24 Jun 2010:
Correction: Identification of Novel Pathogenicity Loci in
Type A Clostridium perfringens causes poultry necrotic enteritis (NE), an enteric disease of considerable economic importance, yet can also exist as a member of the normal intestinal microbiota. A recently discovered pore-forming toxin, NetB, is associated with pathogenesis in most, but not all, NE isolates. This finding suggested that NE-causing strains may possess other virulence gene(s) not present in commensal type A isolates. We used high-throughput sequencing (HTS) technologies to generate draft genome sequences of seven unrelated C. perfringens poultry NE isolates and one isolate from a healthy bird, and identified additional novel NE-associated genes by comparison with nine publicly available reference genomes. Thirty-one open reading frames (ORFs) were unique to all NE strains and formed the basis for three highly conserved NE-associated loci that we designated NELoc-1 (42 kb), NELoc-2 (11.2 kb) and NELoc-3 (5.6 kb). The largest locus, NELoc-1, consisted of netB and 36 additional genes, including those predicted to encode two leukocidins, an internalin-like protein and a ricin-domain protein. Pulsed-field gel electrophoresis (PFGE) and Southern blotting revealed that the NE strains each carried 2 to 5 large plasmids, and that NELoc-1 and -3 were localized on distinct plasmids of sizes ~85 and ~70 kb, respectively. Sequencing of the regions flanking these loci revealed similarity to previously characterized conjugative plasmids of C. perfringens. These results provide significant insight into the pathogenetic basis of poultry NE and are the first to demonstrate that netB resides in a large, plasmid-encoded locus. Our findings strongly suggest that poultry NE is caused by several novel virulence factors, whose genes are clustered on discrete pathogenicity loci, some of which are plasmid-borne.
Citation: Lepp D, Roxas B, Parreira VR, Marri PR, Rosey EL, Gong J, et al. (2010) Identification of Novel Pathogenicity Loci in Clostridium perfringens Strains That Cause Avian Necrotic Enteritis. PLoS ONE 5(5): e10795. doi:10.1371/journal.pone.0010795
Editor: Holger Bruggemann, Max Planck Institute for Infection Biology, Germany
Received: March 24, 2010; Accepted: April 30, 2010; Published: May 24, 2010
Copyright: © 2010 Crown. This is an open-access article distributed under the terms of the Re-use of Public Sector Information Regulations 2005, which permit unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This work was funded by Agriculture and Agri-Food Canada (AAFC) A-base Fund (WBSE# F.1208.FC) (JG and DL), the Ontario Ministry of Agriculture, Food and Rural Affairs (OMAFRA) and the Poultry Industry Council (PIC) (JFP, VRP and DL), United States Department of Agriculture (USDA) grant no. 2007-35204-18362 and Pfizer Animal Health (JGS). Apart from the input provided by ELR in the conception of the research and review of the manuscript, the funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: One of the several funders (Pfizer; E. L. Rosey) had a role in the study design, and in the decision to publish. The authors do not believe that this competing interest affects the objectivity, analyses, interpretations or validity of the study. This does not alter the authors' adherence to all the PLoS ONE policies on sharing data and materials.
Avian necrotic enteritis (NE) is caused by specific strains of Clostridium perfringens and costs the worldwide poultry industry an estimated $2 billion annually, largely due to costs of antimicrobial prophylaxis and inefficient feed conversion , . Elimination of routine antibiotic use (e.g., in the European Union) has been associated with increased incidence of NE. Considerable effort is being invested in research on both pathogenesis as well as immunity to infection as a basis for development of alternate control methods. The alpha toxin (CPA) of C. perfringens was long assumed to play a central role in the pathogenesis of NE, but Keyburn and others  showed that challenge of broiler chicks with a CPA mutant yielded lesions comparable to those in birds inoculated with the wildtype strain. In counterpoint, other studies have revealed that immunization of birds with CPA toxoid protects against virulent C. perfringens challenge , . Furthermore, the intestines of birds challenged with the CPA mutant contained CPA , likely produced by the resident C. perfringens strains and contributing to gut lesions. Subsequently, a new toxin, NetB, was discovered that is related to the pore-forming alpha hemolysin of Staphylococcus aureus and beta toxin of C. perfringens. NetB is common in strains originating from chickens with NE and in those strains, is essential for NE development . Recent work has largely confirmed the critical association of netB with NE strains , although netB-negative C. perfringens may also sometimes produce NE .
Comparative genomics is one approach to determine whether C. perfringens isolates from chickens with NE have a genetic “signature” distinct from either commensal strains, or those causing enteric disease in other hosts. Complete genome sequences are available for a gas gangrene isolate, a soil isolate, and a human food-poisoning isolate of C. perfringens , . Comparison of the three genomes revealed considerable genomic diversity, with more than 300 unique “genomic islands”; furthermore, there is evidence of even greater diversity in other strains . The genetic features in these islands may contribute to the varied disease manifestations that are characteristic of the diversity of C. perfringens as a pathogen . Much of the virulence of toxinotypes A–E of C. perfringens depends upon genes harboured on large plasmids. The enterotoxin gene (cpe) may sometimes be plasmid-borne, and genes for beta2 toxin (cpb2) and other potential virulence determinants – are also found on large plasmids currently being characterized .
In this study, we sequenced multiple C. perfringens NE isolates, and used comparative genomics to identify NE-specific DNA sequences. We found both plasmid-borne and genomic loci containing multiple genes whose predicted products shared high similarity to bacterial virulence and virulence-associated factors.
Sequencing and de novo assembly of C. perfringens genomes
Draft genome sequences for eight type A C. perfringens poultry isolates were generated via 454 and Illumina/Solexa sequencing technologies. Seven of these strains (CP4, JGS4143, JGS4140, JGS1521, JGS5252, JGS5621, JGS1651) were recovered from NE field cases from Canada and the US, and one (JGS1473) from the intestinal tract of a healthy bird (Table 1). The virulence of each NE isolate was subsequently confirmed in an experimental model (data not shown); two strains, CP4 and JGS4143, have been more extensively characterized , , –. Antibiotic resistance profiles of all strains revealed variable resistance to bacitracin and tetracycline (Table 1).
The Roche 454 GS FLX system was used to generate sequences for seven of the strains, which were then assembled with the Newbler assembler (Roche). One strain (CP4) was sequenced in a single-end (2 lanes) and a paired-end (1 lane) run with the Illumina/Solexa Genome Analyzer, and the reads were assembled de novo with Velvet . Both sequencing approaches yielded assemblies of similar quality (Table 2). The 454 sequencing runs generated average coverages ranging from 13.3× to 32.7× based on an estimated 3.2 Mb genome, while the combination of Solexa runs resulted in 262× coverage. The total number of contigs and N50 values associated with these draft genomes were comparable to that of the six publicly available draft C. perfringens genomes generated via a traditional WGS sequencing approach (Table 2).
Optical mapping was performed on CP4 genomic DNA to provide independent verification of the assembly and assist in contig scaffolding. The CP4 genome size estimated from the optical map was 3.4 Mb, while the combined length of the contigs aligned to the map was 3.26 Mb, indicating that ~96% of the genome was correctly accounted for in this draft (Figure S1). In addition, ~375 kb of sequence not aligned to the optical map, distributed into 37 contigs, was presumably composed of mainly extrachromosomal sequences.
Identification of NE-associated loci
We used two complementary approaches to identify and confirm sequences that were both conserved among the various NE strains but absent or significantly divergent from the nine publicly available C. perfringens strains (ATCC 13124, SM101, Str. 13, NCTC 8239, F4969, ATCC 3626, JGS1495, JGS1721, JGS1987) (Table 1). In the first approach, the sequence alignment algorithm Mummer  was used to identify sequences unique to each NE strain by aligning the draft genomes of CP4 and JGS4143 with the nine reference genomes. Unique sequences from the two NE strains were then aligned with each other to identify those conserved between them, and the ORFs present in these unique regions were retrieved. CP4 and JGS4143 were found to contain 435 and 173 genes not found in any of the reference strains, of which 31 were common to both NE strains. The novelty of these ORFs among the NE strains was further confirmed using OrthoMCL , which predicts orthologs based on clustering of reciprocal best BLAST hits.
