The Complete Genome Sequence of the Pathogenic Intestinal Spirochete Brachyspira pilosicoli and Comparison with Other Brachyspira Genomes

Background The anaerobic spirochete Brachyspira pilosicoli colonizes the large intestine of various species of birds and mammals, including humans. It causes “intestinal spirochetosis”, a condition characterized by mild colitis, diarrhea and reduced growth. This study aimed to sequence and analyse the bacterial genome to investigate the genetic basis of its specialized ecology and virulence. Methodology/Principal Findings The genome of B. pilosicoli 95/1000 was sequenced, assembled and compared with that of the pathogenic Brachyspira hyodysenteriae and a near-complete sequence of Brachyspira murdochii. The B. pilosicoli genome was circular, composed of 2,586,443 bp with a 27.9 mol% G+C content, and encoded 2,338 genes. The three Brachyspira species shared 1,087 genes and showed evidence of extensive genome rearrangements. Despite minor differences in predicted protein functional groups, the species had many similar features including core metabolic pathways. Genes distinguishing B. pilosicoli from B. hyodysenteriae included those for a previously undescribed bacteriophage that may be useful for genetic manipulation, for a glycine reductase complex allowing use of glycine whilst protecting from oxidative stress, and for aconitase and related enzymes in the incomplete TCA cycle, allowing glutamate synthesis and function of the cycle during oxidative stress. B. pilosicoli had substantially fewer methyl-accepting chemotaxis genes than B. hyodysenteriae and hence these species are likely to have different chemotactic responses that may help to explain their different host range and colonization sites. B. pilosicoli lacked the gene for a new putative hemolysin identified in B. hyodysenteriae WA1. Both B. pilosicoli and B. murdochii lacked the rfbBADC gene cluster found on the B. hyodysenteriae plasmid, and hence were predicted to have different lipooligosaccharide structures. Overall, B. pilosicoli 95/1000 had a variety of genes potentially contributing to virulence. Conclusions/Significance The availability of the complete genome sequence of B. pilosicoli 95/1000 will facilitate functional genomics studies aimed at elucidating host-pathogen interactions and virulence.


Introduction
Spirochetes of the genus Brachyspira are anaerobic bacteria that colonize the large intestine of animals and birds [1]. There are currently seven officially named species in the genus, including both pathogenic and commensal representatives. The names of the species and their phylogenetic relationships based on their 16S rRNA gene sequences are shown in Figure 1. The two main pathogenic species are Brachyspira hyodysenteriae, the agent of a major pig disease called swine dysentery, and Brachyspira pilosicoli, the cause of a condition known as intestinal (or colonic) spirochetosis. B. pilosicoli has a wider host range than B. hyodysenteriae, and colonizes a variety of species, including human beings [2][3][4][5][6]. Infections with B. pilosicoli are particularly common in intensively housed pigs and chickens: colonized individuals may develop focal areas of inflammation in the large intestine, with diarrhea and reduced rates of growth and production. Colonization with B. pilosicoli also occurs at a high prevalence rate in human beings living in crowded and unhygienic conditions, particularly in developing countries [7][8][9][10][11], as well as amongst homosexual males [12]. Despite the potential importance of B. pilosicoli as a pathogen, it has not been extensively studied. Progress has been hampered by the spirochete's specialised growth requirements and slow growth rate, as well as by a lack of genomic information and an absence of means for genetic manipulation.
A characteristic feature of infection with B. pilosicoli is the attachment of the spirochete by one cell end to the luminal surface of enterocytes in the large intestine; with time, dense mats of these attached bacteria may form a ''false brush-border'' covering the whole luminal surface of the enterocytes [12,13]. Experimentally, following attachment to cultured Caco-2 cells B. pilosicoli has been shown to induce apoptosis, actin rearrangement and increased expression of interleukin-1ß (IL-1ß) and IL-8 [14]. Such pathological changes may contribute to the local colitis and diarrhea that is observed in vivo.
The recent publication of the genome sequence of B. hyodysenteriae strain WA1 [15] and the availability of a nearcomplete genome sequence of Brachyspira murdochii type strain 56-150 T (DSM 12563) in GenBank represent important opportunities to advance research on these Brachyspira species. B. murdochii is generally considered to be a harmless commensal in pigs, chickens and rats [16,17]; however, there have been recent reports that it may have some potential to cause colitis in pigs [18,19]. In the current study the genome of B. pilosicoli strain 95/1000 was sequenced and subjected to comparative genomic analysis, particularly in relation to the sequences of WA1 and 56-150 T . As the sequence of the latter was incomplete some comparisons were not appropriate. The overall aim of this work was to enhance knowledge about B. pilosicoli and its relationships to the other Brachyspira species, particularly in regard to identifying the genetic basis of their different ecologies and pathogenic potentials.

