Skip to content
TheChatArchiveThe Chat Archive
gpt-4o-mini-2024-07-18· OpenAIAi2 WildChat-4.8M (AI2)

translate the content after ------------------------------- into plain Chinese. Species…

  1. User

    translate the content after ------------------------------- into plain Chinese. Species name, scientific terms should not be translated.

    Genomic analysis of the multi-host pathogen Erysipelothrix rhusiopathiae reveals extensive recombination as well as the existence of three generalist clades with wide geographic distribution

    Taya Forde, Roman Biek, Ruth Zadoks, Matthew L. Workentine, Jeroen De Buck, Susan Kutz, Tanja Opriessnig, Hannah Trewby, Frank van der Meer & Karin Orsel 
    

    BMC Genomics volume 17, Article number: 461 (2016) Cite this article

    4484 Accesses
    
    36 Citations
    
    2 Altmetric
    
    Metrics details
    

    Abstract Background

    Knowledge about how bacterial populations are structured is an important prerequisite for studying their ecology and evolutionary history and facilitates inquiry into host specificity, pathogenicity, geographic dispersal and molecular epidemiology. Erysipelothrix rhusiopathiae is an opportunistic pathogen that is currently reemerging in both the swine and poultry industries globally. This bacterium sporadically causes mortalities in captive marine mammals, and has recently been implicated in large-scale wildlife die-offs. However, despite its economic relevance and broad geographic and host distribution, including zoonotic potential, the global diversity, recombination rates, and population structure of this bacterium remain poorly characterized. In this study, we conducted a broad-scale genomic comparison of E. rhusiopathiae based on a diverse collection of isolates in order to address these knowledge gaps. Results

    Eighty-three E. rhusiopathiae isolates from a range of host species and geographic origins, isolated between 1958 and 2014, were sequenced and assembled using both reference-based mapping and de novo assembly. We found that a high proportion of the core genome (58 %) had undergone recombination. Therefore, we used three independent methods robust to the presence of recombination to define the population structure of this species: a phylogenetic tree based on a set of conserved protein sequences, in silico chromosome painting, and network analysis. All three methods were broadly concordant and supported the existence of three distinct clades within the species E. rhusiopathiae. Although we found some evidence of host and geographical clustering, each clade included isolates from diverse host species and from multiple continents. Conclusions

    Using whole genome sequence data, we confirm recent suggestions that E. rhusiopathiae is a weakly clonal species that has been shaped extensively by homologous recombination. Despite frequent recombination, we can reliably identify three distinct clades that do not clearly segregate by host species or geographic origin. Our results provide an essential baseline for future molecular epidemiological, ecological and evolutionary studies of E. rhusiopathiae and facilitate comparisons to other recombinogenic, multi-host bacteria. Background

    Uncovering population structure and its determinants is essential for understanding bacterial ecology and evolution [1]. Key questions concerning host specificity and generalism [2], global patterns of gene flow [3], and the genetic basis for clinical disease manifestations [1] can be addressed by examining the relationships among strains within a species. Additionally, molecular epidemiological studies rely on having an understanding of the population structure as a framework within which to interpret the genomic diversity of the target organism [4], and the identification of host- or geography-associated lineages can be helpful in source attribution [5]. Whole genome sequencing is providing new opportunities to address questions related to population structure within a phylogenetic framework; however, this undertaking is complicated by the need to consider the potential influence of recombination [4, 6].

    Erysipelothrix rhusiopathiae, a Gram positive, facultative intracellular bacterium, is an important opportunistic pathogen for both humans and animals. Zoonotic infections with E. rhusiopathiae typically manifest as erythematous skin lesions known as erysipeloid, and tend to be occupationally associated (e.g. slaughterhouse workers, butchers, fishermen, etc.) [7]. E. rhusiopathiae has been documented in a wide range of wild and domestic species, including birds, mammals, reptiles, fish and arthropods [8]. Best known as the causative agent of swine erysipelas, E. rhusiopathiae can cause significant economic losses in swine production systems due to sporadic cases of acute septicemia, subacute cutaneous lesions, or chronic arthritis, which may be punctuated by larger outbreaks [9]; it is also among the most common causes of carcass condemnation for swine in the United States [10]. In recent years, the incidence of E. rhusiopathiae infection in swine has increased significantly in the mid-western United States, Japan and China [10–12]. Erysipelas is also reemerging in European poultry productions, likely in association with changes in housing systems [13, 14]. In captive marine mammals, E. rhusiopathiae is known to cause serious and often life-threatening infections [15, 16], while recent die-offs involving hundreds of muskoxen in the Canadian Arctic Archipelago have sparked interest in the potential conservation importance of this bacterium [17]. The broad ecological and geographic distribution of E. rhusiopathiae has been attributed to its ability to infect multiple host species which may act as healthy carriers, in combination with its long environmental persistence [18]. However, despite its ubiquity, importance for multiple host species including humans, and a highly variable clinical presentation, little is known about the genetic diversity, population structure, and host specificity of E. rhusiopathiae.

    Recent whole genome sequencing projects have facilitated the taxonomic classification of the genus Erysipelothrix [11, 19]. A member of the phylum Firmicutes, the class Erysipelotrichia has the single order Erysipelotrichales and family Erysipelotrichaceae, the latter comprising 10 genera [20]. Within the genus Erysipelothrix, other recognized species are E. tonsillarum, E. inopinata, E. sp. strain 1, and E. sp. strain 2 [21, 22], as well as the recently identified E. larvae sp. nov. [23]. E. rhusiopathiae and E. tonsillarum have long been recognized as distinct species based on differences in pathogenicity, phenotypic characteristics, and serotype groups [21, 24], supported by DNA-DNA hybridization studies [25]. E. inopinata appears to have diverged prior to the split between E. rhusiopathiae and E. tonsillarum based on 16S rRNA gene sequence phylogeny [22]. The relationship of E. sp. strain 1 and E. sp. strain 2 to other Erysipelothrix species has not been explored.

    The intraspecific classification of E. rhusiopathiae strains is less clearly defined, with a variety of different tools used for categorizing isolates. Serotyping, which involves testing for agglutination with specific antisera recognizing different peptidoglycan antigens of the cell wall [26], is one of the approaches that has been most frequently implemented. Although it has diagnostic value, as the different Erysipelothrix species have distinct sets of serotypes [27], serotyping is an inappropriate tool for inferring evolutionary relatedness among isolates due to the frequent horizontal exchange of capsule-specific genes in many bacterial species [28]. More recently, differences in immunogenic proteins known as surface protective antigens (Spa) have also been used to distinguish between strains of E. rhusiopathiae [13, 29]. The genetic relationship among E. rhusiopathiae isolates has also been examined using various comparative genotyping methods such as pulsed-field gel electrophoresis (PFGE) [30]. However, sequence-based typing methods, which allow for easier inter-laboratory comparison and investigation of functional differences, have only very recently been applied to E. rhusiopathiae: a comparison was made among the three E. rhusiopathiae whole genome sequences (two complete and one draft) available on GenBank at the time of writing [11], and the population structure of this bacterium was examined using multi-locus sequence typing (MLST), focusing primarily on European poultry isolates [13]. Little is known about the importance of recombination in E. rhusiopathiae, although the latter study found the species to be weakly clonal [13], suggesting recombination may contribute significantly to genetic variability in this species. Until now there have been no large-scale genomic comparison studies to fully investigate the genetic diversity of E. rhusiopathiae.

    The objectives of this study were to describe the global genomic diversity of E. rhusiopathiae and to examine its population structure while accounting for the presence of recombination, in order to provide an essential baseline for future studies into the epidemiology and ecology of this multi-host pathogen. A secondary objective was to examine the phylogenetic relationship among different Erysipelothrix species based on whole genome sequence data. Results Sequencing and pan-genome statistics

    Eighty-three newly sequenced E. rhusiopathiae isolates were included in this analysis, with representation from North America (Canada and US), South America (Argentina), Europe (Belgium, Hungary and the UK), Asia (Japan), South Africa, and Australia, with the majority originating from Canada (n = 37), US (n = 20) and Belgium (n = 17) (see Additional file 1: Table S1 for a table of the isolates included in this study). Dates of isolation ranged from 1958 to 2014. Host species of origin encompassed swine (n = 18), poultry (n = 14), captive marine mammals (dolphins and beluga whales; n = 8), fishes (n = 5), wild birds (primarily water fowl; n = 7), wild ungulates (muskoxen, moose and caribou; n = 28), and a wolf. Background information regarding the clinical manifestations associated with the isolates was only available for 12 isolates (Additional file 1: Table S1), but cases of acute and subacute septicemia, as well as E. rhusiopathiae isolated from skin lesions were included. Serotype was known for 16 isolates previously described in the literature (Additional file 1: Table S1); the newly serotyped isolate VI11-2_lu was found to belong to serotype 5.