In the second approach, the Rapid Annotation using Subsystem Technology (RAST) system was used to identify NE-specific sequences . The C. perfringens draft genome sequences were uploaded to the RAST server and automatic annotation was performed. GenBank files of the nine publicly-available C. perfringens genomes were also uploaded using original gene coordinates and names. RAST's sequence-based comparison tool was used to compare NE strain JGS4140 to all other C. perfringens strains. Protein identity values from the comparison were tabulated and genes conserved in the NE strains (>95%) but not in the non-NE strains (<90%) were filtered into a new table. Sequences of contigs containing NE-specific genes were further analysed using the BLAST algorithm  to confirm uniqueness to the NE strains.
The majority of NE-specific genes identified using the methodology above were clustered on a small number of contigs, indicating that these genes were not dispersed randomly throughout the genome. A hybrid assembly was thus generated using the contigs from multiple NE strains to determine their absolute location and contiguity, and this revealed three unique DNA segments. First, a 46.6 kb scaffold was generated by this approach consisting of 37 genes, 25 of which had no predicted orthologs in the nine reference genomes, as determined by OrthoMCL (Table 3). The integrity of this scaffold was experimentally verified in two strains, CP4 and JGS4140, by closing all gaps and verifying anomalous regions through sequencing of PCR products. This resolved a novel ~42 kb “pathogenicity” locus, which we designated NE locus 1 (NELoc-1). This locus harboured netB, as well as genes for other predicted proteins likely involved in virulence (Figure 1A and described below).
The genetic organization of (A) NELoc-1, (B) NELoc-2 and (C) NELoc-3 is shown, each arrow representing a predicted gene and the total size given below each locus. Predicted functional annotations and locus tags are shown above and below each gene, respectively. Genes are colour-coded by their putative role based upon sequence analyses.
Second, another locus of 11.2 kb, NELoc-2 (Table 4; Figure 1B) was identified that consisted of 11 contiguous genes that were absent from eight of the nine reference genomes. This locus was found intact on contigs ranging in size from 22 kb–248 kb in five of the NE strains, while in the other three strains, it was separated into multiple contigs separated by gaps of 52 bp–162 bp, based on the sequence of strain CP4. This locus was also found in the non-NE, ovine enteritis strain JGS1721, in contigs of 11 kb and 120 kb. NELoc-2 shared >99.8% nucleotide identity among all NE strains while that of the JGS1721 strain was 99.1% identical.
Third, a conserved 5.6 kb NE-specific locus (Table 5; NELoc-3) was identified that harboured five genes, three of which were either absent or significantly divergent from equivalent regions in the reference genomes (Figure 1C). This locus was found on a single contig ranging from 7.4 kb–9.9 kb in four of the NE strains, and on multiple contigs separated by small gaps, based on the CP4 sequence, in the remaining strains.
Features of NE-associated loci
Sequence analyses of the genes in the three loci were performed using BLAST, the Conserved Domains Database (CDD), , SignalP , pSortB , and InterProScan  (Tables 3–5). As noted, NELoc-1 is the largest locus of the three, consisting of 37 predicted genes (CP4_3442–CP4_3478); four of which (CP4_3447, 3448, 3460 and 3476) had predicted proteins <100 amino acids (aa) and no significant BLAST hits in GenBank;these were thus excluded from further analyses.
NELoc-1 bears several hallmarks of a horizontally-acquired element. Specifically, we identified two degenerate transposases: one truncated ORF (CP4_3451) with 59% identity to the Tn1546 transposase  and another two ORFs (CP4_3464 and CP4_3465) that, when joined, produce a contiguous alignment (85% identity) with a resolvase/recombinase. CP4_3452 and CP4_3453 also appeared to have been originally joined into a single ORF. The predicted full-length protein encoded by these two ORFs shared 92%–95% identity with a putative transposase/invertase-encoding gene found in the conjugative plasmids pCW3, pCPF4969, pCPF8533etx and pCPF5603. Furthermore, CP4_3477 encoded an apparently truncated swim zinc finger domain protein also present in plasmids pCP8533etx and pCPF5603. Taken together, the presence of these sequences suggested that this locus might be associated with a transposable element and/or plasmid.
Fifteen of the NELoc-1 genes were predicted by pSortb and SignalP to encode cell-surface associated or extracellular proteins, indicating that nearly half of the predicted protein complement of this locus may be exposed to the extracellular environment. Three genes found clustered together near the 5′ end of NELoc-1 (CP4_3471, CP4_3472 and CP4_3473) encode predicted proteins with 66%–76% identity to putative cell surface bound proteins present in several of the C. perfringens draft genomes, and/or contain a Cna-like B-region domain. This domain is found in the Staphylococcus aureus collagen-binding protein Cna and serves to present the collagen-binding domain away from the cell surface . Located immediately downstream from this cluster is a gene (CP4_3470) encoding a putative sortase, an enzyme responsible for the covalent anchoring of surface proteins to the peptidoglycan layer . Bordering these four predicted surface-associated proteins are two genes (CP4_3478 and CP4_3469) that constitute a putative novel cyclic dimeric guanosine-monophosphate (c-di-GMP) signalling system . Specifically, CP4_3478 encodes a predicted protein containing both GGDEF and EAL domains, responsible for the catalytic activity of diguanylate cyclase (DGC) and phosphodiesterase (PDE) enzymes, respectively , while CP4_3469 encodes a protein containing the GGDEF domain only. This system regulates virulence and adhesion in several bacterial pathogens, including Vibrio cholerae, Pseudomonas aeruginosa and Staphylococcus aureus by modulating intracellular levels of the c-di-GMP second messenger molecule. , .
Just 507 bp downstream from CP4_3469 is a gene that is closely related (64% identity over 1521 aa) to the hypothetical protein CPE1281 of C. perfringens strain 13, and more distantly to the discoidin domain protein CPF_1073 (44% identity over 1413 aa) of C. perfringens strain ATCC13124. Kulkarni et al.  identified a CPE1281 homolog that was recognized by serum from immune broiler chickens. Further analysis of the original protein ID data (obtained from Kulkarni et al) showed that four of the peptides were identical to the predicted protein from NELoc-1 and divergent from the other two chromosomally-encoded homologs. Only one peptide (ITDQNIWENNTYPK) was identical to CPE1281, and this sequence was also shared by the NE homolog (Figure S2).
Of particular interest in NELoc-1 were the two putative leukocidins situated near the middle of the locus (CP4_3459, CP4_3458), which shared 44% and 50% amino acid identity with NetB, as well as similarity with other beta-channel pore-forming toxins that carry the leukocidin domain, including the C. perfringens delta-toxin, C. botulinum alpha-hemolysin, and the S. aureus HlyA and bi-component leukocidins , . However, the CP4_3459 and CP4_3458 predicted proteins were only 96 and 114 residues, respectively, while the size of the related pore-forming toxins ranged from 271 (S. aureus PVL toxin) to 322 (NetB) amino acids. Altering a single nucleotide to eliminate the intervening stop codon between the two ORFs resulted in a single ORF encoding a predicted 253 amino acid protein which then yielded a contiguous, nearly full-length alignment with NetB. Thus, these two ORFs likely originated from a single leukocidin gene that subsequently acquired a nonsense mutation.
In addition to these putative toxins, a ricin-domain containing protein (CP4_3450) and an internalin-like protein (CP4_3446) were located near the 3′ end of NELoc-1, flanking netB (CP4_3449). It was recently reported that netB contains an upstream VirR-box regulatory sequence, and is under direct control of the VirR response regulator . The VirR/VirS two-component signal transduction system regulates several virulence genes in C. perfringens, including alpha-toxin, perfringolysin O and alpha-clostripain . By integrating the new sequence information from the element upstream from netB, a modified VirR-binding consensus sequence was designed and used to identify additional putative VirR/VirS-regulated genes in the CP4 genome (Figure 2A). One novel putative VirR-box was found 67–69 bp upstream from the putative internalin in all seven NE strains, suggesting that this gene is co-regulated with netB by the VirR/VirS system (Figure 2B).
(A). A modified VirR-box consensus sequence was produced based on the alignment of known VirR-boxes with the VirR-box upstream from netB. The VirR-box consensus previously reported by Cheung et al. (2004)  is shown in bold, and absolutely conserved nucleotides are in boxes. (B) VirR-boxes identified in CP4, including one novel element upstream of the internalin-like protein gene. Nucleotides that conform with the previously reported VirR-box consensus are in green, and those that do not conform in red.