Spirochete strain and growth conditions
Brachyspira pilosicoli strain 95/1000 (ATCC BAA-1826) was originally isolated in Western Australia from the diarrheic feces of a commercial pig with porcine intestinal spirochetosis [13]. The spirochete was purified by repeated subculture, grown to mid-log phase in pre-reduced anaerobic broth [20], and a cell pellet prepared.

Genomic DNA preparation, library construction and sequencing
Preparation of genomic DNA, library construction and sequencing was as previously described for B. hyodysenteriae [15]. Sequencing was undertaken at the Australian Genome Research Facility. The first round of sequencing was performed via Sanger sequencing, with a total of 42,565 reads generated. The second round was performed using a pyrosequencing approach on a Roche-454 GS20 instrument, generating more than 25 times coverage of the genome. The quality filtered reads were assembled into contiguous sequences using the Newbler Assembler software (http://www.454.com/). To finish the genome sequence, remaining gaps were closed by PCR walking between un-linked contiguous sequences [21].

Sequence analysis and annotation
For both the newly sequenced B. pilosicoli 95/1000 and for the incomplete genome sequence of B. murdochii 56-150 T obtained from GenBank (ABTG00000000), sequence analysis and annotation were as previously described for B. hyodysenteriae WA1 [15]. Dot matrix plots comparing the genomes of 95/1000, 56-150 T and WA1 were generated using Freckle, an in-house development of the Dotter tool [22]. The minimum size of matched sequences was set to 20 base pairs (bp).

General genome features
The general features of the genome of B. pilosicoli 95/1000, together with those of B. hyodysenteriae WA1 and the near-complete genome of B. murdochii 56-150 T are summarised in Table 1  smallest of the three Brachyspira genomes analysed to date (with B. murdochii being the largest). The overall G+C content of the B. pilosicoli genome was 27.9 mol%. Unlike B. hyodysenteriae WA1, B. pilosicoli 95/1000 did not contain a ,36 Kb plasmid. Based on the available sequence it also appeared unlikely that B. murdochii contained a similar plasmid. The B. pilosicoli genome encoded 2,338 genes, with an overall 85% coding region. In comparison, B. murdochii 56-150 T had 3,055 predicted genes in the seven contigs that were available. The percent coding region was similar across the three genomes. The average size of the predicted genes in B. pilosicoli 95/1000 was 997 bp, and tentative function was assigned to 1,645 of these. The B. pilosicoli strain had 569 more functionally assigned genes than did B. murdochii 56-150 T and 282 less than B. hyodysenteriae. B. pilosicoli 95/1000 had 655 genes with unknown function and this was somewhat fewer than for B. hyodysenteriae (704), but much fewer than the extremely high number of unmatched genes for B. murdochii (1,925 (Table 1). As with B. hyodysenteriae [15], only single copies of the 16S, 23S, and 5S ribosomal RNA genes were found in B. pilosicoli and B. murdochii. In B. pilosicoli 95/ 1000, the rrs and rrl genes were closely linked, with the rrl (16S) gene being about 645 Kb from the other two genes. This rRNA gene organisation has been noted earlier; it distinguishes the Brachyspira species from other spirochetes, and presumably pre-dates speciation in the Brachyspira genus [23]. Although spirochetes have a monophyletic origin, the copy number and organisation of rRNA genes differ in the different genera. For instance, the Treponema pallidum rRNA genes appear to be arranged in two typical rrn operons [24,25]. A single rRNA locus is found in most Borrelia species, with rrs separated from rrl and rrf by a small segment of DNA (,4 Kb). In Borrelia burgdorferi the rrf-rrl cluster is duplicated and is found immediately adjacent to the rrs-rrl-rrf cluster [26]. Pathogenic Leptospira species possess two copies each of rrs and rrl and one copy of rrf [27]. The non-pathogenic Leptospira biflexa contains two copies of each rRNA gene [24], and these are dispersed around the genome. Differences in the sequences of the 16S rRNA genes have been used to examine phylogenetic relationships between spirochete genera and species [28]. On this basis, in the case of the Brachyspira species, B. pilosicoli is more distantly related to B. hyodysenteriae than is B. murdochii ( Figure 1).
