enoriega/keyword_pubmed · Datasets at Hugging Face

sentence stringlengths 1 5.95k	pmcid stringlengths 9 10	keyword_rank int32 -1 169k	keyword_indices sequence
The anthropogenic fixation of dinitrogen ( N ) , by means of the Haber-Bosch process , has led to the overuse of synthetic nitrogen-based fertilizers in agriculture ( 1 , 6 ) .	PMC4049103	-1	[]
As a consequence of the increased nitrogen ( N ) content of soils , atmospheric N O concentrations have risen about 20 % relative to preindustrial-era levels ( 2 ) .	PMC4049103	-1	[]
N O emissions are largely the result of bacterial pathways controlling the nitrogen cycle .	PMC4049103	-1	[]
In particular , N O is generated primarily as a product of incomplete classic denitrification ( i.e ., NO reduction to N O via NO and NO ) and secondarily as a by-product of dissimilatory nitrate reduction to ammonia ( DNRA ) and oxidation of ammonium to nitrite ( nitrification ) ( 7 , 8) .	PMC4049103	-1	[]
Besides bacterial activities , abiotic processes and fungal denitrification are thought to be sources of N O ( 9 , 10 ) .	PMC4049103	-1	[]
Model predictions of N O consumption in terrestrial environments focus primarily on the N O-to-N reduction step , presently attributed to classical denitrifiers possessing nitrous oxide reductase ( NosZ ) ( 7 ) .	PMC4049103	-1	[]
Our previous work has revealed the existence of two phylogenetically distinct NosZ clades , one encompassing typical Z-type NosZ proteins , which are commonly found in the Alpha - , Beta - , and Gammaproteobacteria , and the other encompassing atypical NosZ proteins present in diverse organisms representing different phyla .	PMC4049103	-1	[]
Further analysis of sequenced genomes revealed that most of the typical nosZ genes are found in bacteria capable of complete denitrification ( i.e ., encoding all the enzymes for converting NO / NO to N ) , whereas atypical nosZ genes are found in bacteria with more-diverse N metabolism , including those performing DNRA and missing the NO-generating nitrite reductase genes nirK and nirS ( 11 , 12 ) .	PMC4049103	-1	[]
Notably , atypical NosZ proteins have been shown to function as nitrous oxide reductases in several bacteria , such as Wolinella succinogenes ( 13 , 14 ) , Geobacillus thermodenitrificans ( 15 ) , the soil isolate Anaeromyxobacter dehalogenans ( 11 ) , and several Bacillus species isolated from soils ( 16 , 17 ) .	PMC4049103	28,139	[ 37 ]
Examination of the potential of microbial communities to reduce N O to N has traditionally been performed by evaluating nosZ gene and / or transcript presence or abundance by PCR ( 18 , 19 ) .	PMC4049103	-1	[]
Primers targeting nosZ genes , however , were designed according to characterized typical nosZ gene sequences and therefore missed the bulk of divergent atypical genes ( 11 , 12 ) .	PMC4049103	-1	[]
Furthermore , measured N O emissions from soils were frequently lower than predictions based on ( typical ) NosZ transcript abundance and dynamics ( 20 , 21 ) .	PMC4049103	-1	[]
Therefore , it is likely that atypical NosZ abundance accounts , at least in part , for the discrepancy between predicted and observed N O fluxes.To circumvent the limitations and explore the total natural diversity of nosZ genes in the environment , we analyzed short-read metagenomic data sets from various soils and locations .	PMC4049103	-1	[]
Even though metagenomics can provide a relatively unbiased , PCR-independent view of the diversity and abundance of individual genes present in a sample , several technical challenges must first be addressed .	PMC4049103	-1	[]
For instance , in metagenomes of highly diverse microbial communities , such as those obtained from soils , the rates of false positives and false negatives when using similarity searches to detect individual genes in assembled contigs or unassembled short reads has not been rigorously evaluated , with the probable exception of error rates assessed for the purpose of taxonomic classification , i.e ., assigning a sequence to a taxon without necessarily evaluating its potential function and sequence diversity ( 22 , 23 ) .	PMC4049103	-1	[]
Cutoffs that might minimize the number of false-positive matches have not been determined for short-read metagenomes ; instead , arbitrary , predetermined cutoffs based on E values ( i.e ., the likelihood of finding a match by chance ) represent the common practice ( 24 ) .	PMC4049103	-1	[]
The objective of the present study was to analyze the diversity and abundance of both typical and atypical nosZ genes in soils with contrasting physicochemical properties .	PMC4049103	-1	[]
To this end , we first developed a strategy based on similarity searches to determine appropriate cutoffs for accurately detecting metagenomic nosZ fragments by analyzing in silico metagenomes of known sequence composition .	PMC4049103	-1	[]
Subsequently , we applied this strategy and derived cutoffs to detect nosZ reads in metagenomes from two agricultural soils in the U.S. Midwest that have been subjects of an ongoing multiyear study to assess nitrogen cycling processes , as well as in publicly available metagenomes from various soil ecosystems .	PMC4049103	-1	[]
Our metagenomic , PCR-independent approach provided a comprehensive and quantitative examination of the diversity and abundance of both typical and atypical nosZ genes in soils .	PMC4049103	-1	[]
RESULTS .	PMC4049103	-1	[]
Evaluating search algorithms and cutoffs for detecting nosZ genes in metagenomes ..	PMC4049103	-1	[]
To determine the best algorithm and parameters for detecting nosZ reads in 100-bp-long read metagenomes , a reference database of manually verified nosZ genes that preclustered at 95 % sequence identity was queried against two such in silico-generated data sets , libraries I ( representing the whole genome of 122 NosZ-encoding organisms ) and II ( representing the whole genome of 1,081 bacteria , including those in library I ) .	PMC4049103	-1	[]
From a receiver operating characteristic ( ROC ) curve analysis of true - and false-positive rates , the most appropriate bit score cutoff values were 107 and 52.2 for the BLASTn and BLASTx searches , respectively .	PMC4049103	-1	[]