Downstream of the two putative leukocidins (CP4_3459, CP4_3458) were two putative chitinases, which catalyze the breakdown of chitin, a long-chain homopolymer of N-acetylglucosamine. Finally, four additional genes were found at the 3′ end of the locus: two putative transcriptional regulators (CP4_3444 and CP4_3442), a radical SAM-domain protein (CP4_3445), and a metallo-beta-lactamase domain protein (CP4_3443). The metallo-beta-lactamase domain-containing protein superfamily contains ~6000 members with diverse enzymatic functions, including glyoxylases, flavoproteins, and arylsulfatases, in addition to the beta-lactamases . Antibiotic susceptibility tests revealed sensitivity to beta-lactam antibiotics (Table 1), suggesting the protein encoded by this gene does not possess beta-lactamase activity.
Most genes identified in NELoc-2 are predicted to encode membrane or cell wall associated proteins. One gene cluster of interest was the predicted VTC domain – tubulin – resolvase – VTC domain – tubulin sequence (CP4_0460–CP4_0464). BLAST analyses of the regions flanking the resolvase failed to reveal a repeat region between the two genes that would have been essential for transposition. Also of particular interest was the gene encoding spore coat protein CotH, which may serve as a general marker for C. perfringens enteritis strains, since it is also found in the ovine enteritis strain JGS1721. A chitin synthase-encoding ORF may functionally complement two chitinase-encoding ORFs found in NELoc-1; however its relevance is unclear since information about chitin synthases is limited to yeast and higher eukaryotes, where the balance between synthesis and degradation of chitin is essential for cytokinesis .
BLAST analyses of NELoc-2 flanking regions indicated this locus is chromosomally located, and identified a 1374 bp region corresponding to genome coordinates 447880–449255 of C. perfringens strain ATCC 13124 that had been replaced with the 11.2 kb sequence of NELoc-2. Strain 13 and SM101 BLAST results revealed the replacement of smaller regions of 165 bp and 168 bp, respectively, at the same location in the genome. The presence of a resolvase gene (CP4_0462) in the centre of this locus suggested that it may have originated from a mobile element.
Analyses of NELoc-3 revealed a putative hdhA gene (CP4_3572) that encodes 7-alpha-hydroxysteroid dehydrogenase (7-HSDH), a bile-acid inducible enzyme that converts the primary bile acid, chenodeoxycholic acid, into a secondary potentially toxic bile acid, 7-keto-lithocholic acid. A homolog has also been found in Clostridium difficile, and its presence may confer a survival advantage through the suppression of commensal bacteria by the HSDH-mediated production of a toxic metabolite . Another gene of interest was that predicted to encode an extracellular solute binding protein (ESB) (CP4_3571). The predicted protein was found to be 100% conserved among NE strains, compared to a maximum of 78% identity to the C. perfringens strain ATCC 13124.
The predicted protein encoded by CP4_3570 shared significant similarity with hypothetical proteins pCW3_0046 (79% identity) and pCPF5603_58 (76% identity) encoded by C. perfringens plasmids pCW3 and pCPF5603, respectively. This finding, along with the identification of a putative resolvase (CP4_3569) in this locus, suggested NELoc-3 may have arisen from a transposable element and is plasmid-encoded. Unlike NELoc-2, we could not identify flanking genomic sequences on any contigs carrying NELoc-1 or -3, and neither locus aligned with the optical map generated for the CP4 chromosome.
Characterization of large plasmids of NE strains
The identification of plasmid sequences in NELoc-1 and -3 prompted efforts to identify and characterize the plasmids of our NE strains. A number of C. perfringens toxin genes are associated with large plasmids, including beta-toxin (cpb), epsilon toxin (etx), beta2-toxin (cpb2), iota toxin (iap/ibp) and, variably, C. perfringens enterotoxin (cpe) , , –. Recent characterization of toxin plasmids of C. perfringens type A, B, D and E strains – have revealed a toxin plasmid family, designated the pCPF5603-like toxin plasmids , that share a similar structure; this consists of a conserved backbone region of ~30 kb carrying the transfer of clostridial plasmid (tcp) locus for conjugative transfer, and a ~20–40 kb variable region that encodes toxins , , . The complete sequences of three members of this family (pCPF5603, pCPF4969, pCP8533etx) have been determined , . In addition, the tetracycline resistance plasmid, pCW3, is closely related to this plasmid family, sharing the conserved core but carrying a resistance locus in place of the toxin-encoding loci , , .
To first determine if our NE strains carried large plasmids, DNA from our eight sequenced poultry isolates, as well as three additional virulent NE isolates (CP1–3) and one avirulent NE isolate (CP6), were subjected to PFGE. In silico restriction endonuclease analysis of the genomes of SM101 and ATCC13124, as well as pCPF5603, pCPF4969 and pCP8533etx, revealed that NotI cleaved the genomes at no more than one location, whereas the plasmids were cleaved exactly once; this restriction enzyme was therefore chosen to linearize the plasmids prior to PFGE. The PFGE profiles of the virulent NE type A strains digested with NotI revealed two to five large plasmids ranging in size from ~45 kb–90 kb in all strains (Figure 3A). Only one NE strain, JGS5252, carried just two large plasmids, as did the healthy chicken isolate, JGS1473, and the avirulent NE isolate, CP6. Three of the NE strains (CP2, CP4 and JGS5621) also carried a larger plasmid of ~130 kb–150 kb. Further PFGE experiments using undigested DNA, which allows plasmid but not chromosomal DNA to enter a pulsed field gel , confirmed the PFGE results (data not shown).
(A). Agarose plugs containing DNA isolated from the eight sequenced poultry strains were digested with NotI and subjected to PFGE. (B). Southern blotting was performed with DIG-labelled probes for netB and hdhA. Results from both netB and hdhA probes are shown overlayed. In all lanes with two bands, the upper band represents netB and the lower band hdhA. Both probes hybridized to the same band in CP2. In CP6, only the hdhA probe hybridized, while in JGS4143, only the netB probe hybridized.
To determine if the identified plasmids were related to the pCPF5603-like plasmids of C. perfringens, digoxygenin (DIG) -labelled probes for three genes (cna, dcm, tcpF) conserved among all sequenced plasmids of this family were used in Southern blotting experiments of the pulsed-field separated plasmids from five NE strains (CP1–4, CP6). We observed that when NotI-digested genomic DNA was probed, all bands, with the exception of the larger ~150-kb bands, hybridized specifically with one or more of the cna-, dcm-, or tcpF-specific probes, confirming the plasmid identity of these PFGE bands (data not shown). In particular, the tcpF-probe, which is specific for the tcp conjugation locus, hybridized to all the ~45 kb–90 kb plasmids in each strain examined, indicating that they carry this locus and may therefore be conjugative.
Surprisingly, many genes that make up the conserved region in this plasmid family, including constituents of the tcp locus, could not be found in the CP4 draft genome. Repetitive sequences can confound assembly algorithms, so we reasoned that these sequences might be misassembled if they were harboured on multiple plasmids in a single C. perfringens strain. Mapping of CP4 Solexa reads to the three sequenced plasmids using the reference assembler, Maq (http://maq.sourceforge.net), revealed that CP4 does in fact carry sequences corresponding to all of the conserved regions, including the tcp locus (Figure S3). This provides further evidence that the plasmids carried by our NE strains are closely related to the previously characterized C. perfringens toxin-encoding plasmids.
Plasmid localization of NELoc-1 and 3
To test the hypothesis that NELoc-1 and -3 were located on one of the large plasmids identified in the NE strains, DIG-labelled PCR probes based on netB (NELoc-1) and hdhA (NELoc-3) were used in Southern blotting experiments to probe membranes of transferred PFGE-separated plasmids from all 12 poultry strains. Hybridization of the netB-probe to a single ~80 kb–90 kb plasmid was observed in all NE strains, but not in the avirulent strains CP6 and JGS1473 (Figure 3B). The hdhA probe hybridized to a different plasmid of ~70 kb–80 kb in nine of the ten virulent NE strains, as well as CP6, but not in the healthy chicken isolate JGS1473. Five of the strains (CP1–4, CP6) were also examined in Southern blotting experiments with a probe based on cpb2, and an identical hybridization pattern to that of the hdhA probe was observed in these strains (data not shown). Interestingly, the netB, hdhA and cpb2 probes co-hybridized to the same ~80 kb band in strain CP2, suggesting that NELoc-1 and -3 may either be on the same plasmid in this strain, or alternatively, on separate but similarly sized plasmids that are not resolvable under the PFGE conditions used here. Thus, NELoc-1 and -3 are harboured on separate large plasmids in most NE strains, and the latter plasmid also carries cpb2.