The origin and terminus of replication in the genome of B. pilosicoli 95/1000 were predicted based on the position of dnaA (BP951000_0595), as bacterial chromosome replication origins are typically located near this gene. The putative oriC origin of replication was identified in an AT-rich intergenic region upstream of dnaA, in the vicinity of a cluster of hypothetical DnaA boxes. Similar putative oriCs with comparable DnaA boxes have been identified in other spirochete genomes ( Figure 3). The original of replication in B. murdochii was not found due to the genome sequence being incomplete. The putative origins of replication are located centrally within the most highly conserved and synteneous regions of the various spirochete genomes ( Figure 3). B. pilosicoli contained the genes grpE, dnaK, hyp, ark, hyp, arg and gyrA downstream of the dnaA, an arrangement similar to B. hyodysenteriae WA1 [15]. B. pilosicoli had a unique hypothetical coding region immediately upstream of the dnaA gene, whereas there were three hypothetical protein encoding genes upstream of dnaA in B. hyodysenteriae. This finding suggests that in B. pilosicoli 95/ 1000 and B. hyodysenteriae WA1 oriC relocated during evolution, presumably as the result of a DNA rearrangement. The differences in the origins of replication compared to other spirochetes suggest that there could be different mechanisms for replication of the spirochete chromosomes. Experimental studies will be required to verify the origin of replication in B. pilosicoli, as has been accomplished for Borrelia burgdorferi [29]. Whole-genome alignment A dot plot comparison of the genomes of B. pilosicoli, B. hyodysenteriae and B. murdochii showed no large scale conservation of gene order, but only conservation of some genes in clusters ( Figure 4). Gene cluster conservation appeared to be greatest between B. hyodysenteriae and B. murdochii, consistent with these two species being phylogenetically more closely related to each other than to B. pilosicoli. A 26 Kb region of unknown significance that is partially conserved in B. hyodysenteriae WA1, B. pilosicoli 95/1000, Enterococcus faecalis and Escherichia coli [30], also was identified in B. murdochii. Alignment of the two complete Brachyspira genomes using the Artemis Comparison Tool [31] identified extensive gene Figure 2. Genome of B. pilosicoli 95/1000. Circles outer to inner: Genes, forward strand, COG coded; Genes, reverse strand, COG coded; tRNA; rRNA; GC skew; AT skew. All genes are color-coded according to Cluster of Orthologous Gene Categories (COG) functions: violet for translation, ribosomal structure and biogenesis; plum for RNA processing and modification; pink for transcription; deep pink for DNA replication, recombination and repair; hot pink for chromatin structure and dynamics; wheat for cell division and chromosome partitioning; light salmon for nuclear structure; yellow for defence mechanisms; gold for signal transduction mechanisms; pale green for cell envelope biogenesis, outer membrane; spring green for cell motility and secretion; lawn green for cytoskeleton; yellow green for extracellular structures; aquamarine for intracellular trafficking, secretion, and vesicular transport; medium aquamarine for posttranslational modification, protein turnover, chaperones; cyan for energy production and conversion; deep sky blue for carbohydrate transport and metabolism; sky blue for amino acid transport and metabolism; light slate blue for nucleotide transport and metabolism; orchid for coenzyme metabolism; medium orchid for lipid metabolism; dark orchid for inorganic ion transport and metabolism; blue violet for secondary metabolites biosynthesis, transport and catabolism; slate grey for general function prediction only; grey for function unknown; grey for not in COGs; black for tRNA. doi:10.1371/journal.pone.0011455.g002 rearrangements between them ( Figure S1 in supporting information), consistent with the findings of an earlier comparison of partial physical maps of the two species [32].
An exception to the apparent general lack of conservation of gene order between the three Brachyspira genomes was a high conservation of the operons encoding ribosomal proteins. B. pilosicoli 95/1000 contained genes for 33 r-proteins organised in a 18 Kb region, whereas the ribosome clusters in the B. hyodysenteriae and B. murdochii strains included a total of 32 and 36 genes encoding r-proteins organised in 15 Kb and 18 Kb regions, respectively ( Figure 5). The sequences of the B. pilosicoli r-proteins in this cluster were highly similar to the B. hyodysenteriae and B. murdochii homologs. Ribosomal protein genes are interesting because of their ubiquity and similar conservation rate such that horizontal transfer between lineages is unlikely. Apparently there has been a positive selection for clustering of these physically interacting proteins in the sequenced Brachyspira species, whilst there has been no absolute requirement for juxtaposition of other genes in the genomes. Synteny therefore has been lost at a much faster rate than is useful for prediction of gene function. This is consistent with the situation in other bacterial species, where orthologous genes are not necessarily located at the same relative position in the genomes, and only certain gene clusters are syntenic [33]. Genomes of closely related species typically maintain a high degree of synteny [34], whereas genomes of moderately distant species, such as these Brachyspira species, usually have no significant overall synteny [35].

Functional predictions amongst the Brachyspira species
The COG categories of the protein-coding genes are shown in Table 2. Generally there were few significant differences amongst the three genomes, although the smaller genome of B. pilosicoli 95/ 1000 contained more genes than the other two species in five categories: (N) Cell motility, (U) Intracellular trafficking, secretion, and vesicular transport, (C) Energy production and conversion, (H) Co-enzyme transport and metabolism, and (I) Lipid transport and metabolism. On the other hand, B. hyodysenteriae had more genes functioning in (K) Transcription, (V) Defense mechanisms, (T) Signal transduction mechanisms, (O) Posttranslational modification, protein turnover, chaperones, (G) Carbohydrate transport and metabolism, (E) Amino acid transport and metabolism, (P) Inorganic ion transport and metabolism, and (Q) Secondary metabolites biosynthesis, transport and catabolism. B. murdochii had more genes involved with (L) Replication, recombination and repair, and (M) Cell wall/membrane/envelope biogenesis. It was noteworthy that B. pilosicoli had the largest number of genes involved in energy production and conversion, and this capacity may enhance its potential to colonize various hosts.
