The Average Effect of Positive Charge on Ribosomal Densities”), calculating the average relative change in ribosomal density within a transcript (rpos/rprec30) after positively charged residues (lysine, arginine, or histidine) are added to a nascent peptide chain. The effect, note, should be a stalling after the codon specifying the charged amino acid as the stalling process is hypothesized to be an interaction between the charged amino acid and the charged exit tunnel [26], [39]. We find that a single positive charge will slow the ribosome relative to the preceding sequence (Figure 5), regardless of whether the codon encoding the residue is A/G- or C-rich (Figure S11). Our findings show that at maximum (in real transcripts), ribosomes are more than twice as likely to be found at a given region of the transcript as before the addition of the cation to the polypeptide (Figure 5). The higher the density of positive charges in a peptide, the proportionally greater the effect (Figure 3C), in agreement with experimental findings that increasing the number of positive charges locally correspondingly increases ribosomal dwell time [26]. Our estimation of charge-induced pausing is conservative since some ribosomal density after charges is not included in the analysis if the mean ribosomal occupancy of the 30 codons preceding a charged cluster is 0 for a given transcript (our method in this case would require division by 0). We can also test whether charge is responsible for slowing by noting that the pKa, and hence overall net charge, of histidine is lower than that of either arginine or lysine at physiological pH. Thus we should expect a weaker slowing effect due to histidine residues being added to the polypeptide. When we re-calculate the slowing effect after a single positive charge (as shown in Figure 5, first panel), but separate the single charges according to whether or not they are histidine, we indeed observe that histidine causes weaker slowing (Figure S12). The slowing effect after a single histidine residue, as calculated using the area under the curve method, is anywhere from 25%–78% (95% CI) of the slowing found after a single lysine or arginine. As histidine is used much less frequently than either of the other positively charged residues, we consider slowing after single positive charges to be the best comparator due to the larger sample sizes available. When we separate larger positive charge clusters according to their histidine content (at least one histidine in the two- or three-charge clusters, and at least two histidines in the four- or five-charge clusters), we note that the slowing due to the histidine-enriched group is always lesser than that after the histidine-free group (Figure S12). If charge is a major determinant of ribosomal slowing, then it should be capable of explaining the regions of greatest translational pausing within transcripts (see Methods, “The Relative Contributions of Charge, Folding, and Codon Usage to Extremes of Slowing within Transcripts”). We find this is indeed the case. Of all the putative slowing features we consider, only positive charge is more often associated with the higher occupancy window within each transcript (Note S2). Breaking the comparisons into quantiles according to the magnitude of difference in ribosomal occupancy between each pair of windows further reveals that positive charge is the feature most often responsible for not just slowing when comparing between transcripts, but the greatest magnitude of slowing within any given mRNA. As the difference in ribosomal occupancy between the two windows increases, the window with the higher ribosomal occupancy tends increasingly to be the one with more positive charges (Table 1D). In fact the only clearly significantly overused amino acid in the higher occupancy windows is lysine, which is positively charged (Figure S13). This increase in ribosomal occupancy cannot be explained by physiochemical properties of other amino acids, namely hydropathy, negative charge, or polarity (Tables S6, S7, S8, S9). We note that even when both windows in a transcript have the same number of charges each, there is no predominant influence of tAI, rare codon pairs, or RNA structure on ribosomal slowing (Tables S10, S11, S12). The positive charge effects seen above could potentially be explained as covariate to codon usage bias, were, for example, codons specified by rare tRNAs especially abundant near those specifying positively charged residues. Given the absence of evidence for codon usage bias to affect translation rates, this now seems unlikely. To nonetheless test whether this is the case, we examined patterns of codon usage in the vicinity of positive charges similarly to the manner in which we investigated changes in ribosomal occupancy after positively charged clusters above. Thus if nonoptimal codon usage were causing the slowing patterns after encoded positive charges observed in Figure 1, we should see, on average, a relative decrease in tAI in those sites with elevated ribosomal occupancy. Contrary to this expectation, however, the trend for ribosomal occupancy to increase after positive charges (Figure 5) is independent of patterns of codon usage (Figure S14). It is also possible the slowing effects observed after positive charge clusters in Figure 5 occur ancillary to mRNA secondary structure, as such structure may have some slowing effect (Figure 4). Again we allow for mRNA folding to impede the flow of ribosomes starting either locally or 10 codons upstream (in the case that local double-strandedness creates a structure within the transcript that sterically occludes ribosomes from progressing further toward codons within the folded structure). We find that patterns of transcript secondary structure near positive charge clusters are unable to explain the pausing after translation of positive charges (Figure S14). Hence we argue that mRNA folding cannot explain the slowing seen in Figure 5, which is perhaps not surprising given its apparently weak effect on the whole (Figure 4). Given that positive charge slows ribosomes, we should expect that some of the (relatively weaker and/or inconsistent) ribosomal slowing at rare codon clusters or transcript secondary structure might in fact be due to the presence of uncontrolled-for positive charge. We find this to be the case. When groups of rare codons that are followed by either a lesser or greater number of positive charges are plotted separately, it is clear that rare codon clusters do not in and of themselves slow ribosomes (Figure 2B) but that the apparent (yet unsystematic) slowing in Figure 2A is in fact due to the presence of positive charge after some of the codon clusters (Figure 2C). Similarly, sorting by the number of positive charges present after a cluster reveals that some of the slowing observed at structured regions of transcript is likely due to previously unaccounted-for positive charge (Figure 4). We find that codon usage and transcript secondary structure do not substantially affect ribosomal velocities systematically across endogenously occurring transcripts. Although it has been suggested that amino acid starvation might increase the ability of codon usage to modulate ribosomal speed [37], we find no such effect upon examination of ribosomal footprints taken from amino-acid-starved yeast (Figures S8, S9, S10 and Table S5). We do not, however, wish to assert that codon usage and RNA structure can never affect translation rates. Certain secondary structure configurations may substantially impact ribosomal flow. As regards codon usage, if we return to the original logic by which codon usage was thought to affect translation rates, we can both see where the prior logic was misleading and in turn can predict when codon usage should slow ribosomes. The classical logic supposes that because common codons are specified by abundant tRNAs, the waiting time for the ribosome to capture the necessary tRNA must be lower for “optimal” or common codons. The key parameter, however, to determine waiting time is not the absolute tRNA abundance (as often considered) but the tRNA availability. We note, similarly to Qian et al. [16], that if codons are used in proportion to tRNA availability [40], then this could dampen any pausing effect, since rare codons matching rare tRNAs will not be as rate-limiting as if they were used more often. Put differently, if highly abundant transcripts all require the same tRNA, then this acts as a drain on the availability of that tRNA. This can be described in terms of supply and demand economics. In the case of rare codons in lowly expressed transcripts, the supply (the pool of tRNA) is small and the demand (number of codons requiring that tRNA at any given time) low. For a common codon in an abundant transcript, the supply (tRNA pool) is large but the demand is also large. We can then imagine an equilibrium situation in which the ribosome waiting time is the same for all codons as the demand (absolute codon abundance in transcripts) and supply of tRNAs are balanced. This is consistent with our observation that, under normal growth conditions, codon usage does not predict ribosome occupancy. However, the same model can predict that under abnormal conditions, we might see an effect as the situation has been forced far out of supply–demand equilibrium. Greatly overexpressing a transcript rich in rarely used codons should slow the ribosome as the demand for the rare tRNAs now exceeds supply. Likewise, we expect that gross modification of tRNA pools should have gross effects on translational speed as the system has been shifted away from the demand–supply equilibrium. This distinction between normal (equilibrium) and experimentally forced (nonequilibrium) conditions makes good sense of the prior literature, where reports of an effect of codon usage on translational velocity involved experimentally forced conditions (for review, see Note S1). Further evidence that the impact of codon/tRNA abundance is buffered comes from the report that some codons whose aminoacyl-tRNAs are selected either intrinsically rapidly or slowly by the ribosome have either low or high tRNA concentrations within the cell, respectively [41], suggesting that intrinsic differences in the translation speeds of certain codons are not accentuated but rather compensated for. The evidence for codon usage/tRNA buffering indirectly suggests either that some property other than speed causes selection on codon usage (e. g. , accuracy of translation [18]–[21]) or that selection for speed occurs when the demand–supply balance is perturbed, for example when selection acts on growth rates and favor duplications of tRNAs. That codon usage also has little or no effect on ribosome velocity in mammals [15] as well as yeast is then, in retrospect, perhaps not so unexpected. Our results are consistent with the interaction of the cations in the protein with the ribosomal exit tunnel [25], [26], a model supported by the stalling being displaced from the location on the mRNA of the codons specifying the positive charge. Our results also indicate that positive charge, more than other chemical or biophysical properties of amino acids (see Tables S6, S7, S8, S9), is key. While some highly conserved amino acid sequences have been shown to interact with the ribosomal tunnel to stall translation in order to regulate the specific gene product they control (see, e. g. , [42]–[44]), our results suggest a fundamental feature of proteins that slows ribosomes regardless of sequence context (either the local amino acid sequence or the gene in which they reside) and without the addition of trans acting factors. A general slowing of translation due to positive charge has ramifications for the evolution of the poly-A tail. If translated, the poly-A tails results in a long run of positively charged lysines. This is expected to stall run-on ribosomes [39]. This stalling may glue the aberrantly translated peptide to the ribosome, preventing potentially toxic products from diffusing into the cell and/or permit tagging of the peptide in the nascent chain–ribosome complex with a signal for degradation, as observed [26], [39]. Our results are consistent with translation of poly-A tails stalling ribosomes. Extrapolating the linear trend for larger clusters of positive charges to additively slow ribosomes (reported in Figure 3C), we note that a poly-A tail of 80 consecutive adenines (∼27 lysines) in yeast [45] should slow translation at least 4-fold more than that observed in clusters of six or more positive charges (Figure 3C), probably halting it. This is in line with experimental work showing that while nonstop mRNAs without poly-A tails are efficiently translated [46], translation of polyadenylated mRNAs lacking stop codons or full 3′UTRs is repressed after initiation [47]. Similarly, inserting a poly-A tract into a coding sequence represses translation post-initiation, but not on account of rapid mRNA decay [39]; a similar finding was reported for 3′ poly-A tails [48]. Recently, it was shown that translation of 12 consecutive basic amino acids inserted into a reporter gene causes not only translation arrest but degradation of the polypeptide [49]. Why is the tail poly-lysine if any positive charge will do? The reason is likely to be found at the DNA sequence level. Of all codons encoding positive charges, only lysine possesses a codon that is a triplet repeat of a single nucleotide (AAA) and therefore may be added simply and sequentially by a single enzyme. Moreover, the triplet repeats form a homogenous run of adenines, meaning that positive charges will still be added to the nascent chain (and hence stall ribosomes) no matter how the stop codon is missed, be it by failure to interpret the stop when in-frame or owing to frame-shifting. This may have less relevance in species with long 3′UTRs, in which an alternative stop may be found with the UTR, but in the ancestor in which the poly-A tail evolved, if 3′UTRs were short, then this sandtrap for ribosomes may have been of considerable benefit. It is noteworthy that bacteria, which for the most part lack poly-A tails, have an alternative mechanism (tmRNA) to tag and destroy proteins resulting from frameshifting or stop codon readthrough [50]. Stalling initiated by positive charges resulting from translation of poly-A tails in eukaryotes and tmRNA system in prokaryotes may be functionally equivalent modes of error correction [51]. Both sequenced ribosomally protected fragments and sequenced fragmented total mRNA for S. cerevisiae dataset GSE13750 [29] were downloaded from the NCBI Gene Expression Omnibus at www. ncbi. nlm. nih. gov/projects/geo. The rich media and amino-acid-starved sets were considered separately. Annotations of the S. cerevisiae S288C genome as available on June 22,2008 (the build used by Ingolia et al. [29]) were obtained from the Saccharomyces cerevisiae Genome Database (www. yeastgenome. org). Only protein-coding sequences of nondubious classification were considered, giving 6,262 genes for potential analysis. Any sequences containing nonsense codons or that were not multiples of three were excluded. The sequences were further filtered to only allow the standard or alternative start codons indicated in NCBI genetic code Table 1 from http: //www. ncbi. nlm. nih. gov/Taxonomy/taxonomyhome. html/index. cgi? chapter=tgencodes, leaving 6,215 sequences for analysis. The chromosomal location and coordinates of the sequenced fragments given in the original dataset were used in combination with the start and stop coordinates of genes from the annotations to determine which fragments map to which genes, and in the case of the footprint fragments, where along the coding sequence the protected area lies. Since the probability of sequencing error in a stretch of ∼28 nucleotides is quite low (for runs <50 bp on Genome Analyzer 2, error rates are expected to be around 1%), only one mismatch between the sequenced fragment and reference genome sequence was allowed. All fragment counts were taken as the average value of the two experimental replicates. Fragments that were sequenced at least once in one replicate but not listed in the other were marked as having an expression count of 0 at analogous positions in the latter replicate. In the case of fragments that map to more than one possible genomic location, it is impossible to tell which are the true areas covered by ribosomes. In order to avoid the introduction of false-positive ribosomal occupancies en masse into the dataset, which could systematically bias the types of sequences that are occluded, only footprints that mapped uniquely to one location in the reference genome were considered. In line with Ingolia et al. [29], we assigned footprints to protein-coding genes of nondubious classification if the first base of the footprint mapped to 16 nt before the first base or 14 nt before the last base of the gene, in order to take account of which area of the footprint is likely in the ribosomal active site. Since the chance of sequencing another fragment from a stretch of coding sequence increases as a function of gene length, mRNA fragment counts were normalized by dividing by gene length for the relevant gene. In addition, the footprint counts were then divided by the normalized mRNA counts mapping to that gene to obtain per-transcript ribosomal densities (indexed by location along the transcript). We performed this normalization by mRNA to ensure that differences in occupancies we calculate (see Methods, “The Average Effect of Positive Charge on Ribosomal Densities”) are not an artefact of mRNA levels. This left us with a final 5,430 filtered genes with footprint coverage mapped per codon pair per transcript. We note that there are two previous studies [27], [28] that examined this ribosomal footprint data [29] and found a role for codon usage in modulating ribosomal speeds, mainly by detecting a correlation between the local codon optimality along a transcript and the corresponding local density of ribosomal footprints. We cannot offer a reason why these studies produce such a finding, namely because they do not detail their methodology concerning the ribosomal footprint data, including whether they used all or just a subset of all the sequenced footprints (e. g. , dependent on footprint length, the number of mismatches to the genome reference sequence allowed, or the number of places in the genome to which a single footprint could simultaneously map). Another study [16] that examined the same data contradicted the finding that codon usage affects ribosome velocity, highlighting the importance of methodology in the analysis of ribosomal profiling data. This opposing study [16], however, may have mapped footprints to multiple genomic locations and also considered only footprints 28 nt in length in an attempt to precisely map which codon is in the A-site and hence selecting an aminoacylated tRNA. We are not confident that the interpretation of ribosomal footprint data allows for such specificity, as the interpretation of where the A- and P-sites are along the ribosomal footprint seems to be inferred from the average footprint length obtained during initiation and termination [29], whereas the conformation of the ribosome and hence footprint obtained during elongation may differ. Also, it has since been noted that elongation inhibitors, such as cyclohexamide, which was used in the creation of the dataset under consideration [29], alter