We next sought to determine the insertion sites of NELoc-1 and -3 within their respective plasmids. Outward-facing PCR primers were designed at both ends of NELoc-1 based on the CP4 sequence, which were used in combination with primers designed from various genes common to pCPF5603 and pCPF4969 to amplify flanking junction regions via long-range PCR. A 9 kb product was amplified using primers sigP-F and dcm-F, linking the signal peptidase I (CP4_3474) with dcm. At the opposite end of NELoc-1, use of primer pair blac-F and lexA-F amplified a 2.7 kb fragment, thereby linking this end to the lexA repressor gene. These PCR products were subsequently sequenced and used to extend the CP4 NELoc-1 sequence at both ends. Alignment of this extended locus to the three sequenced pCPF5603-like plasmids revealed a structure and insertion site most similar to the variable region of pCP8533etx, which carries both a cpb2 and etx locus (Figure 4A). In pCP8533etx, the etx locus can be delineated from the rest of the variable region by three genes that are directly repeated on either end; these genes are present as single copies in pCPF5603, pCPF4969, and pCW3. A similar duplication is observed in NELoc-1, effectively delimiting this locus into one ~30 kb unit corresponding to the etx locus, extending from the DGC to chitinase B- encoding genes, and another ~12 kb unit corresponding to the cpb2 locus, extending from the Tn1546-like transposase to the MarR transcriptional regulator-encoding genes, and including netB.
Nucleotide sequence alignments of pCPF5603, pCP8533etx and (A) NELoc-1 or (B) NELoc-3 were generated with Artemis Comparison Tool (ACT). The small boxes represent ORFs, with toxin genes colored in red and IS1151 or IS1469 genes in yellow. Grey boxes connecting the different sequences represent regions with sequence similarity. Colored bars underneath each sequence represent different loci as follows: blue, tcp; green, cpe; red, cpb2; black, pCPF5603-specific; light blue, etx. Orange bars represent the sequence found duplicated in pCP8533etx and NELoc-1 and -3 sequences.
For NELoc-3, sufficient flanking plasmid sequence was present in the contig from JGS4140 to define its putative insertion site (Figure 4B). This insertion site also corresponded to that of the etx locus of pCP8533etx and was bordered by the same direct repeats described above. Notably, neither locus contained IS1151 or IS1470-related sequences, in contrast to previously described loci containing cpe, etx, iap/ibp and cpb , , , .
Diversity analysis of the three NE loci among poultry strains
Multiple alignments of each locus from our seven sequenced NE strains revealed a striking level of conservation among these sequences between the different strains (Figure S4). To exclude the possibility that this conservation was due to a clonal relationship, PFGE profiles of genomic DNA were determined for the eight sequenced strains (Figure S5A). In addition, a phylogenetic tree based on the alignment of the total predicted proteome sequence of each strain was constructed via CVtree  (Figure S5B). Three of the strains (JGS4140, JGS1651 and JGS5621) were found to be closely related by both methods, but none appeared to be clonally related and the remaining five strains were significantly divergent.
In most of the sequenced strains, the three NE loci were fragmented into multiple contigs and it was therefore not possible to assess their uniformity based on sequence data alone. An overlapping PCR assay was therefore developed  to assess the diversity of the three loci in our sequenced strains and the four additional poultry isolates, and at the same time confirm the sites of insertion. A set of fifteen PCR reactions were developed for NELoc-1, five reactions for NELoc-2 and four reactions for NELoc-3, all based upon CP4 and JGS4140 sequences and spanning each entire locus. The reactions at either end of each locus were designed to extend into the flanking sequences to verify the site of insertion (Figures S6, S7, and S8).
PCR reactions for NELoc-1 amplified most of the expected products from the ten virulent NE strains (Figure S6), indicating near-uniformity of this locus across these strains. In CP1, reaction NEL1-6, encompassing leukocidin I and II, was ~1.2kb larger than expected, and sequencing of this amplicon revealed an additional ~1kb ORF encoding an intact ISCpe7 transposase upstream from leukocidin I. Four of the reactions for CP2 were negative, indicating that the region encompassing the chitinase and leukocidin genes was absent in this strain. Two of the reactions from CP6, spanning the chitinase and leukocidin genes and one reaction from JGS1473, encompassing the CPE1281 homolog, amplified the expected product. The latter result was consistent with the findings from the comparative genome analyses, which identified a region from NELoc-1 to be partially intact in JGS1473, consisting of the sortase and cell-wall surface protein-encoding genes, in addition to the CPE1281 homolog. This suggests that portions of NELoc-1 may be present in these avirulent strains, possibly on a modified netB plasmid, or chromosomally located.
The pathogenesis of poultry NE has been the subject of considerable investigation in recent years, following reassessment of the role of CPA and subsequent identification of the novel pore-forming toxin NetB , . We show here for the first time that netB resides on a ~42 kb plasmid-encoded pathogenicity locus (NELoc-1) harbored specifically by NE strains. In addition, we identified two other loci associated with NE, NELoc-2 and -3, the latter of which also resides on a plasmid that is similar to, but distinct from, that carrying netB. Sequence alignments of NELoc-1 to -3 from seven strains revealed their striking uniformity among diverse NE isolates (Figure S4). The high level of sequence conservation was surprising and suggests that these plasmids arose from a recent evolutionary event. Southern blotting results with a tcpF probe, together with the detection of the tcp locus sequence in the CP4 Solexa reads, indicate that the plasmids carrying NELoc-1 and -3 also carry this conjugation locus. The observation that these unrelated NE strains carry nearly identical plasmid loci is consistent with the hypothesis that the plasmids were acquired through conjugative transfer. This study provides further evidence of the remarkable contribution of closely related multiple plasmids to the infection biology of C. perfringens in its ability to cause distinct and serious enteric infections in different animal species –, 
While functional studies are ultimately required to elucidate the contribution of these genes to NE pathogenesis, sequence analysis can assist greatly in this effort. NELoc-1 encodes 37 putative proteins, of which 25 have no apparent counterpart in any of the sequenced C. perfringens strains. An internalin-like protein is located immediately upstream from netB containing a leucine-rich repeat (LRR) domain (CP4_3446) with 30–35% identity to Internalin F, D and A of Listeria monocytogenes and putative internalins of Bacillus spp.. Originally identified in L. monocytogenes as surface-anchored proteins required for attachment and invasion , internalins now constitute a multi-gene family, with 25 members identified in the L. monocytogenes genome alone . Genome sequencing projects have also revealed internalin-like proteins in C. botulinum, C. perfringens and C. tetani, though none have been functionally characterized , . The hallmark LRR domain, responsible for the horseshoe-like conformation of the protein, is found in a functionally diverse set of proteins, and is typically involved in protein-protein interactions . The presence of a VirR-box upstream of the gene suggests that it may be co-regulated with netB, and therefore also involved in NE pathogenesis.
Immediately downstream from netB is a gene encoding a 391 aa protein (CP4_3450) containing a beta-trefoil domain similar to that of the carbohydrate-binding ricin B subunit . BLAST searches revealed similarity to the ricin-like domains of a diverse set of proteins, including pierisin from Pieris rapae (27% identity over 334 aa) and MTX from Bacillus sphaericus (30% over 286 aa), which are related ADP-ribosylating toxins , the Cyt1Ca insecticidal pore-forming toxin from Bacillus thuringiensis (28% over 264 aa) , and the HA-1 non-toxin hemagglutinin of C. botulinum (33% identity over 279 aa) , . These proteins share a similar secondary structure consisting of the C-terminal ricin-like domain responsible for carbohydrate binding and, with the exception of HA1, a distinct N-terminal active-site domain. The ~130 aa N-terminal region predicted in CP4_3450 did not share sequence similarity with any protein found in GenBank, and a putative function could therefore not be assigned.
CP4_3446, netB and CP4_3450 are oriented divergently from the majority of the other genes in NELoc-1. Given this common orientation and evidence for co-regulation, it is plausible that these three genes are co-transcribed, and may act in unison. One possible scenario is that CP4_3446 and CP4_3450 act directly with NetB to enhance its pore-forming activity, although this seems unlikely as the cytotoxicity of purified recombinant NetB was found to be comparable to that of the native protein . Alternatively, they may instead be involved in recognition of eukaryotic cell surface receptors, as suggested by the presence of carbohydrate- (CP4_3450) and protein- (CP4_3446) binding domains and similarity to proteins known to function in this capacity.