    A total of 1137 core genes were present among all the E. rhusiopathiae isolates included in this analysis, representing 67 % of the coding sequences present in the Fujisawa reference genome (Fig. 1a and Additional file 2: Figure S1). A total of 512 singleton genes were identified (i.e. present in only one isolate), with an average of 6.6 new unique genes discovered per additional genome sequenced (Fig. 1b). Among the core genes were the ComEC (membrane pore) and dprA (recombination mediator) genes, as well as putative Type II/IV secretion system proteins (Additional file 2: Figure S1), the presence of which is highly suggestive of natural competency [31]. Based on the pairwise comparisons made in the program LS-BSR, no major differences in overall gene content were detected among host species or geographic origin, or among clades that were identified during analyses described below. Fig. 1 figure 1

    Erysipelothrix rhusiopathiae pan-genome. Plots were generated in LS-BSR and visualized using PanGP. a Convergence of the core genome with concurrent accumulation of coding sequences in the pan-genome in relation to the number of genomes analyzed. b Number of unique coding sequences for each additional genome analyzed. Publicly available E. rhusiopathiae isolates and de novo assembled isolates whose average depth of coverage was greater than 15X were included in this analysis Full size image A high proportion of the E. rhusiopathiae core genome has experienced recombination

    The core genome alignment generated using Parsnp was 1,049,431 base pairs, representing 58 % of the reference genome. Overall there was a high degree of sequence similarity, with 99.3 % pairwise identity. The BratNextGen analysis, which included all isolates, inferred that 58 % of the overall core genome alignment had experienced recombination (Additional file 2: Figure S2). There were major differences among the clades in terms of the proportion of the alignment with inferred recombination (Table 1, Additional file 2: Figures S2 and S3). Both recombination detection methods estimated that a greater proportion of the genome of Clade 2 isolates has experienced recombination. BratNextGen did not detect any recombinogenic segments in Clade 1, while this was also the clade with the least recombination across the core genome as detected by Gubbins. Although a large proportion of the alignment was influenced by recombination, average recombination to mutation (r/m) rates were moderate, at 0.96, 2.18 and 0.55 in Clades 1, 2 and 3 respectively. Table 1 Recombination detected in the core genome of Erysipelothrix rhusiopathiae Full size table Phage and plasmid-related sequences were detected in several isolates

    Putative prophage sequences were detected in 47 of the 86 isolates (55 %), representing all host groups and geographic locations (Additional file 1: Table S1). Up to three such sequences were detected per isolate, with a mean length of 22 KB (range: 12.5–47 KB). Twenty-eight of the isolates had phage sequences with a high level of homology (≥96.5 % pairwise identity) with the annotated bacteriophage from the Fujisawa genome (36.5 KB; Additional file 2: Figure S1) across at least an 8 KB segment (designated as P1 in Additional file 1: Table S1). Two of these 28 isolates were inferred to be intact prophages by PHAST. Isolates with P1 phage sequences were present in all three clades, suggesting that the presence of these similar phage sequences is the result of either multiple introductions or ancestral acquisition with subsequent loss along various branches of the phylogeny. A second group of phage sequences (n = 15; designated as P2 in Additional file 1: Table S1) shared 95.9 % pairwise identity across about 12 KB. Four of these were classified as intact by PHAST. There is a strong possibility that some of these P2 sequences have been inherited through vertical descent (e.g. in the five Belgian swine isolates), while the other sequences were identified in unrelated isolates, indicating horizontal transfer. Eight isolates had phage sequences that clustered into a third group (P3) that had at least 12 KB of homologous sequence with 94 % pairwise identity. None of these sequences were designated as intact by PHAST. These sequences all belonged to swine isolates from two distinct clusters, although not all isolates in each cluster had phage sequences detected. Among the other phage sequences detected, most shared a smaller portion of sequence homology with one of the three groups of sequences (designated as ‘partial’ in Additional file 1: Table S1). Only one unique phage sequence was identified among our collection using PHAST: the 22 KB incomplete phage sequence in the poultry isolate G11. Based on BLAST comparisons, this sequence shares a 13 KB segment with an integrative conjugative element of Streptococcus pyogenes with 95.4 % pairwise identity, while the rest of the sequence shared 96 % pairwise similarity to plasmids carrying coding sequences for Type IV secretory pathway components.

    The BLAST search against the publicly available E. rhusiopathiae plasmid sequence [GenBank: NC_002148] found that one isolate in our collection (‘Ery Afrika 1’, Clade 2) had a similar sequence, with 95.2 % pairwise identity and 100 % coverage. During the BLAST search of contigs from each isolate that did not align to the reference genome, one poultry isolate had a hit to a conjugal transfer protein, while five swine isolates had hits to plasmid sequences. Three of these, from swine from Belgium, were hits to the same plasmid sequence; searches for similar sequences among the other three Belgian swine isolates were negative. Population structure of E. rhusiopathiae

    The phylogenetic tree generated in PhyloPhlAn using > 400 conserved bacterial proteins, rooted to other genera of the family Erysipelotrichaceae, places E. tonsillarum at a position basal to E. rhusiopathiae and E. sp. strain 2 in the phylogeny (Fig. 2). The differentiation between clades is more easily visible when zooming in to the within-species level (Fig. 3). Consistent with this phylogeny, the analysis using ChromoPainter and fineSTRUCTURE (Fig. 4 and Additional file 2: Figure S4), and network analysis implemented in SplitsTree (Fig. 5) supported the existence of three distinct clades within the species E. rhusiopathiae. Seven isolates belonged to Clade 1 as supported by all three methods. These isolates originated from two captive marine mammals (beluga and dolphin) from the US, one dolphin from an Australian aquarium, one red wolf from the US, one fish from Japan, one caribou from Canada, and from a sheep dip from Argentina. Fourteen isolates belonged to Clade 2 as supported by all three methods. These consisted of isolates from marine mammals (n = 5) from the US, Belgium and South Africa, from fish (n = 3) from Japan and the US, from poultry (n = 3) and swine (n = 1) from Belgium, from one Canadian caribou and the ATCC19414 isolate from GenBank (host species and geographic origin unknown; see Additional file 1: Table S1). One additional poultry isolate from Belgium (red arrow Figs. 3, 4 and 5) was grouped with the Clade 2 isolates by ChromoPainter, but was classified as an “intermediate” isolate between Clades 2 and 3 using phylogenetic analysis. At least five isolates clustered in this intermediate group based on all three methods: the two whole genome sequences from swine available on GenBank (SY1027 from China and Fujisawa from Japan), and two caribou and one moose isolate from Canada. By the chromosome painting method, two additional isolates clustered within this intermediate clade, both from wild birds from the US; these isolates fell within Clade 3 using the phylogenetic approach (blue and black arrows in Figs. 3, 4 and 5). The three isolates whose clade designation differed between the phylogenetic and chromosome painting methods all fell outside of the main clusters using network analysis (Fig. 5, shown by arrows). The other 57 isolates constituted the dominant Clade 3, composed of all of the Canadian swine (n = 8) and poultry (n = 6) isolates tested, most of the Belgian swine isolates (6/7) and half of the Belgian poultry (4/8) isolates, three additional swine isolates (one from the US and two of unknown origin), isolates from five different North American wild bird species, isolates from North American caribou (n = 11), moose (n = 4) and muskoxen (n = 8), and one fish isolate from Hungary. Fig. 2 figure 2

    Relationship among the species of the genus Erysipelothrix. This phylogenetic tree is based on > 400 conserved bacterial protein sequences, generated using PhyloPhlAn. Members of other genera of the family Erysipelotrichaceae were used as the outgroup for rooting the tree. The scale bar represents the expected number of nucleotide substitutions per sequence position Full size image Fig. 3 figure 3

    Population structure of Erysipelothrix rhusiopathiae based on phylogenetic inference. This tree is based on > 400 conserved bacterial protein sequences, generated using PhyloPhlAn. Other Erysipelothrix spp. were used as the outgroup for rooting the tree (clade collapsed). Concentric rings illustrate host species of origin (colored squares) and geographic origin (grey-scale). Arrows indicate isolates whose correct clade association is not resolved between the phylogenetic and chromosome painting approaches. Presence of the surface protective antigen type B (SpaB) gene was exclusive to Clade 1, while the SpaA gene was found among the two other clades and intermediate isolates Full size image Fig. 4 figure 4

    Population structure of Erysipelothrix rhusiopathiae based on in silico chromosome painting. Tree and heat map illustrate the relatedness among E. rhusiopathiae isolates based on chromosome painting using ChromoPainter and model-based clustering using Bayesian Markov chain Monte Carlo (MCMC) analysis in fineSTRUCTURE. The color scale represents the number of ‘chunks’ shared between populations of donors (x-axis) and recipients (y-axis). Population subgroup assignment is shown in Figure S4 in Additional file 2. Arrows indicate isolates whose correct clade association is not resolved between the phylogenic and chromosome painting approaches Full size image Fig. 5 figure 5

    Population structure of Erysipelothrix rhusiopathiae based on network analysis. A phylogenetic network of E. rhusiopathiae was estimated using Neighbor-Net analysis as implemented in SplitsTree. Arrows indicate isolates whose correct clade association is not resolved between the phylogenic and chromosome painting approaches Full size image

    Less concordance was observed among the different methods with respect to the relationships of isolates within the dominant Clade 3. This is illustrated in Fig. 6, which shows the maximum likelihood (ML) phylogeny generated using non-recombinogenic core single nucleotide polymorphisms (SNPs) inferred by Gubbins, with clusters supported by either PhyloPhlAn or fineSTRUCTURE superimposed. In general there was stronger support for more recent nodes based on higher bootstrap values in the ML tree, as well as more frequent support by the other two methods. No temporal signal was detected in the root-to-tip analysis of Clade 3 isolates using Path-O-Gen (correlation < 0.001). Fig. 6 figure 6

    Relationship among Clade 3 isolates. This maximum likelihood tree (mid-point rooted) was generated using PhyML based on the curated set of single nucleotide polymorphisms (SNPs) found to be outside recombinant segments as determined using Gubbins. Bootstrap values with >70 % support are shown, in addition to support for clusters of isolates by PhyloPhlAn and fineSTRUCTURE Full size image

    A consistent relationship was found between Spa-type and clade: all isolates in Clade 1 had a single copy of the SpaB gene, while all other E. rhusiopathiae isolates had a single copy of SpaA (Fig. 3). Conversely, among those isolates whose serotype had been defined, significant homoplasy was observed (Additional file 2: Figure S5). Serotype 5 isolates were found in all three clades, serotype 2 isolates were found in Clades 2 and 3, and serotype 1a isolates fell into both intermediate and Clade 3. Limited evidence for host or geographic specificity

    All three clades included isolates from various host species and from multiple continents. However, all marine mammal isolates (n = 8) fell into Clades 1 or 2 and none within the dominant Clade 3, while no swine or poultry isolates were found to belong to Clade 1. Within Clade 3, clustering by host species, geographic location, and submitting laboratory were all found to be significant using the program BaTS (Additional file 3: Table S2). Although not fully supported by the other two methods, the phylogenetic tree based on non-recombinant SNPs (Fig. 6) grouped the majority of livestock isolates within Clade 3 (21/27 swine and poultry isolates) into a single sub-group with 96 % bootstrap support, whereas the majority of isolates of wildlife origin fell into a separate sub-group within this clade. Discussion Homologous recombination has occurred extensively throughout the core genome of E. rhusiopathiae

    This is the first large-scale genomic study of the multi-host pathogen E. rhusiopathiae. Our results indicate that homologous recombination plays an important role in generating diversity within this species, confirming recent findings by an MLST study [13]. Based on our data, several mechanisms may contribute to recombination in this species: the uptake of genetic material from the external environment, phages, and plasmids. Bacteria capable of importing foreign DNA across the cell envelope (transformation) are said to be ‘naturally competent’ and possess specific machinery for this task [31]. The presence of the ComEC (membrane pore) and dprA (recombination mediator) genes within the core genome, as well as coding sequences for putative Type II/IV secretion system proteins, is highly suggestive that E. rhusiopathiae has this capacity [31, 32]. Transduction and conjugation may also be important mechanisms for horizontal gene transfer in E. rhusiopathiae, since phage-related sequences were abundant among the isolates in our study and some plasmid-related sequences were also identified, despite the fact that the sensitivity for detecting such sequences from draft assemblies may be low [33]. In previous studies, up to a third of isolates tested were found to harbor plasmids [34, 35], and several mobile genetic elements in the available reference genomes bore signatures of having plasmid or prophage origin [11]. To our knowledge, this is the first study to have examined the prevalence and diversity of phage-related sequences in a large collection of E. rhusiopathiae isolates. Although limited data were available regarding the clinical presentation associated with many of the isolates, no clear association was found between the presence of phage sequences and pathogenicity. The presence of particular phage sequences did not appear to be a phylogenetically informative marker for epidemiological inference, since related phage sequences were found throughout the different clades, and in some cases, phage sequences were not conserved within groups of highly similar isolates. Similarly, the inconsistent presence of plasmid sequences among closely related isolates suggests that conjugation may occur frequently.

