Citation Context

...was provided by the Sanger Center. Gene family construction The procedure used for defining gene families is summarized as follows (Figure 6). First, we performed in each clade all-against-all BLASTP =-=[65]-=- comparisons, inter- and intra- genomically. We set the e-value cutoff to 1027, required the hits to be at least 100 aa in length and the length of the BLASTP hit to span at least 70% of the length of...

Citation Context

...genetic analyses The reference phylogenetic tree of each clade was reconstructed from the concatenated alignments of genes comprising their core genome. Protein alignments were generated using MUSCLE =-=[73]-=- and then back-translated to DNA. Tree-Puzzle [74] was used to generate the matrix of distances by maximum likelihood with the HKY+C model and exact parameter estimates. The trees were then computed f...

Citation Context

...alogs and xenologs. Non-synonymous (dN) and synonymous (dS) substitution rates in paralogs (blue; dashed linear fit) and xenologs (red; solid linear fit) in all clades computed using Codeml from PAML =-=[76]-=- (model = 1, fix_omega = 0). doi:10.1371/journal.pgen.1001284.g005 HGT Drives Protein Family Expansions PLoS Genetics | www.plosgenetics.org 7 January 2011 | Volume 7 | Issue 1 | e1001284 Importantly,...

Citation Context

...uzzle [74] was used to generate the matrix of distances by maximum likelihood with the HKY+C model and exact parameter estimates. The trees were then computed from these distance matrices using BIONJ =-=[75]-=-. We performed 100 bootstrap experiments on the concatenated sequences to assess the robustness of the topology. Ancestral state reconstruction We used the reference phylogeny and maximum likelihood (...

Citation Context

...pecies for which there is sufficient genome-wide experimental data. We first tested if the two classes of protein families have different gene expression levels using the codon adaptation index (CAI; =-=[44]-=-). CAI is higher for paralogs than for xenologs (0.68 and 0.48, Mann-Whitney-Wilcoxon test, p,0.0001) and smallest for genes in families without expansions (0.42, Figure 4). Similar results were found...

Citation Context

...ntroduced at high rates by HGT. These families were excluded from the analysis. Given the pattern of presence/absence of genes and a reference tree, we inferred gene expansion events with BayesTraits =-=[33]-=-. Families without protein expansions within the lineages were excluded from further analysis. We also removed IS and phage sequences, but not the corresponding cargo regions, because these elements c...

Citation Context

...cs | www.plosgenetics.org 8 January 2011 | Volume 7 | Issue 1 | e1001284 in preparation) and removed from the analysis. We verified that for E. coli our databank contained all IS present in IS Finder =-=[67]-=-. We used Phage_finder to predict prophages [68]. IS are frequently pseudogenized. To remove small fragments of IS elements we searched for regions of homology with known IS in the genome using Repeat...

Citation Context

...f each clade was reconstructed from the concatenated alignments of genes comprising their core genome. Protein alignments were generated using MUSCLE [73] and then back-translated to DNA. Tree-Puzzle =-=[74]-=- was used to generate the matrix of distances by maximum likelihood with the HKY+C model and exact parameter estimates. The trees were then computed from these distance matrices using BIONJ [75]. We p...

Citation Context

...ing itself). Since we are interested in multi-gene families (i.e. at least one genome has two or more genes), we clustered pairwise BLASTP hits into multiple relationships using mclblastline from MCL =-=[66]-=-, with parameters: --blast-m9 --blast-score= e --blast-sort =a --mcl-I = 2.0 --mcl-scheme=7. After inspection of the results using inflation parameter values from 1.0 to 10.0 in increments of 0.2, we ...

Citation Context

...f homology with known IS in the genome using Repeatoire [69]. Core genome construction Since genomes in each clade have diverged recently, the orthologs should be highly similar. We enlisted OrthoMCL =-=[70]-=- to build the groups of orthologs inside of the gene families via allagainst-all BLASTP comparisons in a clade of interest. OrthoMCL defines putative orthologous relationships between a pair of genome...