These bit scores provided sensitivities ( the fraction of correctly classified positive BLAST matches ) of 100 % and 98.4 % for BLASTn and BLASTx , respectively , and a specificity ( the fraction of correctly classified negative BLAST matches ) of 99.9 % for both algorithms .	PMC4049103	-1	[]
A hidden Markov model ( HMM ) search resulted in a sensitivity of 46 % and a specificity of 99.9 % ( Table 1 ) .	PMC4049103	-1	[]
Additional HMM searches , using models that included more typical and atypical NosZ sequences ( built from reference sequences in Table S1 in the supplemental material ) improved the number of nosZ reads retrieved from both in silico libraries ( ~ 56 % ) .	PMC4049103	-1	[]
Irrespective of the HMM model employed , a lower fraction of nosZ reads was captured when the HMM was used than when BLAST searches were performed .	PMC4049103	-1	[]
Most of the reads missed by the HMM-based approach lacked highly conserved amino acid residues , and this accounted for the lower performance of HMM searches , consistent with expectations ( HMM models rely heavily on conserved residues ) .	PMC4049103	-1	[]
Therefore , remaining analyses were performed using the BLAST algorithms.To test the limitations in retrieving metagenomic nosZ reads , single typical and atypical representative NosZ protein or nucleotide sequences were independently queried against library II. Although BLASTn had a specificity similar to that of BLASTx , the latter algorithm was able to capture 735 % and 270 % more reads ( i.e ., reads annotated as nosZ with a bit score greater or equal to the calculated cutoff for true positives ) of the typical and the atypical references , respectively .	PMC4049103	-1	[]
Therefore , BLASTx was used in the remaining analyses .	PMC4049103	-1	[]
Using typical NosZ from Bradyrhizobium japonicum strain USDA 110 as a single reference , reads derived from 74 out of 127 different alleles encoding NosZ were captured and found to be enriched in nosZ sequences closely related to the reference sequence .	PMC4049103	-1	[]
In contrast , reads for only 32 out of 127 alleles encoding NosZ were captured when atypical NosZ from Anaeromyxobacter sp. strain Fw 109-5 was used as a reference in the analysis ( Fig. 1 , right panel ) .	PMC4049103	28,139	[ 19 ]
This atypical NosZ reference does not exhibit sequence identity to other target sequences in the 54 to 82 % amino acid identity range , and thus , the lack of moderately related sequences among the target sequences accounts for the results obtained ( Fig. 1 , left panel ) .	PMC4049103	-1	[]
More importantly , a linear relationship was observed between the fraction of total reads detected and the level of divergence between the full-length reference and target sequences ( Fig. 1 ) , where 50 % or more of the reads were detected when the two sequences shared more than 64 or 68 % sequence identity for typical or atypical NosZ reference sequences , respectively.Further examination of the reads recruited along the typical NosZ reference ( Fig. 2 ) showed that the true-positive matches ( i.e ., reads derived from nosZ genes with a bit score greater than 52.2 ) were evenly distributed along the NosZ reference sequence .	PMC4049103	-1	[]
The N terminus of the reference sequence ( 1 to 60 amino acid positions ) was rarely covered by either true - or false-positive matches , suggesting that this part of the gene should be avoided when assessing nosZ abundance in metagenomes .	PMC4049103	-1	[]
Abundance of nosZ genes in sandy and silty soils ..	PMC4049103	-1	[]
A characterization of the taxa , coverage , and assembly statistics of the two soil metagenomes is described in the supplemental material .	PMC4049103	-1	[]
Both soil metagenomes were queried against a 95 % - identical preclustered set of reference NosZ sequences .	PMC4049103	-1	[]
All matches having a bit score greater than the calculated cutoff determined based on the in silico library analysis were identified as nosZ reads and classified as typical or atypical depending on their best match .	PMC4049103	-1	[]
Atypical nosZ reads were clearly the most abundant , comprising 72.9 % and 89.6 % of the total nosZ reads found in the sand and silt loam soil metagenomes , respectively ( Fig. 3 ) .	PMC4049103	-1	[]
Further , 97 % of the nosZ reads found in both soil metagenomes ( 4,929 and 7,280 , total , in the sandy and silty loam soils , respectively ) were recruited by 72 of the 105 NosZ reference sequences , revealing that most of the diversity covered by the references was represented in both soils ( Table S2 ) .	PMC4049103	-1	[]
In addition , both soil samples showed similar estimated absolute abundances for nosZ reads : ~ 1.4 × 10 and 2.1 × 10 reads of all reads for the Havana sand and Urbana silt loam , respectively ( Fig. 3 ) .	PMC4049103	-1	[]
The ratio of nosZ reads to single-copy housekeeping gene reads indicated that approximately 16 % of the soil bacterial genomes harbored a nosZ gene ( Table S3 ) .	PMC4049103	-1	[]
Phylogenetic analysis of the atypical nosZ reads showed that closely related genes found in members of the Anaeromyxobacter , Gemmatimonas , Opitutus , and Hydrogenobacter genera were the most abundant in both soil samples ( Fig. 4 ) .	PMC4049103	28,139	[ 17 ]
Additionally , less abundant genera , such as Bradyrhizobium and Rhodopseudomonas , both known to harbor typical nosZ genes , were also present in both soils .	PMC4049103	-1	[]
Remarkably , atypical nosZ reads affiliated with Anaeromyxobacter represented 12.7 % and 15.2 % of the total nosZ reads found in both sand and silt loam soils , respectively .	PMC4049103	28,139	[ 7 ]
The most abundant typical nosZ reads were assigned to the Ralstonia ( 3.2 % ) , Thiobacillus ( 3.1 % ) , Bradyrhizobium ( 1.6 % ) , and Rhodopseudomonas ( 1.7 % ) genera ( Table S2 ) .	PMC4049103	-1	[]
nosZ diversity and abundance in other soil metagenomes ..	PMC4049103	-1	[]
In general , atypical nosZ reads were more abundant than other reads in the soil metagenomes evaluated ( Fig. 5 ) .	PMC4049103	-1	[]