    Although a high proportion of the core genome has experienced recombination events, the r/m rates we estimated for each Clade (between 0.55 and 2.18) were only moderate in comparison to other bacterial species [36]. This, along with the high pairwise identity (99.3 %) found within the core genome, suggests that within the E. rhusiopathiae core genome, alleles are highly conserved, although its organization is not. This may explain the lack of concordance found between MLST and PFGE [13], since this latter technique is sensitive to structural rearrangements. Within each clade, a larger proportion of the core genome was estimated to have experienced recombination based on the Gubbins output in comparison to BratNextGen. Gubbins is based on the method initially cited in Croucher et al. [37], and has since been employed in several other studies of bacterial pathogens [38, 39]. Inference of recombinant segments is based on the detection of areas of higher SNP density in comparison to the background threshold level. Since areas of elevated mutations are not necessarily specific for recombination and could instead represent regions of the genome with naturally higher mutation rates or that are under less purifying selection, this could be a potential source of false positives using this method [40]. BratNextGen, on the other hand, uses a Bayesian change-point clustering model to detect evolutionarily distinct lineages; these clusters are taken as the putative ancestral origins when estimating recombination probabilities in each isolate. This program has been frequently cited in intraspecific bacterial population studies based on whole genome data [41, 42]. Despite the differences in the extent of recombination found between the two programs used, both suggest that recombination has occurred most frequently in Clade 2, where the r/m rate was two to four times higher compared to the other clades and a greater proportion of the genome was found to have been implicated. Clade-associated differences in recombination rates have been previously observed in other bacterial species such as Listeria monocytogenes [43]. The population structure of E. rhusiopathiae was consistent across multiple inference approaches

    Given the uncertainties associated with whole genome phylogenetic reconstruction in the presence of recombination, we took three conservative, independent approaches for inferring the population structure of E. rhusiopathiae. The use of a set of conserved protein sequences has been shown to be robust to horizontal gene transfer [44]. Chromosome painting does not rely on phylogenetic inference but rather reconstructs the chromosome haplotype of “recipient” individuals as a composition of recombination-derived segments from the other “donor” individuals [45]. Finally, network analysis provides a means by which to visualize alternative phylogenetic histories in organisms that are not strictly clonal [46]. Using these three approaches, good concordance was found for assigning isolates into clades. The fact that these methods were based on input from both de novo genome assembly (conserved protein tree and network analysis) and a reference-based mapping approach (chromosome painting), provides support for the robustness of these results.

    Clade 1 could be distinguished from the other two clades and intermediate isolates by the presence of a SpaB gene, as opposed to SpaA found in the other isolates. These surface protective antigens are potent immunogens with potential relevance for vaccine development [29]. SpaA is among the more well-characterized genes that have been proposed to be associated with E. rhusiopathiae pathogenicity [19], with a role in endothelial adherence [47] and resistance to phagocytosis [48]. The SpaB gene shares approximately 60 % amino acid similarity with SpaA and is antigenically distinct [49]. It was previously found that E. rhusiopathiae strains of aquatic origin were able to express more than one Spa-type [29], although an underlying genetic basis for this was not observed among the isolates in our collection. Not surprisingly, serotype was not associated with the phylogenetic relatedness among isolates [50, 51]. Since homoplasy of serotype is likely commonplace (Additional file 2: Figure S5), it should not be used to infer evolutionary relatedness among isolates, and its use as an epidemiological tool (e.g. for deciding whether multiple isolates belong to the same outbreak) would require additional information on the expected frequency of serotype switching.

    Despite removing recombinogenic sites prior to constructing the phylogeny of the dominant Clade 3, basal relationships among these isolates were difficult to elucidate, with low bootstrap values obtained for most of the deeper nodes (Fig. 6). It is possible that not all recombination events were detected and that this still had a confounding influence; a PHI-test revealed a strong signal of recombination even in this curated SNP set. The lack of a detectable clock-like signal could be partially due to residual recombinogenic segments in this dataset [4], although given the lack of correlation between the date of isolation and root-to-tip divergence, it seems probable that additional factors are contributing. Hypothetically, highly variable generation times (e.g. little bacterial replication during environmental phases or carriage in comparison to during active infection) could be a plausible explanation for the lack of temporal structure. Failing to calibrate the molecular clock, we were unable to estimate a global or clade-specific substitution rate for E. rhusiopathiae based on whole genome sequence data. E. rhusiopathiae shows limited host association

    Bacteria show remarkable variability in the extent to which they specialize to colonize specific host species, the determinants of which are thought to be the result of complex host-pathogen interactions [52]. Many bacterial species have fairly distinct host-adapted strains [2, 53], while other bacterial species are characterized by a mix of host-adapted and generalist sub-types [42, 54]. The determinants of host specificity can be seen along a broad continuum from changes in a single amino acid residue, to changes in gene content (presence/absence) or the acquisition of genomic islands [55]. In our study, no consistent differences in gene content were found between isolates from different host species, geographic locations or clades.

    It is known from experimental infection studies that multiple host species are susceptible to the same strains of E. rhusiopathiae; isolates from one host species that are subsequently inoculated into another species often result in infection and clinical disease in the recipient species [56], although susceptibility varies [57]. We found that all three clades comprised isolates from a variety of taxonomically very different host species. It is, therefore, probable that all E. rhusiopathiae lineages are infectious for a wide range of species, although possibly with varying levels of infectiousness and pathogenicity. Our data also contained strong indications of cross-species transmission given that some isolates from wildlife were nested within lineages dominated by domestic hosts and vice versa (Figs. 3 and 6), suggesting host or ecological predilection exists, but not strict host specificity. Similar findings were reported using MLST, wherein the dominant clonal complex encompassed isolates from multiple host species, including poultry, pigs, sheep and humans [13].

    Although the different clades do not appear to be strictly limited to specific hosts, the observation that isolates from captive marine mammals fell exclusively into Clades 1 and 2 is significant (p < 0.0001), especially given that these are from aquariums distributed globally. Whether this finding is associated with differences in susceptibility of marine mammals to these strains, or that these strains are differentially distributed due to adaptions to marine environments warrants further investigation. Both Clade 1 and Clade 2 E. rhusiopathiae isolates were associated with fatal infections in marine mammals (Additional file 1: Table S1). However, no Clade 1 isolates, and therefore no strains carrying the SpaB gene, were found in any of the swine or poultry isolates in our collection. Since it has been previously shown that strains of marine origin carrying the SpaB gene can experimentally cause at least mild to moderate lesions in swine [56], the lack of Clade 1 strains in swine and poultry is more likely due to lack of exposure than a lack of pathogenic potential.

    Within the dominant Clade 3, although clustering by host species was detected, equally strong support for clustering based on geographic origin and source laboratory was found (Additional file 3: Table S2), suggesting that these associations cannot be separated from sampling bias related to the opportunistic nature of our isolate collection. This represents a common challenge in phylogenetic studies of pathogens [58]. Further examination into host and geographic clustering within this clade using a broader set of sample sources, including more detailed metadata in order to rule out clustering due to epidemiologically linked isolates, would be of value. Despite this limitation, there appears to be a non-random segregation of Clade 3 isolates between those of livestock and wildlife origin (Fig. 6); swine and poultry isolates from three independent sources of both North American and European origin tended to cluster independently of those from northern ungulates. Among the livestock isolates, there was evidence of regular cross-continental exchange.