the conformation of the ribosome, leading to advised caution in determining position-specificity from individual footprints [15]. For these reasons, we consider it optimal to stringently map footprints to a single location in the genome, thus preventing the introduction of a false correlation between certain codons and ribosomal density, and to consider all of the sequence that is occluded by the footprint instead of attempting to pinpoint the location of a structural site in the ribosome from the artefact of the footprint. To determine what determines occupancy, we could consider some general linear model in which we employ multiple parameters (local codon usage, local RNA stability, and local charge density) to predict occupancy on a codon-by-codon basis. However, such models assume that the data points are independent. Owing to the nature of the data (ribosomes sit over spans of sequence), the occupancy seen at one codon by necessity is nonindependent of that seen at neighboring codons. Thus, such methods are not generally valid. To overcome the nonindependence problem, we do not consider each codon as a separate data point. Rather we consider the dimensions of the spans of increased relative occupancy and consider how trends in the dimensions of these spans correlate with the density of the potentially slowing feature in question, an approach that we outline in Figure 1 and below. We consider for any feature (e. g. , a cluster of codons or positive charges) the start position of this feature (position x = 0). We then define for each codon at and after the start of the feature (x≥0) how the occupancy is related to the mean occupancy of the 30 codons upstream of x = 0 within the same mRNA as the feature. We define the relative occupancy of any given codon (rpos/rprec30) as its occupancy (rpos) divided by the mean occupancy of the 30 codons prior to the considered feature (rprec30). For plotting purposes, we also normalize the occupancy of all codons 5′ of the focal position at x = 0 by the same rprec30 value. Dividing by pre-cluster ribosomal densities to obtain a ratio normalizes for differences between transcripts such as expression level, accommodates and normalizes for differences in ribosomal density that may be caused by characteristics of upstream sequence, and allows for comparisons of the relative change in ribosomal movement across different mRNAs. These relative occupancy ratios, which we calculate surrounding every identified feature, thus represent the speeding (if the ratio is <1) or slowing (if the ratio is >1) of ribosomes after a given feature as they translate that portion of that gene. After calculating the relative ribosomal occupancy ratios surrounding a feature within all available mRNAs, we then align these mRNAs by the start of that feature and calculate the average relative occupancy ratios across transcripts. We plot these averages such that y at codon x = 1 is the mean ratio of all the observations across multiple RNAs at x = 1, codon x = 2 is the mean ratio across multiple RNAs at x = 2, and so on. These plots then present a span of increased relative occupancy or of decreased relative occupancy following the start of the feature at x = 0 across all instances of that feature available to our analysis. As tRNA gene copy number has been shown to strongly correlate with tRNA abundance [27], [52], preferential use of codons that base-pair to the anticodons of high-copy tRNAs is taken to reflect adaptation of coding sequence to the tRNA pool and hence optimal codon usage for translational efficiency and/or accuracy. Each codon can then be ascribed an adaptiveness value (Wi) [32]. The tRNA adaptation index (tAI) is the geometric mean of the scores for the constituent codons and is then a measure of the degree to which protein-coding genes use codons corresponding to tRNA isoacceptors with high gene copy numbers within a given genome (although, see also Figures S1, S2, S3 and Table S1 for analyses of rare codons where “rare” is defined as genomically infrequent) [32]. The codonR package to calculate tAI was downloaded from http: //people. cryst. bbk. ac. uk/~fdosr01/tAI/index. html on May 7,2011. Yeast tRNA genes were obtained from the UCSC Table Browser [53] at http: //genome. ucsc. edu/cgi-bin/hgTables. Statistical calculations for the tAI (and for other analyses generally) were done in R [54]. As a gene with just one codon would have a tAI value equal to Wi of that codon, we refer in the text to a codon' s tAI value. In the main text we define rare codons to be those in the lowest quartile of Wi values as derived for yeast (CGA, ATA, CTT, CTG, CTC, CGG, AGT, CCC, GCG, AGC, CCT, TCG, TGT, ACG, and GTG). We interchangeably use “rare” for “non-optimal” as the frequency of codon usage in yeast is roughly proportional to the numbers of tRNAs that can decode them [40]. Protein-coding sequences were scanned for single rare codons; two rare codons anywhere within a five-codon stretch, three rare codons within eight codons, four or five within 10, and six or more within 16. The cluster specifications outlined here were chosen to maximize cluster sample sizes while incorporating the following caveat: we required that a block of 30 non-rare codons had to precede the identified codon clusters and that no other rare codons could be present in the next 30 codons apart from those in the identified cluster. When investigating consecutive rare codon clusters in a parallel analysis, we required that no consecutive rare codons could be present in the surrounding ∼60 codons apart from those in the identified cluster (single rare codons were permitted as otherwise sample sizes would be far too small). As noted above, the first rare codon in the cluster is always considered to be at position x = 0. As there are not enough rare codon clusters that are isolated from the ribosome-slowing effects of positive charges, we were unable to introduce the requirement that no positive charges be present in the vicinity of the rare codon cluster. Instead, we split the rare codon clusters into two groups—those that had two or more positive charges coded for in the sequence following the rare cluster, and those that had either zero or one positive charge—and plotted the results for these groups separately. As noted above, we perform a normalization with respect to the local ribosomal occupancy. Within a given mRNA, the relative increase or decrease in ribosomal density (rpos/rprec30) at each position surrounding a rare cluster was calculated by dividing the measured ribosomal density at each codon position (rpos) by the average ribosomal occupancy of the thirty codons preceding the first rare codon in the cluster (at position x = 0) within that same mRNA (rprec30). The average relative change in ribosomal occupancy (mean rpos/rprec30) at a given position during/after a cluster was then calculated by aligning all identified regions of a given cluster size according to the first codon present in each cluster and calculating the average ratio (i. e. , increase or decrease in measured ribosomal occupancy) in positions increasingly distant from the aligned clusters (a schematic of this approach is contained in Figure 1). We used experimentally and not computationally determined RNA structure data. By exposing transcripts independently to endonucleases specific for single- and double-stranded RNA, the degree to which individual nucleotides of an mRNA are involved in intramolecular secondary structure has been experimentally quantified [38]. The resulting metric is “parallel analysis of RNA structure” (PARS) values, with higher values (positive) indicating a propensity for secondary structure and lower (negative) values signifying lack thereof. The authors show the PARS metrics along transcripts in yeast globally correlate with the degree of single- versus double-strandedness predicted by the Vienna Package. PARS values for yeast transcripts were downloaded at http: //genie. weizmann. ac. il/pubs/PARS10/pars10_catalogs. html. PARS values corresponding to CDS regions were determined relative to the local coordinate file from the same website. S. cerevisiae protein-coding sequences were scanned for stretches 30 codons in length whose average PARS value was 0 or negative (and hence tending to be single-stranded), which were immediately followed by a block 31 codons in length whose average PARS value was positive (i. e. , with propensity for double-strandedness). To ensure a clear transition from single- to double-stranded structure upon averaging across transcripts, we added the requirement that the last codon in the first 30-block have a negative PARS value and that the first codon in the subsequent 31-block have a positive PARS value. Only nonoverlapping blocks (61 codons in length) were retained, with priority given to those with the highest combined number of negative PARS value in the first 30 codons and positive PARS value in the latter 31 codons. The general contribution of folding to slowing was then examined by calculating rpos/rprec30 (as described above) at each position and then taking the average across aligned single-stranded into double-stranded blocks. In the first instance of such a test, we investigated the hypothesis that the ribosome closely approaches the base of the double-stranded structure such that the ribosome is positioned closely over the first double-stranded ribonucleotide (i. e. , at the beginning of the 31st codon out of 61) by the time slowing occurs. Here the first 30 codons in the identified block are classed as the preceding 30 codons before slowing might occur. We then repeated the analysis examining whether pausing of the ribosome might occur somewhat further upstream—for example, if the mass of the ribosome sterically hinders it from progressing at its normal rate even before the double-stranded ribonucleotide approaches the active site. In this second analysis, we used the same identified blocks as above, but moved the potential point of slowing to 10 codons upstream of the first codon with a positive PARS score. Hence the preceding 30 codons used in this case to normalize nearby ribosomal densities were also shifted 10 codons upstream as well. Changes in rates of translation were measured by calculating the relative change in ribosomal densities that occurs within a transcript, on average, after positively charged residues (lysine, arginine, or histidine) are added to the nascent peptide chain. Such an effect should be observed at or after the encoded charge (s) in the mRNA as the positively charged amino acid travels down the exit tunnel. To test for an additive effect of charge on ribosomal density, S. cerevisiae protein-coding sequences were scanned for single positively charged amino acids, two positively charged residues anywhere within five amino acids, three positively charged residues within eight amino acids, four or five positively charged amino acids within 10 amino acids, and six or more positive charges within 16 amino acids, with the first positively charged residue always considered to be at x = 0. As in the case of the rare codon cluster analyses, these loosely defined cluster specifications were chosen to maximize the sample sizes available of clusters containing different numbers of positive charges. To eliminate interference from charged amino acids outside these charged clusters, we required that a block of 30 non-positively charged amino acids precede the identified positive-charge clusters, and that no other positively charged amino acids be present in the next 30 amino acids apart from those in the identified cluster. Thirty residues were chosen as this is approximately the length of extended peptide that the ribosomal exit tunnel can accommodate [55], [56]. Thirty non-basic residues therefore should provide a baseline ribosomal occupancy reading, and hence inference of the speed of translation, before the positively charged residues are added to the peptide chain and enter the exit tunnel. The relative increase or decrease in ribosomal density at each position (rpos/rprec30) was calculated for each transcript with an encoded positive-charge cluster. The average relative change in ribosomal occupancy (mean rpos/rprec30) at a given position during/after a cluster was then calculated across regions aligned by similar-sized clusters (see also Figure 1 for a visual of this approach). Regarding our methodology, we find that noise in footprint density is not a problem for our analysis as we see similar findings when we consider genes with either low or high footprint coverage (Figure S15). The above methods start by locating the appropriate putative ribosome-slowing feature within transcripts and then measures changes in ribosomal occupancy surrounding them. A complementary approach is to look to large changes in ribosomal density and then ask whether positive charges or rare codons are more often associated with the denser, putatively more slowly translated regions. Such an approach is best carried out on a within-mRNA level, as this normalizes for differences in overall expression levels across genes. Within each gene for which we retained ribosomal protection data (see Methods, “Ribosomal Density Data”), we located the two nonoverlapping 10-codon windows (approximately the length of RNA a ribosome footprint spans [29]) with the highest and lowest average ribosomal occupancy in that transcript. To circumvent the arbitrariness of choosing the location of the low-occupancy 10-codon window in a transcript for which there may be multiple possible windows with no footprint data available (i. e. , a footprint count of 0), we added the requirement that experimental protection data exist for each codon in the window. For each window in the pair, we recorded the average ribosomal occupancy as well as (1) the tAI; (2) the number of adjacent, nonoverlapping pairs of rare codons; (3) the number of positive charges encoded within and up to five codons upstream of the window (since a charge added while the ribosome was a few codons upstream should still be present within the exit tunnel); (4) the number of rare 6-mers (defined to be the lowest 10% of all possible in-frame 6-nt sequences within open reading frames); and (5) propensity for transcript secondary structure. We included 6-mers here, as while individual codons may be rare, it does not necessarily follow that two adjacent rare codons are just as rare of a combination, and thus examining the contribution of rare 6-mers provides an extra level of stringency in assessing the role of codon usage. As secondary structure either at the codon in question or downstream of the codon in question might pause ribosomes (see Results), we considered the PARS values not only within but also for an additional 10 codons downstream of each originally identified 10-codon window. If codon usage bias is modulating ribosomal speed, we expect to observe the main effect over and locally surrounding the codon in question, whereas we expect, if charge indeed is influencing ribosomal velocity, to observe a downstream effect of positive charge on ribosomal density as the cation travels further down the negatively charged exit tunnel. Thus we also counted any positive charges encoded in the five codons immediately preceding the identified windows since a charge added while the ribosome was a few codons upstream should still be present within the exit tunnel. Conversely, as secondary structure either at the codon in question or downstream of the codon in question might pause ribosomes (see Results), we considered the PARS values not only within but also for an additional 10 codons downstream of each originally identified 10-codon window. Since the interpretation of PARS values may be somewhat more labile (as not only the magnitude but the sign of the values may have meaning), we tested whether double-stranded structure might associate with the more dense window in two different ways. We used two methods of measuring propensity for transcript structure. In the first method (“PARS score”), an average PARS value for a window ≤0 means the window is single-stranded, and an average PARS value >0 means the window is double-stranded; the exact magnitude of the PARS value is disregarded beyond this. In the second method (“conservative PARS score”), smaller changes in the magnitude of PARS values count more: the mean PARS value is calculated for each extended (20-codon) window, and the means are then compared. The ability of each metric to explain the difference in average ribosomal occupancies between the two windows was then assessed by asking how often the window with more of a potentially ribosome-slowing feature was also the window with greater occupancy. For example, if the window with the higher ribosomal occupancy paired with the less optimal (lower) tAI—which would be expected if less optimal codons do in fact slow ribosomes—then the gene was assigned a tAI score of 1; if the higher occupancy window paired with the more optimal tAI, indicating tAI is not a good predictor of increased occupancy, the a tAI score of −1 was assigned; and if the tAI was the same in the two windows, a score of 0 was given. Similar tests were performed independently on the number of rare codon pairs, PARS metrics, and number of positive charges associated with each window, with more rare pairs/more positive charges in the more occupied window—and hence potentially capable of explaining the elevated ribosomal density—each being scored 1, fewer being scored −1, and the same number in each window scored 0. There are two potential complications of this method that we are able to address and dismiss. Firstly, although there is a tendency for ribosomal occupancy to decrease, on average, along the length of transcript [29], the correlations we report hold when we test only transcripts in which the high ribosomal occupancy windows are downstream of the low ribosomal occupancy windows (higher occupancy and increased positive charge, Spearman rho 0. 12, p = 0. 00031; higher occupancy and an excess of rare pairs, Spearman p = 0. 62; an excess of rare 6-mers, rho = −0. 06, p = 0. 07; higher occupancy and lower tAI, Spearman rho −0. 15, p = 2. 4e-05). This, along with the additive pattern in Figure 3C, shows that the correlation between positive charge and increased ribosomal density is not methodological artefact. Secondly, a window might have an apparently low average ribosomal occupancy if in fact there were ribosomal footprints that should have been assigned to that region of the transcript, but which was ultimately excluded from the analysis if the footprint mapped to multiple genomic locations. To this we note that the same analysis, when redone allowing the low-occupancy window to have a footprint count of zero, still gives similar results (Table S13). To further test this possibility, we created a list of nonredundant locations. These are sites in each transcript for which all mapping footprints were uniquely mapping to that location. In other words, no footprint data were excluded from being mapped to these sites because it also mapped somewhere else in the genome. Redoing the window comparisons analysis using the nonredundant locations, we find the results (Figures S16, S17 and Table S14) qualitatively match our original results using the dataset described above (see Methods, “Ribosomal Density Data”). Hence, we consider that our method fairly infers the contribution of different sequence features to ribosomal slowing.