The frozen deep-soil permafrost metagenomes ( core 1 sample in reference 25 ) showed a greater abundance for typical nosZ reads ( ~ 80 % of total NosZ ) ; however , atypical nosZ reads predominated in the upper or active layer ( ~ 74 % of total nosZ sequences ) .	PMC4049103	-1	[]
Interestingly , after induced thawing , the microbial communities at both depths showed a small increase in the relative abundance of atypical nosZ reads .	PMC4049103	-1	[]
Furthermore , except with the boreal forest biome , several biomes studied by Fierer and colleagues ( 26 ) , including tropical forest , polar and hot desert , arctic tundra , and temperate grassland , showed higher abundances of atypical than of typical nosZ reads .	PMC4049103	-1	[]
DISCUSSION .	PMC4049103	-1	[]
Importance of atypical NosZ ..	PMC4049103	-1	[]
The discovery of functional atypical NosZ proteins has opened the possibility that a much larger number of microorganisms with previously unaccounted N O-reducing potential contribute to lessening the N O emissions to the atmosphere than previously expected ( 12 ) .	PMC4049103	-1	[]
The abundance and diversity of atypical nosZ genes were likely missed in previous PCR-based surveys because typical nosZ sequences presented the basis for primer design ( 11 , 12 ) and the two nosZ types share only 60.9 % ± 8.2 % nucleotide identity , on average .	PMC4049103	-1	[]
In the present PCR-independent metagenome analysis , atypical NosZ sequences were more abundant ( > 73 % of total nosZ reads ) than their typical counterparts , not only in two agricultural soils differing in physicochemical properties representative of many regions in the Midwest United States but also in soils from distant geographic locations representing a variety of habitats .	PMC4049103	-1	[]
Our results were also consistent with the widespread presence of atypical nosZ genes , previously hypothesized based on the number of genomes found to encode atypical NosZ proteins among the available genome sequences ( 11 ) .	PMC4049103	-1	[]
Therefore , these findings reveal an unexpectedly high potential for N O reduction mediated by atypical NosZ in a variety of soil habitats.It is important to note that our study , being solely based on DNA sequences , evaluated N O reduction potential as opposed to the specific in situ activity of NosZ enzymes , typical or atypical .	PMC4049103	-1	[]
Since negative ( purifying ) selection efficiently removes unused genes from genomes in microbial populations , the high abundance of atypical nosZ sequences found in different soil samples underpins their functional and / or ecological potential ( e.g ., Fig. 5 ) .	PMC4049103	-1	[]
Given also that N O reduction is the only known biochemical function carried by NosZ ( 11 ) , our results collectively suggest that atypical NosZ proteins are as important , if not more important , than their typical counterparts in controlling N O fluxes in soils and likely other environments .	PMC4049103	-1	[]
Our study also provided the means ( e.g ., gene sequences for primer design and a bioinformatics strategy ) to facilitate future studies of the effect of environmental conditions on NosZ activity and dynamics toward a more predictive understanding of the nitrogen cycle.The most abundant nosZ genes in the agricultural soils studied here are affiliated with the Anaeromyxobacter genus ( Fig. 4 ) .	PMC4049103	28,139	[ 57 ]
Members of this genus are widely distributed in soils with different physical and chemical characteristics as well as soils from a variety of geographic locations ( 11 , 27 , 28 ) .	PMC4049103	-1	[]
The high abundance of nosZ genes affiliated with members of the nondenitrifying Anaeromyxobacter genus is consistent with recent PCR surveys that employed primers targeting atypical nosZ sequences ( 11 , 12 ) and A. dehalogenans 16S rRNA gene sequences ( 11 ) .	PMC4049103	28,139	[ 12 ]
Further , a high phylogenetic congruence between typical nosZ and 16S rRNA gene phylogenies was previously reported ( 29 , 30 ) .	PMC4049103	-1	[]
Therefore , abundant atypical nosZ metagenomic sequences ( Fig. 4 and Table S2 ) that have distant matches to homologs of known NosZ-encoding taxa may be harbored by novel taxa .	PMC4049103	-1	[]
The sequences reported here should facilitate the identification of new taxa , expanding our understanding of phylogenetic diversity in NosZ-encoding soil organisms .	PMC4049103	-1	[]
The majority of the abundant atypical nosZ reads that were assignable to known taxa were found in potentially nondenitrifying genomes of genera such as Anaeromyxobacter , Ignavibacterium , Opitutus , Dyadobacter , and Gemmatimonas , which were overlooked in previous PCR surveys targeting typical nosZ genes .	PMC4049103	28,139	[ 24 ]
Therefore , the inclusion of these unaccounted N O reducers in future environmental studies may help bridge the gap between measured N O emissions and denitrification potential based solely on typical nosZ gene or nosZ transcript measurements.Our results for the Illinois soils are based on a composite sample comprised of equal DNA quantities extracted from multiple subsamples , intended to capture spatial heterogeneity at each field site ( at both centimeter depths and meter landscape scales ) .	PMC4049103	-1	[]
Since the soils were taken at a single time point , it is likely that some of the trends reported here for these soils ( e.g ., abundance of specific NosZ-encoding taxa , average coverage of metagenomes , etc. ) might differ temporally , given that agricultural soils typically receive seasonal management inputs .	PMC4049103	-1	[]
Nonetheless , the high abundance of atypical nosZ sequences found in these agricultural soil metagenomes and in soils from different locations and of diverse physicochemical characteristics ( e.g ., Fig. 5 ) emphasize their potential importance for nitrogen cycling .	PMC4049103	-1	[]
A bioinformatics methodology to detect target genes ..	PMC4049103	-1	[]