    Although efforts were made to include a global collection of isolates from a representative variety of host species, most isolates originated from North America and Europe, and some important hosts including humans, reptiles and arthropods were not part of our collection. The characterization of additional isolates, particularly from under-represented continents and host species, will help build our understanding of the global population structure of E. rhusiopathiae. Whole genome sequencing supports the current division among Erysipelothrix species

    We have shown E. sp. strain 2 to be phylogenetically distinct from both E. rhusiopathiae and E. tonsillarum based on the conserved protein phylogeny. Its distinction from E. rhusiopathiae is further supported by the fact that its level of sequence similarity was insufficient for mapping of reads to the full genome sequences of either Fujisawa or SY1027, or for inclusion in the core genome alignment generated through Parsnp, which requires ≥ 97 % average nucleotide identity. Based on the conserved protein phylogeny, E. rhusiopathiae and E. sp. strain 2 appear to have evolved from the less pathogenic species E. tonsillarum. This relationship was not observed in previous phylogenetic studies that combined the output from multiple phylogenetic approaches [20]. Further investigation into the genetic differences separating E. rhusiopathiae and E. sp. strain 2 from E. tonsillarum could potentially provide insights into how these species have acquired a higher level of pathogenicity. The recent detection of Erysipelothrix spp. in the subsurface biosphere may represent an interesting opportunity for examining the evolution of this genus in a broader context [59]. Conclusions

    Using different approaches, we were able to confidently determine the population structure of the multi-host pathogen E. rhusiopathiae, despite the important role that recombination has had in its evolutionary history. Evidence was found to suggest that novel DNA may be acquired by this bacterium through transformation, transduction and conjugation, lending plasticity to the genome. The species comprises three major clades that are found across multiple continents and host species representing both livestock and wildlife, with some indication that clades or subclades may differ in their host predilection and recombination rate. Epidemiological inference was hampered by the opportunistic nature of the isolate collection available for genome sequencing and future studies would benefit from targeted sample collection. Nonetheless, our results provide an essential framework for supporting future in-depth epidemiological and evolutionary studies involving this species and comparative studies with other recombinogenic, multi-host bacteria. Methods Bacterial isolates

    In order to examine the intraspecific genomic diversity of E. rhusiopathiae, isolates were opportunistically collected so as to encompass a broad range of geographic locations, host species, clinical manifestations, and years of initial isolation. E. rhusiopathiae isolates, as well as isolates from other Erysipelothrix spp. (E. tonsillarum and E. sp. strain 2) were kindly provided by various collaborators (Additional file 1: Table S1). Upon receipt, isolates were sub-cultured onto Columbia Agar (CA) with 5 % sheep blood (BD-Canada, Mississauga, ON, Canada) for morphological characterization; if colony morphology was characteristic of E. rhusiopathiae (clear to pale blue in color, circular, small diameter, often with alpha hemolysis [18]), a single colony was re-streaked to obtain a clonal population for DNA extraction. In addition to these archived isolates, further isolates were obtained through sample testing associated with wildlife health surveillance projects. Various tissue samples from wild ungulates (moose, caribou and muskoxen) were selectively cultured for E. rhusiopathiae: 2 g of tissue were mechanically homogenized in 20 ml of brain heart infusion (BHI) broth with 5 % serum using a Stomacher 80 Biomaster (Seward, Port Saint Lucie, FL, USA), incubated overnight at 37 °C with 5 % CO2, followed by 48 h incubation in selective medium containing kanamycin (40 μg/ml), neomycin (50 μg/ml) and vancomycin (25 μg/ml) [60, 61] and sub-culture to agar plates of the same selective medium for 48-72 h at 37 °C with 5 % CO2. Colonies were further sub-cultured on CA plates as described for the other isolates. DNA extraction and PCR confirmation

    DNA was extracted from clonal populations by suspending 1 loopful of colonies in 200 μl of phosphate buffered saline (PBS), and then using the DNeasy Blood and Tissue Kit (Qiagen, Mississauga, ON, Canada) following manufacturer’s instructions, with the exception that DNA was eluted in a smaller volume (50 μl) of the provided elution buffer to yield a higher DNA concentration. Extracted DNA was confirmed to be from E. rhusiopathiae by qPCR using previously described primers and a species-specific probe targeting the 3’ non-coding region of the rRNA gene cluster [27]. The 20 μl PCR reaction consisted of 10 μl of TaqMan Fast Advanced Master Mix (Applied Biosystems, Carlsbad, CA), 10 pmol of each primer, 1 pmol of probe, and 2 μl of template DNA. Reactions were performed using a CFX96 thermocycler (Bio-Rad, Mississauga, Ontario, Canada) with the following cycling conditions: 50 °C for 2 min, 95 °C for 20 s, then 40 cycles of 95 °C for 3 s and 57 °C for 30 s. E. tonsillarum isolates were also confirmed by probe-based qPCR using a different probe with this same set of primers and reaction conditions [27]. The identity of E. sp. strain 2 was confirmed in silico by identifying the specific primer sequences for the E. sp. strain 2 23S rRNA gene in the draft assembly [62]. Library preparation and sequencing

    Library preparation was performed using the Nextera XT v2 kit (Illumina, San Diego, CA) following manufacturer’s instructions, including a PhiX control spiked in at 1 %. Multiplex sequencing was performed on an Illumina MiSeq machine at the University of Calgary, resulting in 250 base pair paired-end reads. Read mapping and variant detection

    The program ConDeTri was used to trim raw reads, extract those of high quality, and remove duplicates [63] (see Additional file 2: Figure S6 for a schematic diagram of the analysis pipeline, and Additional file 4: File S1 for specific commands and parameters used for each program). Unique trimmed reads from each isolate were mapped against the complete E. rhusiopathiae Fujisawa genome [Genbank:NC_015601] using BWA-MEM with default settings [64]. Sequencing and mapping parameters (number of reads, number of mapped reads, mean coverage, mean mapping quality) were extracted from .bam files using Qualimap [65] (Additional file 1: Table S1). All sequenced isolates had a minimum mean coverage of 9X, with an average mean coverage of 33X. Median mapping quality was consistently high, with a mean value of 57.

    Variants were detected using the mpileup command in SAMtools [66]. A list of high quality SNP sites across all E. rhusiopathiae genomes was generated using custom python scripts (written by HT; available upon request) that filtered on base quality, mapping quality, read depth, and heterozygosity. A site was included in the variant list if it had consensus base quality ≥ 40, mapping quality ≥ 40, at least three reads mapping to that site on each strand, and the majority base present in >95 % of reads at that site for at least one isolate. Sites were excluded where more than one alternate allele was found since these positions were believed to be less phylogenetically reliable due to possible sequence saturation or mapping error. Alleles at each SNP site were called across all isolates if the consensus base quality was ≥ 30, mapping quality was ≥ 30, a minimum of 2 reads were present on each strand, and >95 % of reads supported the same allele in the isolate. Additional whole genome sequences of E. rhusiopathiae available on GenBank at the time of writing, strain SY1027 [Genbank:NC_021354] and the draft sequence of the strain ATCC19414 [Genbank:NZ_ACLK00000000] were included by simulating reads using the wgsim script in SAMtools and mapping these back to the reference genome in the same way as the other isolates for allele calling. Mobile and repetitive elements were identified in the annotated reference genome, as well as using the repeat-match command in MUMmer [67] and were removed from the variant list. A total of 32,148 high quality unique variant sites compared to the Fujisawa reference genome were identified across all 85 E. rhusiopathiae isolates (including the two GenBank sequences) once SNPs within repetitive regions and mobile elements were removed. A list of sites where alleles were correctly called in 100 % of isolates (n = 6078) was then generated. A total of 142 E. rhusiopathiae isolates were sequenced [Genbank BioProject: PRJNA288715], however in cases where multiple isolates were considered to be epidemiologically linked based on metadata and sequence similarity (i.e. <15 high quality SNP differences across the whole genome), only the isolate with the highest coverage was selected for inclusion in this analysis. De novo assembly

    Unique, trimmed reads were assembled using SPAdes [68] applying a k-value of 55; this assembler was selected as the best compromise between contig number/L50 and fewer misassemblies based on metrics generated for candidate assemblers using QUAST [69]. SPAdes was run using the built-in BayesHammer for further read error correction [70] and Mismatch Corrector, a post processing tool that uses BWA. The PAGIT suite of programs was used to order and extend contigs and correct errors in the consensus sequence [71]. Output assembly metrics pre-PAGIT (number of contigs, N50) and post-PAGIT (number of scaffolds and GC %) were determined using QUAST (Additional file 1: Table S1). The average number of scaffolds per isolate after the PAGIT improvement pipeline was 13 (range: 4–89). The average GC% ranged from 35.9–36.5 %, with the mean across all isolates being 36.3 %. Annotation

    RAST (Rapid Annotation using Subsystem Technology) was used to annotate and predict coding sequences in each isolate [72]. Putative prophage sequences were detected using PHAST [73]. To identify potential plasmid-associated sequences, contigs that were not aligned to the reference genome in PAGIT were searched against the NCBI database using Megablast. Additionally, a BLAST search of the E. rhusiopathiae plasmid sequence available on GenBank was performed against all assemblies. Hits to known plasmids or to genes encoding conjugal transfer proteins were considered suggestive of the presence of plasmids. To determine the Spa-type of each isolate, a custom BLAST database including representative sequences from the three recognized Spa types (A, B and C) [49] was searched against each E. rhusiopathiae isolate using Geneious version 7.1.8 [74]. The serotype of isolate VI11-2_lu was determined using methods previously described [56]. The genetic basis for differences in serotype has not been previously described in the literature, therefore it was not possible to determine the serotype of unknown isolates based on the genomic sequence data. Comparative genomics

    Pan genome statistics were generated using LS-BSR [75] and visualized using PanGP [76]. To improve sensitivity in detecting core proteins that might be missed due to inferior assembly quality in isolates with a lower depth of coverage, isolates with coverage less than 15X (n = 8) were excluded from this analysis. The LS-BSR script compare_BSR.py was used to look for differences in gene content between groups of isolates. Specifically, pairwise comparisons in gene content (genes present in all members of one group that are absent in all members of a second group) were made among the different Clades (as determined by PhyloPhlAn and fineSTRUCTURE), between isolates originating from specific host species (swine, poultry, marine mammal and muskox isolates compared to all other isolates), as well as geographic origin (North American vs. European). Unfortunately, given the lack of metadata on clinical manifestation associated with our collection of isolates, it was not possible to assess genetic differences that might be associated with pathogenicity in this study. The core genetic content was visualized in GView using isolates with coverage ≥ 15X [77]. Tests for recombination

    The program Parsnp [78] was used to generate a core genome alignment (i.e. conserved orthologous regions present in all included genomes). Input for this alignment was the 83 PAGIT-improved E. rhusiopathiae de novo assemblies, as well as the three E. rhusiopathiae sequences available from GenBank. Putative prophage sequences detected with PHAST were masked using the BEDTools maskfasta command [79] prior to creating the alignment. This core nucleotide alignment was used as the input for both BratNextGen [80] and Gubbins [40]. Since Gubbins is best suited for detecting recombination in closely-related groups of isolates [40], separate analyses were run for each clade of isolates as later determined using PhyloPhlAn and fineSTRUCTURE. BratNextGen was run setting the hyperparameter α to 4 and using 20 iterations of the recombination estimation algorithm. The statistical significance was estimated using 100 permutations of the algorithm, setting significance at α = 0.05. Gubbins was run within the publicly available virtual machine using default settings. Recombination to mutation rates (r/m; number of SNPs in recombinogenic segments: number of SNPs inferred to be the result of spontaneous mutation) were calculated from Gubbins output by taking the sum of mutations inside and outside recombinant segments along all internal branches leading to each terminal node of the output tree, starting from the node of the inferred common ancestor of that clade. Examination of the population structure

    Since we found that recombination has played an important role in generating the diversity observed within the species E. rhusiopathiae, three techniques that are robust to recombination were used to examine the population structure. First, PhyloPhlAn was used to estimate a phylogenetic tree based on a set of > 400 conserved protein sequences common to most bacteria [44]. The amino acid fasta files generated using RAST were used as the input for the program. The following isolates were included: 83 de novo and three E. rhusiopathiae strains available on GenBank, three de novo assembled E. spp. isolates (Additional file 1: Table S1), as well as the draft sequence of E. tonsillarum [NZ_AREO00000000]. As an outgroup, draft whole genome sequences of a selection of members from closely related genera [20] were retrieved from GenBank. These were of Holdemania filiformis DSM 12042 [NZ_ACCF00000000], [Clostridium] spiroforme DSM 1552 [NZ_ABIK00000000], Erysipelatoclostridium ramosum DSM 1402 [NZ_ABFX00000000], Solobacterium moorei F0204 [NZ_AECQ00000000], and Bulleidia extructa W1219 [NZ_ADFR00000000]. A circularized version of this tree was generated in the Interactive Tree of Life (iTOL) v3.0 to illustrate the relationship among the E. rhusiopathiae isolates and their host species and geographic origin [81].