Citation Context

...variation both in Neisseria and in Helicobacter [40,41]. Interestingly, both clades naturally transform conspecific DNA and are known to have extremely high rates of intra-species horizontal transfer =-=[42,43]-=-. The high frequency of HGT within the species is liable to produce a large fraction of very similar xenologs scattered in the genome. This factor is presumably corrected by the use of the co-localiza...

Citation Context

...ions of protein families and increased genome size. Expansions of protein families arise most frequently by HGT Duplication processes in prokaryotes produce tandem, strictly identical copies of genes =-=[8,38,39]-=-. At the evolutionary distances considered in this work, this implies that paralogs arising in the lineages are co-localized because of the low rearrangement rates and are highly similar in sequence b...

Citation Context

...in interactions To assess the differences between the processes of protein family expansions in terms of cellular networks we used a set of 74,776 protein-protein interactions (PPI) in E. coli MG1655 =-=[50]-=-. We computed the fraction of interactions shared by pairs of proteins in families with expansions (DPPI). As expected, paralogs share more interactions than xenologs (DPPI 1.0 and 0.30, p,0.01, MannW...

Citation Context

...ve expansions of protein families in prokaryotes. As such, expansions are a particular case of the processes leading to the horizontal transfer of genetic information that is known to shape metabolic =-=[59]-=-, genetic [53] and interaction networks [60,61] in prokaryotes. Importantly, recent works have shown that extensive horizontal transfer also exists among eukaryotes [62–64]. These results may thus als...

Citation Context

... might have been affected by HGT [26–28], it has been estimated that HGT contributes at best to 25% of all expansions of protein families [26,29–32] (22% overall but 60% for large protein families in =-=[23]-=-). These contradictory conclusions require an explanation. Prior analyses on the relative abundance of paralogs over xenologs were performed several years ago using the available distant genomes. Howe...

Citation Context

...new genetic information, liberating genomic evolutionary processes from tinkering exclusively with pre-existing genes [22– 24]. HGT occurs at very high rates leading to very large species pan-genomes =-=[25]-=-. Yet, while up to 96% of the genes in a given prokaryote genome might have been affected by HGT [26–28], it has been estimated that HGT contributes at best to 25% of all expansions of protein familie...

Citation Context

...ences, but not the corresponding cargo regions, because these elements constitute a large fraction of repeats in genomes [15], are known to be horizontally transferred and are in general quickly lost =-=[34,35]-=-. Our main analysis includes the remaining 3190 families. These families have few members (0–3) in each genome (Figure 1) showing that expansions are rare at these narrow evolutionary scales. This is ...

Citation Context

... expansion [8,15–19]. Furthermore, models aiming at explaining the patterns of biological networks are in general based on the preconception that protein families expand by gene duplication processes =-=[20,21]-=-. Horizontal gene transfer (HGT) results in the acquisition of radically new genetic information, liberating genomic evolutionary processes from tinkering exclusively with pre-existing genes [22– 24]....

Citation Context

...variation both in Neisseria and in Helicobacter [40,41]. Interestingly, both clades naturally transform conspecific DNA and are known to have extremely high rates of intra-species horizontal transfer =-=[42,43]-=-. The high frequency of HGT within the species is liable to produce a large fraction of very similar xenologs scattered in the genome. This factor is presumably corrected by the use of the co-localiza...

Citation Context

...e 7 | Issue 1 | e1001284 in preparation) and removed from the analysis. We verified that for E. coli our databank contained all IS present in IS Finder [67]. We used Phage_finder to predict prophages =-=[68]-=-. IS are frequently pseudogenized. To remove small fragments of IS elements we searched for regions of homology with known IS in the genome using Repeatoire [69]. Core genome construction Since genome...

Citation Context

...th dN and dS were higher in xenologs than in paralogs (p,0.01, Mann-Whitney-Wilcoxon tests). We also tested if higher evolutionary rates in xenologs were caused by amelioration to the host GC content =-=[47]-=- (see Materials and Methods). As expected, xenologs have higher G+C deviations to the core genome than paralogs (0.05 vs. 0.04, p,0.01, Wilcoxon test). This difference is small, as found previously in...

Citation Context

...ences, but not the corresponding cargo regions, because these elements constitute a large fraction of repeats in genomes [15], are known to be horizontally transferred and are in general quickly lost =-=[34,35]-=-. Our main analysis includes the remaining 3190 families. These families have few members (0–3) in each genome (Figure 1) showing that expansions are rare at these narrow evolutionary scales. This is ...