Ribosomes do not synthesize protein at a constant rate along transcripts, and changes in translation speed can have knock-on consequences for the expression of that protein, even altering its folding or subcellular localization. It has long been thought that RNA-level features modulate translation rates, whether by delays incurred through the presence of codons that require relatively rare tRNAs, or by regions of mRNA folding that physically impede ribosomal progression. We find on the contrary that it is not RNA-level features but positive charges in the already translated protein that most retard ribosomes, possibly by interacting with the negatively charged ribosomal exit tunnel. We show that positive charge explains the sites where ribosomes stall most commonly within transcripts. We also show why, if protein charge were not considered, one could be misled into suspecting a role for non-optimal codons. Finally, we observe that the poly-A tail provides a massively positively charged terminus no matter in which frame it is translated. A missed stop codon or frameshifting would then lead to a stalled ribosome, which is consistent with experimental data.

Abstract Introduction Results Discussion Methods

sequence analysis systems biology genome expression analysis genomics functional genomics gene expression genetics molecular genetics biology computational biology genetics and genomics

2013

Positively Charged Residues Are the Major Determinants of Ribosomal Velocity

Pheromones form an essential chemical language of intraspecific communication in many animals. How olfactory systems recognize pheromonal signals with both sensitivity and specificity is not well understood. An important in vivo paradigm for this process is the detection mechanism of the sex pheromone (Z) -11-octadecenyl acetate (cis-vaccenyl acetate [cVA]) in Drosophila melanogaster. cVA-evoked neuronal activation requires a secreted odorant binding protein, LUSH, the CD36-related transmembrane protein SNMP, and the odorant receptor OR67d. Crystallographic analysis has revealed that cVA-bound LUSH is conformationally distinct from apo (unliganded) LUSH. Recombinantly expressed mutant versions of LUSH predicted to enhance or diminish these structural changes produce corresponding alterations in spontaneous and/or cVA-evoked activity when infused into olfactory sensilla, leading to a model in which the ligand for pheromone receptors is not free cVA, but LUSH that is “conformationally activated” upon cVA binding. Here we present evidence that contradicts this model. First, we demonstrate that the same LUSH mutants expressed transgenically affect neither basal nor pheromone-evoked activity. Second, we compare the structures of apo LUSH, cVA/LUSH, and complexes of LUSH with non-pheromonal ligands and find no conformational property of cVA/LUSH that can explain its proposed unique activated state. Finally, we show that high concentrations of cVA can induce neuronal activity in the absence of LUSH, but not SNMP or OR67d. Our findings are not consistent with the model that the cVA/LUSH complex acts as the pheromone ligand, and suggest that pheromone molecules alone directly activate neuronal receptors. Pheromones—“carriers of excitement” in Greek—are chemical signals produced by an individual and recognized by conspecifics to induce a defined behavioral response [1]. Because pheromones are the basis of communication controlling diverse social interactions in many organisms (e. g. , aggregation, kin recognition, sexual and competitive behaviors), understanding how these chemicals are detected has long been a fundamental question [1]. Many animal pheromones are long-chain hydrocarbons [2], whose structural versatility provides a large chemical repertoire for the generation of species- and behavior-specific signals [3], [4]. In mammals, pheromones can also be peptides, such as rodent Exocrine gland-secreting peptide 1, which is released in male tears and enhances female sexual receptivity [5], [6], or small proteins, such as major urinary proteins [7]. Pheromone detection has been intensively studied in insects [8]–[12]. These volatile chemicals are captured from the air on the surface of insect antennae, head appendages that are covered with porous, cuticular sensory hairs, called sensilla. Pheromone molecules pass, probably by diffusion [13], through the pores to the interior of sensilla, which house the ciliated dendritic endings of olfactory sensory neurons (OSNs) (Figure 1A). These cilia are bathed in lymph fluid rich in odorant binding proteins (OBPs) (also known as pheromone binding proteins [PBPs]) and odorant degrading enzymes (ODEs), which are secreted from auxiliary cells that flank OSN somata. Three transmembrane receptors localize to pheromone-sensing OSN cilia: one member of the odorant receptor (OR) repertoire, which is thought to be the principal determinant of pheromone response specificity [14], [15], the co-receptor ORCO [16], [17], and a CD36-related protein called sensory neuron membrane protein (SNMP) [18]–[21]. One of the best-studied pheromone detection systems is that of the D. melanogaster sex pheromone (Z) -11-octadecenyl acetate (cis-vaccenyl acetate [cVA]), which evokes neural activity in OR67d-expressing OSNs (Figure 1A). Genetic analyses demonstrated that cVA detection requires OR67d [22], [23], ORCO [20], [21], and SNMP [20], [21], as well as the OBP LUSH [24]. Unexpectedly, loss of LUSH also results in decreased spontaneous firing of OR67d neurons—in the absence of pheromone—leading to the proposition that this OBP plays a direct role in cVA-evoked neuronal activity, rather than acting simply as a solubilizer or carrier of this pheromone [24]. In vitro, LUSH binds cVA as well as several short-chain n-alcohols [25]–[27], and phthalates [28]. X-ray crystal structures for apo (unliganded) LUSH, various alcohol/LUSH complexes, and the cVA/LUSH complex have been reported [25]–[27], [29]. A notable common feature of complexes of LUSH with either cVA or alcohols is that the ligand is almost completely encapsulated within the six α-helical bundle—similar to many other ligand/OBP complexes [30], [31]—making it unlikely for ORs to be able to recognize directly the ligand within this complex (Figure 1B) [29]. Importantly, the cVA/LUSH complex is structurally distinct from apo LUSH or alcohol/LUSH complexes [25]–[27], [29], in particular within the C-terminal tail (amino acids 113–124 in the processed LUSH protein) (Figure 1B and 1C) [29]. These observations led to the suggestion that this unique conformation of the cVA/LUSH complex is important for the stimulation of OR67d neurons [29]. The functional significance of these cVA-evoked structural changes has been tested by recombinant expression of site-directed mutant versions of LUSH that are predicted to exhibit different degrees of this conformational change [29]. These purified proteins were infused into individual sensilla housing OR67d OSNs in lush mutant animals, which lack the endogenous OBP, via a glass recording electrode. Basal and cVA-evoked activity of these neurons was then measured by extracellular electrophysiological recordings. A complementary pair of mutants, LUSHF121A and LUSHF121W, predicted to diminish or enhance the conformational change, respectively, led to decreased or increased cVA sensitivity. Notably, one mutant, LUSHD118A, which disrupts a predicted salt bridge suggested to be present in apo but not cVA-bound LUSH [29], induced increased firing of OR67d neurons in the absence of cVA. Presentation of cVA did not further increase this neuronal activation. These observations led to a model in which cVA induces “conformational activation” of LUSH, and that it is this cVA/LUSH complex—and not free cVA—which is recognized (through an undefined mechanism) by the neuronal pheromone receptors [29]. This model contrasts with the widely held idea that pheromone molecules must ultimately directly bind and stimulate pheromone-sensing ORs [10], [11], [14], [15], although pheromone/OR interactions have never been shown biochemically. However, the role of an extracellular protein, LUSH, in pheromone neuron activation in insects provided an interesting parallel with the discovery that small protein pheromones can stimulate olfactory neurons in mammals [7]. We have tested this model [29] by transgenic expression of the same LUSH mutants predicted to affect the cVA-induced conformational changes. We find that in vivo-expressed LUSH mutants do not recapitulate the effects observed with recombinant LUSH. We also show that LUSH, but not SNMP or OR67d, is dispensable for pheromone-evoked activity at high cVA concentrations. These results do not support the proposition of the cVA/LUSH complex as the pheromone-sensing neuron ligand. To test the activity of in vivo-expressed mutant LUSH proteins, we first generated a wild-type genomic lush rescue construct, which spans the entire transcription unit and flanking intergenic sequences (Figure 2A). This construct, referred to hereafter as LUSHwt, is expected to contain all regulatory sequences necessary to recapitulate endogenous lush expression. By site-directed mutagenesis, we generated three additional lush constructs encoding proteins equivalent to the recombinantly expressed LUSH variants previously analyzed (LUSHF121A, LUSHF121W, and LUSHD118A) [29]. Each transgene was integrated in the same genomic location by phiC31-mediated germline transformation [32] and crossed into a lush null mutant background [33] to generate flies that are genetically identical except for the missense mutations within the transgenic lush coding sequence. We first compared LUSH expression in these fly strains by Western blot analysis of antennal protein extracts using an anti-LUSH antibody. In all genotypes, we observed a single prominent band, which corresponds to LUSH as it co-migrates with endogenous LUSH in wild-type extracts and is absent in lush mutant extracts (Figure 2B). Relative quantification of LUSH levels showed that there is no statistical difference in the expression of the mutant or wild-type transgenic LUSH proteins when compared to endogenous LUSH (Figure 2B). There is some variability between extracts from the same genotype, which is probably due to the difficulty of reproducible protein extraction from the small, cuticular, antennal structures. Immunofluorescence for LUSH and ORCO on antennal sections confirmed that all transgenic LUSH variants are expressed in auxiliary cells surrounding ORCO-positive OSNs (immunofluorescence detection of secreted LUSH in sensillar lymph is difficult, probably because this extracellular fluid is largely lost during tissue preparation and staining). These LUSH-expressing cells are located within the distal region of the antenna where pheromone-sensing trichoid sensilla are found (including those housing OR67d neurons), in a pattern indistinguishable from endogenous LUSH (Figure 2C). In cVA/LUSH complexes, the pheromone directly interacts with F121 in the C-terminal tail, suggesting that this residue has a central role in triggering the cVA-induced conformational change of LUSH [29]. It was hypothesized that substitution of F121 by a smaller residue (such as alanine) or a larger residue (such as tryptophan) might reduce or enhance, respectively, this conformational change. Indeed, recombinant LUSHF121A—infused through the recording electrode into OR67d sensilla (also referred to as T1 sensilla [29]) of lush mutants—restored sensitivity to cVA that is ∼50-fold lower than that conferred by recombinant wild-type LUSH across a 1,000-fold range of pheromone concentrations [29]. Conversely, infusion of recombinant LUSHF121W conferred a 5-fold increase in sensitivity of OR67d neurons to cVA [29]. Together, these observations were consistent with the idea that cVA-induced conformational changes in LUSH underlie OR67d neuron activation. We tested whether these properties could be recapitulated by transgenic LUSH variants by measuring cVA-evoked responses in OR67d neurons in LUSHwt, LUSHF121A, and LUSHF121W animals in single sensilla electrophysiological recordings. All three transgenes restore sensitivity to this pheromone in a lush mutant background (Figure 3A). However, in contrast to the observations with recombinant LUSH [29], neuronal responses across a 10,000-fold range of cVA concentrations were very similar across all three genotypes (Figure 3B). The only statistically significant difference between these strains was observed at the highest stimulus concentration, where LUSHF121W flies exhibited slightly lower cVA sensitivity than LUSHwt (Figure 3B). This result is contrary to the report that recombinant LUSHF121W enhances cVA sensitivity [29]. A second residue in LUSH implicated in cVA-evoked conformational activation is D118, which is predicted to form a salt bridge with K87 in apo LUSH that is disrupted upon cVA binding (Figure 1C) [29]. It was hypothesized that this salt bridge maintains LUSH in an “inactive” state, leading to the prediction that mutation of D118 would produce an “activated” form of LUSH. Consistently, recombinant LUSHD118A was found to induce—in the absence of cVA—increased activity in OR67d neurons compared to recombinant wild-type LUSH, up to the level observed in wild-type OR67d neurons stimulated with 1% cVA [29]. This LUSHD118A-evoked activity depended upon both OR67d and SNMP [29]. To test whether D118 is critical in transgenically expressed LUSH, we measured spontaneous activity in OR67d neurons in LUSHwt and LUSHD118A flies. We also tested the LUSHF121A and LUSHF121W lines, because recombinant LUSHF121A was unable to rescue the loss of spontaneous activity in OR67d neurons observed in lush mutants thereby providing a second piece of evidence linking LUSH conformational changes and OR67d neuronal firing [24], [29]. We observed that all LUSH transgenes restored spontaneous firing of OR67d neurons (Figure 4A). Quantification of this activity revealed minor variation in their mean firing frequencies (∼1–3 spikes/s), but these differences were not statistically significant across genotypes (Figure 4B). Importantly, LUSHD118A flies did not exhibit the elevated level of spontaneous activity reported for recombinant LUSHD118A. Surprisingly, spontaneous firing was also observed in the LUSHF121A genotype (Figure 4B), which is inconsistent with the activity of recombinant LUSHF121A but consistent with our observation that the transgenic protein supports cVA-evoked activity as effectively as LUSHwt (Figure 3B). The loss of basal activity in OR67d neurons in the absence of LUSH provided an initial hint for a direct role for this OBP in promoting OR67d activity [24]. However, while spontaneous activity is highly reduced in lush mutants, it is not completely abolished (0. 05±0. 01 spikes/s mean ± standard error of the mean (SEM); n = 19) (Figure 4A and 4B). Moreover, OR67d/ORCO evokes robust spontaneous activity in the absence of LUSH when mis-expressed [14], [21] and in OR67d sensilla lacking both LUSH and SNMP [21]. The origin of this lush phenotype and its relevance for cVA signal transduction therefore remains unclear. One possible explanation is that loss of LUSH from the sensillum lymph changes the physiological properties of this medium with indirect effects upon OR67d neuron excitability. One important difference between the experiments with recombinant LUSHD118A [29] and transgenic LUSHD118A (Figure 4) is that the former protein was provided acutely to sensilla during electrophysiological recordings, while the latter is present continuously in the lymph. We considered the possibility that transgenic LUSHD118A is also constitutively active but that its constant presence leads to desensitization of OR67d neurons. If this were the case, we would not expect these OSNs to be able to respond to cVA. However, we observed that LUSHD118A transgenic flies still display robust responses to a 10,000-fold range of cVA concentrations (Figure 5A and 5B). These responses are statistically indistinguishable from LUSHwt, except at the highest dose presented (Figure 5B). Analysis of the temporal dynamics of cVA-evoked neuronal activity revealed that both the onset and the decay of cVA responses are very similar for LUSHD118A and LUSHwt (Figure 5C). These observations indicate that the transgenic expression of LUSHD118A does not lead to desensitization of OR67d neurons. Rather, they highlight an important functional difference between recombinant and transgenic LUSHD118A: recombinant LUSHD118A was reported to be unable to support further activation of OR67d neurons when cVA is presented [29]. This observation was suggested to reflect the fact that the “conformationally activated” LUSHD118A could not be further activated by cVA. By contrast, our results indicate that transgenic LUSHD118A supports cVA-evoked activity essentially identically to wild-type LUSH. Because of these differences, we analyzed an independently generated LUSHD118A transgenic line (Figure S1) [34]. Spontaneous activity in OR67d neurons in these flies was not elevated (indeed, it was slightly lower) compared to a control wild-type rescue transgenic strain (Figure S1A and S1B), and cVA-evoked responses were essentially normal (Figure S1C and S1D), consistent with the properties of our LUSHD118A transgene. Given the discrepancies between the properties of transgenic and recombinant LUSH proteins, we re-examined the published LUSH structures to determine the relationship between the presence and type of ligand and the conformation of this OBP (Table 1). We first superposed the 12 available structures of LUSH (Figure S2A) [25]–[27], [29]. Within each asymmetric unit in LUSH crystals (except one: apo LUSH, Protein Data Bank [PDB] ID 1OOI), there are two protein molecules, referred to as monomer A and B in cVA/LUSH [29] (dimerization of other OBPs has been observed in crystal structures [30], and there is no evidence that this reflects an in vivo biochemical property of LUSH). All non-cVA bound LUSH structures (except butanol/LUSHT57A PDB 3B87) possess similar conformations of loop 116–122 (Figure S2A), and the cVA/LUSH complex is distinct from these structures within this region. However, at least three cVA/LUSH conformations have been observed in the crystal: monomer A of cVA/LUSH exists in two different conformations, and the structure of monomer B is distinct from either of these (Figure S2A) [29]. Thus, interaction of cVA with LUSH, as captured by X-ray crystallography, reveals substantial conformational heterogeneity rather than a single “activated” form of this complex that was postulated to be recognized by pheromone receptors [29]. We re-assessed the possible presence of a salt bridge between D118 and K87 in all LUSH structures by measuring minimal distances between the D118 anionic carboxylate group (Oδ1/Oδ2) and the K87 cationic ammonium K87 (Nζ) (Table 1); this value must be less than 4 Å if such an ionic bond forms [35]. In the cVA/LUSH complex (PDB 2GTE), D118 and K87 are sufficiently distant in any conformation to be unable to form a salt bridge (Table 1), as described [29]. However, this salt bridge is not a consistent feature of the structures of apo or alcohol-bound LUSH (Table 1): in two cases (PDB 1OOI and 3B87), no salt bridge is observed in either monomer in the asymmetric unit, while in eight other cases, the salt bridge is established reliably in only one monomer. These observations are inconsistent with the idea that a D118-K87 salt bridge maintains LUSH in an inactive state and that cVA binding specifically disrupts this bond to activate LUSH [29]. The model that the cVA/LUSH complex directly activates OR67d neurons predicts that LUSH should be essential for cVA-evoked activation of this receptor. However, previous studies demonstrated that OR67d (with or without SNMP) can respond weakly to cVA in the absence of LUSH when mis-expressed in a neuron normally insensitive to this pheromone [14], [21], [29]. As heterologous receptor expression may not necessarily reflect activity of the endogenous signal transduction pathway, we examined the responses of OR67d neurons to cVA in the presence or absence of LUSH by adapting a close-range stimulation assay [14]. Typically, odors are presented to antennae by delivering (and diluting) in an airstream the headspace of a filter paper impregnated with the stimulus odor [36]. In our close-range assay, we presented cVA by approaching the antenna (within ∼0. 