    ChromoPainter and fineSTRUCTURE were implemented as an alternative method for examining population structure without relying on phylogenetic inference [45]. This analysis assesses each genome (recipient) as the sum of segments that could have been received from any of the other genomes in the analysis (donors). These inferred shared segments are summarized in a ‘co-ancestry matrix’ through Principal Component Analysis which is then used as the input for Bayesian Markov chain Monte Carlo (MCMC) model-based clustering. We used the SNP alignment obtained from the reference-based mapping approach as the input for ChromoPainter, including all 86 E. rhusiopathiae isolates. Instructions in the example provided on the official program web page were followed [82]. A uniform recombination map was generated using a perl script available with this program, and was used in ChromoPainter’s E-M procedure for estimating the parameters of effective population size and global mutation rate (estimated to be 2379.89 and 1.17 E-3 respectively). These parameters were then used to run ChromoPainter, followed by ChromoCombine which calculates the variance expected in the data, needed for running fineSTRUCTURE. This was run using 1,000,000 MCMC iterations, half of which were burn-in, sampling every 250 iterations. A heat map representing the inferred relationships was generated using the fineSTRUCTURE graphical user interface.

    Finally, a phylogenetic network was inferred using Neighbor-Net [46], implemented within SplitsTree using a core nucleotide alignment of all 86 E. rhusiopathiae genomes without masked phage sequences generated in Parsnp. To achieve greater resolution among the isolates of Clade 3 which was found to be the dominant clade, the SNP sites determined to be outside of recombinant segments by Gubbins (n = 7580) were used to generate a maximum likelihood tree of the isolates within this clade using PhyML [83], using a generalized time-reversible (GTR) model of nucleotide substitution with a gamma distribution, and performing 1000 bootstrap replicates. To test whether a temporal signal could be detected in Clade 3 isolates once recombinant segments had been removed, this maximum likelihood tree was examined in Path-O-Gen, which looks for correlation between year of isolation and root-to-tip divergence [84]. Clustering within Clade 3 by host species (swine/poultry/wild birds/wild ungulates/other), geographic origin (Europe, central North America, Northern Canadian provinces, Arctic, other/unknown) or submitting laboratory was examined in BaTS [85]. BaTS analysis was run on a set of trees generated in MrBayes [86], using the curated set of non-recombinant SNPs from Gubbins as the input alignment, performing 10 million iterations, sampling every 1000 samples, and taking the first half as burn-in. A Fisher’s exact test was used to assess the statistical significance of the lack of Clade 3 isolates found among marine mammals.

  2. Assistant

    多宿主病原体Erysipelothrix rhusiopathiae的基因组分析揭示了广泛的重组现象以及三种广泛分布的通用克隆的存在

    Taya Forde, Roman Biek, Ruth Zadoks, Matthew L. Workentine, Jeroen De Buck, Susan Kutz, Tanja Opriessnig, Hannah Trewby, Frank van der Meer & Karin Orsel

    BMC Genomics第17卷,文章编号:461(2016)引用此文章

    4484次访问

    36次引用

    2个Altmetric

    指标详细信息

    摘要 背景

    了解细菌种群的结构是研究其生态学和进化历史的重要前提,可促进对宿主特异性、致病性、地理扩散和分子流行病学的探讨。Erysipelothrix rhusiopathiae是一种机会性病原体,目前在全球的养猪和家禽产业中重新出现。这种细菌偶尔会导致圈养海洋哺乳动物的死亡,并最近被认为与大规模野生动物死亡有关系。然而,尽管其经济相关性和广泛的地理及宿主分布,包括人畜共患病潜力,全球这个细菌的多样性、重组率和种群结构仍然很少被描述。本研究根据多样的分离株收集进行了E. rhusiopathiae的广泛基因组比较,以解决这些知识空白。 结果

    本研究对来自不同宿主物种和地理来源的83个E. rhusiopathiae分离株进行测序并组装,分离时间在1958年至2014年之间,采用了参考基于映射和新装配的方式。我们发现核心基因组中有很大一部分(58%)经历了重组。因此,我们使用了三种独立的方法来定义这一物种的种群结构,这些方法对重组的存在具有鲁棒性:基于一组保守蛋白序列的系统发育树,计算机染色体着色法和网络分析。所有三种方法基本一致,支持E. rhusiopathiae物种内存在三种不同的克隆。尽管我们发现了一些宿主和地理聚类的证据,但每个克隆中都包括来自不同宿主种类和多个大陆的分离株。 结论

    通过全基因组测序数据,我们确认了最近的建议,即E. rhusiopathiae是一种弱克隆物种,且受到同源重组的广泛影响。尽管频繁的重组,我们仍然可以可靠地识别出三种不同的克隆,这些克隆并不明确地按宿主物种或地理来源划分。我们的结果为未来E. rhusiopathiae的分子流行病学、生态学和进化研究提供了重要的基础,并便于与其他具有重组能力的多宿主细菌进行比较。

    背景

    揭示种群结构及其决定因素对于理解细菌的生态学和进化至关重要。关于宿主特异性与通用性的关键问题、全球基因流动模式以及临床疾病表现的遗传基础可以通过 examining 物种内的株系关系进行探讨。此外,分子流行病学研究依赖于对种群结构的理解,此结构为解读目标生物的基因组多样性提供了框架,而识别与宿主或地理相关的谱系有助于源归属的确立。全基因组测序为解决与种群结构相关的问题提供了新的机会;然而,这一工作复杂且需考虑重组的潜在影响。

    E. rhusiopathiae是一种革兰阳性、兼性细胞内细菌,是人类和动物的重要机会性病原体。与E. rhusiopathiae的动物感染通常表现为称为红斑性皮肤损害的黄痂,往往与职业有关(例如屠宰场工人、肉商、渔民等)。E. rhusiopathiae已在多种野生和家养物种中被记载,包括鸟类、哺乳动物、爬行动物、鱼类和节肢动物。该病原体作为猪红斑病的病原,被认为是造成猪产业中显著经济损失的因素,可能伴随大规模的爆发。近年来,在美国中西部、日本和中国,E. rhusiopathiae感染的发生率显著上升。红斑病在欧洲家禽生产中也再次出现,可能与养殖系统的变化有关。在圈养的海洋哺乳动物中,E. rhusiopathiae已知会导致严重且常常危及生命的感染,而最近有数百只麝牛在加拿大北极群岛的死亡事件引发了对这一细菌潜在保护价值的关注。E. rhusiopathiae的广泛生态和地理分布被归因于其能够感染多种宿主物种,这些宿主可能是健康携带者,加之其在环境中持久存在的能力。然而,尽管它的普遍性和对包括人类在内的多种宿主物种的重要性,以及临床表现的高度变异性,关于E. rhusiopathiae的遗传多样性、种群结构和宿主特异性知之甚少。

    最近的全基因组测序项目促进了Erysipelothrix属的分类。作为放线菌门的一个成员,Erysipelotrichia类只有一个目Erysipelotrichales和一个科Erysipelotrichaceae,后者包含10个属。在Erysipelothrix属中,其他已认识的物种包括E. tonsillarum、E. inopinata、E. sp. strain 1和E. sp. strain 2,以及最近识别的E. larvae sp. nov。E. rhusiopathiae和E. tonsillarum长期以来被认为是不同的物种,基础是它们在致病性、表型特征和血清型群上的差异,DNA-DNA杂交研究对此提供了支持。16S rRNA基因序列系统发育显示,E. inopinata似乎在E. rhusiopathiae和E. tonsillarum之间的分化之前就产生了。E. sp. strain 1和E. sp. strain 2与其他Erysipelothrix物种之间的关系尚未探讨。

    E. rhusiopathiae分离株的种内分类不太明确,使用各种工具对分离株进行分类。血清分型是一种最常用的方法,它涉及测试与相应细胞壁的特定多肽糖抗原进行凝聚反应。尽管不同的Erysipelothrix种具有不同的血清型,因而具有诊断价值,但由于许多细菌物种中存在频繁的胶囊特定基因的横向转移,血清分型并不适合推断分离株之间的进化关系。最近,已知的免疫原性蛋白质表面保护抗原(Spa)的差异也被用于区分E. rhusiopathiae的分离株。使用脉冲场凝胶电泳(PFGE)等各种比较分型方法也对E. rhusiopathiae分离株之间的遗传关系进行了研究。然而,基于序列的分型方法(这些方法允许进行更易于的实验室间比较和功能差异研究)在最近才应用于E. rhusiopathiae:对当时GenBank中可用的三条E. rhusiopathiae全基因组序列(两条完整序列和一条草图序列)进行了比较,并使用多位点序列分型(MLST)研究了该细菌的种群结构,主要聚焦于欧洲家禽的分离株。目前尚不清楚重组在E. rhusiopathiae中的重要性,尽管最近的研究发现该物种是弱克隆的,表明重组可能在该物种的遗传变异中起重要作用。迄今为止,还没有大规模的基因组比较研究来全面调查E. rhusiopathiae的遗传多样性。