Our evaluation of in silico-generated data sets of known species and gene composition showed that both BLASTn and BLASTx algorithms represent reliable means of detecting reads encoding nosZ ( or other target ) genes , albeit with their own strengths and limitations .	PMC4049103	-1	[]
The selection of the most appropriate algorithm should consider the computational resources available .	PMC4049103	-1	[]
For example , BLASTx is more computationally demanding than BLASTn but can capture more-divergent sequences if more distantly related sequences / organisms are targeted .	PMC4049103	-1	[]
However , BLASTn similarity searches are less affected by frameshift-introducing sequencing errors , which might be significant in short-read data even after stringent quality read trimming .	PMC4049103	-1	[]
Frameshift correction tools such as FrameBot ( 31 ) , HMM-Frame ( 32 ) , and FragGeneScan ( 33 ) are available to correct these sequencing artifacts and also predict protein-coding regions in short reads.The low performance of the profile-based approach ( HMM ) versus that of BLASTx ( Table 1 ) is presumably attributable to the lower sensitivity of the former with ( i ) short sequences , ( ii ) sequencing errors creating frameshifts , and ( iii ) a reduced fraction of highly conserved amino acid residues specific to the protein of interest , as suggested previously ( 32 , 34 ) .	PMC4049103	1,397	[ 5 ]
Regardless of these limitations , HMM-based searches are preferable when targeting distantly related homologs and using full-length sequences ( e.g ., targeting complete gene sequences recovered in assembled contigs ) .	PMC4049103	-1	[]
Recommendations for the study of other genes ..	PMC4049103	-1	[]
The aforementioned approach based on in silico metagenomes , the BLAST algorithm , and ROC analysis can be modified for other functional genes of interest .	PMC4049103	-1	[]
Special attention should be given to conserved domains in the target gene or protein that are shared with other nontarget genes / proteins .	PMC4049103	-1	[]
As shown in Fig. 2 , no false-positive matches were observed for the B. japonicum strain USDA 110 NosZ reference sequence for bit score values above the calculated cutoff .	PMC4049103	-1	[]
The latter finding indicates that no high-identity domains or motifs are shared with other non NosZ sequences .	PMC4049103	-1	[]
Other genes may deviate from this pattern , and a case-by-case evaluation ( e.g ., Fig. S1 ) is recommended .	PMC4049103	-1	[]
Our approach , when modified to use sliding windows along the sequence of the reference gene , can also determine appropriate cutoffs for different regions of the sequence and identify regions that represent reliable targets for further analyses ( e.g ., low abundance of false-positive matches and PCR primer design ) .	PMC4049103	-1	[]
In silico data sets simulating different error rates , insert sizes , and coverage can easily be constructed to mimic different short-read sequencing technologies or methodologies .	PMC4049103	-1	[]
Nonetheless , simulating the diversity and variable abundances of individual taxa of real soil metagenomes remains challenging ( e.g ., our in silico data sets had substantially less diversity than the real metagenomes used in the study ) .	PMC4049103	-1	[]
The expansion of in silico library I by 13.6 million reads from 959 sequenced genomes not containing a nosZ gene ( i.e ., in silico library II ) did not increase the number of false-positive matches obtained for nosZ ( Table 1 ) , suggesting that a small number of , if any , false-positive matches should be expected for real metagenomes .	PMC4049103	-1	[]
In addition , having a comprehensive and well-curated set of protein or gene reference sequences is a key requirement for robust assessment of the best cutoffs and parameters to effectively retrieve reads encoding the gene ( s ) of interest.In conclusion , we developed a bioinformatics approach for the detection of target genes in short-read metagenomes .	PMC4049103	-1	[]
This methodology can be extended to the study of any other genes or proteins of interest .	PMC4049103	-1	[]
The high abundance of the previously unaccounted atypical nosZ genes in the soil samples suggests that nondenitrifiers and denitrifiers that harbor atypical nosZ genes may contribute more than previously thought to the reduction of N O to innocuous N gas .	PMC4049103	-1	[]
MATERIALS AND METHODS .	PMC4049103	-1	[]
Samples , DNA extraction , and sequencing ..	PMC4049103	-1	[]
In November 2011 , agricultural soil samples were collected from two sites with long histories of commercial corn and soybean production in the U.S. Midwest corn belt : ( i ) Havana , IL ( 93 % sand ; 7 % clay ; lat 40.296 , long − 89.944 ; elevation , 150 m ) , and ( ii ) Urbana , IL ( 21 % sand ; 69 % silt ; 10 % clay ; lat 40.075 , long − 88.242 ; elevation , 222 m ) .	PMC4049103	-1	[]
In order to provide a metagenome representative of the total soil profile and minimize the effect of sample heterogeneity , soil was collected as three replicate cores ( 2.5 cm by 30 cm ) taken at three locations 30 m apart within each field plot ( 9 cores , total , per field ) , with each core partitioned into four depths ( 0 to 5 cm , 5 to 10 cm , 10 to 20 cm , and 20 to 30 cm ) .	PMC4049103	-1	[]
Soil physicochemical characteristics were measured at each depth ( A & L Laboratories , Ft. Wayne , IN ) ( Table S4 ) .	PMC4049103	-1	[]
DNA was extracted from ~ 0.5 g of soil from each fraction according to a previously described phenol-chloroform extraction and purification protocol ( 35 ) , and approximately equal quantities of DNA from each fraction based on agarose gel quantification were pooled to create one composite sample for each soil type .	PMC4049103	-1	[]
The Illumina TruSeq and Nextera DNA library preparation protocols were used for the Havana sand and Urbana silt loam samples , respectively .	PMC4049103	-1	[]
Sequencing of composite DNA samples was performed using the Illumina HiSeq 2000 platform , resulting in 38.4 and 40.2 Gbp of 100-bp long paired-end reads for the Havana sand and Urbana silt loam samples , respectively .	PMC4049103	-1	[]