1 mm) with a filter paper strip on which 2 µl of a 10% solution of cVA (or the pure paraffin oil solvent) was spotted. Contact between the filter paper and antenna was completely avoided. For each sensillum tested, we approached twice (for ∼1 s each, separated by a ∼4–5 s interval) with a filter paper containing paraffin oil and twice with a filter paper containing cVA. Wild-type OR67d sensilla responded consistently and strongly to presentation of cVA (24/24 approaches; mean corrected responses upon first stimulus approaches ± SEM = 112. 2±8. 4 spikes/s, n = 12) (Figure 6A). No responses were observed to paraffin oil indicating that there is no mechanosensory artifact due to close approach of the filter paper. In lush mutants, we also observed robust, albeit lower, responses to presentation of cVA but not paraffin oil (24/24 approaches; mean corrected responses upon first stimulus approaches ± SEM = 60. 2±6. 7 spikes/s, n = 12) (Figures 6A and S3). The existence of these cVA-evoked responses is consistent with observations that delivery of cVA in an airstream can evoke weak activity in OR67d neurons in lush mutant flies at the highest stimulus concentrations (Figure 3B and [29]) and suggests that our close-approach assay merely achieves higher stimulus concentrations than those possible by airstream delivery. These observations indicate that while LUSH is important for high sensitivity responses to cVA, it is not essential for cVA-evoked activity in OR67d neurons. To demonstrate that LUSH-independent activation of OR67d neurons by cVA is specific and not an artifact of high stimulus concentrations, we also tested mutant flies lacking the neuronal receptors OR67d or SNMP in our close-range assay. Like lush mutants, Or67d mutant sensilla exhibit low spontaneous firing frequency, but no responses to cVA were observed (0/20 approaches) (Figures 6A and S3). As observed previously [20], [21], snmp mutant flies display—for unclear reasons—elevated spontaneous firing of OR67d neurons, but these sensilla also failed to exhibit evoked activity upon close-range stimulation by cVA (0/24 approaches) (Figures 6A and S3). Together, these results confirm that both neuronal membrane receptors are, in contrast to LUSH, essential for cVA-evoked activity in endogenous OR67d neurons [20]–[23]. Moreover, they indicate that the responses observed in lush mutant sensilla are unlikely to be due to non-specific activation of these neurons by high concentrations of pheromone. We further verified that close-range stimulation of OR67d neurons by cVA is specific by recording responses in non-OR67d trichoid sensilla. These sensilla can be easily distinguished from those housing OR67d OSNs because they contain two or three different neurons that exhibit multiple distinct spike amplitudes and have higher basal activity levels than OR67d OSNs [24]. While 20/20 OR67d sensilla respond to close approach with cVA, only 1/20 non-OR67d sensilla displayed a detectable response (Figures 6B and S3). The sole responding sensillum could potentially house an OR65a neuron, which is weakly sensitive to cVA [14]. The discrepancy between our findings and those previously reported [29] are likely to reflect the difference in how wild-type and mutant LUSH proteins are provided to cVA-sensing sensilla. Recombinant LUSH proteins were expressed (lacking the N-terminal signal peptide) in Escherichia coli, and purified and re-folded from inclusion bodies [24]–[26], [29]. These proteins were then introduced into individual sensilla via a glass electrode, which was simultaneously used for extracellular recording of action potentials of the OR67d neurons [24], [29]. It is presumed that LUSH passes from the electrode to the sensillar lymph by diffusion, although the precise diffusion rate and the final concentration of LUSH in the lymph at equilibrium are difficult to determine. Nevertheless, this method provides a way to assess the effect of acute (i. e. , within a timescale of 5–30 min) delivery of LUSH on spontaneous and cVA-evoked activity [24], [29]. We provided LUSH via a transgenic genomic rescue construct that appears to contain all necessary transcriptional and translational regulatory sequences to recapitulate endogenous LUSH expression. Importantly, we have compared the function of mutant LUSH transgenes with a wild-type transgene inserted at the same genomic location; these flies are genetically identical except for the desired single amino acid substitutions in LUSH. As transgenically expressed LUSH is supplied both during development and in the mature olfactory system, it is conceivable that tonic exposure of OR67d neurons to mutant versions of LUSH changes their activity. This could account for the higher basal activity observed in OR67d neurons after acute delivery of the presumed constitutively active LUSHD118A compared to the effect of the same protein expressed transgenically [29], [34]. We believe that this explanation is unlikely, however, as transgenic LUSHD118A fully supports cVA-evoked activity, indicating that OR67d neurons have not adapted to tonic exposure to this mutant protein. Indeed, our finding that LUSHD118A rescues the ability of OR67d neurons to respond to cVA conflicts with the previous demonstration that recombinant LUSHD118A does not confer cVA sensitivity [29]. We argue that the transgenically expressed LUSH is more likely to reflect the activity of the wild-type and mutant variants than when provided recombinantly. Like other OBPs, LUSH is expressed in auxiliary cells that flank OSNs and must pass through their endomembrane system before being secreted into the sensillar lymph [37], [38]. During this transit, LUSH presumably forms (like other OBPs) intramolecular disulfide bridges in the lumen of the endoplasmic reticulum; it is unknown whether the protein undergoes other post-translational modifications. The transgenic approach allows stable expression of LUSH proteins that are likely to follow this same transport pathway, and avoids the technical challenges of delivering purified, recombinant LUSH. Several functions of LUSH have been described since the identification of this OBP gene in an enhancer-trap screen [33]. lush mutants were initially shown to display behavioral defects in avoidance of high concentrations of short-chain n-alcohols [33]. Subsequently, biochemical and structural studies demonstrated that LUSH binds alcohols, with encapsulation of single alcohol molecules in the binding pocket serving to conformationally stabilize LUSH [25], [26]. Although another study suggested that LUSH binds aromatic compounds and not alcohols in vitro [28], electrophysiological analysis showed that lush mutants display defects in alcohol-evoked inhibition of neural activity in certain classes of (non-cVA sensing) trichoid OSNs [24]. However, the identity of these alcohol-sensitive neurons, the role of LUSH in regulating their activity, and how this relates to the alcohol-avoidance defects of lush mutants remain unknown. The dramatic reduction in cVA-sensitivity in OR67d neurons in lush mutants provided the first evidence for a role for this OBP in olfactory transduction in vivo. LUSH is highly expressed in most, if not all, trichoid sensilla in the antenna [39], which house nine OSN classes expressing different ORs [40]. Two of these receptors, OR67d and OR65a, respond to cVA [14] but the ligands for the remaining ORs are unknown. The presence of LUSH in the lymph fluid surrounding these receptors opens the possibility that this OBP—like SNMP [20], [21]—acts in the detection of multiple pheromones. Indirect evidence for this possibility comes from consideration of the behavioral phenotype of lush mutants. Loss of LUSH was initially linked to defects in cVA-induced aggregation behavior but not sexual behaviors [24]. More recently, cVA (or artificial) stimulation of OR67d neurons has been shown to be necessary and sufficient both to inhibit male courtship behavior and to promote female receptivity [22], [34]. The absence of reported courtship defects in lush mutants could be explained, for example, if LUSH was involved in recognition of other pheromones by different ORs with opposite functions in regulating sexual behavior. While cVA is the only volatile pheromone identified so far in Drosophila, LUSH can bind other insect pheromones, suggesting its ligand-binding pocket can accommodate structurally diverse, long-chain hydrocarbons [41]. In sum, evidence from behavioral, biochemical, structural, and expression studies suggests that LUSH plays several roles in chemosensation, with a variety of odor ligands and different ORs. According to the proposed model [29], the unique ability of a cVA/LUSH complex to activate OR67d neurons is a reflection of the distinct conformation of the complex. However, our re-analysis of the available X-ray crystal structures argues that current data for LUSH do not unambiguously support this possibility for two reasons. First, as previously noted [29], cVA/LUSH complexes are structurally heterogeneous, and it remains unclear which—if any—structure reflects the in vivo conformation of cVA/LUSH. Second, the presumed D118-K87 salt bridge that was suggested to maintain LUSH in an “inactive” state in OR67d sensilla [29] is not a consistent feature of either apo LUSH or complexes of this OBP with non-pheromonal ligands. Indeed, because a LUSHD118A crystal structure was reported to resemble the cVA/LUSH structure [29], our demonstration that LUSHD118A behaves similarly to wild-type LUSH in vivo questions the functional significance of the observed conformational changes. It is possible that mutagenesis of other (or additional) residues within LUSH may more precisely mimic the pheromone-bound state (s), but it is not obvious which residues should be targeted. It also remains conceivable that cVA binding induces structural alterations in LUSH that are not revealed by X-ray crystallography but which are relevant for pheromone signaling; investigation of this possibility would require new ways of visualizing the interaction of cVA with LUSH. Our genetic, electrophysiological, and structural analyses of LUSH fail to find evidence in support of the conformational activation model [29], raising the question of how LUSH participates in cVA detection. The close-approach stimulation assay reveals that cVA can activate OR67d neurons in the absence of LUSH. While the quantity of pheromone entering the sensilla in this assay is likely to be in excess of that encountered in nature [29], our control experiments show that this stimulation is specific to OR67d neurons and depends upon the neuronal receptor components SNMP and OR67d. Moreover, the LUSH-independent stimulation of OR67d is in accord with several previous analyses of this receptor [14], [21], [29]. Together, these observations argue for a mechanism in which cVA must directly activate OR67d, and indicate that the specificity of pheromone detection resides principally in this pheromone ligand/OR interaction. This proposition is inline with observations that chemically diverse hydrocarbon pheromones in other insects can, with varying degrees of efficiency, activate their cognate ORs in the absence of OBPs in several different in vivo and in vitro heterologous expression systems [15], [22], [42]–[48]. The major loss of sensitivity to cVA in lush mutants [24] indicates an important role for this OBP in pheromone detection, and likely an essential one at ecologically relevant pheromone concentrations. While our experiments do not address how LUSH functions mechanistically, we suggest that encapsulation of cVA by LUSH helps first to solubilize the hydrophobic pheromone molecules in the aqueous sensillar lymph and to protect them from degradation by ODEs [49], but that this OBP must ultimately deliver and release cVA to the neuronal pheromone receptors. The combined effect of these actions of LUSH would be to produce a concentration of pheromone available to bind OR67d that is several orders of magnitude greater than would be achieved without this OBP. These ideas are consistent with the in vivo analysis of cVA detection in wild-type and lush mutants ([24], [29] and this work), but also derive from studies on pheromone-binding OBPs in heterologous systems. For example, several moth pheromone receptors, when expressed in cultured mammalian cells, can respond to high concentrations of their cognate ligand solubilized in dimethyl sulfoxide but that provision of the pheromone together with an OBP/PBP negates the requirement for this organic solvent [45], [46], [50]. How LUSH might release cVA to the pheromone receptors is unclear, but would presumably require a reversal of the cVA-evoked conformational change. In other OBPs/PBPs, several lines of evidence support a pH-dependent conformational regulation that controls ligand release, at least in vitro (reviewed in [11]). We also suggest that binding of LUSH with one or more of the neuronal membrane receptors (SNMP, OR67d, and/or ORCO) could trigger cVA release. OBPs were first identified over 30 years ago [51]. The size of these repertoires in insect genomes and the diversity in their sequences, expression patterns, and in vitro biochemical properties argue for a widespread role in chemical detection in insects [11], [30], [52], [53]. To date, the loss-of-function genetic analysis of LUSH [24] is the most compelling demonstration of the importance of this role. Defining the precise mechanism by which LUSH and other OBPs act in vivo, however, still awaits. Drosophila stocks were maintained on conventional food medium under a 12 h light∶12 h dark cycle at 25°C. The wild-type genotype (Figures 2,6, and S3) was w1118. Previously described mutant alleles and transgenic lines used were lush1 [33], Or67dGAL4 [22], snmp1 [21], a lush rescue transgene (referred to here as lushwt-DS) [33], and a lushD118A transgene (referred to here as lushD118A-DS) [34] (kindly provided by Dean Smith, UT-Southwestern). New transgenic lines were generated with the phiC31-based integration system [32], using the attP40 landing site [54], by Genetic Services Inc. The identity of lush transgenic flies was re-verified by PCR amplification of the transgenic lush coding sequence and sequencing. Gene-specific primers were designed using Primer3 (http: //frodo. wi. mit. edu/) to amplify a genomic region of 3,228 bps, starting 958 bp upstream of the lush start codon to 1,693 bp downstream of the lush stop codon; this region includes the entire lush transcription unit and flanking intergenic non-protein coding sequences. The wild-type lush transgene (LUSHwt) was generated by PCR amplification with these primers on genomic DNA prepared from the genome-sequenced strain [55], using the KAPA HiFi PCR kit (Kapa Biosystems); the resulting PCR product was T: A cloned into pGEM-T Easy (Promega), sequenced, and subcloned with restriction enzymes (XhoI/XbaI), whose recognition sites were incorporated into the PCR primers, into the pattB vector [32]. Point mutations in the lush coding sequence were introduced by site-directed mutagenesis of pGEM-T lushwt, which was fully sequence-verified both before and after subcloning into pattB. Immunofluorescence on antennal cryosections was performed as described [56]. Approximately 200 third antennal segments were harvested by snap-freezing flies in a mini-sieve (Scienceware, Bel-Art Products) with liquid nitrogen and gently shaking them to break off and collect the appendages in a Petri dish under the sieve containing 0. 1% Triton X-100. Antennal segments were selected under a binocular microscope and transferred with a pipette into an Eppendorf tube. After centrifugation at 12,000 g for 2 min, the liquid phase was removed, and the antennae disrupted using a TissueLyzer (Qiagen). Protein extracts were made by incubating lyzed antennae in 150 µl of extraction buffer (20 mM Tris [pH 7. 5], 100 mM NaCl, 5 mM KCL, 1. 5 mM MgCl2,4% glycerol, 0. 02% n-dodecyl-D-maltoside) for 1 h at 4°C, followed by centrifugation at 12,000 g for 15 min at 4°C. 30 µl of extract were separated on 4%–20% precast gels (NuSep) and transferred to Hybond-ECl membrane (Amersham), which were then probed with primary antibodies against LUSH (described above; diluted 1∶4,000) or IR8a (diluted 1∶5,000) [56]. Goat anti-IgG rabbit (Promega) or donkey anti-IgG guinea pig (Fitzgerald Industries International) secondary antibodies coupled to HRP were used to detect LUSH and IR8a proteins, respectively. Blots were developed with medical X-RAY films (Fujifilm) using the ECL Plus Western blotting detection system (GE Healthcare). The resulting films were scanned without any automatic gain and bands were quantified in ImageJ. A rectangular box with a width slightly smaller than the narrowest band was defined on the first lane of the image and then used to measure densities of all lanes from the same film. The background was calculated for each lane and subtracted from the density of the corresponding bands as described [57]. Extracellular recordings of OSN activity in individual sensilla of 4- to 8-day-old male flies were performed as described [21], [36].