    本研究的目标是描述E. rhusiopathiae的全球基因组多样性,并在考量重组存在的情况下检验其种群结构,以为未来对该多宿主病原体流行病学和生态学的研究提供基本依据。第二个目标是基于全基因组序列数据考察不同Erysipelothrix物种之间的系统发育关系。 结果 测序和泛基因组统计

    本研究分析了83个新测序的E. rhusiopathiae分离株,代表了来自北美(加拿大和美国)、南美(阿根廷)、欧洲(比利时、匈牙利和英国)、亚洲(日本)、南非和澳大利亚的分离株,大部分来自加拿大(n=37)、美国(n=20)和比利时(n=17)(见附加文件1:表S1)。分离日期范围从1958年到2014年。来源的宿主物种包括猪(n=18)、家禽(n=14)、圈养海洋哺乳动物(海豚和白鲸;n=8)、鱼(n=5)、野生鸟类(主要是水鸟;n=7)、野生有蹄类动物(麝牛、驼鹿和驯鹿;n=28),还有一只狼。关于与这些分离株相关的临床表现的信息仅在12个分离株中可用(附加文件1:表S1),但 incluidos了急性和亚急性败血症的病例,以及从皮肤损害中分离的E. rhusiopathiae。已有文献中已知16个分离株的血清型(附加文件1:表S1);新分型的分离株VI11-2_lu被发现属于血清型5。

    在本次分析中,所有E. rhusiopathiae分离株中共有1137个核心基因,代表了Fujisawa参考基因组中67%的编码序列。共识别出512个单独的基因(即仅存在于一个分离株中),每增加一个测序的基因组,平均发现6.6个新的独特基因。在核心基因组中包括ComEC(膜孔)和dprA(重组介导因子)基因,以及假定的II/IV型分泌系统蛋白,表明了自然适应性的高度可能性。根据在程序LS-BSR中进行的成对比较,未发现宿主物种或地理起源或在后续分析中识别的克隆之间的整体基因组成存在重大差异。

    E. rhusiopathiae的高比例核心基因组经历了重组

    使用Parsnp生成的核心基因组比对总长度为1,049,431个碱基对,占参考基因组的58%。总体序列相似性程度高,成对同一性为99.3%。BratNextGen分析包含所有分离株,推理出58%的整体核心基因组比对经历了重组。克隆之间在推断的重组比例方面存在显著差异。两个重组检测方法都估计克隆2的基因组中经历重组的比例更高。BratNextGen未在克隆1中检测到任何重组片段,而这是在核心基因组中重组最少的克隆。尽管比对的很大部分受到重组影响,但平均的重组与突变(r/m)比率适中,在克隆1、2和3中分别为0.96、2.18和0.55。

    在几个分离株中检测到了与噬菌体和质粒相关的序列

    在86个分离株中发现了47个(55%)有潜在的前噬菌体序列,涵盖了所有宿主组和地理位置(附加文件1:表S1)。每个分离株最多发现三个这样的序列,平均长度为22 KB(范围:12.5–47 KB)。28个分离株与Fujisawa基因组注释的噬菌体具有高水平的同源性(成对同一性≥96.5%),在至少8 KB长的片段中(附加文件1:表S1中以P1标识)。这些28个分离株中有两个被PHAST推测为完整的前噬菌体。所有三个克隆中都有P1噬菌体序列的分离株,这表明这些相似的噬菌体序列的存在是多次引入或祖先获得后的丢失的结果。第二组噬菌体序列(n=15;附加文件1:表S1中标识为P2)在约12 KB的区域共享了95.9%的成对同一性。这其中四个被PHAST分类为完整的。强烈表明其中一些P2序列可能是通过垂直传递遗传的,而其他序列则在无关的分离株中被识别,显示出横向转移。八个分离株的噬菌体序列形成了第三组(P3),在具有94%配对同一性的至少12 KB的同源序列中。none 这些序列均未被PHAST认定为完整。这些序列全部来自两个不同集群的猪分离株,虽然并不是每个集群中的所有分离株都有检测到噬菌体序列。在其他检测到的噬菌体序列中,大部分仅与三组序列中的一组共享较小部分的序列同源性(附加文件1:表S1中标识为“部分”)。在我们的收集中,仅发现了一个独特的噬菌体序列:来自家禽分离株G11的22 KB不完整噬菌体序列。基于BLAST比对,这个序列在与Streptococcus pyogenes的一种整合共轭元素共享了13 KB的片段,成对同一性为95.4%,而其余序列与携带IV型分泌途径组分编码序列的质粒共享96%的配对相似性。

    对公开可用的E. rhusiopathiae质粒序列(GenBank: NC_002148)的BLAST搜索发现,来自我们收集的一个分离株(‘Ery Afrika 1’,克隆2)具有相似序列,成对同一性为95.2%且涵盖100%。在从每个分离株中未与参考基因组比对的contig进行BLAST搜索时,一个家禽分离株与共生转移蛋白有比对,而五个猪分离株与质粒序列有比对。这其中三个来自比利时的猪分离株与同一质粒序列有关,其他三个比利时猪分离株未发现相似序列的搜索结果。

    E. rhusiopathiae的种群结构

    在PhyloPhlAn中生成的系统发育树使用超过400个保守细菌蛋白,根植于Erysipelotrichaceae科的其他属,将E. tonsillarum置于E. rhusiopathiae和E. sp. strain 2的基部。通过放大到物种内层次,克隆之间的差异更为明显。与此系统发育一致,使用ChromoPainter和fineSTRUCTURE的分析,及在SplitsTree中实施的网络分析都支持E. rhusiopathiae物种内存在三种不同的克隆。7个分离株被支持为克隆1的成员,这些分离株来自美国的两头圈养海洋哺乳动物(白鲸和海豚)、一条来自澳大利亚水族馆的海豚、一只来自美国的红狼、一条来自日本的鱼、一头来自加拿大的驯鹿以及阿根廷的一个羊沐浴池。14个分离株被支持为克隆2的成员,这些分离株来自美国、比利时和南非的5头海洋哺乳动物、来自日本和美国的3条鱼、来自比利时的3个家禽和1个猪、来自加拿大的一头驯鹿以及GenBank中的ATCC19414分离株(宿主物种和地理来源未知;见附加文件1:表S1)。此外,来自比利时(红箭Figs. 3, 4和5)的一个家禽分离株被ChromoPainter与克隆2的分离株分组,但根据系统发育分析被归类为克隆2和克隆3之间的“中间”分离株。根据所有三种方法,至少有五个分离株归入这一中间组:来自GenBank的两个全基因组序列(来自中国的SY1027和来自日本的Fujisawa),以及来自加拿大的两头驯鹿和一头驼鹿。根据染色体着色法,另外两个分离株聚成该中间克隆,均来自美国的野生鸟类;这些分离株在系统发育方法中被归入克隆3(蓝色和黑色箭头显示于Figs. 3, 4和5)。在系统发育和染色体着色法下,克隆指示不同的三个分离株通过网络分析都归属于主要集群外(图5中用箭头表示)。其他57个分离株构成了占优势的克隆3,包括所有经过测试的来自加拿大的猪(n=8)和家禽(n=6)分离株,大多数比利时猪分离株(6/7)和一半的比利时家禽(4/8)分离株,另外还有三个其他猪分离株(一个来自美国,两个来源不明),以及来自五种不同北美野生鸟类、北美驯鹿(n=11)、驼鹿(n=4)和麝牛(n=8),以及一个来自匈牙利的鱼分离株。

    在不同方法之间,克隆3内的分离株之间的关系观察到较少的一致性。图6显示了使用Gubbins推断的非重组核心单核苷酸变异(SNP)的最大似然(ML)系统发育树,均嵌合有PhyloPhlAn或fineSTRUCTURE支持的分组。总体而言,基于更高的bootstrap值对较早的节点的支持较强,且其他两种方法支持的情况下更为频繁。在根至尖图分析克隆3的分离株中未检测到时间信号(相关性<0.001)。

    在Spa类型与克隆之间发现了一致关系:克隆1中的所有分离株都有一份SpaB基因,而所有其他E. rhusiopathiae分离株都有一份SpaA基因。相反,在那些已定义血清型的分离株中观察到了显著的同源性。血清型5的分离株出现在所有三个克隆中,血清型2的分离株出现在克隆2和克隆3中,而血清型1a的分离株则出现在中间和克隆3中。

    宿主或地理特异性的证据有限

    所有三个克隆包括来自不同宿主物种和多个大陆的分离株。然而,所有海洋哺乳动物分离株(n=8)都属于克隆1或克隆2,在占优势的克隆3中没有发现,而克隆1中也未发现猪或家禽分离株。在克隆3中,通过宿主物种、地理位置和提交实验室的聚类都被发现是显著的,使用程序BaTS进行分析(附加文件3:表S2)。尽管其他两种方法没有充分支持,从非重组SNP的系统发育树(图6)看,绝大多数家畜分离株在克隆3中(21/27的猪和家禽分离株)被分组为具有96%的bootstrap支持的单一子群,而大多数野生动物来源的分离株则位于该克隆中的一个独立子群。

    讨论 E. rhusiopathiae核心基因组中经历了广泛的同源重组

    这是对多宿主病原体E. rhusiopathiae的首次大规模基因组研究。我们的结果表明,同源重组在生成该物种的多样性中发挥了重要作用,确认了最近MLST研究的发现。根据我们的数据,几种机制可能导致该物种中的重组:从外部环境中摄取的遗传物质、噬菌体和质粒。能够通过细胞膜导入外源DNA(转化)的细菌被称为“自然能力”,并具备特定的机器来完成这一任务。核心基因组中存在的ComEC(膜孔)和dprA(重组介导因子)基因,以及假定的II/IV型分泌系统蛋白,表明E. rhusiopathiae可能具备这一能力。转导和接合可能也是E. rhusiopathiae中横向基因转移的重要机制,因在我们的研究中分离株中丰富的噬菌体相关序列表明,尽管从草图组装中检测这些序列的灵敏度可能较低,以前的研究发现,高达三分之一的测试分离株包含质粒,几个可用参考基因组中的可移动遗传元件显示出质粒或前噬菌体来源的特征。根据我们所知,这是首次在大量E. rhusiopathiae分离株中检查噬菌体相关序列的流行程度和多样性。尽管关于与许多分离株相关的临床表现的数据有限,但在噬菌体序列的存在与致病性之间未发现明确的关联。特定噬菌体序列的存在似乎并不是进行流行病学推断的系统发育信息性标记,因为在不同的克隆中发现了相关的噬菌体序列,而且在某些情况下,噬菌体序列在高度相似的分离株群体内并不保持。类似地,各种相关分离株间质粒序列的不一致存在表明接合可能经常发生。