The anthropogenic fixation of dinitrogen ( N ) , by means of the Haber-Bosch process , has led to the overuse of synthetic nitrogen-based fertilizers in agriculture ( 1 , 6 ) .

PMC4049103

-1

[]

As a consequence of the increased nitrogen ( N ) content of soils , atmospheric N O concentrations have risen about 20 % relative to preindustrial-era levels ( 2 ) .

PMC4049103

-1

[]

N O emissions are largely the result of bacterial pathways controlling the nitrogen cycle .

PMC4049103

-1

[]

In particular , N O is generated primarily as a product of incomplete classic denitrification ( i.e ., NO reduction to N O via NO and NO ) and secondarily as a by-product of dissimilatory nitrate reduction to ammonia ( DNRA ) and oxidation of ammonium to nitrite ( nitrification ) ( 7 , 8) .

PMC4049103

-1

[]

Besides bacterial activities , abiotic processes and fungal denitrification are thought to be sources of N O ( 9 , 10 ) .

PMC4049103

-1

[]

Model predictions of N O consumption in terrestrial environments focus primarily on the N O-to-N reduction step , presently attributed to classical denitrifiers possessing nitrous oxide reductase ( NosZ ) ( 7 ) .

PMC4049103

-1

[]

Our previous work has revealed the existence of two phylogenetically distinct NosZ clades , one encompassing typical Z-type NosZ proteins , which are commonly found in the Alpha - , Beta - , and Gammaproteobacteria , and the other encompassing atypical NosZ proteins present in diverse organisms representing different phyla .

PMC4049103

-1

[]

Further analysis of sequenced genomes revealed that most of the typical nosZ genes are found in bacteria capable of complete denitrification ( i.e ., encoding all the enzymes for converting NO / NO to N ) , whereas atypical nosZ genes are found in bacteria with more-diverse N metabolism , including those performing DNRA and missing the NO-generating nitrite reductase genes nirK and nirS ( 11 , 12 ) .

PMC4049103

-1

[]

Notably , atypical NosZ proteins have been shown to function as nitrous oxide reductases in several bacteria , such as Wolinella succinogenes ( 13 , 14 ) , Geobacillus thermodenitrificans ( 15 ) , the soil isolate Anaeromyxobacter dehalogenans ( 11 ) , and several Bacillus species isolated from soils ( 16 , 17 ) .

PMC4049103

28,139

[ 37 ]

Examination of the potential of microbial communities to reduce N O to N has traditionally been performed by evaluating nosZ gene and / or transcript presence or abundance by PCR ( 18 , 19 ) .

PMC4049103

-1

[]

Primers targeting nosZ genes , however , were designed according to characterized typical nosZ gene sequences and therefore missed the bulk of divergent atypical genes ( 11 , 12 ) .

PMC4049103

-1

[]

Furthermore , measured N O emissions from soils were frequently lower than predictions based on ( typical ) NosZ transcript abundance and dynamics ( 20 , 21 ) .

PMC4049103

-1

[]

Therefore , it is likely that atypical NosZ abundance accounts , at least in part , for the discrepancy between predicted and observed N O fluxes.To circumvent the limitations and explore the total natural diversity of nosZ genes in the environment , we analyzed short-read metagenomic data sets from various soils and locations .

PMC4049103

-1

[]

Even though metagenomics can provide a relatively unbiased , PCR-independent view of the diversity and abundance of individual genes present in a sample , several technical challenges must first be addressed .

PMC4049103

-1

[]

For instance , in metagenomes of highly diverse microbial communities , such as those obtained from soils , the rates of false positives and false negatives when using similarity searches to detect individual genes in assembled contigs or unassembled short reads has not been rigorously evaluated , with the probable exception of error rates assessed for the purpose of taxonomic classification , i.e ., assigning a sequence to a taxon without necessarily evaluating its potential function and sequence diversity ( 22 , 23 ) .

PMC4049103

-1

[]

Cutoffs that might minimize the number of false-positive matches have not been determined for short-read metagenomes ; instead , arbitrary , predetermined cutoffs based on E values ( i.e ., the likelihood of finding a match by chance ) represent the common practice ( 24 ) .

PMC4049103

-1

[]

The objective of the present study was to analyze the diversity and abundance of both typical and atypical nosZ genes in soils with contrasting physicochemical properties .

PMC4049103

-1

[]

To this end , we first developed a strategy based on similarity searches to determine appropriate cutoffs for accurately detecting metagenomic nosZ fragments by analyzing in silico metagenomes of known sequence composition .

PMC4049103

-1

[]

Subsequently , we applied this strategy and derived cutoffs to detect nosZ reads in metagenomes from two agricultural soils in the U.S. Midwest that have been subjects of an ongoing multiyear study to assess nitrogen cycling processes , as well as in publicly available metagenomes from various soil ecosystems .

PMC4049103

-1

[]

Our metagenomic , PCR-independent approach provided a comprehensive and quantitative examination of the diversity and abundance of both typical and atypical nosZ genes in soils .

PMC4049103

-1

[]

RESULTS .

PMC4049103

-1

[]

Evaluating search algorithms and cutoffs for detecting nosZ genes in metagenomes ..

PMC4049103

-1

[]

To determine the best algorithm and parameters for detecting nosZ reads in 100-bp-long read metagenomes , a reference database of manually verified nosZ genes that preclustered at 95 % sequence identity was queried against two such in silico-generated data sets , libraries I ( representing the whole genome of 122 NosZ-encoding organisms ) and II ( representing the whole genome of 1,081 bacteria , including those in library I ) .

PMC4049103

-1

[]

From a receiver operating characteristic ( ROC ) curve analysis of true - and false-positive rates , the most appropriate bit score cutoff values were 107 and 52.2 for the BLASTn and BLASTx searches , respectively .