The presence of two putative chitinases (CP4_3454 and CP4_3455) is intriguing and suggests a possible role for chitin hydrolysis in NE pathogenesis. This abundant polysaccharide is an insoluble homopolymer of N-acetylglucosamine (GlcNac) found in the exoskeletons of invertebrates and the cell walls of fungi. Chitinases have been detected in other clostridial species, including C. paraputrificum ,  and C. botulinum , both of which have chitinolytic activity, but to our knowledge this is the first example of a chitinase gene identified in C. perfringens. These enzymes are typically produced by bacteria that use chitin as a carbon source, or are pathogens of insects or fungi . The incorporation of chitin in the diets of chickens has been shown in some cases to improve growth performance , . The putative chitin synthase (CP4_0468) in NELoc-3 may work in conjunction with the two putative chitinases (CP4_3454 and CP4_3455) in NELoc-1 to take advantage of the chitin as an extra carbon source. The degradation of chitin into GlcNac may also confer a survival advantage to the NE strains since it inhibits the adhesion of probiotic lactobacilli to chicken intestinal mucus . Chitin synthase may be needed to convert excess GlcNac back into chitin if the need arises. Intriguingly, in the intestinal bacterial pathogen V. cholerae the product of a putative chitinase gene with GlcNAc-binding activity has been shown to act as a common adhesion molecule for both chitinous and intestinal surfaces . GlcNAc forms the bulk of the carbohydrates present in intestinal mucin and this protein has been shown not only to be involved in intestinal colonization in mice but also to increase the quantity of intestinal mucin produced , suggesting a possible role for putative chitinases in intestinal colonization in NE.
Also of interest were two putative leukocidins (CP4_3458 and CP4_3459) that exhibited close similarity to NetB, and to a lesser extent to the related C. perfringens pore-forming beta and delta toxins , hemolysin II of Bacillus spp., and the bi-component leukocidins of S. aureus. The two genes are in the same reading frame; elimination of the intervening stop codon forms a single gene with an intact leukocidin domain, and the predicted protein forms a nearly full-length alignment with NetB. These genes may therefore represent an evolutionary relic of a previous netB duplication event, or even a now-defunct second member of an original bi-component hetero-oligomeric pore-forming toxin.
The discovery of a potential novel c-di-GMP signaling system, composed of putative PDE (CP4_3469) and DGC (CP4_3478) genes, as well as several nearby genes for putative cell surface proteins (CP4_3471, CP4_3472 and CP4_3473) was of particular interest. This signaling system has recently been shown to play a central role in governing adhesion, motility, and virulence in pathogenic bacteria such as P. aeruginosa, Salmonella Typhimurium, S. aureus and V. cholerae , , , . The c-di-GMP nucleotide acts as a second messenger by binding and regulating downstream effectors that are only now being elucidated . Intracellular c-di-GMP levels are modulated through the opposing activities of a DGC and PDE, responsible for its synthesis and degradation, respectively . Both CP4_3469 and CP4_3478 contain the GGDEF domain (responsible for DGC activity), while CP4_3469 also contains an EAL domain (responsible for PDE activity). It is therefore likely that CP4_3478 functions as a DGC, while CP4_3469 may have both PDE and DGC activities, although it is common for PDEs to also carry a non-functional GGDEF domain . CP4_3478 also contains a signal peptide and several N-terminal transmembrane regions, making it a good candidate for an integral membrane DGC. The cluster of four surface-associated protein genes found between CP4_3469 and CP4_3478 may encode adhesins under the control of this system. Two of the genes were also found in avirulent strain JGS1473, however, suggesting a limited role in virulence. It is well-recognized that large numbers of C. perfringens will coat the damaged intestinal epithelial surface of chickens with NE , but little to no information exists on the specificity of this adherence.
The recognition of a homolog of the large hypothetical protein CPE1281 from Strain 13 was among many of the unexpected discoveries reported here. Earlier work based on mass spectrometry of proteins expressed by virulent but not avirulent C. perfringens recovered from birds with NE  mis-identified the protein as CPE1281, instead of the NE-specific homolog. Immunization of chickens with the most immunogenic part of the CE1281 protein gave excellent protection of birds against experimentally-induced NE  Epitope-mapping of the CPE1281 protein (not the NE 1281 homolog described here) identified immunodominant regions of the CPE1281 protein, which include the possible zinc-binding signature region of this hypothetical protease in both proteins (G—HELGHNF), which were used in immunization when expressed from a Salmonella vaccine vector . Further work is required to determine whether the efficacy of this protein in protecting birds against NE is the result of cross-protection against the NE CPE1281 homologue.
Two homologs of tubulin/FtsZ (CP4_0461 and CP4_0464) are found in NELoc-2 sharing 33% and 36% protein identity, respectively, with tubulin/FtsZ common to other C. perfringens strains. The homologs might be similar in function to the Bacillus thuringiensis tubulin homolog TubZ, which is proposed to facilitate plasmid segregation by coupling with plasmid DNA, via DNA-binding proteins, and causing plasmids separation . In this case, NELoc-2 may not be specifically involved in virulence, but might instead contribute to the preservation of the NELoc-1 and -3 virulence plasmids via the two tubulin/FtsZ homologs. This may help explain the frequency of plasmid occurrence in NE stains from different sources.
Extracellular solute-binding proteins (ESBs) of bacteria may serve as chemoreceptors, recognition constituents of transport systems, and initiators of signal transduction pathways, suggesting several possible roles for the conserved ESB gene in NELoc-3 . The NADP-dependent 7-alpha-hydroxysteroid dehydrogenase found in NELoc-3 shares 64.3% protein identity with the 7-HSDH of C. difficile. It may similarly be responsible for converting the primary bile acid, chenodeoxycholic acid, into the secondary bile acid, 7-keto-lithocholic acid, which is potentially toxic to other members of the microbiota ., though it is not clear how NE strains are protected.
PFGE analyses revealed the presence of two to five plasmids, ranging in size from ~45 to 150 kb, in ten NE isolates of known virulence. Southern blotting of pulsed-field gels revealed netB (NELoc-1) on a ~80 kb–90 kb plasmid in virulent isolates and hdhA (NELoc-3) on a separate ~70 kb–80 kb plasmid. Only in CP2 were netB and hdhA apparently co-localized to the same plasmid. Almost all C. perfringens toxins are now known to be transcribed from genes on large plasmids. These plasmids share a conserved backbone carrying a tcp conjugation locus, and it has been suggested that they be referred to as the pCPF5603-like toxin plasmids . Alignment of NELoc-1 and -3 and flanking sequences with members of this family reveal a shared structure and site of insertion, most closely resembling pCP8533etx, suggesting that the NELoc-1 and -3 plasmids are genuine members of this growing plasmid family. Toxin genes present in other pCPF5603-like toxin plasmids, including cpe , etx  and iap/ibp , are in close proximity to IS1151 sequences, and the detection of circular transposition intermediates carrying both IS1151 and toxin gene sequences , ,  has led to the suggestion that these elements mobilize many of the C. perfringens toxin genes . Contrary to this, related sequences were not found near either NELoc-1 or NELoc-3, which may indicate that these loci were acquired through a different mechanism, possibly involving a distinct transposable element.
Interestingly, the plasmid on which NELoc-3 resided also contained cpb2, which has been previously identified in NE isolates . The role of the Cpb2 toxin in enteric disease including NE is unclear. Since cpb2 is widespread in C. perfringens, the sequencing approach taken here failed to identify this gene as part of the “NE signature”, but does not preclude the involvement of it, or other toxins such as CPA, in NE.
Overlapping PCR studies of NELoc-1, -2 and -3 demonstrated a conserved organization and site of insertion among 10 type A NE strains (Figures S6, S7, and S8). Only CP2 had a different profile, with the absence of ~6kb encompassing the chitinase B and leukocidin I and II genes. This may be indicative of the stepwise evolution of the virulence loci of C. perfringens plasmids through addition or deletion of specific segments bordered by IS elements.
In contrast to NELoc-1, both NELoc-2 and -3 were also found in the avirulent strain CP6. Interestingly, this strain was originally recovered from a chicken with NE, but was subsequently avirulent in experimental challenge studies . The presence of intact NELoc-2 and -3 in this strain suggests that it may have lost at least part of NELoc-1; anecdotally, loss of virulence is not uncommon in NE isolates and the recognition of the plasmid basis of virulence provides a simple explanation for this well-recognized phenomenon. The conservation of the NELoc-3 plasmid in NE isolates suggests that both plasmids contribute to the virulence of NE strains. Further work is required to characterize this plasmid.