    尽管核心基因组的很大比例经历了重组事件,但我们为每个克隆估算的r/m率(在0.55到2.18之间)与其他细菌物种相比仅处于适中水平。这结合核心基因组中发现的高成对同一性(99.3%),提示在E. rhusiopathiae核心基因组内,等位基因高度保守,尽管其组织不然。这可能解释了MLST与PFGE之间未能达成一致性,因为后者对结构重排敏感。在每个克隆内,基于Gubbins的输出估计的经历重组的核心基因组比例。此外,利用BratNextGen的研究分别发现,重组主要在克隆2发生,其中r/m比率比其他克隆高两到四倍,且更大比例的基因组被发现涉及重组。克隆相关的重组率差异在其他细菌物种如兰斯氏单胞菌中也有先例。

    E. rhusiopathiae的种群结构在多重推理方法中一致

    考虑到在存在重组的条件下全基因组系统发育重建的复杂性,我们采用了三种独立保守的方法来推断E. rhusiopathiae的种群结构。使用保守蛋白序列集合的方法已被证明对横向基因转移颇具鲁棒性。染色体绘制方法不依赖于系统发育推理,而重建“受体”个体的染色体单倍体型作为“供体”个体的重组衍生片段的总和。最后,网络分析提供了系统发育历史可视化的手段以用于非严格克隆的生物。根据这三种方法,分离株被分配到克隆的良好一致性。这些方法均根据新组装的基因组(保守蛋白树和网络分析)与参考基础映射方法(染色体绘制)采集结果,证明这些结果的鲁棒性。

    通过存在于克隆1中的SpaB基因与其他两个克隆和中间分离株中的SpaA基因的存在,我们能区分出克隆1。表面保护抗原是一种有效的免疫原,对疫苗开发有潜在的相关性。SpaA在已有文献中是与E. rhusiopathiae致病性关联的较为明确的基因之一,扮演内皮细胞附着和抗吞噬的角色。SpaB基因与SpaA有约60%的氨基酸相似性并且在抗原性上有所区别。先前发现,来自水域的E. rhusiopathiae菌株能够表达多种Spa类型,然而在我们收集的分离株中并未观察到此潜在的遗传基础。不意外,血清型与分离株之间的系统发育相关性无关。由于血清型广泛同源,因此不应用于推断分离株之间的进化关系,作为流行病学工具的用途(例如决定多个分离株是否属于同一疫情)需要更多关于血清型切换预期频率的信息。

    尽管在构建占优势的克隆3系统发育图之前已去除了重组位点,但基于bootstrap值的对大多数更早节点的支持难以阐明。在去除重组片段后,对克隆3分离株进行的根至尖分析未能检测到可辨识时间信号。未检测到时间信号的部分原因可能是由于此数据集中残留重组片段的影响。尽管没有检测到相关性,但假设的代谢周期高度变动(例如在环境阶段期间几乎没有细菌复制,或相对于主动感染时期的携带)可能是缺乏时间结构的合理解释。没有校准分子时钟,我们无法基于全基因组序列数据为E. rhusiopathiae估计全球或克隆特定的替换率。

    E. rhusiopathiae显示有限的宿主关联

    细菌在特定宿主物种的定殖程度上表现出显著的变异性,这种特定性被认为是由复杂的宿主-病原体相互作用所造成的。许多细菌种类都有相对明确的宿主适应株,而其他细菌物种则表现出适应宿主和通用型亚型的混合特征。宿主特异性的决定因素沿着从单个氨基酸残基到基因内容的变化(存在/缺失)或基因组岛的获取的广泛梯度分布。在我们的研究中,不同宿主物种、地理位置或克隆之间未发现基因内容的一致差异。

    通过实验感染研究已知,许多宿主物种对相同的E. rhusiopathiae株系是易感的;来自一种宿主物种的分离株接种至另一种宿主时,通常会导致接受宿主的感染和临床病症,尽管易感性有所不同。我们发现,所有三个克隆都包含来自不同宿主物种的分离株。因此,E. rhusiopathiae的所有谱系值得认为是广泛感染多种物种,尽管可能具有不同的感染性和致病性。由于一些野生动物的分离株嵌套在以家畜为主的谱系内,反之亦然,我们的数据也强烈表明了跨物种传播的可能性,暗示宿主或生态偏好存在,但不严格特定宿主。类似的发现被MLST研究证实,在这个占主导地位的克隆复合物中,包含了多个宿主物种的分离株,包括家禽、猪、羊和人类。

    尽管不同克隆之间未能严格限制于特定宿主,但观察到来自圈养海洋哺乳动物的分离株仅存在于克隆1和克隆2中显着(p<0.0001),尤其由于这些样本来自全球分布的水族馆。这一发现是否与海洋哺乳动物对这些株型的易感性差异有关,或这些株型因为适应海洋环境而不同分布仍需进一步探讨。克隆1和克隆2的E. rhusiopathiae分离株均与海洋哺乳动物的致命感染相关。然而,在我们收集的猪或家禽分离株中没有发现克隆1分离株,因此也没有携带SpaB基因的株系。既然已知带有SpaB基因的海洋来源株系能够在猪中致使至少轻微到中度的损伤,因此克隆1株在猪和家禽中缺乏的更可能是由于缺乏接触,而不是缺乏致病潜力。

    在主导克隆3内,尽管检测出宿主物种的聚类,但基于地理来源和源实验室的聚类也得到了同样强的支持,似乎表明这些关联与我们分离株收集的机会性质相关的采样偏倚无法被区分。这在病原体的系统发育研究中是一个常见的挑战。对克隆3中宿主和地理聚类的进一步审查,将有助于使用更广泛的样本来源进行更详细的元数据采集,以排除由于流行病学相关分离株造成的聚类。尽管存在这一限制,但克隆3分离株在家畜和野生动物来源之间似乎存在非随机隔离;来自北美和欧美三处独立来源的猪和家禽分离株,往往独立于来自北方有蹄类动物的分离株聚集。在家畜分离株中,证据显示定期的跨洲际交流。

    尽管努力包括来自多样宿主物种的全球性收集,绝大多数分离株来自北美和欧洲,有些重要宿主(包括人类、爬行动物和节肢动物)不在我们收集的范畴内。将来自代表性不足的大陆和宿主物种的额外分离株的图片化,将有助于增进我们对E. rhusiopathiae全球种群结构的理解。

    全基因组测序支持Erysipelothrix物种间的现有划分

    我们已经证明E. sp. strain 2在系统发育上与E. rhusiopathiae和E. tonsillarum有明显的区别,基于保守蛋白的系统发育。其与E. rhusiopathiae的区别得到进一步支持,因为其序列相似度不足以将读长映射到Fujisawa或SY1027的完整基因组序列,或作为输入到通过Parsnp生成的核心基因组比对,该比对要求平均核苷酸同一性≥97%。根据保守蛋白的系统发育,E. rhusiopathiae和E. sp. strain 2似乎是从致病性较低的E. tonsillarum演化而来的。在先前结合多种系统发育方法输出的系统发育研究中未观察到这一关系。进一步研究E. rhusiopathiae和E. sp. strain 2与E. tonsillarum之间的遗传差异,可能会为这些物种获得更高水平的致病性提供见解。最近在亚表层生物圈中检测到Erysipelothrix spp.可能代表着一个有趣的机会,以更广泛的视角探讨这一属的演化。

    结论

    采用不同的方法,我们能够有信心地确定多宿主病原体E. rhusiopathiae的种群结构,尽管重组在其进化历史中发挥了重要作用。证据表明,该细菌可通过转化、转导和接合获取新DNA,从而让其基因组具有可塑性。该物种由三大克隆组成,遍布多个大陆和宿主物种,包括家畜和野生动物,并表明克隆或亚克隆可能在宿主偏好和重组率上存在差异。由于我们用于基因组测序的分离株收集的机会性特性,流行病学推断受到制约,未来研究应在目标样本收集上进行更大努力,然而我们的结果为支持未来深入的流行病学和进化研究提供了重要框架,并为其他具有重组能力的多宿主细菌的比较研究提供了参考。 方法 细菌分离株

    为了检查E. rhusiopathiae的种内基因组多样性,分离株依据广泛的地理位置、宿主物种、临床表现和首次分离年份进行了机会收集。E. rhusiopathiae分离株以及来自其他Erysipelothrix spp.(E. tonsillarum和E. sp. strain 2)的分离株由多位合作伙伴慷慨提供(附加文件1:表S1)。收到分离株后,将其接种于富含5%羊血的哥伦比亚琼脂(BD-Canada, Mississauga, ON, Canada)以进行形态学表征;若所分离的菌落形态特征符合E. rhusiopathiae(无色至浅蓝色、圆形、小直径、经常伴有α溶血),则重新划单一菌落以获取用于DNA提取的克隆群体。除这些存档分离株外,还通过与野生动物健康监测项目相关的样本测试获得了进一步的分离株。选择性培养的野生有蹄动物(驼鹿、驯鹿和麝牛)组织样本中的E. rhusiopathiae:用Stomacher 80 Biomaster(Seward, Port Saint Lucie, FL, USA)将2克组织在20毫升含5%血清酸奶心脏熏煮液(BHI)中进行机械均化,孵育于37 °C、5% CO2的氛围中培养过夜,随后在含有氨基糖苷类抗生素的选择性培养基中再孵育48小时(氨基苷的浓度分别为:40 μg/ml的链霉素、50 μg/ml的新霉素和25 μg/ml的万古霉素)并于37 °C,5% CO2的相同选择性培养基中进一步接种至少48-72小时。再将所获得的菌落如同对其他分离株所述继续进行重新接种于哥伦比亚琼脂上。 DNA提取和PCR确认