PMC4049103

-1

[]

These bit scores provided sensitivities ( the fraction of correctly classified positive BLAST matches ) of 100 % and 98.4 % for BLASTn and BLASTx , respectively , and a specificity ( the fraction of correctly classified negative BLAST matches ) of 99.9 % for both algorithms .

PMC4049103

-1

[]

A hidden Markov model ( HMM ) search resulted in a sensitivity of 46 % and a specificity of 99.9 % ( Table 1 ) .

PMC4049103

-1

[]

Additional HMM searches , using models that included more typical and atypical NosZ sequences ( built from reference sequences in Table S1 in the supplemental material ) improved the number of nosZ reads retrieved from both in silico libraries ( ~ 56 % ) .

PMC4049103

-1

[]

Irrespective of the HMM model employed , a lower fraction of nosZ reads was captured when the HMM was used than when BLAST searches were performed .

PMC4049103

-1

[]

Most of the reads missed by the HMM-based approach lacked highly conserved amino acid residues , and this accounted for the lower performance of HMM searches , consistent with expectations ( HMM models rely heavily on conserved residues ) .

PMC4049103

-1

[]

Therefore , remaining analyses were performed using the BLAST algorithms.To test the limitations in retrieving metagenomic nosZ reads , single typical and atypical representative NosZ protein or nucleotide sequences were independently queried against library II. Although BLASTn had a specificity similar to that of BLASTx , the latter algorithm was able to capture 735 % and 270 % more reads ( i.e ., reads annotated as nosZ with a bit score greater or equal to the calculated cutoff for true positives ) of the typical and the atypical references , respectively .

PMC4049103

-1

[]

Therefore , BLASTx was used in the remaining analyses .

PMC4049103

-1

[]

Using typical NosZ from Bradyrhizobium japonicum strain USDA 110 as a single reference , reads derived from 74 out of 127 different alleles encoding NosZ were captured and found to be enriched in nosZ sequences closely related to the reference sequence .

PMC4049103

-1

[]

In contrast , reads for only 32 out of 127 alleles encoding NosZ were captured when atypical NosZ from Anaeromyxobacter sp. strain Fw 109-5 was used as a reference in the analysis ( Fig. 1 , right panel ) .

PMC4049103

28,139

[ 19 ]

This atypical NosZ reference does not exhibit sequence identity to other target sequences in the 54 to 82 % amino acid identity range , and thus , the lack of moderately related sequences among the target sequences accounts for the results obtained ( Fig. 1 , left panel ) .

PMC4049103

-1

[]

More importantly , a linear relationship was observed between the fraction of total reads detected and the level of divergence between the full-length reference and target sequences ( Fig. 1 ) , where 50 % or more of the reads were detected when the two sequences shared more than 64 or 68 % sequence identity for typical or atypical NosZ reference sequences , respectively.Further examination of the reads recruited along the typical NosZ reference ( Fig. 2 ) showed that the true-positive matches ( i.e ., reads derived from nosZ genes with a bit score greater than 52.2 ) were evenly distributed along the NosZ reference sequence .

PMC4049103

-1

[]

The N terminus of the reference sequence ( 1 to 60 amino acid positions ) was rarely covered by either true - or false-positive matches , suggesting that this part of the gene should be avoided when assessing nosZ abundance in metagenomes .

PMC4049103

-1

[]

Abundance of nosZ genes in sandy and silty soils ..

PMC4049103

-1

[]

A characterization of the taxa , coverage , and assembly statistics of the two soil metagenomes is described in the supplemental material .

PMC4049103

-1

[]

Both soil metagenomes were queried against a 95 % - identical preclustered set of reference NosZ sequences .

PMC4049103

-1

[]

All matches having a bit score greater than the calculated cutoff determined based on the in silico library analysis were identified as nosZ reads and classified as typical or atypical depending on their best match .

PMC4049103

-1

[]

Atypical nosZ reads were clearly the most abundant , comprising 72.9 % and 89.6 % of the total nosZ reads found in the sand and silt loam soil metagenomes , respectively ( Fig. 3 ) .

PMC4049103

-1

[]

Further , 97 % of the nosZ reads found in both soil metagenomes ( 4,929 and 7,280 , total , in the sandy and silty loam soils , respectively ) were recruited by 72 of the 105 NosZ reference sequences , revealing that most of the diversity covered by the references was represented in both soils ( Table S2 ) .

PMC4049103

-1

[]

In addition , both soil samples showed similar estimated absolute abundances for nosZ reads : ~ 1.4 × 10 and 2.1 × 10 reads of all reads for the Havana sand and Urbana silt loam , respectively ( Fig. 3 ) .

PMC4049103

-1

[]

The ratio of nosZ reads to single-copy housekeeping gene reads indicated that approximately 16 % of the soil bacterial genomes harbored a nosZ gene ( Table S3 ) .

PMC4049103

-1

[]

Phylogenetic analysis of the atypical nosZ reads showed that closely related genes found in members of the Anaeromyxobacter , Gemmatimonas , Opitutus , and Hydrogenobacter genera were the most abundant in both soil samples ( Fig. 4 ) .

PMC4049103

28,139

[ 17 ]

Additionally , less abundant genera , such as Bradyrhizobium and Rhodopseudomonas , both known to harbor typical nosZ genes , were also present in both soils .

PMC4049103

-1

[]

Remarkably , atypical nosZ reads affiliated with Anaeromyxobacter represented 12.7 % and 15.2 % of the total nosZ reads found in both sand and silt loam soils , respectively .

PMC4049103

28,139

[ 7 ]

The most abundant typical nosZ reads were assigned to the Ralstonia ( 3.2 % ) , Thiobacillus ( 3.1 % ) , Bradyrhizobium ( 1.6 % ) , and Rhodopseudomonas ( 1.7 % ) genera ( Table S2 ) .

PMC4049103

-1

[]

nosZ diversity and abundance in other soil metagenomes ..