Animals produce chemical signals, called pheromones, to communicate with other members of the same species to regulate social behaviors. How pheromones are detected is a fundamental question. One important model system for pheromone recognition is in Drosophila, where the sex pheromone cis-vaccenyl acetate (cVA) activates olfactory sensory neurons expressing the odorant receptor OR67d. cVA neuronal stimulation requires not only OR67d but also an extracellular odorant binding protein called LUSH. Previous work showed that cVA association with LUSH induces a conformational change in the protein. A mutant version of LUSH that mimics this structural alteration—when expressed in vitro and infused around OR67d neurons—stimulates these neurons in the absence of cVA, suggesting that the ligand that activates OR67d neurons is a cVA/LUSH complex rather than free cVA. By contrast, we show that the same mutant, when expressed in vivo, does not activate OR67d neurons, questioning the significance of the conformational change. We also show that high concentrations of cVA can induce neuronal activity in the absence of LUSH, but not OR67d, indicating that LUSH is important but not essential for pheromone detection. Together, these findings challenge the established model of cVA signaling and suggest that this pheromone directly activates OR67d.

Abstract Introduction Results Discussion Materials and Methods

molecular neuroscience cellular neuroscience model organisms sensory systems biology sensory perception molecular cell biology neuroscience

2013

Ligands for Pheromone-Sensing Neurons Are Not Conformationally Activated Odorant Binding Proteins