    通过将1个涂抹环的菌落悬浮于200μl的磷酸盐缓冲液(PBS)中,使用DNeasy血液和组织试剂盒(Qiagen, Mississauga, ON, Canada)按照制造商说明提取DNA,但DNA洗脱量处于较小体积(50μl)的提供洗脱缓冲液中,以获得更高的DNA浓度。提取的DNA通过QT-PCR确认为E. rhusiopathiae,使用先前描述的引物和定位于rRNA基因簇3'非编码区的物种特异性探针进行确认。20μl的PCR反应包括10μl的TaqMan快速高级混合液(Applied Biosystems, Carlsbad, CA)、10 pmol的每个引物、1 pmol的探针和2μl的模板DNA。反应使用CFX96热循环仪(Bio-Rad, Mississauga, Ontario, Canada)进行,循环条件如下:“50 °C反应2分钟,95 °C反应20秒,然后进行40个循环的95 °C 3秒和57 °C 30秒”。还使用不同的探针对E. tonsillarum分离株进行了确认,采用与此同一组引物和反应条件的探针基础QT-PCR。E. sp. strain 2的身份在草图组装中通过识别特定的E. sp. strain 2的23S rRNA基因的引物序列进行确认。 文库准备和测序

    文库准备使用Nextera XT v2试剂盒(Illumina, San Diego, CA)进行,按照制造商说明,包含1%的PhiX对照。进行多重测序,在卡尔加里大学的Illumina MiSeq设备上获得250个碱基对的配对末端读取数据。 读取映射和变异检测

    程序ConDeTri用于裁剪原始读取数据,提取高质量的读取数据,并消除重复数据。每个分离株的唯一裁剪读取数据与完整E. rhusiopathiae Fujisawa基因组(Genbank: NC_015601)使用BWA-MEM进行比对(默认设置)。通过Qualimap提取测序和映射参数(读取数、映射读取数、平均覆盖率、平均映射质量)。所有测序的分离株至少拥有9X的平均覆盖率,平均覆盖率为33X,映射质量中位数保持在“均值57”之高。

    使用SAMtools中的mpileup命令检测变异。使用自定义的Python脚本(由HT编写,索取可用)生成贯穿所有E. rhusiopathiae基因组的高质量SNP位点列表,该脚本基于碱基质量、映射质量、读取深度和杂合性进行过滤。若任何位点的共识碱基质量≥40,映射质量≥40,至少在每条链上有三次读取比对到该位点,并且在至少一个分离株中,有超过95%的读取支持相同的极大碱基,则将该位点包含在变异列表中。如果发现多个备选等位基因,则将这些位点排除,因为由于可能的序列饱和或映射错误,这些位置被认为在系统发育上较不可靠。通过实例确认,若每个SNP位点的共识碱基质量≥30,映射质量≥30,且在每条链上至少有2次读取且超过95%的读取支持相同的等位基因,则各个SNP位点的等位基因在所有分离株中被调用。去除重复区域和可移动元件后,聚集在Fujisawa参考基因组上发现的总共32148个高质量独特变异位点被识别于85个E. rhusiopathiae分离株中(包括兄弟GenBank序列)。生成了一个在100%分离株中等位基因被正确调用的位点列表(n=6078)。共测序142个E. rhusiopathiae分离株,但在基于元数据和序列相似性(即,<15个高质量SNP的全基因组差异)考虑为流行病学相关联的情况下,仅选择具有最高覆盖率的分离株纳入本次分析。 新装配

    使用SPAdes对独特的裁剪读取进行组装,k值为55;此组装程序是基于QUAST为候选组装器生成的指标,取得了更好的比对数量L50与较少的错误组装之间的权衡,SPAdes使用了内置的BayesHammer进行进一步的读取错误校正,以及一个使用BWA的后处理工具Mismatch Corrector。程序PAGIT用于订购与扩展contig并纠正共识序列中的错误。使用QUAST(附加文件1:表S1)确定PAGIT前后输出组装指标(contig数量、N50)与后续结果(支架长度和GC%);应用PAGIT改进管道后,分离株平均支架数量为13(范围:4-89)。平均GC包含范围为35.9-36.5%,在所有分离株中均值为36.3%。 注释

    RAST(利用子系统技术的快速注释)被用于注释并预测每个分离株中的编码序列。利用PHAST识别潜在的前噬菌体序列。为了鉴定质粒相关序列,PAGIT中未与参考基因组比对的contig与NCBI数据库进行比对。此外,对可在GenBank上获得的E. rhusiopathiae质粒序列进行BLAST搜索,已知质粒或编码合并转移蛋白的基因的命中被视为质粒存在的指示。为了确定每个分离株的Spa型,使用包括三种已知Spa类型(A、B和C)代表序列的自定义BLAST数据库,利用Geneious版本7.1.8检索每个E. rhusiopathiae分离株。使用先前描述的方法确定分离株VI11-2_lu的血清型。由于文献中尚未描述未知分离株在血清型上的遗传基础,因此无法通过基因组序列数据确认未知分离株的血清型。 比较基因组学

    使用LS-BSR生成泛基因组统计,并使用PanGP进行可视化。为了提高在较低覆盖的分离株中可能错过的核心蛋白水平检测的灵敏度,在分析中排除了覆盖低于15X(n=8)的分离株。LS-BSR脚本compare_BSR.py用于查找两个组分离株之间基因内容的差异。具体而言,在不同克隆之间(由PhyloPhlAn和fineSTRUCTURE确定)进行基因内容(在一个组的所有成员中存在的基因而在第二组的所有成员中缺失)之间的成对比较,其中涉及特定宿主物种(猪、家禽、海洋哺乳动物和麝牛分离株与其他所有分离株比较),以及地理来源(北美与欧洲)。不幸的是,由于缺乏与我们分离株集合相关的临床表现的元数据,无法评估与致病性相关的遗传差异。核心遗传内容使用GView进行了可视化,筛选覆盖≥15X的分离株。 重组检测

    程序Parsnp用于生成核心基因组比对(即所有包含的基因组中存在的保守正交区域)。此比对的输入为83个PAGIT改进的E. rhusiopathiae新装配与三个可用的E. rhusiopathiae序列。PHAST检测出的潜在前噬菌体序列在创建比对之前使用BEDTools的maskfasta命令进行了屏蔽。该核心核苷酸比对作为BratNextGen和Gubbins的输入。由于Gubbins最适合于检测紧密相关分离株的重组,因此为各个分离株的克隆进行了即可分析。BratNextGen使用设置超参数α为4并运行20次重组估计算法的迭代来运行。使用100次算法的置换估计统计显著性,设置显著性为α=0.05。Gubbins在公共虚拟机上以默认设置运行。重组与突变率(r/m;重组片段中的SNP数:推测为自发突变的SNP数)由Gubbins的输出计算,基于所有内部分支的突变总和和连接到每个终端节点的输出树中的可推测的共同祖先节点开始。 种群结构考察

    由于我们发现重组在E. rhusiopathiae中发挥了重要作用,因此使用了三种对重组鲁棒的技术来检查种群结构。首先,PhyloPhlAn用超过400个保守蛋白序列估计系统发育树。使用RAST生成的氨基酸fasta文件作为程序的输入。包括了83个新装配和三条可在GenBank上找到的E. rhusiopathiae菌株,以及三个新构装的E. spp.分离株(附加文件1:表S1),以及E. tonsillarum的草案序列[NZ_AREO00000000]。为树根选取的外群包括来自密切相关属的一些草案全基因组序列。除Holdemania filiformis DSM 12042 [NZ_ACCF00000000]外,Clostridium sp. DSM 1552 [NZ_ABIK00000000]和Erysipelatoclostridium ramosum DSM 1402 [NZ_ABFX00000000]、Solobacterium moorei F0204 [NZ_AECQ00000000]和Bulleidia extructa W1219 [NZ_ADFR00000000]以外。该树的环状版本于生命互动树(iTOL)v3.0中生成,以展示E. rhusiopathiae分离株及其宿主物种和地理起源之间的关系。

    接下来,通过使用ChromoPainter和fineSTRUCTURE作为替代方式来检验种群结构,而不依赖于系统发育推理。该分析评估每个基因组(接收者)作为可以从分析中其他任何基因组(供体)接收的重组衍生片段的总和。推断的共享片段通过主成分降维分析概述,随后用作Bayesian Markov chain Monte Carlo(MCMC)模型基础的聚类分析的输入。使用基于参考的映射方法得到的SNP比对作为ChromoPainter的输入,包含所有86个E. rhusiopathiae分离株。遵循程序网页上提供示例中的指示,生成均匀重组的图谱,后用于ChromoPainter的E-M过程,估计有效种群大小和全球突变率(分别估计为2379.89和1.17E-3)。这些参数用于运行ChromoPainter,随后运行ChromoCombine计算数据中预期的方差,该步骤所需的运行fineSTRUCTURE。此步骤运行100万个MCMC迭代,前半部分为burn-in,每250次采样一次,生成代表推断关系的热图。

    最后,使用Neighbor-Net推断出系统发育网络,在SplitsTree中实现,基于不掩盖的F的核心核苷酸比对序列生成E. rhusiopathiae的网络。为了实现对克隆3分离株的更好分辨率,使用Gubbins确定不在重组片段外的SNP位点(n=7580)生成克隆内分离株的最大似然树,应用PhyML和引导分布的总体时间可逆(GTR)核苷酸替代模型,执行1000次bootstrap复制。为了检测克隆3分离株的时间信号,在去除重组片段后,该最大似然树在Path-O-Gen中检查其是否能够识别,并在分离株的年份和根至尖偏差之间寻找相关性。通过BaTS检验克隆3内根据宿主物种(猪/家禽/野生鸟类/野生有蹄类/其他)、地理起源(欧洲、北美中部、加拿大北省、北极及其他/未知)或提交实验室的聚类。BaTS分析是在MrBayes中生成树的集合上运行的,使用来自Gubbins的非重组SNP的策划集作为输入比对,执行1000万次迭代,采样每1000个样本,并将前半部分设为burn-in。使用Fisher精确检验评估在海洋哺乳动物中发现的缺乏克隆3分离株的统计显著性。

Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.