PMC4049103

-1

[]

In general , atypical nosZ reads were more abundant than other reads in the soil metagenomes evaluated ( Fig. 5 ) .

PMC4049103

-1

[]

The frozen deep-soil permafrost metagenomes ( core 1 sample in reference 25 ) showed a greater abundance for typical nosZ reads ( ~ 80 % of total NosZ ) ; however , atypical nosZ reads predominated in the upper or active layer ( ~ 74 % of total nosZ sequences ) .

PMC4049103

-1

[]

Interestingly , after induced thawing , the microbial communities at both depths showed a small increase in the relative abundance of atypical nosZ reads .

PMC4049103

-1

[]

Furthermore , except with the boreal forest biome , several biomes studied by Fierer and colleagues ( 26 ) , including tropical forest , polar and hot desert , arctic tundra , and temperate grassland , showed higher abundances of atypical than of typical nosZ reads .

PMC4049103

-1

[]

DISCUSSION .

PMC4049103

-1

[]

Importance of atypical NosZ ..

PMC4049103

-1

[]

The discovery of functional atypical NosZ proteins has opened the possibility that a much larger number of microorganisms with previously unaccounted N O-reducing potential contribute to lessening the N O emissions to the atmosphere than previously expected ( 12 ) .

PMC4049103

-1

[]

The abundance and diversity of atypical nosZ genes were likely missed in previous PCR-based surveys because typical nosZ sequences presented the basis for primer design ( 11 , 12 ) and the two nosZ types share only 60.9 % ± 8.2 % nucleotide identity , on average .

PMC4049103

-1

[]

In the present PCR-independent metagenome analysis , atypical NosZ sequences were more abundant ( > 73 % of total nosZ reads ) than their typical counterparts , not only in two agricultural soils differing in physicochemical properties representative of many regions in the Midwest United States but also in soils from distant geographic locations representing a variety of habitats .

PMC4049103

-1

[]

Our results were also consistent with the widespread presence of atypical nosZ genes , previously hypothesized based on the number of genomes found to encode atypical NosZ proteins among the available genome sequences ( 11 ) .

PMC4049103

-1

[]

Therefore , these findings reveal an unexpectedly high potential for N O reduction mediated by atypical NosZ in a variety of soil habitats.It is important to note that our study , being solely based on DNA sequences , evaluated N O reduction potential as opposed to the specific in situ activity of NosZ enzymes , typical or atypical .

PMC4049103

-1

[]

Since negative ( purifying ) selection efficiently removes unused genes from genomes in microbial populations , the high abundance of atypical nosZ sequences found in different soil samples underpins their functional and / or ecological potential ( e.g ., Fig. 5 ) .

PMC4049103

-1

[]

Given also that N O reduction is the only known biochemical function carried by NosZ ( 11 ) , our results collectively suggest that atypical NosZ proteins are as important , if not more important , than their typical counterparts in controlling N O fluxes in soils and likely other environments .

PMC4049103

-1

[]

Our study also provided the means ( e.g ., gene sequences for primer design and a bioinformatics strategy ) to facilitate future studies of the effect of environmental conditions on NosZ activity and dynamics toward a more predictive understanding of the nitrogen cycle.The most abundant nosZ genes in the agricultural soils studied here are affiliated with the Anaeromyxobacter genus ( Fig. 4 ) .

PMC4049103

28,139

[ 57 ]

Members of this genus are widely distributed in soils with different physical and chemical characteristics as well as soils from a variety of geographic locations ( 11 , 27 , 28 ) .

PMC4049103

-1

[]

The high abundance of nosZ genes affiliated with members of the nondenitrifying Anaeromyxobacter genus is consistent with recent PCR surveys that employed primers targeting atypical nosZ sequences ( 11 , 12 ) and A. dehalogenans 16S rRNA gene sequences ( 11 ) .

PMC4049103

28,139

[ 12 ]

Further , a high phylogenetic congruence between typical nosZ and 16S rRNA gene phylogenies was previously reported ( 29 , 30 ) .

PMC4049103

-1

[]

Therefore , abundant atypical nosZ metagenomic sequences ( Fig. 4 and Table S2 ) that have distant matches to homologs of known NosZ-encoding taxa may be harbored by novel taxa .

PMC4049103

-1

[]

The sequences reported here should facilitate the identification of new taxa , expanding our understanding of phylogenetic diversity in NosZ-encoding soil organisms .

PMC4049103

-1

[]

The majority of the abundant atypical nosZ reads that were assignable to known taxa were found in potentially nondenitrifying genomes of genera such as Anaeromyxobacter , Ignavibacterium , Opitutus , Dyadobacter , and Gemmatimonas , which were overlooked in previous PCR surveys targeting typical nosZ genes .

PMC4049103

28,139

[ 24 ]

Therefore , the inclusion of these unaccounted N O reducers in future environmental studies may help bridge the gap between measured N O emissions and denitrification potential based solely on typical nosZ gene or nosZ transcript measurements.Our results for the Illinois soils are based on a composite sample comprised of equal DNA quantities extracted from multiple subsamples , intended to capture spatial heterogeneity at each field site ( at both centimeter depths and meter landscape scales ) .

PMC4049103

-1

[]

Since the soils were taken at a single time point , it is likely that some of the trends reported here for these soils ( e.g ., abundance of specific NosZ-encoding taxa , average coverage of metagenomes , etc. ) might differ temporally , given that agricultural soils typically receive seasonal management inputs .

PMC4049103

-1

[]

Nonetheless , the high abundance of atypical nosZ sequences found in these agricultural soil metagenomes and in soils from different locations and of diverse physicochemical characteristics ( e.g ., Fig. 5 ) emphasize their potential importance for nitrogen cycling .

PMC4049103

-1

[]

A bioinformatics methodology to detect target genes ..