In conclusion, this study has contributed significantly to our understanding of the pathogenetic basis of NE in chickens. The majority of virulence-associated genes identified here are carried on different plasmids in virulent NE strains of C. perfringens, and loss of the locus encoding netB appears to attenuate virulence. The findings of this study also support previous evidence for a common ancestor of the C. perfringens toxin plasmids. Functional characterization of putative virulence genes identified here may provide significant insights into the mechanism of NE pathogenesis.
Genes identified as potential virulence factors can be cloned into suitable vectors for expression of recombinant proteins. It may be determined that codon optimization is necessary to improve levels of protein expressed. Following isolation and purification, the proteins could be combined with an appropriate adjuvant, and used to immunize chickens, providing protection against infection caused by Clostridium perfringens. Alternatively, they could be cloned into vectors suitable for direct vaccination of birds as nucleic acid vaccines. They could also be cloned into viral vectors, which could be transfected into host cells for the production and purification of recombinant virus, which could then be used to immunize birds. Finally, they could be cloned into bacterial expression systems, which could then be isolated as recombinant bacteria, which could also be used to immunize birds.
Materials and Methods
Bacterial strains and genomic DNA isolation
Clostridium perfringens strains used in this study are described in Table 1. CP1–4 and CP6 are field isolates from NE cases in Ontario, and were obtained from D.A. Barnum, Department of Pathobiology, University of Guelph. The virulence of strains CP1–4 has been confirmed, while CP6 is known to be avirulent . Genomic DNA was isolated from 5 ml of overnight culture in Brain Heart Infusion (BHI) at 37°C under anaerobic conditions. Bacteria were pelleted and lysed for 5 min in lysis buffer (5mg lysozyme/ml, 50 mM Tris, pH 8, 25% sucrose), treated with RNase A (40 ug/ml) at 37°C for 30 min followed by proteinase K (300 ug/ml) overnight at 37°C. Samples were extracted with equal volumes of phenol, phenol:chloroform:isoamyl alcohol (25:24:1), and chloroform:isoamyl alcohol (24:1), precipitated with isopropanol, and then resuspended in TE buffer. Following a second precipitation with 2 volumes of 95% ethanol, the pellets were washed twice with 70% ethanol and resuspended in water. This crude DNA extract was further purified with the DNeasy kit (Qiagen, Mississauga, Canada) according to the manufacturer's instructions. The quality of the genomic DNA was assessed by agarose gel electrophoresis and the identity confirmed by PCR amplification of cpa and netB.
High-throughput sequencing, assembly, and annotation
CP4 was sequenced using the Illumina/Solexa Genome Analyzer System in both single-end and paired-end runs, each generating 34–36-bp reads. Each run was assembled separately with the de novo short-read assembler Velvet v0.7.48 , using empirically optimized parameters, and the two assemblies subsequently combined using the minimus2 script from AMOS v2.0.8  The contigs were oriented and ordered according to the finished reference genome for ATCC 13124 using Mummer v3.21  and custom perl scripts. Additional joining of contigs was accomplished by searching for short overlapping sequences between neighbouring contigs. In addition, an optical map of CP4 (OpGen, Inc.) was used to identify assembly errors, which were manually corrected where possible. Contigs that did not align to either the reference genome or the optical map were placed at the end of the sequence.
The contigs were joined into a pseudomolecule with the linker sequence NNNNNCACACACTTAATTAATTAAGTGTGTGNNNNN and submitted to the JCVI Annotation Service, where it was run through the prokaryotic annotation pipeline. Included in the pipeline are gene finding with Glimmer, Blast-extend-repraze (BER) searches, HMM searches, TMHMM searches, SignalP predictions, and automatic annotations from AutoAnnotate. Putative domains were predicted with the Conserved Domains Database (CDD)  and InterProScan . Manual correction of the annotations was performed for the genes found in the three NE loci.
NE locus regions identified as potentially misassembled or containing sequencing errors were resolved by capillary sequencing of PCR products. Primers used for PCR and sequencing are listed in Table S1.
Identification of NE-associated genes
Sequences common to NE strains JGS4143 and CP4, but absent from nine reference genomes, were identified via MUMmer v3.21, Bedtools v2.1 , and custom perl scripts. Briefly, the draft genome sequence of CP4 was aligned to each reference genome using NUCmer with the parameters -l 10 -c 30 -b 500 -maxgap 1000. The resulting coordinate files were converted to Bed format and alignments <50bp were removed using custom perl scripts. The coordinates were inverted using the complementBed script from the Bedtools v2.1 package, producing CP4 genomic regions that did not align to each reference genome. These were successively intersected using intersectBed to give the regions found only in CP4, and finally intersected with the regions that aligned between CP4 and JGS4143 to output those common to the two NE strains, but not found in any of the reference genomes. Custom perl scripts were used to convert the coordinate files to EMBL format for viewing in Artemis v11 , and to extract the genes overlapping these intervals from the annotated genome. To confirm that these genes were unique to the NE strains, the nine reference genomes were searched for potential orthologs using OrthoMCL v2.0 , which identifies orthologs based on clustering of reciprocal best BLAST hits, using cut-offs of E-value <1e-5 and percent match >50%.
The three finished genomes were downloaded from the NCBI ftp site (ftp://ftp.ncbi.nih.gov/genomes/Bacteria/), while the contigs for the six draft genomes were downloaded separately from GenBank and combined into a single multi-fasta file.
Antimicrobial susceptibility testing
C. perfringens strains were tested for antimicrobial susceptibility by the agar disc diffusion method or E-test. The following antimicrobial agents (Oxoid, Hampshire, UK) were tested: ampicillin (10 mg), bacitracin (E-test), chloramphenicol (30 mg), clindamycin (2 mg), erythromycin (15 mg), nalidixic acid (30 mg), teicoplanin (30 mg), tetracycline (30 mg), vancomycin (5 mg). MIC values were analysed after incubation at 37°C overnight in anaerobic conditions.
Pulse Field Gel Electrophoresis (PFGE)
PFGE was performed to analyze genetic diversity and the presence of plasmids in 12 poultry C perfringens strains, as described . Briefly, DNA plugs for PFGE were prepared from overnight cultures of C. perfringens grown in TGY and the bacterial pellets incorporated into a final agarose concentration of 1% in PFGE-certified agarose (Bio-Rad Laboratories, Hercules, CA). Plugs were incubated overnight with gentle shaking at 37°C in lysis buffer (0.5M EDTA pH8.0, 2.5% of 20% sarkosyl (Fisher Scientific, Fair Lawn, NJ), 0.25% lysozyme (Sigma-Aldrich Co., St. Louis, MO) and subsequently incubated in 2% proteinase K (Roche Applied Science) buffer for 2 days at 55°C. One third of a plug per isolate was equilibrated in 200 µL restriction buffer at room temperature for 20 min and then digested with 10 U of NotI (New England Biolabs, Ipswich, MA) at 37°C overnight. Electrophoresis was performed in a 1% PFGE-certified gel and separated with the CHEF-III PFGE system (Bio-Rad Laboratories, Hercules, CA) in 0.5× Tris-borate-EDTA buffer at 14°C at 6 V for 19 h with a ramped pulse time of 1 to 12 s. Gels were stained in ethidium bromide and visualized by UV light. Mid-Range II PFG markers (New England Biolabs) were used as molecular DNA ladder.
Preparation of DIG probes and Southern blot hybridizations of PFGEs
DNA probes for all Southern blots steps were labelled by PCR amplification in the presence of digoxigenin-11-dUTP (DIG; Roche Applied Science) according to the manufacturer's recommendation. DNA probes were amplified from C. perfringens strain CP4. DNA probes for netB and hdhA genes were prepared with specific internal primers (Table S1). DNA from PFGE gels was transferred to nylon membranes (Hybond-N; Amersham). DNA hybridizations and detection were performed by using the DIG labelling and CSPD substrate according to the manufacturer's recommendation (DIG system user's guide for filter hybridization, Roche). For Southern blot hybridizations, nylon membranes were prehybridized for at least 2 h at 42°C in hybridization solution without labelled probe and then hybridized separately at 42°C with specific DNA probes for 16 h. The membranes were washed at 68°C under high-stringency conditions. For each different DIG- labelled probe, the membrane was first stripped with 0.2 N NaOH and 0.1% sodium dodecyl sulfate, incubated with prehybridization solution, and then reprobed.