We demonstrate for the first time in vertebrates, that alternative splicing of interferon (IFN) genes can lead to a functional intracellular IFN (iIFN). Fish IFN genes possess introns and in rainbow trout three alternatively spliced transcripts of the IFN1 gene exist. Two of the encoded IFNs are predicted to lack a signal peptide. When overexpressed these iIFNs induce antiviral responses. Variants of the two IFNR receptor chains (IFNAR1 and IFNAR2) lacking a signal peptide are also present in trout. Transfection of HEK 293T cells with the iIFN and iIFNR molecules results in STAT phosphorylation and induction of antiviral genes. These results show that fish possess a functioning iIFN system that may act as a novel defence to combat viral infection. Interferons (IFN) are a key antiviral defence in jawed vertebrates, with many well characterised IFN-induced genes expressed following their release [1], [2], that participate in various ways to try to inhibit virus replication and modulate immune responses [3]–[6]. Three types of IFNs are known in tetrapods [7], [8], distinguished by amongst other things the stimuli that induce their expression, their receptor usage and the responses they evoke [9], [10]. In bony fish two types of IFNs are present, that appear to be regulated and function in a manner similar to their higher vertebrate counterparts [11]–[16]. Whilst there has been some debate as to whether fish possess type I or III IFN (in addition to type II), partly due to the presence of introns in the fish genes, phylogenetic and structural analysis suggest these genes encode type I IFNs [7], [17]. In cartilaginous fish only type I IFNs have been found to date [18]. Multiple genes are a feature of the type I and III IFN families, where in humans the genes are clustered on chromosome 9 and 19 respectively. Similarly in Xenopus multiple type I and III genes are present in the genome, with five and four genes located on scaffolds 48 and 389 respectively [7]. In bony fish there are also multiple type I IFN genes that appear to be mostly clustered together. For example, in zebrafish (Danio rerio) 3 genes are co-located on chromosome 3, with a fourth on chromosome 12 [19]. In Atlantic salmon (Salmo salar) 11 IFN genes have been found co-located in sequenced BAC clones, with 8 linked in a clone containing the growth hormone gene [20], and even more copies are present in rainbow trout (Oncorhynchus mykiss, unpublished). However, relatively few IFN genes have been found in some of the more advanced Acanthopterygian teleost fish [8]. Analysis of the cyprinid and salmonid genes reveals that two major families exist, characterized by the presence of 2 (group I) or 4 (group II) cysteine residues in the mature protein [18], [21]. Curiously the 2 cysteine form retains C1 and C3 unlike mammalian IFN-β/ε which retain C2 and C4. The salmonid genes cluster into multiple sub-groups (termed a, b, c, etc), with at least 2–3 subgroups of type I IFN genes present within each of the group I and II families [20], [22, unpublished]. Gain- and loss- of function studies in zebrafish have shown that the group I and II IFNs in this species signal via different heterodimeric receptors, that have one receptor chain in common [19]. This contrasts to the situation in mammals, where despite their high sequence divergence all type I IFNs mediate their antiviral responses by binding to a common receptor complex containing the IFNAR1 and IFNAR2 chains, although the ternary binding affinity of this complex can vary [23]. Hints that fish IFNs may have other novel features have come from analysis of their transcripts. Initially reported as having an unexpectedly long 5′-UTR containing numerous start codons with downstream in-frame stops, the salmonid IFNa genes were found to be produced as short and long transcripts, each proposed to contain a different promoter region immediately upstream of the apparent transcription starts [24]. Subsequent analysis with homozygous trout revealed the two forms were the result of alternative splicing, with the longer form generating a protein that was 16 aa smaller than that produced by the short transcript, and lacking a signal peptide [25]. Both transcripts were expressed equally in unstimulated tissues, and both forms were induced by viral infection (Infectious Hematopoietic Necrosis Virus, IHNV) although more highly for the short (presumed secreted) form. However, curiously the sum of the short and long forms detected by PCR was less than the amount of total mRNA copies detected using primers that amplified both forms, and the authors suggested additional transcripts may be present. IFN variants lacking a signal peptide have also been found in IFNa genes from catfish [26], [27], where they are constitutively expressed in cells unlike the signal peptide containing transcripts. This promoted us to search for further splice variants of the IFNa transcripts in rainbow trout, to establish whether the different encoded proteins are made intracellularly within fish cells, and to determine functionally whether such intracellular IFNs (iIFN) can induce antiviral defences and protect cells from viral infection. Having established this was the case, we have searched for IFNR variants that may be present within trout cells, and show that such variants exist for two IFN receptor chains that potentially represent the group I IFN receptor in trout. Lastly, we have transfected HEK 293T cells with the iIFN plus the iIFNR and show it is possible to induce the expression of downstream IFN-inducible genes and STAT phosphorylation. Thus fish appear to have a functioning iIFN system that may act as a novel defence to combat viral infection. Further analysis of the splice variants of the trout IFN1 gene (the first member of the “a” subgroup found in trout) discovered a third variant, which we term intracellular (i) IFN1b, produced by splicing of a second intron within the 5′UTR (Figure 1A, Supplementary Data File S1). Thus, the secretory (s) IFN1 cDNA conserves the first and second introns, and the start codon ATG is located within the second intron. In the variant identified by Purcell et al. [25], which we term iIFN1a, the cDNA lacks the second intron and the ATG is located in the third exon. Lastly, the iIFN1b cDNA lacks the first and second introns, and the ATG is located at the junction of the first and second exons. The three transcripts potentially encode three IFN proteins that vary at the N-terminus of the molecule (Figure 1A). Only the sIFN1 contains a conventional signal peptide, as identified by SignalP 3. 0 analysis, whilst both iIFN1a and iIFN1b appear to lack a signal peptide. iIFN1a is 7 amino acids (aa) longer at the N-terminus relative to the mature sIFN1 peptide, and iIFN1b differs from iIFN1a in that it contains an additional 24 aa at the N-terminus (8 aa more than the full length sIFN1). Constitutive expression of trout IFN1 variants was studied in RTG-2 cells, a trout fibroblast cell line known to express IFNs, by real time PCR. The expression of all three variants was detectable, with iIFN1a having the highest expression level in RTG-2 cells, as also seen in a macrophage cell line (RTS-11 cells, unpublished) (Figure 2A). Expression of the trout IFN1 variants was studied after stimulating RTG-2 cells with the double stranded synthetic RNA polyI∶C, for 4 h and 24 h. PolyI∶C stimulation significantly increased the expression of sIFN1 at both 4 h and 24 h, with a >100-fold increase at the latter timing (Figure 2B). iIFN1a and iIFN1b were also increased significantly at 24 h ∼10-fold. As a positive control for the stimulation, expression of the IFN-induced gene Mx was also analysed and was found to be significantly increased (∼100-fold) at both time points examined post-stimulation. To examine the effect of intracellular polyI∶C on sIFN1 and iIFN1 expression in RTG-2 cells, the cells were electroporated with polyI∶C and RNA extracted 24 h later. sIFN1, iIFN1a and iIFN1b were all induced significantly by transfection with polyI∶C, but by far the largest increase was seen with sIFN1 where an approx. 420-fold increase occurred (Figure 2C). In contrast, iIFN1a and iIFN1b were increased 5–6-fold. Again Mx expression was also analysed and was increased ∼10-fold by this treatment. Lastly, the expression of the trout IFN1 variants was studied in head kidney tissue after infection of trout with viral hemorrhagic septicemia virus (VHSV), a pathogenic rhabdovirus of salmonids. At day one post-infection, a statistically significant increase in expression level of all three transcripts was already apparent (Figure 2D). For sIFN1, the induced expression was highest at day three post-infection (789-fold) and decreased thereafter. For iIFN1a and iIFN1b, the induced expression was highest at day one post-infection (95- and 457-fold respectively), and reduced thereafter, although much more rapidly for iIFN1a during the first week post-infection. VHSV infection also increased significantly the expression of the Mx gene, with highest levels apparent at day 7 post-infection, and remaining relatively high to the last sampling time at day 12 (Figure 2D). To study whether iIFN1a and iIFN1b are translated into intracellular IFN proteins, the respective 5′ non-coding region and the open reading frame region of iIFN1a and iIFN1b was cloned and inserted into the pTurboGFP-N expression vector that is expected to generate an IFN-GFP fusion protein if translated. RTG-2 cells transfected with plasmids pTurbo-iIFN1a-GFP or pTurbo-iIFN1b-GFP were examined under a fluorescent microscope 24 h after transfection. A global cytosolic distribution of fluorescence was seen for both the iIFN1-GFP fusion proteins (Supplementary Figure S1). To further confirm that the iIFN1 transcripts are translated into intracellular IFN proteins, the protein extracted from RTG-2 cells transfected with pTurbo-iIFN1b-GFP was analysed by Western blotting, using an anti-TurboGFP antibody. A protein band of approx. 30 kDa was observed in the samples from cells transfected with the control plasmid pTurbo-GFP, correlating with the expected size of GFP (Fig. 1b). In contrast, the protein detected in cells transfected with pTurbo-iIFN1a-GFP was approx. 43 kDa and that detected with pTurbo-iIFN1b-GFP was approx. 46 kDa, the predicted sizes for iIFN1 (iIFN1a, 18 kDa; iIFN1b, 21 kDa) plus pTurbo-GFP (25 kDa). This confirmed that the iIFN1a and iIFN1b transcripts can be translated into iIFN protein. The open reading frame of the IFN1 variants was inserted into the pQE-30 expression vector, and the recombinant proteins were produced and purified under native conditions (Supplementary Figure S2). Incubation of RTG-2 cells with the three proteins increased the expression of Mx and all three IFN1 variants, as detected by real time PCR, although the sIFN1 transcript was always more highly induced than the two iIFN1 variants (Figures. 3A–C). A clear dose-dependent increase was seen with sIFNa and iIFN1b but was less apparent for iIFN1a. In addition, all three proteins significantly induced two of the TLRs studied (TLR3 and TLR8a1) and the cytosolic pattern recognition receptors (PRR) RIG-I, MDA5 and LGP2 that detect intracellular virus (Figure 3D). The two iIFN1 splice variants were also able to induce TLR9 expression. The open reading frame of the IFN1 variants was inserted into the ptGFP1 vector, that drives the expression of the inserted gene and GFP independently. Stable cell lines were established by selection with G-418 (Figure 4A). When green fluorescent cells represented >80% of total cells the cell lines were used in experiments designed to establish whether overexpression of the different splice variants conferred antiviral resistance. Initially RNA was extracted to determine whether the IFN1 variants were indeed overexpressed in the cell lines. The plasmid-derived and endogenous expression of each variant was determined by real time PCR in the transfected cells, and compared to that of the respective IFN1 transcript level seen in cells transfected with ptGFP1 (Figure 4B). The endogenous and exogenous IFN1 transcripts were detected using primers specific to each IFN1 variant (endogenous IFN1) or using a forward primer specific to the iIFN gene and a reverse primer at the junction of IFN1 and the SV40 3′ untranslated region in the vector (exogenous IFN1). Over-expression of one of the IFN variants (iIFN1b) was also confirmed by Western blotting using a polyclonal antibody that specifically detects the iIFN1b protein (Figure 4C). The higher plasmid-derived expression of the IFN variants showed that these genes were indeed overexpressed in the three cell lines. In addition, it was apparent that overexpression of sIFN also impacted on the level of endogenous sIFN expression, similar to the results seen with the recombinant protein experiment. Overexpression of the three variants also increased significantly the expression of the Mx gene (Figure 4D), indicating that the IFN1 variants may be up-regulating antiviral pathways within each of the cell lines. Since the expression of the Mx gene was increased significantly in sIFN1, iIFN1a and iIFN1b transfected cells (P<0. 5), these cells were used to examine their resistance against Viral Hemorrhagic Septicemia Virus (VHSV) infection. The transfected cells were infected with VHSV for two weeks and fixed. The cells were then stained with crystal violet and the ratio of OD562 between infected or uninfected transfected cells was calculated to compare the cytopathic effect. The cells transfected with sIFN1, iIFN1a and iIFN1b showed a significantly higher OD562 ratio than the cells transfected with control plasmid (ptGFP1) after infection with the same amount of virus, indicating they were more resistant to VHSV infection (Figures 4E, F). Having established that overexpression of the three variants induced antiviral protection within the cells, we undertook in silico analysis to see whether splice variants of the IFN receptors (IFNR) might also exist, as one possible explanation for the results. Initially we identified both trout IFNR chains, IFNAR1 and IFNAR2 (Supplementary Data Files S2, S3, S4), equivalent to the CRFB5 and CRFB3 chains, respectively, described in the genomes of other teleost fish species [28]. The identified trout IFNAR1 transcript is 1618 bp in length, and encodes a protein of 405 aa (GenBank accession No. GU319961). In phylogenetic tree analysis, where the tree was constructed using MEGA4 based on class II cytokine receptor genes from teleosts and selected tetrapods, the trout IFNAR1 molecule groups with other IFNAR1/CRFB5 sequences (Figure 5A). SignalP 3. 0 analysis indicated that mIFNAR1 possesses a putative signal sequence of 20 aa in length, that is predicted to be cleaved between Gla20 and Glu21, and on motif analysis revealed two Ig domains in common with other fish IFNAR1 but in contrast to mammalian IFNAR1 where four Ig domains are present (Figure 5C). We termed this molecule membrane associated IFNAR1 (mIFNAR1). 5′RACE and 3′RACE also identified another IFNAR1 receptor transcript (Figure 5B, Supplementary Data Files S2), which encodes a protein of 388 aa (GenBank accession No. GU319962). This protein was predicted to be a non-secretory protein, in that no signal peptide could be detected, and therefore may represent an intracellular form of this IFN receptor chain (named iIFNAR1). The transcript for iIFNAR1 is 2272 bp in length, and differs from mIFNAR1 in the 5′ untranslated region (UTR) and in having a longer 3′ UTR, and is likely generated by a distinct gene. The Tandem Repeats Finder Program found that a 43 bp DNA sequence (AGGGTTGGCCTGAAAACCCACAGGACGGTAGATCTCCAGGAAG) is repeated 8 times in the 3′ UTR of iIFNAR1. The full length aa sequence of mIFNAR1 and iIFNAR1 had 47–57% aa similarity with teleost CRFB5. The two receptors have 7 different aa residues in addition to the 17 aa deletion at the N terminus that accounts for the lack of a signal peptide. The identified trout IFNAR2 transcript is 1411 bp in length, and encodes a protein of 268 aa (GenBank accession No. JX532086). It contains an identifiable signal peptide, and two Ig domains (Figure 5C), as in mammalian IFNAR2 molecules, and was termed mIFNAR2. Relative to the trout IFNAR1 molecule it has a relatively short intracellular domain of 24 aa vs 152 aa in IFNAR1, but nevertheless it contains a conserved JAK1 binding site (LPKT (S) L) and an overlapping JAK2 site (XPXP) (See Supplementary Data File S4). The trout IFNAR2 groups with other teleost fish CRFB1-3 molecules in phylogenetic tree analysis (Figure 5A), which themselves group with tetrapod IFNAR2 with high bootstrap support. A second transcript was discovered by cloning of the full length cDNA using a single pair of primers located at the 5′ and 3′ UTR and is apparently generated from alternative splicing of the transcript of a single gene (Figure 5B, Supplementary Data File S3). Compared with mIFNAR2, this transcript is 105 aa shorter at the N terminal region due to a reading frame shift caused by deletion of an exon in the coding region, that gives a molecule with no detectable signal peptide and only one Ig domain (Figure 5B, Supplementary Data File S3). We term this molecule iIFNAR2 (GenBank accession No. JX532087). Constitutive expression of trout IFN receptors was also detected in trout cell lines by real time PCR, as seen in RTG-2 cells (Figure 2A). In relation to the comparative expression of the receptor variants, the expression of mIFNAR1 and mIFNAR2 was higher than that of iIFNAR1 and iIFNAR2 in RTG-2 and other cell lines (Figure 2A). Expression of trout IFN receptors was studied in RTG-2 cells after stimulation with polyI∶C for 4 h and 24 h. PolyI∶C stimulation significantly increased (∼10-fold) the expression of mIFNAR1 and iIFNAR1 at 4 h, but only increased mIFNAR1 expression at 24 h post-stimulation (Figure 2B). PolyI∶C stimulation also increased significantly the expression of mIFNAR2 and iIFNAR2, but in contrast to IFNAR1 expression was higher at 24 h post-stimulation and a larger increase was seen with both transcripts. To examine the effect of intracellular polyI∶C on IFNR expression, RTG-2 cells were electroporated with polyI∶C. All four IFNAR transcripts were induced significantly by transfection with polyI∶C, with mIFNAR2 and iIFNAR2 again having greater increases in expression level relative to IFNAR1 (Figure 2C). Lastly, IFNAR1 expression was examined in trout infected with VHSV. However, no statistically significant effect of infection on either mIFNAR1 or iIFNAR1 transcript expression levels was detectable in the head kidney at any of the sampling times (Figure 2D). To determine whether intracellular proteins of IFNAR1 and IFNAR2 are made from the transcripts iIFNAR1 and iIFNAR2, the respective open reading frame region of iIFNAR1 and iIFNAR2 was inserted into the pTurboGFP-N expression vector and the resultant plasmids were transfected into RTG-2 cells. Western blotting analysis of the cell lysate of the transfected cells using anti-GFP antibody revealed two protein bands of approx. 