Citation Context

...cs | www.plosgenetics.org 8 January 2011 | Volume 7 | Issue 1 | e1001284 in preparation) and removed from the analysis. We verified that for E. coli our databank contained all IS present in IS Finder =-=[67]-=-. We used Phage_finder to predict prophages [68]. IS are frequently pseudogenized. To remove small fragments of IS elements we searched for regions of homology with known IS in the genome using Repeat...

Citation Context

...enerates xenologs that potentially have highly different regulatory dependencies. This picture is in agreement with the frequently observed co-transfer of transcription factors and their target genes =-=[52]-=-, and subsequent slow integration of these sub-networks in the larger genetic network [53]. Discussion Contrary to previous studies, we found a high rate of HGT in all eight clades of prokaryotes anal...

Citation Context

...ions of protein families and increased genome size. Expansions of protein families arise most frequently by HGT Duplication processes in prokaryotes produce tandem, strictly identical copies of genes =-=[8,38,39]-=-. At the evolutionary distances considered in this work, this implies that paralogs arising in the lineages are co-localized because of the low rearrangement rates and are highly similar in sequence b...

Citation Context

...ender the two copies adaptive when selection for higher gene dosage ceases (reviewed in [9–12]). There is ample evidence that intra-chromosomal gene duplication (IGD) has an adaptive role in bacteria =-=[13,14]-=-, e.g. in antigenic variation, antibiotic resistance or in genome expansion [8,15–19]. Furthermore, models aiming at explaining the patterns of biological networks are in general based on the preconce...

Citation Context

...ture is in agreement with the frequently observed co-transfer of transcription factors and their target genes [52], and subsequent slow integration of these sub-networks in the larger genetic network =-=[53]-=-. Discussion Contrary to previous studies, we found a high rate of HGT in all eight clades of prokaryotes analyzed in this study. At least 88% of all Figure 4. Gene expression differs according to gen...

Citation Context

...onal Biology, University of Maryland, College Park, Maryland, United States of America Introduction Prokaryotes have highly variable gene repertoires, varying from just over 100 genes to nearly 10000 =-=[1,2]-=-. Such variations in genome size are typically associated with expansions and contractions of protein families. Expansions of protein families are associated with the acquisition of novel functions, n...

Citation Context

...ontribution of IGD (resp. accounting for 78% and 45%). It is well-established that such repeated elements often engage in gene conversion for antigenic variation both in Neisseria and in Helicobacter =-=[40,41]-=-. Interestingly, both clades naturally transform conspecific DNA and are known to have extremely high rates of intra-species horizontal transfer [42,43]. The high frequency of HGT within the species i...

Citation Context

...ncrease in size of the largest bacterial genomes [54]. The important contribution of xenology to family expansions is also consistent with data showing E. coli’s promiscuity to most transferred genes =-=[55]-=- and data highlighting the impact of transferred genes on the evolution of Salmonella [56]. It has been proposed that due to deletion biases in prokaryotes, gene duplications are not afforded with suf...

Citation Context

...etter/best hits. To limit false positives this list was further refined by combining information on the distribution of similarity of these putative orthologs with gene order conservation data (as in =-=[71]-=-). At these distances, gene order conservation is high [58], and positional information can significantly decrease classification errors [72]. Each ortholog pair was then tested for gene order conserv...

Citation Context

...onal Biology, University of Maryland, College Park, Maryland, United States of America Introduction Prokaryotes have highly variable gene repertoires, varying from just over 100 genes to nearly 10000 =-=[1,2]-=-. Such variations in genome size are typically associated with expansions and contractions of protein families. Expansions of protein families are associated with the acquisition of novel functions, n...

Citation Context

...in each genome (Figure 1) showing that expansions are rare at these narrow evolutionary scales. This is in agreement with high rates of change but low rates of fixation of changes in gene repertoires =-=[36,37]-=-. The most frequent event found in our data was the gain of a paralog or xenolog by the largest genomes in the clade (Table S2), such as S. agalactiae NEM316 (20% of all expansions in Streptococcus). ...