Overlapping PCR analysis of NE locus 1–3
A battery of PCR reactions was performed to assess the conservation of NELoc-1–3 among 12 poultry isolates. For NELoc-1 and 2 reactions, a ready-to-use PCR mixture of Platinum PCR SuperMix high-fidelity kit (Invitrogen, Burlington, ON, Canada) was used in a 25 µl reaction containing 0.8 µM of each primer. A touch-down PCR program was used: 94°C for 3 min, 35 cycles of 94°C for 15 s, 65°C to 50°C for 15 s/cycle (the annealing temperature is decreased by 1°C every cycle until 50°C), extension at 68°C for 5 min, and finally, 68°C for 10 min. For longer range fragments the extension time was increased to 15 min. For the NELoc-3 reactions, each PCR mixture contained approximately 50 ηg of template DNA, 0.5 U of Platinum Taq High Fidelity (Invitrogen), 0.2 mM dNTPs, 1× PCR buffer, 2 mM MgSO4, and 0.2 µM of each primer in a 25µl reaction. The reaction mixtures were subjected to the following amplification conditions: one cycle of 95°C for 5 min; 35 cycles of 95°C for 30s, 53°C for 30s, and 68°C for 5 min; followed by one cycle of 72°C for 10 min. All primers used are described in Table S2. PCR product sizes were determined by agarose gel electrophoresis and visualization by ethidium bromide staining and fragments that did not match the expected size were sequenced.
Mapping of Solexa sequencing reads to plasmid sequences
In order to detect plasmid sequences in the CP4 genome that could not be assembled by Velvet, the Solexa reads were mapped to each of the published C. perfringens plasmid sequences using Maq v0.7.1 (http://maq.sourceforge.net/) and visualized with Circos v0.51 .
Optical mapping of CP4 chromosome. Optical mapping of NcoI-digested genomic DNA isolated from strain CP4 was performed by OpGen, Inc. This technique is limited to the mapping of chromosomal fragments. The lower bar represents the optical map and the upper bar the in silico NcoI-digested pseudomolecule generated from the CP4 draft genome sequence. Lines linking the two bars represent NcoI-restriction sites matched by the optical mapping software. The black bar represents the sequence from the CP4 draft genome that did not align with the CP4 optical map.
(0.05 MB PDF)
F5/8 Type C Domain Protein Alignment and Peptide ID Matching. Multiple alignment of protein sequences from three paralogous F5/8 Type C domain proteins, two chromosomal (B and C) and one located in NELoc-1 (A), was used to generate a phylogenetic tree. Peptide sequences obtained from Kulkarni et al.  were compared against each of the three paralogs and amino acid differences are indicated in red.
(0.30 MB PDF)
Mapping of CP4 Solexa reads to plasmid sequences. CP4 Solexa reads were mapped to C. perfringens virulence plasmids A) pCPF5603, B) pCPF4969 and C) pCP8533etx using Maq, and the depth of coverage along the plasmid visualized with Circos. From inner to outer: ring 1, plasmid sequence with the conserved backbone region in red, variable region in blue, and the putative Tn916 region highlighted in orange; ring 2, open-reading frames, with forward in green and reverse in red; ring 3, coverage of CP4 Solexa reads, red peaks indicate >400× coverage.
(0.42 MB PDF)
Polymorphism maps of NE loci. Green vertical lines indicate approximate position of polymorphisms between the seven NE strains. The shortest green line represents a single nucleotide polymorphism (SNP) in one strain and the longest green line indicates a SNP in three strains.
(0.17 MB PDF)
Phylogenetic analysis of sequenced C. perfringens strains. The phylogenetic relationship among our eight sequenced isolates was assessed by (A) PFGE analysis of SmaI-digested genomic DNA and (B) sequence alignment of the whole proteome from each strain using CVtree. Additional eight publicly available C. perfringens genomes were also included for comparison.
(0.72 MB PDF)
Overlapping PCR analysis of NE locus 1. PCR products spanning the entire locus are represented by black bars and the PCR results for each strain tested are given below as follows: +.PCR product was of expected size; −, no PCR product produced. Where the PCR product did not match the expected size, the actual size is given.
(0.08 MB PDF)
Overlapping PCR analysis of NE locus 2. PCR products spanning the entire locus are represented by black bars and the PCR results for each strain tested are given below as follows: +.PCR product was of expected size; −, no PCR product produced.
(0.07 MB PDF)
Overlapping PCR analysis of NE locus 3. PCR products spanning the entire locus are represented by black bars and the PCR results for each strain tested are given below as follows: +.PCR product was of expected size; −, no PCR product produced.
(0.02 MB PDF)
Primers used for PCR DIG labeling and sequencing.
(0.04 MB PDF)
Primers used for overlapping PCR analysis of NELoc-1, -2 and -3.
(0.05 MB PDF)
We thank JCVI for providing the JCVI Annotation Service which provided us with automatic annotation data and the manual annotation tool Manatee, Xiangdong Liu and Xuejiang (Roger) Shi at the Analytical Genetics Technology Centre at Princess Margaret Hospital, Toronto, for Solexa sequencing and bioinformatics support, the Genomic Analysis and Technology Core at Arizona Research Laboratories for 454 sequencing support and C.-S. Jonathan Liu at SoftGenetics LLC, Pennsylvania, for assistance with the Solexa assembly. This work was made possible by the facilities of the Shared Hierarchical Academic Research Computing
Performed the experiments: DL BR VRP PRM. Analyzed the data: DL BR VRP. Contributed reagents/materials/analysis tools: JG JGS JFP. Wrote the paper: DL BR VRP JGS GV JFP. Conceived the research: JFP DL ELR JG JGS. Designed the experiments: DL BR VRP.
- 1. Van Immerseel F, Rood JI, Moore RJ, Titball RW (2009) Rethinking our understanding of the pathogenesis of necrotic enteritis in chickens. Trends in Microbiology 17: 32–36.
- 2. Cooper KK, Songer JG (2009) Necrotic enteritis in chickens: A paradigm of enteric infection by Clostridium perfringens type A. Anaerobe 15: 55–60.
- 3. Keyburn AL, Sheedy SA, Ford ME, Williamson MM, Awad MM, et al. (2006) Alpha-Toxin of Clostridium perfringens Is Not an Essential Virulence Factor in Necrotic Enteritis in Chickens. Infect Immun 74: 6496–6500.
- 4. Kulkarni RR, Parreira VR, Sharif S, Prescott JF (2007) Immunization of Broiler Chickens against Clostridium perfringens-Induced Necrotic Enteritis. Clin Vaccine Immunol 14: 1070–1077.
- 5. Cooper KK, Trinh HT, Songer JG (2009) Immunization with recombinant alpha toxin partially protects broiler chicks against experimental challenge with Clostridium perfringens. Vet Microbiol 133: 92–97.
- 6. CF C, Trinh H, Glock R, Songer JG (2009) Alpha toxin accumulates in the gut of broiler chicks challenged with a cpa mutant of Clostridium perfringens type A; 2009; Chicago, Ill.
- 7. Keyburn AL, Boyce JD, Vaz P, Bannam TL, Ford ME, et al. (2008) NetB, a new toxin that is associated with avian necrotic enteritis caused by Clostridium perfringens. PLoS Pathogens 4:
- 8. Keyburn AL, Yan X-X, Bannam TL, Van Immerseel F, Rood JI, et al. (2010) Association between avian necrotic enteritis and Clostridium perfringens strains expressing NetB toxin. Vet Res 41: 21.
- 9. Cooper KK, Songer JG (2009) Virulence of Clostridium perfringens in an experimental model of poultry necrotic enteritis. Veterinary Microbiology. In Press, Corrected Proof.
- 10. Shimizu T, Ohtani K, Hirakawa H, Ohshima K, Yamashita A, et al. (2002) Complete genome sequence of Clostridium perfringens, an anaerobic flesh-eater. Proceedings of the National Academy of Sciences 99: 996–1001.
- 11. Myers GSA, Rasko DA, Cheung JK, Ravel J, Seshadri R, et al. (2006) Skewed genomic variability in strains of the toxigenic bacterial pathogen, Clostridium perfringens. Genome Research 16: 1031–1040.
- 12. Songer J (1996) Clostridial enteric diseases of domestic animals. Clin Microbiol Rev 9: 216–234.
- 13. Miyamoto K, Fisher DJ, Li J, Sayeed S, Akimoto S, et al. (2006) Complete Sequencing and Diversity Analysis of the Enterotoxin-Encoding Plasmids in Clostridium perfringens Type A Non-Food-Borne Human Gastrointestinal Disease Isolates. Journal of Bacteriology 188: 1585–1598.
- 14. Li J, Miyamoto K, McClane BA (2007) Comparison of Virulence Plasmids among Clostridium perfringens Type E Isolates. Infect Immun 75: 1811–1819.