Previously it has been shown that B. hyodysenteriae lacks many genes for the biosynthesis of small molecules, and therefore must   acquire these from the environment [15]. Interestingly, B. pilosicoli was predicted to contain substantially fewer genes involved in inorganic ion transport and metabolism (P) than both B. hyodysenteriae and B. murdochii (53 versus 74 and 73, respectively; Table 2). Furthermore, it is not known which B. pilosicoli genes may act to compensate for the shortage of such biosynthetic pathways, since the substrates of many of the genes regarded as encoding transporters due to their possession of the motif sequences were unknown. These differences were unexpected, given that B. pilosicoli has a wider host range than the other two Brachyspira species, and hence might be predicted to have a greater metabolic capacity and flexibility in order to survive in these more varied nutritional environments.

Global gene comparisons between the Brachyspira species
As expected, comparative analysis of the B. pilosicoli genome across available microbial genomes in the non-redundant (nr) database at NCBI identified greatest similarities with B. hyodysenteriae and B. murdochii. However, as with B. hyodysenteriae [15], the next highest levels of similarities at the protein level were with Clostridium species (,10%) and E. coli (,5%). The three sequenced Brachyspira species contained a total of 1,087 conserved or ''core'' genes ( Figure 6). It would be instructive to determine whether these same genes are conserved in other Brachyspira species. The majority of genes in B. pilosicoli (1,769, 77%) were conserved in B. hyodysenteriae; therefore, most of the genome-inferred metabolic potential of B. hyodysenteriae described previously can be extrapolated to B. pilosicoli. Only an additional 99 genes were conserved between B. pilosicoli and B. murdochii, whereas an additional 311 were conserved between B. hyodysenteriae and B. murdochii. B. hyodysenteriae had 1,014 genes that were not shared with B. pilosicoli, whereas B. murdochii contained a remarkable 1,900 genes not found in B. pilosicoli. On this basis, B. pilosicoli seemed to be more similar to B. hyodysenteriae than to B. murdochii. A large proportion of the unique genes in both B. hyodysenteriae and B. murdochii were of unknown function (Table 1).
Although the B. hyodysenteriae genome was ,450 Kb larger than that of B. pilosicoli, both species contained roughly the same number of gene across the identified functional gene categories, and only a relatively small number of genes in B. pilosicoli were obviously different from those in B. hyodysenteriae. Of the 525 potential genes that were specific for B. pilosicoli, 319 were of unknown function. The remaining 206 functionally annotated B. pilosicoli-specific genes included those predicted to be involved in energy production, carbohydrate metabolism, amino acid metabolism, capsule biosynthesis (although B. pilosicoli is not known to have a capsule), or encoding transcriptional regulators, transporters and predicted surface antigens, as well as forming a bacteriophage region. Selected differences found in B. pilosicoli that appear to be potentially significant are described below. Bacteriophage genes. Brachyspira pilosicoli 95/1000 contained a 25 Kb region (location 1,595,515-1,615,675 bp) encoding components of an integrated bacteriophage (Figure 7). The 29 associated genes in the region were absent from B. hyodysenteriae and B. murdochii, apart from the OrfG VSH-1 protein homolog that has been associated with a gene transfer agent in B. hyodysenteriae. Sixteen genes were identified as encoding hypothetical proteins, whilst 13 were bacteriophage-associated: putative integrase XerDC family protein (BP951000_1455); OrfG VSH-1 phage protein homolog (BP951000_1459); putative Lys -endolysin; glycoside hydrolase (BP951000_1461); putative RNA polymerase domain-containing protein (BP951000_1463); phage tail tape measure protein homolog, TP901 family (BP951000_1469); bacteriophage TP901-1 like family protein homolog (BP951000_1474); putative DNA packaging, Phage QLRG family protein (BP951000_1476); phage major capsid protein homolog, HK97 family (BP951000_1477); phage prohead protease (BP951000_1478); putative phage portal protein, HK97 family (BP951000_1479); DNA methylase N-4/N-6 domain protein homolog (BP951000_1480); putative phage terminase, small subunit, P27 family protein (BP951000_1481); and putative phage terminase, large subunit (BP951000_1482). The GC content in this bacteriophage region was 29.65%, which was slightly higher than the average for the whole genome (27.90%). It is not yet clear whether this integrated bacteriophage is functional, or capable of transferring genetic material between B. pilosicoli strains. If it is, it could be developed into a tool for genetic manipulation of the spirochete. By analogy with other prophages, it potentially also could be involved in lysogenic conversion, for example modifying the spirochete phenotype to enhance its virulence [36]. It will be important to determine how widespread and conserved this bacteriophage region is amongst B. pilosicoli strains. B. hyodysenteriae WA1 also contained two genes predicted to be bacteriophage-associated: phage terminase large subunit (xtmB, BHWA1_01969) and integrase (BHWA1_02688), but these were distinct from the bacteriophage genes found in B. pilosicoli. A predicted integrase-recombinase protein (4083292.C14.orf00918) was found in B. murdochii, but it only had 25.61% identity at the protein level.