73 kDa and 45 kDa, close to the expected sizes of the iIFNAR1-GFP and iIFNAR2-GFP fusion proteins (Figures 5D, E). The intracellular forms of trout IFN1 and IFNAR1 were cloned into the expression vector pcDNA3. 1 to give pcDNAiIFN1a-V5 or pcDNAiIFN1b-V5 and pcDNA-Flag-iIFNAR1. These constructs were overexpressed in RTG-2 cells and fixed after 48 h for immunostaining using mouse anti-V5 monoclonal antibody and rabbit anti-Flag antibody as the primary antibodies and Alexa FluorR 594 goat anti-mouse IgG and Alexa FluorR 488 goat anti-rabbit IgG as the secondary antibodies. The confocal microscopy images showed that the iIFN1 ligands have a global cytosolic distribution within the cells (Figure 6A), which was consistent with the fluorescent results of the pTurbo-iIFN1a-GFP and pTurbo-iIFN1b-GFP fusion proteins. The iIFNAR1 in contrast showed a mainly perinuclear distribution, although some nuclear and cytoplasmic localisation was also detectable. A clear co-localisation of the ligands and receptor in the perinuclear site was apparent (Figure 6A). To investigate whether iIFN1 can signal via intracellular receptors, iIFN1b, iIFNAR1 and iIFNAR2 were expressed in HEK 293T cells, alone or combination, since fish IFNs would be unlikely to interact with endogenous receptors in this human cell line (Figure 6). Initially we verified that the three genes were expressed as expected following transfection with one, two or all three constructs (Figure 6B). We then studied the expression of three antiviral genes, PKR, Mx1 and viperin, known to be induced by IFNs and found that all three were significantly up-regulated at the transcript level when the HEK 293T cells were transfected with iIFN and both receptor chains (Figure 6C). We confirmed that this also resulted in increased protein expression for viperin, by Western blotting (Figure 6D). Phosphorylation of STAT1 and STAT2 was next studied to get an insight into the signaling events leading to anti-viral gene expression. Co-transfection with all three constructs, iIFN1b/iIFNAR1/iIFNAR2, led to phosphorylation of both STAT1 and STAT2 (Figure 6E). Interestingly, co-expression of iIFN1b and iIFNAR1 also induced phosphorylation of STAT1, albeit to a lower degree than that seen with both iIFNAR1 and iIFNAR2, but not STAT2. Co-expression of iIFN1b with iIFNAR1 or iIFNAR2, or with both receptors, also increased the expression of (unphosphorylated) STAT1 and STAT2. The present study shows for the first time in a vertebrate, that alternative splicing of an IFN gene can lead to a functional intracellular IFN (iIFN). One of the most apparent differences between fish and amphibian type I IFN genes and their counterparts in amniotes is the possession of multiple introns in the former [7]. It seems likely that the intron-lacking type I IFN genes evolved from the intron-containing type I IFN genes via a retrotransposition event which led to the loss of the entire IFN locus and insertion of a type I IFN gene transcript (s). Therefore, fish and amphibian type I IFNs require intron splicing for production of IFN transcripts. In addition, alternative splicing of fish type I IFN genes has been reported previously in several teleost fish, including Atlantic salmon, rainbow trout and catfish [24]–[26], and suggests that RNA may play a role in diversification of IFN function. Our study demonstrates not only that an iIFN is produced but that it can function within cells to generate a signal via intracellular IFN receptors, leading to induction of an antiviral response. All three trout IFN1 variants were an outcome of the alternative splicing of two introns present in the 5′ flanking region that generate transcripts that encode for proteins which vary at the N-terminus, only one of which has an identifiable signal peptide. To date, the alternative splicing of the IFN mRNA has been detected only in the “a” subgroup of the group I (containing 2 Cys – 2C) IFN genes, and it remains to be determined whether other members of the group I or group II (4C) family IFN genes can also produce splice variants. In agreement with the catfish findings [26], the trout iIFN1 splice variants are constitutively expressed to some degree, as seen in the different cell lines studied, with iIFN1a having the highest expression level. On stimulation with polyI∶C all three splice variants were also induced, and whilst the secretory form increased the most (several hundred fold), it was clear that the splicing to generate the iIFN1s was not turned off to achieve this. The three splice variants were also induced upon viral infection, with iIFN1b and sIFN1 following relatively similar kinetics, but with iIFN1a decreasing more rapidly. Such findings suggest that all three forms may play a role in mediating IFN responses in trout. In addition, that the iIFN1 transcripts were inducible in response to stimulation with intracellular (transfected) or extracellular polyI∶C suggests that iIFN production is likely mediated via host pattern recognition receptors on the cell surface or in the cytoplasm. Pattern recognition receptors such as the RIG-I like helicases (RLRs) and Toll like receptors (TLRs) that recognise (in mammals) non-self Pathogen Associated Molecular Patterns (PAMPs) (including polyI∶C and viral double stranded RNAs) have been characterised in many fish species and are known to be required for induced (s) IFN synthesis [29]–[32]. Although conservation of many aspects of the IFN system is evident among vertebrates [30], [33], nevertheless it is possible that some components are specific to fish. For example, TLR22 has been found only in teleost fish species, and is important in triggering IFN synthesis by sensing extracellular RNA PAMPs [34]. The present study has demonstrated for the first time that the alternatively spliced IFN variants in fish are translated into proteins. Cells transfected with a plasmid harbouring the appropriate 5′ untranslated region together with the putative iIFN (a or b) open reading frame linked to the N terminus of GFP were shown to produce GFP by fluorescent microscopic analysis, and the presence of the fusion protein was also confirmed by Western blotting (Figure 1B). The recombinant proteins of the putative iIFNs were produced in bacteria, and when added to RTG-2 cells were shown to induce Mx expression, indicating they were able to activate cellular antiviral responses, presumably via binding to the IFN receptors on the cell surface. The effects of the secreted and non-secreted forms of trout IFNs were in general comparable, although the level of induction using iIFN1b was somewhat lower and perhaps indicative that the long N-terminal may negatively impact receptor binding/signaling. Curiously, sIFN1 was apparently unable to induce TLR9 expression in contrast to the iIFN proteins, although the level of induction was quite small. Interferons, like other cytokines, are usually secreted and need to interact with the receptors on the cell surface to initiate intracellular responses (Figure 7). Potentially one way in which the iIFNs could function is in a paracrine manner once released from cells that are killed or undergo apoptosis during viral infection (Figure 7). However, our studies clearly demonstrate that the trout iIFNs can also function intracellularly. The iIFNs when overexpressed in transfected cells were able to induce Mx protein expression and enhance resistance to viral infection. In mammals, although intracellular forms of IFNs have not been reported under physiological conditions, overexpression of an artificial human IFN-β lacking a signal peptide in transfected cells can activate the JAK/STAT pathway and induce antiviral responses, thought to be mediated via binding to the IFN receptor complex within the cytoplasm [35], [36]. Whilst this could be a mode of action of the fish iIFN ligands, we hypothesized that it was also possible that iIFN receptor variants may exist that are expressed within the cell. This prompted us to search for the existence of such intracellular receptors, and the two putative IFN receptor homologues (IFNAR1 and IFNAR2) were identified by analyzing the rainbow trout EST database. In each case a second transcript containing a form that lacked a signal peptide was also discovered. These two putative iIFNRs are generated in different ways. iIFNAR2 is the result of alternative splicing of the IFNAR2 transcript, whilst iIFNAR1 appears to exist as two genes, with differences in the 5′- and 3′-UTRs as well as the N-terminal region of the ORF. The trout iIFNRs were found to have a mainly perinuclear distribution within the cell, and co-localised with the putative iIFN ligand at this site. This suggests a possible direct interaction between them, which may be responsible for the activation of the IFN regulated genes in the iIFN overexpressing cells. The ability of the iIFNs to interact with the iIFNRs was studied further in HEK 293T cells, since these cells do not express the endogenous mIFNRs, and hence interaction with the mIFNRs within the cell can be excluded as a pathway leading to activation of STATs or induction of antiviral genes. Transfection with iIFN1b alone or together with iIFNAR2 had no effect on STAT1 or STAT2 phosphorylation. However, iIFN1b plus iIFNAR1 did result in STAT1 phosphorylation, whilst transfection with all three molecules gave a stronger effect on STAT1 and phosphorylation of STAT2. This combination also induced significantly the expression of PKR, Mx1 and viperin in the HEK 293T cells. This suggests that an interaction of iIFN1 with the iIFNRs was occurring and was capable of triggering a typical IFN signaling pathway leading to induction of antiviral genes. It also appears that iIFNAR1 was sufficient for at least some phosphorylation of STAT1 when co-transfected with iIFN1b, and therefore presumably some ligand binding was possible by this single receptor chain. This did not happen when using iIFNAR2, which lacks one of the two Ig domains typically present in the fish IFNRs, and perhaps this chain acts more as an accessory chain and carrier of JAK1/Tyk2 to promote downstream signaling. The present experiment does not rule out the possibility that the iIFNs could interact with the mIFNRs within fish cells, as suggested for mammals above, and further studies will be needed to determine if one or other pathway is dominant or more important to induce the antiviral state elicited by trout iIFN. Based on these studies, it is reasonable to postulate that iIFNs in trout may act as effector molecules to swiftly trigger an antiviral response following virus infection, perhaps avoiding some of the many ways in which viruses can disrupt these pathways [37]. In addition, excessive iIFNs could be stored in the cytoplasm of infected cells and released in relatively large amounts should the cells be killed, alerting neighbouring cells through their surface receptors. Finally, although secreted IFN was not detected in the supernatants following overexpression of iIFNs in the RTG-2 cells, it cannot be excluded that they could be transported to the exterior by non-conventional secretion routes, perhaps in certain cell types, to act locally or distantly, and this also requires further investigation. In conclusion, alternatively spliced IFN transcripts exist in fish, generated by splicing of introns located at the 5′ end of the IFN1 gene. The encoded IFNs are translated into proteins lacking a signal peptide, that when overexpressed in cells induce antiviral genes and viral resistance. Variants of the two IFNR receptor chains are also present that lack a signal peptide, and have a perinuclear expression within cells. Transfection of HEK 293T cells with the iIFN and iIFNR molecules results in STAT phosphorylation and induction of antiviral genes. Together these results suggest that fish have functional intracellular IFNs. Animal experiments were conducted in strict accordance with the UK Animals (Scientific Procedures) Act 1986 and Home Office Code of Practice guidance, under project licence number: PPL 60/3965, approved by the Animal Ethics Committees of Marine Scotland, UK. RTG-2 cells were maintained as described previously [22]. HEK 293T cells were cultured in Dulbecco' s modified Eagles' s medium supplemented with 10% FBS and cultured at 37°C in a humidified incubator with 5% CO2. Viral hemorrhagic septicemia virus (VHSV, isolate DK-F1) was prepared in BF-2 cells [18]. pTurbo-iIFN1a-GFP and pTurbo-iIFN1b-GFP were constructed to confirm whether intracellular IFNs translate into proteins. pQE30-iIFN1a and pQE30-iIFN1b were used for production of recombinant proteins in bacteria. For over-expression studies of sIFN1, iIFN1a and iIFN1b, ptGFP1-sIFN1, ptGFP1-iIFN1a and ptGFP1-iIFN1b were constructed. pCDNA-iIFN1a, pCDNA-iIFN1b and pcDNAFlag-iIFNAR1 were used for co-localization studies of intracellular IFNs and receptor. ptGFP1, pcDNAFlag, ptGFP1-iIFN1b, pcDNAFlag-iIFNAR1 and ptGFP1-iIFNAR2 were used for analyzing iIFN1b signal transduction in HEK293T cells. Trout recombinant IFN1 proteins were produced in E. coli M15 cells and purified using Ni-NTA resin (Qiagen). Endotoxins were removed as described previously [18]. Rainbow trout (Oncorhynchus mykiss) (∼15 g) were maintained at the Marine Scotland Science Marine Laboratory in Aberdeen, Scotland. Fish were injected intraperitoneally with 100 µl VHSV (1×107 TCID50 per fish) or control medium [38]. Head kidney from 4 fish per group was sampled for quantitative PCR analysis of gene expression. RTG-2 cells were transfected with polyI∶C or plasmids using an Amaxa Nucleofector II transfection system (Lonza) [22]. Transfection of HEK 293T cells was performed using the LipofectAMINE 2000 transfection reagent (Invitrogen). Total RNA was extracted from RTG-2 cells and HEK 293T cells using TRIzol reagent (Invitrogen). PCR analysis was performed on a Roche LightCyler@ 418 using primers specific to individual genes (Supplementary Table S1). The relative expression of target genes was normalized to the expression of elongation factor (EF) -1α (trout) or β-actin (human) and expressed as arbitrary units or fold change relative to the corresponding control group. The mean of three independent experiments was used for statistical analysis (Student' s t test), with P<0. 05 between sample sets considered significant. RTG-2 cells were transfected with ptGFP1, ptGFP1-sIFN1, ptGFP1-iIFN1a or ptGFP1-iIFN1b and were subject to G418 selection to enrich GFP positive cells. The stably transfected cells grew normally and no obvious cell lysis was observed. VHSV challenge of the cells in the 96 multiwell plates was described previously [18]. The plates were stained with crystal violet (Sigma), photographed and subsequently analysed for quantitation of OD562 using a Bio Lab-Tek plate reader. The OD562 of cells infected with a given dilution of inoculum was normalized to that of corresponding uninfected cells, to minimize variation of the monolayers, and the ratio was used to compare the cells transfected with IFN plasmids to those transfected with ptGFP1. A polyclonal antibody was raised in rabbit against a peptide (TCASPENESPRLRM) located in the extended N terminal region of iIFN1b to specifically detect the iIFN1b protein in RTG-2 cells transfected with ptGFP1-iIFN1b. The lysates of RTG-2 cells (48 h after transfection) and HEK 293T cells (24 h after transfection) were separated by SDS-PAGE, transferred to PVDF membranes, immunoblotted with the primary antibodies against iIFN1b (1∶100); TurboGFP (1∶2000); actin (1∶2000); viperin (1∶2000); STAT1 (1∶600); STAT2 (1∶1000); pSTAT1 (Tyr701) (1∶1000), or p-STAT2 (Tyr690) (1∶1000) ) and then the secondary HRP conjugated goat anti-mouse or goat anti-rabbit antibody (Thermo Scientific). The proteins were visualized using the SuperSignal Western Blotting Kit (Thermo Scientific) according to the manufacturer' s instructions. The RTG-2 cells transfected with GFP or GFP fusion plasmids were examined under a fluorescent microscope and photographed. For confocal microscopic analysis, the RTG-2 cells cultured on the coverslips were fixed at 36–48 h after transfection, incubated with the primary antibodies (mouse anti-V5 or rabbit anti-Flag antibody) (1∶250) and the secondary antibodies (Alexa FluorR 488 goat anti-rabbit IgG or Alexa FluorR 594 goat anti-mouse IgG, Invitrogen) (1∶200). The coverslips were examined using a Zeiss Axioplan 2 fluorescent microscope.