Citation Context

...ontribution of IGD (resp. accounting for 78% and 45%). It is well-established that such repeated elements often engage in gene conversion for antigenic variation both in Neisseria and in Helicobacter =-=[40,41]-=-. Interestingly, both clades naturally transform conspecific DNA and are known to have extremely high rates of intra-species horizontal transfer [42,43]. The high frequency of HGT within the species i...

Citation Context

...als and Methods). As expected, xenologs have higher G+C deviations to the core genome than paralogs (0.05 vs. 0.04, p,0.01, Wilcoxon test). This difference is small, as found previously in Salmonella =-=[48]-=-, possibly because we removed IS and phages. For similar GC deviations we still found that xenologs have higher synonymous and non-synonymous substitution rates (Mann-Whitney-Wilcoxon tests on the rat...

Citation Context

...ogy to family expansions is also consistent with data showing E. coli’s promiscuity to most transferred genes [55] and data highlighting the impact of transferred genes on the evolution of Salmonella =-=[56]-=-. It has been proposed that due to deletion biases in prokaryotes, gene duplications are not afforded with sufficient opportunity for neo- or sub-functionalization, reducing its role to transient gene...

Citation Context

...anges in the results: xenolog expansions vastly outnumber paralog expansions. Many models assume that duplication is the major mechanism underlying the evolution of protein-protein interactions (PPI) =-=[4,20]-=- or regulatory networks [21]. Duplications constrain genomic evolutionary processes to tinker with pre-existing information producing identical genes that are functionally and genetically redundant. I...

Citation Context

...alse classifications in the present work because: (i) dropping the co-localization criterion still results in a preponderance of xenologs, (ii) rearrangements are rare at these evolutionary distances =-=[58]-=-. Second, xenologs can integrate close to their homolog in the genome. This can occur by chance alone, as a rare event, or because transferred genes tend to insert in integration hotspots. Additionall...

Citation Context

...otes. As such, expansions are a particular case of the processes leading to the horizontal transfer of genetic information that is known to shape metabolic [59], genetic [53] and interaction networks =-=[60,61]-=- in prokaryotes. Importantly, recent works have shown that extensive horizontal transfer also exists among eukaryotes [62–64]. These results may thus also be relevant to understand the evolution of bi...

Citation Context

...utative orthologs with gene order conservation data (as in [71]). At these distances, gene order conservation is high [58], and positional information can significantly decrease classification errors =-=[72]-=-. Each ortholog pair was then tested for gene order conservation via the ordered list of putative orthologs between the two genomes (i.e. lateral transfer is discounted). Genes not satisfying the cons...

Citation Context

...least one of them. Homologs with no known interactions were recorded as zero. We used E. coli data [51] on 2020 regulators including regulation by (1) transcription factors, (2) RNA-binding proteins, =-=(3)-=- sigma factors, (4) protein–protein interactions and 5) DNA supercoiling. As expected, paralogs share more regulators than xenologs (0.1 to 0.0, p = 0.06, MannWhitney-Wilcoxon test). The difference is...

Citation Context

...lineages were excluded from further analysis. We also removed IS and phage sequences, but not the corresponding cargo regions, because these elements constitute a large fraction of repeats in genomes =-=[15]-=-, are known to be horizontally transferred and are in general quickly lost [34,35]. Our main analysis includes the remaining 3190 families. These families have few members (0–3) in each genome (Figure...

Citation Context

...than for xenologs (0.68 and 0.48, Mann-Whitney-Wilcoxon test, p,0.0001) and smallest for genes in families without expansions (0.42, Figure 4). Similar results were found using proteomic data (emPAI) =-=[45]-=- (p,0.005). We conclude that paralogs are more expressed than xenologs. While this is at odds with previous work on E. coli, where genes with the lowest CAI were found to be overrepresented among rece...

Citation Context

...largest genomes in each clade are also the ones containing more new xenologs. This is concordant with the proposed role of horizontal transfer in the increase in size of the largest bacterial genomes =-=[54]-=-. The important contribution of xenology to family expansions is also consistent with data showing E. coli’s promiscuity to most transferred genes [55] and data highlighting the impact of transferred ...