ORF grdX preceded the thioredoxin/glycine reductase gene cluster in both B. pilosicoli and B. murdochii, and it is known to be cotranscribed in C. sticklandii [40]. An additional two copies of trxB were identified in the B. pilosicoli genome (BP951000_1651 and BP951000_1853). Other genes in B. pilosicoli and B. murdochii that were outside the cluster and which encoded proteins with predicted selenocysteine sites were thioredoxin (BP951000_0519 and BP951000_1639; 4083292.C42.orf00619 and 4083292. C42.orf00482) and glutathione peroxidase (BP951000_2065 and BP951000_2066; 4083292.C16.orf00268). Given the potential role of these selenoproteins in antioxidant functions, deficiency of selenium may have detrimental effects on their ability to withstand oxidative stress.
The glycine reductase complex catalyses the reductive deamination of glycine to acetylphosphate and ammonia with the generation of ATP from ADP and orthophosphate [41]. B. pilosicoli and B. murdochii therefore were predicted to be able to utilise glycine anaerobically as a sole source of carbon and energy, and hence have a distinct energy-conserving mechanism using an internal reaction in which glycine serves as an electron donor during oxidation by a glycine cleavage system, or as an electron acceptor when being reduced by glycine reductase. In contrast, the lack of this complex in B. hyodysenteriae suggested that it is unable to ferment glycine to act as a carbon and energy source. The difference between the species suggests that high levels of glycine might favor B. pilosicoli and B. murdochii populations over B. hyodysenteriae, either in vitro or in vivo. The glycine reductase complex is predicted to play an important role in allowing B. pilosicoli to successful colonize a broader range of host species than B. hyodysenteriae, and it also may confer advantages to B. murdochii. Clearly, further research is required to determine whether glycine does enhance growth of B. pilosicoli and B. murdochii, as predicted.
The absence of the grd cluster in B. hyodysenteriae may make it unable to synthesise glutathione and glutaredoxins, and therefore it was predicted to be less able to withstand oxidative stress than the other two species. Only two copies of a gene encoding thioredoxin reductase were identified in B. hyodysenteriae (trxB, BHWA1_02087 and BHWA1_00602). The predicted ability of B. pilosicoli to rapidly respond to an oxidative stress or a redox insult was consistent with its reported ability to survive outside the host for a longer period than B. hyodysenteriae [42,43], and also may explain why it is able survive in more oxygenated host tissues than B. hyodysenteriae, such as in the bloodstream of immunocompromised patients [44]. B. murdochii also has been isolated from extraintestinal sites [45]. This predicted ability to withstand oxidative stress also may make B. pilosicoli more adaptable in terms of its abilities to colonize the large intestines of a wide variety of host species, where environmental conditions are likely to vary.
Sulfatase genes. Three copies of genes encoding sulfatases (BP951000_0858, BP951000_0859 and BP951000_0861) were identified in B. pilosicoli 95/1000, whilst one putative copy was found in B. murdochii (4083292.C42.orf00884), and none in B. hyodysenteriae. Sulfatases play important roles in the cycling of sulfur in the environment, in the degradation of sulfated glycosaminoglycans and glycolipids in the lysosome, and in remodeling sulfated glycosaminoglycans in the extracellular space. The sulfatase genes potentially could encode for enzymes that modify gycosaminoglycans to generate binding sites required for attachment of B. pilosicoli [46].
Fucosyltransferase genes. A novel set of two copies of genes encoding alpha-1, 2-fucosyltransferase (fucT, BP951000_1232 and BP951000_1235) and beta-1, 3-galactosyltransferase (BP951000_1768) were identified in B. pilosicoli 95/1000: these are key enzymes in the biosynthesis of Lewis antigens, structures found on the surface of human erythrocytes and epithelial cells. The gastric pathogen Helicobacter pylori can express Lewis and related antigens in the O-chains of its surface lipopolysaccharide [47], and this activity is believed to be important for bacterial persistence, immune evasion, and possibly generation of inflammation [48]. Further work is required to determine whether B. pilosicoli expresses Lewis antigens, and to investigate their potential involvement in the pathogenesis of infection by this spirochete species.