PMC4049103

-1

[]

Our evaluation of in silico-generated data sets of known species and gene composition showed that both BLASTn and BLASTx algorithms represent reliable means of detecting reads encoding nosZ ( or other target ) genes , albeit with their own strengths and limitations .

PMC4049103

-1

[]

The selection of the most appropriate algorithm should consider the computational resources available .

PMC4049103

-1

[]

For example , BLASTx is more computationally demanding than BLASTn but can capture more-divergent sequences if more distantly related sequences / organisms are targeted .

PMC4049103

-1

[]

However , BLASTn similarity searches are less affected by frameshift-introducing sequencing errors , which might be significant in short-read data even after stringent quality read trimming .

PMC4049103

-1

[]

Frameshift correction tools such as FrameBot ( 31 ) , HMM-Frame ( 32 ) , and FragGeneScan ( 33 ) are available to correct these sequencing artifacts and also predict protein-coding regions in short reads.The low performance of the profile-based approach ( HMM ) versus that of BLASTx ( Table 1 ) is presumably attributable to the lower sensitivity of the former with ( i ) short sequences , ( ii ) sequencing errors creating frameshifts , and ( iii ) a reduced fraction of highly conserved amino acid residues specific to the protein of interest , as suggested previously ( 32 , 34 ) .

PMC4049103

1,397

[ 5 ]

Regardless of these limitations , HMM-based searches are preferable when targeting distantly related homologs and using full-length sequences ( e.g ., targeting complete gene sequences recovered in assembled contigs ) .

PMC4049103

-1

[]

Recommendations for the study of other genes ..

PMC4049103

-1

[]

The aforementioned approach based on in silico metagenomes , the BLAST algorithm , and ROC analysis can be modified for other functional genes of interest .

PMC4049103

-1

[]

Special attention should be given to conserved domains in the target gene or protein that are shared with other nontarget genes / proteins .

PMC4049103

-1

[]

As shown in Fig. 2 , no false-positive matches were observed for the B. japonicum strain USDA 110 NosZ reference sequence for bit score values above the calculated cutoff .

PMC4049103

-1

[]

The latter finding indicates that no high-identity domains or motifs are shared with other non NosZ sequences .

PMC4049103

-1

[]

Other genes may deviate from this pattern , and a case-by-case evaluation ( e.g ., Fig. S1 ) is recommended .

PMC4049103

-1

[]

Our approach , when modified to use sliding windows along the sequence of the reference gene , can also determine appropriate cutoffs for different regions of the sequence and identify regions that represent reliable targets for further analyses ( e.g ., low abundance of false-positive matches and PCR primer design ) .

PMC4049103

-1

[]

In silico data sets simulating different error rates , insert sizes , and coverage can easily be constructed to mimic different short-read sequencing technologies or methodologies .

PMC4049103

-1

[]

Nonetheless , simulating the diversity and variable abundances of individual taxa of real soil metagenomes remains challenging ( e.g ., our in silico data sets had substantially less diversity than the real metagenomes used in the study ) .

PMC4049103

-1

[]

The expansion of in silico library I by 13.6 million reads from 959 sequenced genomes not containing a nosZ gene ( i.e ., in silico library II ) did not increase the number of false-positive matches obtained for nosZ ( Table 1 ) , suggesting that a small number of , if any , false-positive matches should be expected for real metagenomes .

PMC4049103

-1

[]

In addition , having a comprehensive and well-curated set of protein or gene reference sequences is a key requirement for robust assessment of the best cutoffs and parameters to effectively retrieve reads encoding the gene ( s ) of interest.In conclusion , we developed a bioinformatics approach for the detection of target genes in short-read metagenomes .

PMC4049103

-1

[]

This methodology can be extended to the study of any other genes or proteins of interest .

PMC4049103

-1

[]

The high abundance of the previously unaccounted atypical nosZ genes in the soil samples suggests that nondenitrifiers and denitrifiers that harbor atypical nosZ genes may contribute more than previously thought to the reduction of N O to innocuous N gas .

PMC4049103

-1

[]

MATERIALS AND METHODS .

PMC4049103

-1

[]

Samples , DNA extraction , and sequencing ..

PMC4049103

-1

[]

In November 2011 , agricultural soil samples were collected from two sites with long histories of commercial corn and soybean production in the U.S. Midwest corn belt : ( i ) Havana , IL ( 93 % sand ; 7 % clay ; lat 40.296 , long − 89.944 ; elevation , 150 m ) , and ( ii ) Urbana , IL ( 21 % sand ; 69 % silt ; 10 % clay ; lat 40.075 , long − 88.242 ; elevation , 222 m ) .

PMC4049103

-1

[]

In order to provide a metagenome representative of the total soil profile and minimize the effect of sample heterogeneity , soil was collected as three replicate cores ( 2.5 cm by 30 cm ) taken at three locations 30 m apart within each field plot ( 9 cores , total , per field ) , with each core partitioned into four depths ( 0 to 5 cm , 5 to 10 cm , 10 to 20 cm , and 20 to 30 cm ) .

PMC4049103

-1

[]

Soil physicochemical characteristics were measured at each depth ( A & L Laboratories , Ft. Wayne , IN ) ( Table S4 ) .

PMC4049103

-1

[]

DNA was extracted from ~ 0.5 g of soil from each fraction according to a previously described phenol-chloroform extraction and purification protocol ( 35 ) , and approximately equal quantities of DNA from each fraction based on agarose gel quantification were pooled to create one composite sample for each soil type .

PMC4049103

-1

[]

The Illumina TruSeq and Nextera DNA library preparation protocols were used for the Havana sand and Urbana silt loam samples , respectively .

PMC4049103

-1

[]

Sequencing of composite DNA samples was performed using the Illumina HiSeq 2000 platform , resulting in 38.4 and 40.2 Gbp of 100-bp long paired-end reads for the Havana sand and Urbana silt loam samples , respectively .

PMC4049103

-1

[]