Citation Context

... expansion [8,15–19]. Furthermore, models aiming at explaining the patterns of biological networks are in general based on the preconception that protein families expand by gene duplication processes =-=[20,21]-=-. Horizontal gene transfer (HGT) results in the acquisition of radically new genetic information, liberating genomic evolutionary processes from tinkering exclusively with pre-existing genes [22– 24]....

Citation Context

...ions of protein families and increased genome size. Expansions of protein families arise most frequently by HGT Duplication processes in prokaryotes produce tandem, strictly identical copies of genes =-=[8,38,39]-=-. At the evolutionary distances considered in this work, this implies that paralogs arising in the lineages are co-localized because of the low rearrangement rates and are highly similar in sequence b...

Citation Context

...e aligned them to the corresponding ortholog, when there was one. We analyzed separately non-synonymous (dN) and synonymous (dS) substitution rates, after excluding highly divergent pairs (if dS.1.5) =-=[46]-=-. As expected we found the highest dN and dS values in xenologs (dSHGT=0.43, dSIGD=0.26, p,0.001; dNHGT= 0.07, dNIGD=0.03, p,0.001, Mann-Whitney-Wilcoxon tests). The lower values of dN and dS reflect ...

Citation Context

...otes. As such, expansions are a particular case of the processes leading to the horizontal transfer of genetic information that is known to shape metabolic [59], genetic [53] and interaction networks =-=[60,61]-=- in prokaryotes. Importantly, recent works have shown that extensive horizontal transfer also exists among eukaryotes [62–64]. These results may thus also be relevant to understand the evolution of bi...

Citation Context

...ender the two copies adaptive when selection for higher gene dosage ceases (reviewed in [9–12]). There is ample evidence that intra-chromosomal gene duplication (IGD) has an adaptive role in bacteria =-=[13,14]-=-, e.g. in antigenic variation, antibiotic resistance or in genome expansion [8,15–19]. Furthermore, models aiming at explaining the patterns of biological networks are in general based on the preconce...

Citation Context

...onclude that paralogs are more expressed than xenologs. While this is at odds with previous work on E. coli, where genes with the lowest CAI were found to be overrepresented among recent duplications =-=[30]-=-, this discrepancy probably results from our explicit removal of IS and phages that have very low CAI because they are A+T rich and are lowly expressed under exponential growth. When such elements are...

Citation Context

...ith our observation that xenologs are more persistent than paralogs. Instead, the results are consistent with the observed negative correlation between expression levels and both dN and dS in E. coli =-=[49]-=-. Paralogs being more expressed than xenologs, they evolve slower in synonymous positions, due to Figure 3. Abundance of IS and prophages and increased inference of IGD events when included in analysi...

Citation Context

...e number of regulators interacting with both homologs over the sum of regulators interacting with at least one of them. Homologs with no known interactions were recorded as zero. We used E. coli data =-=[51]-=- on 2020 regulators including regulation by (1) transcription factors, (2) RNA-binding proteins, (3) sigma factors, (4) protein–protein interactions and 5) DNA supercoiling. As expected, paralogs shar...

Citation Context

...used Phage_finder to predict prophages [68]. IS are frequently pseudogenized. To remove small fragments of IS elements we searched for regions of homology with known IS in the genome using Repeatoire =-=[69]-=-. Core genome construction Since genomes in each clade have diverged recently, the orthologs should be highly similar. We enlisted OrthoMCL [70] to build the groups of orthologs inside of the gene fam...

Citation Context

...in each genome (Figure 1) showing that expansions are rare at these narrow evolutionary scales. This is in agreement with high rates of change but low rates of fixation of changes in gene repertoires =-=[36,37]-=-. The most frequent event found in our data was the gain of a paralog or xenolog by the largest genomes in the clade (Table S2), such as S. agalactiae NEM316 (20% of all expansions in Streptococcus). ...

Citation Context

...roposed that due to deletion biases in prokaryotes, gene duplications are not afforded with sufficient opportunity for neo- or sub-functionalization, reducing its role to transient gene amplification =-=[57]-=-. Xenologs are related, yet different, proteins that can provide a potentially advantageous distinct function immediately upon transfer. Since they tend to be located apart from the native homolog the...