Sialidase genes. Three copies of a gene encoding a sialidase (neuraminidase) family protein homolog (nanA, BP951000_2021, BP951000_2022 and BP951000_2023) were identified in B. pilosicoli 95/1000, but not in B. hyodysenteriae or B. murdochii. NanA proteins are produced by a wide variety of mucosal pathogens, and are a potential virulence factor in bacteria [49,50]. As they are widespread and conserved among a very broad range of important human pathogens, this implies that they have a critical role in microbial ecology [51]. These enzymes may enhance B. pilosicoli colonization or induce tissue damage.
Aconitase gene. B. pilosicoli and B. murdochii were both found to have a gene encoding aconitase (BP951000_0370 and 4083292.C42.orf00530, respectively), and this was not present in B. hyodysenteriae. Aconitase is a tricarboxylic acid (TCA) cycle protein that catalyses the conversion of citrate to isocitrate, and, amongst other things, has been implicated in the virulence of Staphylococcus aureus [52]. It is possible that the aconitase gene may contribute to the control of virulence factor synthesis [53]. As none of the three anaerobic Brachyspira species contained genes for a complete TCA cycle, this suggests that retention of the aconitase gene may have an important functional significance. Two other genes were located adjacent to the aconitase gene in both B. pilosicoli and B. murdochii, these being putative citrate synthase (BP951000_0368 and 4083292.C42.orf00523) and putative isocitrate dehydrogenase (BP951000_0369 and 4083292. C42.orf00525). They were not identified in B. hyodysenteriae WA1. The presence of aconitase implies that B. pilosicoli and B murdochii can utilise the TCA to produce glutamate, and this capacity is discussed later in the section on central metabolic pathways.

Potential virulence factor screening
Independent of the previous analysis, screening of all coding sequences (CDS) of the three species using Blast, SignalP, PSORT and TMHMM to look for similarities to genes reported to be associated with virulence in other bacteria identified a total of 235 genes with putative roles in virulence in B. pilosicoli 95/1000, compared to 303 in B. hyodysenteriae WA1 and 142 in B. murdochii 56-150 T . The number of genes in the different categories for all three Brachyspira species is shown in Table 3. Interpretation of the results for B. murdochii was complicated by the fact that the genome was incomplete, and additional genes are likely to be identified once a complete genome sequence becomes available. Overall, apart from what is shown, the predicted gene products did not have significant similarities with those of other well-characterized toxins or adhesins described in the major species of enteropathogenic bacteria -such as in E. coli or Clostridium species. Nevertheless, it is important to remember that other virulenceassociated genes are likely to exist amongst those that are currently recorded as encoding proteins that are hypothetical or of unknown function.
It was of interest that B. hyodysenteriae, the most pathogenic of the three species, contained more genes involved with lipopolysaccharide biosynthesis, motility and chemotaxis, and adhesion and/ or surface proteins than did the other two species. The latter two did not have more of any category of these potential virulenceassociated genes, apart from the bacteriophage genes in B. pilosicoli.
Lipooligosaccharides. The outer envelope of the Brachyspira species is considered to contain lipoologosaccharide (LOS) rather than lipopolysaccharide, based on the lack of a typical Gramnegative ladder formation of repeated O-sugar components on silver-stained SDS-PAGE gels [54,55]. LOS is considered to be involved in the induction of colonic lesions associated with B. hyodysenteriae [56,57], and hence should be considered as potentially being involved in virulence in other Brachyspira species. LOS is also antigenic, and protective immunity to B. hyodysenteriae is specific to the LOS serogroup [58]. The importance of LOS in relation to protective immunity to B. pilosicoli is less clear, as pigs experimentally infected with B. pilosicoli 95/1000 did not develop a systemic antibody response against the spirochete [59].