The type I interferon (IFN) family consists of multiple members which are encoded by intronless genes in reptiles, birds and mammals but intron-containing genes in amphibians and fish. They coordinate antiviral defence by binding to cell surface receptors. Here, we demonstrate for the first time in vertebrates, that intracellular IFNs can be produced from alternatively spliced IFN transcripts and are able to elicit cellular responses through intracellular IFN receptors. This functional intracellular IFN system in fish may play a significant role in activating antiviral pathways in cells infected with virus or in neighbouring cells, and represents a novel defence to combat viral pathogens.

Abstract Introduction Results Discussion Materials and Methods

2013

Intracellular Interferons in Fish: A Unique Means to Combat Viral Infection

Lipopolysaccharide (LPS) is one of the most important virulence and antigenic components of Burkholderia pseudomallei, the causative agent of melioidosis. LPS diversity in B. pseudomallei has been described as typical, atypical or rough, based upon banding patterns on SDS-PAGE. Here, we studied the genetic and molecular basis of these phenotypic differences. Bioinformatics was used to determine the diversity of genes known or predicted to be involved in biosynthesis of the O-antigenic moiety of LPS in B. pseudomallei and its near-relative species. Multiplex-PCR assays were developed to target diversity of the O-antigen biosynthesis gene patterns or LPS genotypes in B. pseudomallei populations. We found that the typical LPS genotype (LPS genotype A) was highly prevalent in strains from Thailand and other countries in Southeast Asia, whereas the atypical LPS genotype (LPS genotype B) was most often detected in Australian strains (∼13. 8%). In addition, we report a novel LPS ladder pattern, a derivative of the atypical LPS phenotype, associated with an uncommon O-antigen biosynthesis gene cluster that is found in only a small B. pseudomallei sub-population. This new LPS group was designated as genotype B2. We also report natural mutations in the O-antigen biosynthesis genes that potentially cause the rough LPS phenotype. We postulate that the diversity of LPS may correlate with differential immunopathogenicity and virulence among B. pseudomallei strains. Lipopolysaccharide (LPS) is a major component of the outer membrane of Gram-negative bacteria, playing an important role in cell integrity and in signaling host innate immune response [1]. Structurally, LPS is composed of three major components: lipid A, the bacterial endotoxin that is embedded in the phospholipid bilayer of the outer membrane; core-oligosaccharide; and O-antigen. These three components are linked together as a part of the bacterial outer membrane. In a highly pathogenic bacterial species, such as Burkholderia pseudomallei, LPS has a major role in stimulating host innate immune response during infection [2]. B. pseudomallei LPS has been classified as a type II O-polysaccharide (O-PS) and is one of 4 different surface polysaccharides produced by this pathogen [3]. Previous studies have shown that B. pseudomallei LPS is required for serum resistance and virulence [4]. It has been well established in many bacterial diseases that overstimulation of the host cells by LPS can lead to the features of septic shock [5]. Likewise for B. pseudomallei, septicemia is a major cause of death. In Northeast Thailand especially in Ubon Ratchathani Province where melioidosis is highly endemic, the average incidence rate of melioidosis is 12. 7 cases per 100,000 people per year with the average of 42. 6% of mortality rate [6]. Cellular recognition of LPS by the innate immune system triggers the proinflammatory cytokines by host cells, which aids in the clearance of the pathogen. Previous studies have supported a potential role for B. pseudomallei LPS in protective immunity, with high concentrations of antibodies to LPS associated with survival in severe melioidosis [7], [8]. As a result, LPS has been used in vaccine development and provided protective immunity in a murine model of melioidosis [2]. In addition, it was demonstrated that LPS had an important role in bacterial virulence because the LPS mutant B. pseudomallei strain SRM117, which lacked the O-antigenic polysaccharide moiety was more susceptible to macrophage killing during the early phase of infection than its parental strain 1026b [9]. A previous study [10] identified LPS diversity based upon electrophoretic mobility with SDS-PAGE and detection using immunoblot analysis. This diversity included two serotypes (A and B) possessing different electrophoretic ladder profiles and a rough type that did not contain the ladder patterns; all were antigenically distinct [10]. Molecular structure of O-antigen serotype A or typical type has been described as the unbranched heteropolymers consisting of disaccharides repeats of -3) -β-D-glucopyranose- (1-3) -6-deoxy-α-L-talopyranose- (1- in which approx. 33% of the L-6dTalp residues bear 2-O-methyl and 4-O-acetyl substituents whereas the other L-6dTalp residues carry only 2-O-acetyl substituents [11]. We note that the structures are not known for any of the other B. pseudomallei O-antigens. B. thailandensis, a genetically related non-pathogenic species, has LPS that is cross-reactive to sera obtained from B. pseudomallei and B. mallei infections, and this has led to the development of a vaccine for melioidosis using LPS from B. thailandensis [12]. B. mallei, the causative agent of glanders, has O-antigen structure similar to those found in B. pseudomallei and B. thailandensis, except that it has different side-group modifications at the L-6dTalp residues which lack the acetylation at the O-4 position [13]. These structural differences are associated with the absence of oacA gene in B. mallei. oacA encodes for O-antigen acetylase A in B. thailandensis and its homolog in B. pseudomallei K96243 is identified as BPSL1936 [14]. B. pseudomallei genomes are very diverse due to horizontal gene transfer events [15], [16] and dynamic changes in repeated sequences [17]. This results in diverse phenotypic characteristics such as bacterial colony morphotypes [18], and importantly, may be implicated in the diverse clinical manifestations observed among melioidosis patients. The latter range from asymptomatic cases, to localized infections, to whole body sepsis, along with differential seroreactivities [19], all of which may be correlated with the great genomic diversity in this species [15], [17]. Nevertheless, the specific roles of genetic diversity in B. pseudomallei in differential clinical presentations of melioidosis requires further analysis, as clinical studies suggest host risk factors are the major determinant of disease severity [20]. Because LPS phenotypic diversity is important for serology and diagnostics, we investigated the genetic and molecular basis of differential LPS phenotypes in a large B. pseudomallei population. Bioinformatics, phenotypic characterization, as well as, population genetics approach were used in this study to better understand this important trait. Artemis and Artemis Comparison Tool (ACT) software [21] was used to display and compare multiple B. pseudomallei genomes. Genomes and nucleotide sequences used in this study are listed in Table 1. Mutations in O-antigen biosynthesis genes were identified using basic homologous gene based alignments. Multiplex-SYBR-Green PCR assays were designed to target 3 different LPS genotypes. Gene wbiE of B. pseudomallei K96243, gene BUC_3396 of B. pseudomallei 576, and gene BURP840_LPSb16 of B. pseudomallei MSHR840 were used as the PCR markers to investigate frequency of LPS genotypes A, B, and B2, respectively (Figures 1&2). PCR primers used in this study are as follows: wbiE_F, 5′-TCAAACCTATCCGCGTGTCGAAGT-3′; wbiE_R, 5′-TCGTCGTCAAGAAATCCCAGCCAT-3′; BUC3396_ F, 5′-AATCTTTTTCTGATTCCGTCC-3′; BUC3396_R, 5′ -ACCAGAAGACAAGGAGAAAGGCCA-3′; BURP840_LPSb16_F, 5′-AACCGGGTAGTTCGCGATTAC-3′; and BURP840_LPSb16_R, 5′-ATACGCCGGTGTAGAACAGTA-3′. The PCR assay was conducted in 10-µL reaction mixtures containing 1× SYBR-Green master mix (Applied Biosystems, USA), 0. 3 µM of each PCR primer, and 0. 1 to 1. 0 ng of DNA template. Most tested DNA samples were made in collaborative laboratories in Thailand and Australia using various DNA extraction techniques. The reactions were performed on an ABI 7900HT Sequence Detection System (Applied Biosystems, USA) utilizing 40 cycles. Each cycle contained two steps: denaturation at 95°C for 15 s and annealing at 60°C for 30 s. The PCR products were further analyzed by melting them continuously from 60°C to 95°C to generate a dissociation curve. The melting temperatures of PCR amplicons for genes wbiE, BUC_3396, and BURP840_LPSb16 were constant at 87. 0°C, 83°C, and 88. 5°C, respectively. We used this assay to analyze DNA templates from 999 diverse B. pseudomallei strains isolated from clinical, animal, and environmental samples from Australia (n = 600), Thailand (n = 349), Malaysia (n = 27), Vietnam (n = 7), Papua New Guinea (n = 2), and unknown countries in Southeast Asia (n = 14), as well as 77 B. thailandensis strains, 2 B. thailandensis-like spp. strains, and 37 strains of unknown soil bacteria. Whole genome sequencing was performed using 454 sequencing technology (Roche, USA) by US Army Edgewood Chemical Biological Center (ECBC), MD, USA. Artemis –based analysis and BLAST were used to annotate the O-antigen biosynthesis genes of B. pseudomallei strains MSHR840, MSHR139, and MSHR1950. DNA sequencing for wbiI and oacA genes was performed using ABI 3130×l Genetic Analyzer (Applied Biosystems, USA). LPS identification and characterization: Techniques for LPS extraction and SDS-PAGE analysis followed a previous study [10]. Immunoblot analysis was performed using sera from melioidosis patients with known infection with B. pseudomallei LPS genotype A or B strains as the primary antibodies. Horse radish peroxidase (HRP) – conjugated anti-human IgG was used as the secondary antibody in a standard immunoblot analysis. Monoclonal antibody 3D11, the B. mallei LPS-specific mAb (Research Diagnostics Inc. , USA), was used as a primary antibody in the immunoblot analysis of the oacA mutant strains. Select B. pseudomallei strains were tested for growth, multiplication, and survival in the presence of 30% normal human serum (NHS) as previously described [4] with some modifications. Briefly, each B. pseudomallei strain was inoculated in a 2 mL of TSBDC media and incubated overnight at 37°C and 250 rpm in an orbital shaker. The overnight culture (100 µL) was used to inoculate 3 mL of TSBDC media and then incubated at the same conditions for 4 hr to reach mid exponential growth phase. Serum susceptibility tests were performed in 1. 5 mL microfuge tubes containing 100 µL of bacterial culture, 300 µL of NHS (Lonza Inc. , USA), and 600 µL PBS. The mixture was incubated at 37°C for 2 hr, and then the number of viable bacterial cells was determined using plate counting. B. pseudomallei 1026b and E. coli HB101 were used as positive and negative controls in this study, respectively. Nucleotide sequences and annotations of the O-antigen biosynthesis genes in B. pseudomallei strains MSHR840, MSHR139, and MSHR1950LPS were submitted to GenBank under accession nos. GU574442, HM852063, and HM852062, respectively. We compared 27 B. pseudomallei, 10 B. mallei, 3 B. thailandensis, and 2 B. oklahomensis genomes (Table 1) to identify the LPS O-antigen biosynthesis genes. Assuming synteny and common genomic locations, along with known or predicted function, B. pseudomallei O-antigen biosynthesis genes were assigned to two major groups. Group A (LPS genotype A) was identical or very similar to the O-antigen biosynthesis operon observed in B. pseudomallei 1026b [4], whereas group B (LPS genotype B) was found in an atypical LPS strain 576 and also in the species type strain, NCTC13177. LPS genotype A was found in most B. pseudomallei and all B. mallei and B. thailandensis genomes examined. Surprisingly, the more distantly related B. oklahomensis strain EO147 also had LPS genotype A, which was different from the predicted O-antigen biosynthesis gene cluster in other B. oklahomensis strains (C6786, C7532, and C7533; data not shown). This may represent a lateral gene transfer event into EO147 and is deserving of additional study. Furthermore, regions within the two clusters had different levels of sequence conservation. Genes located at the ends of these two clusters (e. g. , wbiGHI, and rmlBAC; Figure 1) had higher sequence similarity than most of the genes in the core of the clusters. Indeed, many of the cluster cores contain distinct gene composition. The conserved genes include those important for oligosaccharide synthesis and O-antigen biosynthesis [4]. LPS genotype frequencies were analyzed across a large strain collection using PCR-based assays. Multiplex-SYBR-Green PCR assays were designed to target a specific gene unique for each genotype. Gene wbiE (BPSL2676) of B. pseudomallei strain K96243 and gene BUC_3396 of strain 576 were used to represent the presence of LPS genotypes A and B, respectively (Figures 1&2). A total of 999 B. pseudomallei strains from different geographic locations and epidemiological origins (e. g. , clinical, animal, and environmental strains) were tested for their LPS genotypes. We noted that 23 B. pseudomallei strains were collected from one melioidosis patient. We found that LPS genotype A was the most common genotype in both Australian and Southeast Asian strain populations (Figure 2). LPS genotype B was relatively rare in Southeast Asian strains (∼2. 3%), but was found in 13. 8% of Australian strains. Five strains from Australia and two strains from Papua New Guinea were non-typeable using these two PCR gene markers. Three of these strains, MSHR840, MSHR1950, and MSHR139 were further analyzed for O-antigen biosynthesis gene identification using whole genome sequencing. The O-antigen biosynthesis gene clusters from these strains were identified and annotated (GenBank accession nos. GU574442, HM852062, HM852063). Comparative genomics demonstrated that many genes in this new cluster were similar to those of the LPS genotype B genes of B. pseudomallei 576 and were distinct from the K96243 LPS genotype A genes. Hence, these newly identified O-antigen biosynthesis gene clusters represent a variant of the LPS genotype B and, consequentially, were designated as LPS genotype B2 (Table 1). Figure 1 shows the genomic comparison of these three different O-antigen biosynthesis gene clusters: A, B, and B2 (from B. pseudomallei strains K96243,576, and MSHR840, respectively). We note that %G+C content of the core of these 3 different clusters is relatively low (∼59–60%) compared to the conserved parts of the O-antigen biosynthesis operon (∼68%). This supports the hypothesis that these genomic differences are due to genetic recombination e. g. , horizontal gene transfer, which is common in B. pseudomallei [15], [16]. Comparative genomics of these three different clusters using homologous-based alignment are summarized in Table S1. Again, we note that genes wbiGHI, and rmlBAC are conserved among these three different clusters. Furthermore, gene BURP840_LPSb16 from strain MSHR840 was selected for use as a PCR marker to represent the LPS genotype B2. PCR genotype analysis (Figure 2) revealed that all seven of the previously non-typeable strains were positive for the LPS genotype B2. The LPS B2 genotype was found only in strains from Australia and Papua New Guinea. It is important to note that there is no known clonal relationship among these seven strains. The LPS B2 genotype genes were also found in a B. thailandensis-like spp. strain MSMB121, which was isolated in Australia (unpublished data). Complete LPS genotypic data are reported in Table S2. LPS genotyping results were further examined by direct comparison to LPS electrophoretic phenotypes [10]. Due to the difficulty of international Select Agent transfer and BSL3 handling, we phenotyped only ∼ 24% of the isolates that were genotyped. We note that this is a limitation of our study. That said, all examined LPS A or B phenotypes were perfectly matched with their LPS A or B genotypes. In addition, 22 strains producing the rough LPS phenotype were all identified as LPS genotype A (Table S2). The genetic basis of the rough phenotype and its derivation from the A phenotype is known for only 16 of these strains (see below). SDS-PAGE revealed that LPS genotype B2 strains produced a distinct ladder pattern, though they were all detectable with type B sera using immunoblot hybridization. The B2 phenotype had a wider range of molecular weights (40–120 kDa) than the LPS types A and B. In total, three LPS banding patterns plus the rough LPS type (no ladder) can be detected (Figure 3). A frame-shift mutation observed in the O-antigen biosynthesis wbiI gene of B. pseudomallei strain MSHR1655 was correlated with its rough phenotype. This is one of nearly 100 strains that were isolated over 8 years from a patient with severe bronchiectasis associated with melioidosis. The mutation was an extra guanine inserted after nucleotide position 815 of the wbiI gene (Figure 4). The wbiI gene encodes an oligosaccharide epimerase/dehydratase and is conserved in all O-antigen biosynthesis gene clusters of B. pseudomallei. A mutation in this gene probably impacts on the synthesis of the O-antigen in this bacterial strain. There were 23 serial B. pseudomallei isolates observed from the chronically infected patient and the wbiI gene sequences were determined in all of them to detect frame shift mutations. The frame-shift mutation occurred in 16 isolates, all of which were collected on or after day 550 of the infection. The wild type sequence was present in the other seven isolates from earlier in the infection (Figure 4). Moreover, phenotypic characterization revealed that LPS samples extracted from the 16 wbiI mutated strains did not have the O-antigen ladder pattern (i. e the rough phenotype) based upon SDS-PAGE and silver straining (Figure 5A). Thus, it seems likely the frame-shift mutation in the wbiI gene blocks synthesis of the O-antigen. A recent study has reported that oacA gene, known to be involved in the acetylation at the O-4 position of the L-6dTalp residues of B. thailandensis O-antigen [14], is mutated in B. pseudomallei MSHR1655. Since MSHR1655 was isolated from the same patient above, we then sequenced the oacA gene in all of these clonal strains. We found that the oacA mutation occurred in the same 16 strains that had the wbiI mutation (Figure 4C). Additional study of the oacA gene in other whole genome sequenced strains determined that B. pseudomallei 112 and B. thailandensis TXDOH also had point mutation in their oacA genes (Table 1; Figure S1). To determine if the oacA gene plays only a single role in the side group modification of the L-6dTalp residues, or a dual role in combination with the synthesis of the O-antigen, both strains were tested for O-antigen production and immunogenic specificity. We found that B. pseudomallei 112 and B. thailandensis TXDOH expressed O-antigen type A ladder pattern and their O-antigen bands were strongly positive with the B. mallei LPS-specific mAb 3D11 (Figure 6) that recognized the lack of 4-O acetylation of the L-6dTalp residues [14]. This suggests the oacA gene in B. pseudomallei and B. thailandensis has a role in the acetylation at the O-4 position of the O-antigen L-6dTalp residues but is not involved in the synthesis of the O-antigen. Thus, we determined that the rough LPS phenotype observed in the 16 clonal chronic lung strains was due to the mutation of their wbiI gene, but not from the effect of the oacA mutation. In this study, we also identified six other independent rough LPS strains, but mutations did not occur in their wbiI or oacA genes. Searching for mutations in other genes of these strains warrants a follow up study to understand alternate mechanisms that generate the rough phenotype. Because LPS is essential for outer membrane integrity and serum resistance, four B. pseudomallei strains from this chronic lung patient were further tested in serum bactericidal assays. Two of the wbiI mutant strains that expressed the rough LPS phenotype (MSHR1655 and MSHR3042) were unable to grow in the presence of 30% normal human serum (NHS). In contrast, two early infection isolates from the same patient expressing the typical LPS A phenotype (MSHR1043 and MSHR1048) were able to resist the inhibitory human serum effect and grow (Figure 5B). Furthermore, we also confirmed that the LPS genotype B2 strains were killed in growth media containing 30% NHS, whereas the LPS genotype B strains were resistant (Figure S2). We believe that this finding of serum susceptibility in LPS genotype B2 is important and deserves further investigation. Despite the fact that genes responsible for the O-antigen biosynthesis in B. pseudomallei 1026b were identified many years ago [4], diversity of these genes across multiple B. pseudomallei strains has not been well studied until now. Advances in genome sequencing and comparative genomics have provided insights into the complexity and diversity of B. pseudomallei genomes. B. pseudomallei genomic studies can now strive for correlations between genomic diversity and differential phenotypes; perhaps the clinical outcomes of individual strains of B. pseudomallei may be predicted using basic genomic analysis. In our current study, we were able to establish a correlation between differential LPS phenotypes and diversity of O-antigen biosynthesis genes or known as LPS genotypes. Three different major LPS genotypes have been identified so far. LPS genotype A was designated to the strains that contained the O-antigen biosynthesis genes that were identical or very similar to those found in a reference strain 1026b [4], whereas the LPS genotype B category is represented by the atypical LPS strain 576. Finally, LPS B2 genotype was identified as a variant of the LPS genotype B because many of its O-antigen biosynthesis genes were similar to those of LPS genotype B, and both groups were serotype B positive. LPS genotype A was the most common genotype in both geographic locations: Southeast Asia and Australia where it accounted for 97. 7% and 85. 3% of the populations, respectively. Interestingly, the frequency of LPS genotype B was relatively high (approx. 13. 8%) in Australian strains, whereas they accounted for only 2. 3% of the strains from Southeast Asia. LPS genotype B2 was found in only 7 strains, 5 of which were from Australia, and the other 2 strains were from Papua New Guinea. In addition, LPS genotype B2 was also found in a member of B. thailandensis-like species which was recently discovered in Australia [22]. This would suggest that the LPS genotype B2 genes in B. pseudomallei may be acquired by horizontal gene transfer from a common soil bacterial species in Australia, or vice versa. Comparative genomics and phenotypic characterization of this LPS genotype B2 in B. pseudomallei and its near-relative species warrants further investigation. Because the LPS genotypes B and B2 were frequently found in Australia but not in Southeast Asia, it is possible that this finding may be due to different therapies used for clinical cases in these 2 endemic locations. We have investigated this and found that the majority of these isolates were obtained before any exposure to antibiotics or treatment therapy. In addition, some of the LPS genotype B strains were collected from soil in Australia, and 2 strains of the LPS genotype B2 were found in animal cases. This confirms that the occurrence of LPS types B and B2 in Australia is not associated with the exposure to antibiotics or treatment therapy. Although, we phenotyped only 24% of the isolates that were genotyped, most tested strains were perfectly matched between their genotypes and phenotypes, except those 16 rough LPS genotype A strains from a single chronic case that had mutations in their wbiI genes (Figure 4). In this current study, we were unable to identify the genetic basis or mutations in 6 independent LPS genotype A strains that did not produce the O-antigen (Table S2). Because the typical LPS was also found in B. thailandensis, the use of anti-LPS antibody based latex agglutination for the identification of B. pseudomallei in environmental specimens was not successful in an early study [23]. B. thailandensis LPS has also been shown to cross-react with rabbit and mouse sera obtained from inoculation with B. pseudomallei or B. mallei suggesting that LPS molecules from B. thailandensis, a non-pathogenic bacterium, may be useful in ongoing efforts to develop novel vaccines and/or diagnostic reagents [24]. This has brought to our attention whether low-grade B. thailandensis infections might naturally provide protection against melioidosis. Although the O-antigen biosynthesis genes in B. pseudomallei and B. thailandensis are similar, a recent study by a Singaporean group has revealed that lipid A components of the LPS from both B. pseudomallei and B. thailandensis must be different; the murine and human macrophages produced lower levels of tumor necrosis factor alpha, interleukin-6 (IL-6), and IL-10 in response to B. pseudomallei LPS than in response to B. thailandensis LPS in vitro [25]. In our current study, the typical LPS was also found in B. oklahomensis strain EO147, formerly known as an American B. pseudomallei strain [26], suggesting that the typical LPS is widely spread in multiple Burkholderia species. This group includes highly pathogenic species such as B. pseudomallei and B. mallei, but also non-pathogenic species: B. thailandensis, B. thailandensis-like species, and B. oklahomensis. The evolution of LPS diversity across these closely related species is likely a function of differential selection and horizontal transfer of genetic elements. This diversity could play a role in frequency and distribution of disease in humans. However, without understanding molecular structures of these O-antigen types, it is difficult to access the phenotypic effects of this genetic diversity. Structural analysis of the O-antigen types B and B2 deserves further investigations. In addition, we have found that the LPS genotype B2 strains were sensitive to 30% NHS, whereas the LPS type B strains were resistant (Figure S2). This finding demonstrates a level of phenotypic differences between these two serologically related groups. We believe that the consequences for case treatment associated with these differential serum susceptibilities also warrant further investigations. A previous study has shown that the two less common LPS phenotypes (smooth type B and rough type) were more prevalent in clinical than environmental isolates and more prevalent in Australian isolates than Thai isolates [10]. In our current study, LPS genotype B was found in both clinical and environmental strains from Australia, whereas the rough LPS was still found only in clinical strains. Based on our description of the molecular basis for LPS phenotypes, it is unlikely that B. pseudomallei will readily switch its LPS phenotype from A to B, or vice versa, as has been suggested previously [10]. The gene compositions of LPS genotypes A and B are very different and a simple switching mechanism is difficult to envision. In addition, we have found that at least some rough LPS strains have mutations in their O-antigen biosynthesis genes. These include 16 clonally related isolates from a single chronic lung infected patient (Table S2). All of these strains were identified as LPS genotype A with mutations in their O-antigen biosynthesis genes. Using Tn5-OT182 mutagenesis, DeShazer and colleagues identified at least seven genes in the O-antigen biosynthesis operon of B. pseudomallei 1026b that were responsible for O-antigen biosynthesis and serum resistance; these included rmlB, rmlD, wbiA, wbiC, wbiE, wbiG, and wbiI [4]. In our current study, we found point mutations in wbiI and oacA genes of B. pseudomallei isolates that were collected from a chronic lung patient (Figure 4). We hypothesize that the frame-shift mutation in the wbiI genes blocks O-antigen biosynthesis in all mutant strains, but not from the effect of the oacA mutation. This is because we observed the oacA mutations in B. pseudomallei 112 and B. thailandensis TXDOH that had normal O-antigen biosynthesis gene cluster (Table 1 and Figure S1). Our study has demonstrated that these two oacA mutant strains expressed O-antigens identical to those found in B. mallei due to lack of the 4-O acetylation of the L-6dTalp residues of the O-antigen. The lack of the 4-O acetylation of the L-6dTalp residues has recently been described in the oacA knock-out mutant B. thailandensis ZT0715 and a wild-type B. mallei ATCC23344 [14]. We have demonstrated that these wbiI mutant strains produced rough LPS and were sensitive to normal human serum suggesting that the wbiI gene encoding for epimerase, or dehydratase, was essential for the biosynthesis of B. pseudomallei O-antigen. Although loss of the O-antigen might compromise serum survival it might also be adaptive in particular niches. B. pseudomallei survival or persistence in the host might be enhanced without the surface presentation of the O-antigenic moiety of the LPS, as it would not be recognized by host immune systems and would, therefore, avoid being killed by antibodies. The O-antigenic polysaccharide of B. pseudomallei modulates the host cell response, which in turn controls the intracellular fate of B. pseudomallei inside macrophage. This was concluded from the observation that the O-antigen mutant B. pseudomallei strain SRM117 was more susceptible to macrophage killing during the early phase of infection than the parental wild-type strain 1026b [27]. This was also confirmed by the same group when they demonstrated the importance of intracellular killing by the human polymorphonuclear cells (PMNs), macrophages (Mφs), and susceptibility to killing by 30% normal human serum [28]. LPS and CPS (capsular polysaccharide) have been used as subunits in immunizing BALB/c mice against B. pseudomallei infection [2]. Mice vaccinated with LPS developed predominantly IgM and IgG3 responses, whereas the mice vaccinated with the CPS developed a predominantly IgG2b response. Furthermore, immunization with the LPS provided an optimal protective response, and the immunized mice challenged by the aerosol route showed a small increase in the mean time to death compared with the unvaccinated controls [2]. Previously, it was shown that B. pseudomallei LPS from strain 1026b signaled through Toll-like receptor (TLR) 2 and not through TLR4 [29]. This was observed in the TLR2 knock-out mutant mice that displayed a markedly improved host defense, but it was not observed in TLR4 knock-out mice [29]. In contrast, a study in HEK293 cells demonstrated that heat-killed B. pseudomallei strains K96243 or BP-1 activated TLR2 and TLR4, and in the presence of MD-2, LPS and lipid A from BP-1 are TLR4 ligands [30]. We note that B. pseudomallei 1026b and K96243 expressed the typical O-antigen type A, but the O-antigen type of BP-1 was not reported in that study. Although there was no report of association between the LPS types and disease severity (e. g. , fatal versus non-fatal, and septicemia versus localized), clinical manifestations (neurologic versus non-neurologic), or underlying risk factors (diabetic versus non-diabetic) observed in a previous study [10], full phenotypic characterization including virulence in animal models, innate immune response, etc of these different LPS types warrants further investigations given the LPS diversity that we have described.

Burkholderia pseudomallei is an environmental Gram-negative bacterium and the cause of melioidosis, an often life-threatening disease affecting people in Southeast Asia and northern Australia. Melioidosis is usually contracted by bacterial inoculation, ingestion or inhalation. Effective vaccines for melioidosis are currently unavailable. This organism contains a large genome, which varies greatly among strains due to a high frequency of genetic recombination. We report here on diversity of lipopolysaccharides (LPS) in this species, a major component of the bacterial outer membrane and a known immunogenic virulence factor. We developed LPS genotyping techniques to study frequency of two major LPS types, known as typical and atypical LPS, in B. pseudomallei strains collected from two endemic regions: Southeast Asia and Northern Australia. LPS genotype variation differed among B. pseudomallei populations. During the investigation, we discovered a new LPS genotype in a sub-population group of B. pseudomallei in Australia. We postulate that such differences are likely to be associated with variable immunopathogenicity and clinical presentation in the human host.

Abstract Introduction Methods Results Discussion

medicine public health and epidemiology epidemiology infectious disease epidemiology genetics immunology biology genomics microbiology genetics and genomics

2012

The Genetic and Molecular Basis of O-Antigenic Diversity in Burkholderia pseudomallei Lipopolysaccharide