B. pilosicoli contained a set of 25 core genes involved in the biosynthesis of LOS (Table 3). This was fewer than the 30 genes identified for B. hyodysenteriae, but more than the 10 so far identified in B. murdochii. The predicted pathways for biosynthesis of LOS and peptidoglycan for B. hyodysenteriae and B. pilosicoli are shown in Figure 9. The main difference between the species related to the presence of an rfbBADC cluster of genes on the ,36 kB plasmid in B. hyodysenteriae WA1 that encode proteins for nucleotide sugar biosynthesis (dTDP-rhamnose) that are likely to be involved in Oantigen assimilation. A similar rfbBADC cluster was not found in B. pilosicoli or in B. murdochii, implying that they were unable to produce these antigens. Similar rfb gene clusters have been linked to virulence in numerous Gram negative bacteria [60,61], and, for example, are plasmid encoded in the case of some Salmonella enterica serovars [62]. Further work is required to examine the potential role of the rfb cluster in the virulence of B. hyodysenteriae. It was interesting that despite the presence of genes encoding glycosyltransferases in this pathway, B. hyodysenteriae does not appear to produce ladder-like O-antigens [54]. This discrepancy requires further investigation. The most precise way to describe the diversity of LPS/LOS in the Brachyspira species will be to make a detailed comparison of polysaccharide content and structures. It is unfortunate that no such structural data currently are available.  Chemotaxis. The capacity for chemotactic responses and motility are extremely important for Brachyspira species, allowing them to colonize specialized niches in the complex nutritional, physical and microbiological environment of the large intestine. Interestingly, a striking difference was found in the numbers of chemotaxis genes. B. hyodysenteriae contained 52 genes associated with chemotaxis, compared to 27 in B. pilosicoli and 39 in the partial genome of B. murdochii [ Table 3]. The distribution of methyl-accepting chemotaxis genes also was different between the species, with B. pilosicoli having no mcpC genes. These differences are likely to be highly significant in relation to the chemotactic signals that the different species can respond to: they may explain the differences that have been observed in the attraction of B. pilosicoli and B. hyodysenteriae to mucin, and their tendency to occupy different local niches within the large intestine [63]. B. hyodysenteriae colonizes the lumen of the colon, but also penetrates deep into the colonic crypts, where it enters goblet cells. Although B. pilosicoli also can enter the crypts, unlike B. hyodysenteriae it penetrates the dense mucus layer at the crypt shoulders, and attaches to the underlying colonic enterocytes. As B. hyodysenteriae has a much more limited host range than B. pilosicoli, it seems paradoxical that it has a greater number and variety of methylaccepting chemotaxis genes.
Motility. Only minor difference were found in the number and types of flagellar-associated genes between the three Brachyspira strains (Table 3). B. hyodysenteriae WA1 and B. murdochii 56-150 T contained four genes for FlaB core proteins while B. pilosicoli 95/1000 contained three. WA1 and 95/1000 contained three FlaA sheath protein genes while 50-150 T contained two (Table 3). 50-150 T also contained two copies of fleN, encoding flagellar synthesis regulator FleN, while the other two strains had single copies. As expected, the three strains had many minor differences in the predicted sequences of the flagellar proteins. When considering the predicted flagellar structure, it should be remembered that B. pilosicoli only has 4-6 periplasmic flagella at each cell end compared to 7-14 for B. hyodysenteriae [1], and this difference could involve the need for some modifications. Furthermore, differences in flagellin protein ratios can affect the stiffness of the periplasmic flagella in spirochetes, and this stiffness directly correlates with their motility [64]. Hence, depending on expression of the flagellin proteins, the species are predicted to have different capacities for motility.
Lipoproteins and surface proteins. As with other bacteria, predicted surface associated proteins and lipoproteins of the Brachyspira species are likely to be important in interactions with the host; for example, cell surface proteins are thought to be involved with the polar attachment of B. pilosicoli to colonic enterocytes [65]. B. pilosicoli contained substantially fewer genes encoding lipoproteins than did B. hyodysenteriae (13 versus 34). Such lipoproteins, where they are surface associated, are likely to be important as potentially targets for immune recognition and could be used as subunit vaccine candidates [66]. An example of the difference between the two fully sequenced Brachyspira species was the presence of a gene encoding a peptidoglycan-associated outer membrane lipoprotein (PAL, BHWA1_02111) in B. hyodysenteriae that was absent in B. pilosicoli. PAL is involved in maintenance of the integrity of the cell envelope [67], and contributes to bacterial virulence and inflammation in Gram-negative sepsis [68]. Examples of lipoprotein-encoding genes that were shared between the species include those for basic membrane lipoprotein, outer membrane lipoprotein, apolipoprotein N acyltransferase, lipoprotein releasing system, transmembrane protein, LolC/E family, and lipoprotein releasing system ATP binding protein. The species had other predicted lipoproteins, but these were of unknown function.
B. hyodysenteriae has been reported to have a set of variable surface proteins of about 39 KDa [69], and strain WA1 contained nine genes encoding these proteins (VspA to VspF, and VspH), with two copies of VspA and VspD. Genes for VspA to VspF were located adjacent to each other (BHWA1_01456 to BHWA1 _01601), whilst those for one copy of VspA and for VspH were located elsewhere on the genome (BHWA1_00889 and BHWA1_02382, respectively). B. pilosicoli contained single copies of the genes for VspD, VspE and VspH located in different regions, while B. murdochii 56-150 T had three copies of vspG, two of vspF and one of vspG. Differential expression of these genes may be involved in immune evasion, allowing persistence of the spirochetes [69].
Amongst the other outer membrane proteins, B. pilosicoli 95/ 1000 contained three copies of bspA, predicted to encode a BspAlike surface protein. BspA-like proteins are expressed on the surface of many pathogenic bacteria: they may bind to fibronectin, stimulate a significant serological response [70], and induce secretion of IL-8 [71].
Hemolysis-associated genes. B. hyodysenteriae is strongly hemolytic, whereas B. pilosicoli and B. murdochii are weakly hemolytic [1]. The strong hemolytic activity of B. hyodysenteriae is thought to be an important virulence trait [72,73]. In addition to the seven genes associated with hemolysis previously identified in the genome of B. hyodysenteriae WA1 [15], an additional putative hemolysin gene was identified (BHWA1_00962).
Only single copies of tlyA, tlyB, tlyC, and acpP (also known as hlyA [73]) were identified in the partial genome of B. murdochii. On the other hand, B. pilosicoli 95/1000 contained eight genes potentially involved with hemolysis, of which seven were similar to those in B. hyodysenteriae. These encoded hemolysin A (tlyA, BP951000_0123), ATP-dependent Clp protease proteolytic subunit (clpP, also known as tlyB, BP951000_1802), hemolysin C (tlyC, BP951000_1288), hemolysin related protein (hly, BP951000_0473), acyl carrier protein: contains beta-hemolysin (acpP or hylA, BP951000_0533), putative hemolysin III (BP951000_0424), and putative channel protein hemolysin III family protein (BP951000_0888). The eighth was a putative hemolysin (BP951000_2207) that was similar to a protein of Parabacteroides distasonis ATCC 8503, with ,42.9% protein identity. B. pilosicoli lacked a gene with similarity to that encoding the new putative hemolysin of B. hyodysenteriae (BHWA1_00962). From this, it is possible that the new putative hemolysin identified in B. hyodysenteriae is responsible for its strongly hemolytic phenotype. Alternatively, it has been suggested that differences between B. hyodysenteriae and the weakly hemolytic Brachyspira species such as B. pilosicoli in the FabF ACP synthase and FabG ACP reductase proteins that flank the acyl carrier protein (HlyA) may result in different lipid moieties being attached to the HlyA proteins [32]. Acyl carrier protein-dependant fatty acylation is known to be important in activation of prohemolysin to active hemolysin in E. coli [74], and such differences therefore may be important for expression of the strongly hemolytic phenotype in B. hyodysenteriae.
Proteases. B. hyodysenteriae WA1, B. pilosicoli 95/1000 and B. murdochii 56-150 T contained similar numbers of genes predicted to encode proteases (15, 15 and 14, respectively). The proteolytic capacity in all three species was linked with the large number of ORFs encoding enzymes involved in uptake and metabolism of amino acids. These proteases are likely to be required for survival in the intestinal environment, but also may be involved in induction of local destruction of host tissues, and hence contribute to virulence. A serine protease previously described in B. pilosicoli (BP951000_0826, BP951000_1141 and BP951000_2083) [75] was found to have .82% sequence identity to proteases of B. murdochii 56-150 T (ZP_04048074) and B. hyodysenteriae WA1 (YP_002721653).
Secretion systems. As with B. hyodysenteriae [15], B. pilosicoli lacked genes for the specialized secretion systems found in pathogenic Gram-negative bacteria, although it had genes for the Sec pathway and ABC-type transporters. Six sec genes were identified: secA (BP951000_0230), secY (BP951000_1006), secF (BP951000_1532), secD (BP951000_1533), secE (BP951000_1559) and secG (BP951000_1896), and there were 36 genes encoding ABC transporters (compared to 35 and 29 for B. hyodysenteriae and B. murdochii, respectively). Ten flagella-associated genes that can form part of a type III secretary system were found in B. pilosicoli 95/1000, but needle associated genes that encode proteins required for the injection of toxins into the host cell were not identified.

Central metabolic pathways
A reconstruction of the central metabolic pathways of B. pilosicoli showed that it shared many metabolic capabilities with B. hyodysenteriae. The pathways for B. pilosicoli are shown in Figure 10, and can be compared with those for B. hyodysenteriae by examining Figure 2 in reference 15. In both species the glycolysis-gluconeogenesis metabolic axis constituted the backbone of energy production and the starting point of many biosynthetic pathways. The biosynthesis of peptidoglycan, phospholipids, aromatic amino acids, fatty acids and cofactors commenced from pyruvate, or from intermediates in the glycolytic pathway. As with B. hyodysenteriae, a complete set of genes for the non-oxidative pentose phosphate pathway, nucleotide metabolism and a respiratory electron transport chain were identified in B. pilosicoli and B. murdochii. It can be seen from the figures that some of the obvious ways in which B. pilosicoli differed from B. hyodysenteriae related to the presence of pathways associated with citrate and 2oxo-glutarate, a pathway from acetyl-CoA to ethanol, and a lack of the rfbBADC genes located on the B. hyodysenteriae plasmid (predicted to be involved in O-antigen biosynthesis).