- 15. Miyamoto K, Li J, Sayeed S, Akimoto S, McClane BA (2008) Sequencing and Diversity Analyses Reveal Extensive Similarities between Some Epsilon-Toxin-Encoding Plasmids and the pCPF5603 Clostridium perfringens Enterotoxin Plasmid. Journal of Bacteriology 190: 7178–7188.
- 16. Sayeed S, Li J, McClane BA (2007) Virulence Plasmid Diversity in Clostridium perfringens Type D Isolates. Infect Immun 75: 2391–2398.
- 17. Sayeed S, Li J, McClane BA (2010) Characterization of Virulence Plasmid Diversity among Clostridium perfringens Type B Isolates. Infect Immun 78: 495–504.
- 18. Thompson DR, Parreira VR, Kulkarni RR, Prescott JF (2006) Live attenuated vaccine-based control of necrotic enteritis of broiler chickens. Vet Microbiol 113: 25–34.
- 19. Kulkarni RR, Parreira VR, Sharif S, Prescott JF (2006) Clostridium perfringens Antigens Recognized by Broiler Chickens Immune to Necrotic Enteritis. Clin Vaccine Immunol 13: 1358–1362.
- 20. Kulkarni RR, Parreira VR, Sharif S, Prescott JF (2008) Oral immunization of broiler chickens against necrotic enteritis with an attenuated Salmonella vaccine vector expressing Clostridium perfringens antigens. Vaccine 26: 4194–4203.
- 21. Jiang Y, Kulkarni RR, Parreira VR, Prescott JF (2009) Immunization of Broiler Chickens Against Clostridium perfringensâ€“Induced Necrotic Enteritis Using Purified Recombinant Immunogenic Proteins. Avian Diseases 53: 409–415.
- 22. Barbara AJ, Trinh HT, Glock RD, Glenn Songer J (2008) Necrotic enteritis-producing strains of Clostridium perfringens displace non-necrotic enteritis strains from the gut of chicks. Vet Microbiol 126: 377–382.
- 23. Zerbino DR, Birney E (2008) Velvet: Algorithms for de novo short read assembly using de Bruijn graphs. Genome Res 18: 821–829.
- 24. Kurtz S, Phillippy A, Delcher AL, Smoot M, Shumway M, et al. (2004) Versatile and open software for comparing large genomes. Genome biology 5:
- 25. Li L, Stoeckert CJ Jr, Roos DS (2003) OrthoMCL: identification of ortholog groups for eukaryotic genomes. Genome Res 13: 2178–2189.
- 26. Aziz R, Bartels D, Best A, DeJongh M, Disz T, et al. (2008) The RAST Server: Rapid Annotations using Subsystems Technology. BMC Genomics 9: 75.
- 27. Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ (1990) Basic local alignment search tool. Journal of Molecular Biology 215: 403–410.
- 28. Marchler-Bauer A, Anderson JB, Chitsaz F, Derbyshire MK, DeWeese-Scott C, et al. (2009) CDD: specific functional annotation with the Conserved Domain Database. Nucl Acids Res 37: D205–210.
- 29. Dyrløv Bendtsen J, Nielsen H, von Heijne G, Brunak S (2004) Improved Prediction of Signal Peptides: SignalP 3.0. Journal of Molecular Biology 340: 783–795.
- 30. Gardy JL, Laird MR, Chen F, Rey S, Walsh CJ, et al. (2005) PSORTb v.2.0: Expanded prediction of bacterial protein subcellular localization and insights gained from comparative proteome analysis. Bioinformatics 21: 617–623.
- 31. Zdobnov EM, Apweiler R (2001) InterProScan–an integration platform for the signature-recognition methods in InterPro. Bioinformatics 17: 847–848.
- 32. Arthur M, Molinas C, Depardieu F, Courvalin P (1993) Characterization of Tn1546, a Tn3-related transposon conferring glycopeptide resistance by synthesis of depsipeptide peptidoglycan precursors in Enterococcus faecium BM4147. J Bacteriol 175: 117–127.
- 33. Deivanayagam CCS, Rich RL, Carson M, Owens RT, Danthuluri S, et al. (2000) Novel fold and assembly of the repetitive B region of the Staphylococcus aureus collagen-binding surface protein. Structure 8: 67–78.
- 34. Marraffini LA, DeDent AC, Schneewind O (2006) Sortases and the Art of Anchoring Proteins to the Envelopes of Gram-Positive Bacteria. Microbiol Mol Biol Rev 70: 192–221.
- 35. Cotter PA, Stibitz S (2007) c-di-GMP-mediated regulation of virulence and biofilm formation. Current Opinion in Microbiology 10: 17–23.
- 36. Schirmer T, Jenal U (2009) Structural and mechanistic determinants of c-di-GMP signalling. Nat Rev Micro 7: 724–735.
- 37. Karaolis DKR, Rashid MH, Chythanya R, Luo W, Hyodo M, et al. (2005) c-di-GMP (3′-5′-Cyclic Diguanylic Acid) Inhibits Staphylococcus aureus Cell-Cell Interactions and Biofilm Formation. Antimicrob Agents Chemother 49: 1029–1038.
- 38. Manich M, Knapp O, Gibert M, Maier E, Jolivet-Reynaud C, et al. (2008) Clostridium perfringens Delta Toxin Is Sequence Related to Beta Toxin, NetB, and Staphylococcus Pore-Forming Toxins, but Shows Functional Differences. PLoS ONE 3: e3764.
- 39. Kaneko J, Kamio Y (2004) Bacterial two-component and hetero-heptameric pore-forming cytolytic toxins: structures, pore-forming mechanism, and organization of the genes. Biosci Biotechnol Biochem 68: 981–1003.
- 40. Cheung JK, Keyburn AL, Bannam TL, Carter GP, Lanckriet A, et al. (2009) Regulation of NetB Production in Clostridium perfringens. 6th ClostPath International Conference. Rome, Italy.
- 41. Ohtani K, Hirakawa H, Tashiro K, Yoshizawa S, Kuhara S, et al. Identification of a two-component VirR/VirS regulon in Clostridium perfringens. Anaerobe. In Press, Corrected Proof.
- 42. Daiyasu H, Osaka K, Ishino Y, Toh H (2001) Expansion of the zinc metallo-hydrolase family of the [beta]-lactamase fold. FEBS Letters 503: 1–6.
- 43. Schmidt M (2004) Survival and cytokinesis of Saccharomyces cerevisiae in the absence of chitin. Microbiology 150: 3253–3260.
- 44. Sebaihia M, Wren BW, Mullany P, Fairweather NF, Minton N, et al. (2006) The multidrug-resistant human pathogen Clostridium difficile has a highly mobile, mosaic genome. Nat Genet 38: 779–786.
- 45. Gibert M, Jolivet-Renaud C, Popoff MR (1997) Beta2 toxin, a novel toxin produced by Clostridium perfringens. Gene 203: 65–73.
- 46. Fisher DJ, Miyamoto K, Harrison B, Akimoto S, Sarker MR, et al. (2005) Association of beta2 toxin production with Clostridium perfringens type A human gastrointestinal disease isolates carrying a plasmid enterotoxin gene. Molecular Microbiology 56: 747–762.
- 47. Cornillot E, Saint-Joanis B, Daube G, Katayama S-i, Granum PE, et al. (1995) The enterotoxin gene (cpe) of Clostridium perfringens can be chromosomal or plasmid-borne. Molecular Microbiology 15: 639–647.
- 48. Bannam TL, Teng WL, Bulach D, Lyras D, Rood JI (2006) Functional Identification of Conjugation and Replication Regions of the Tetracycline Resistance Plasmid pCW3 from Clostridium perfringens. J Bacteriol 188: 4942–4951.
- 49. Miyamoto K, Wen Q, McClane BA (2004) Multiplex PCR Genotyping Assay That Distinguishes between Isolates of Clostridium perfringens Type A Carrying a Chromosomal Enterotoxin Gene (cpe) Locus, a Plasmid cpe Locus with an IS1470-Like Sequence, or a Plasmid cpe Locus with an IS1151 Sequence. J Clin Microbiol 42: 1552–1558.
- 50. Xu Z, Hao B (2009) CVTree update: a newly designed phylogenetic study platform using composition vectors and whole genomes. Nucl Acids Res 37: W174–178.
- 51. Hamon M, Bierne H, Cossart P (2006) Listeria monocytogenes: a multifaceted model. Nat Rev Micro 4: 423–434.
- 52. Bierne H, Sabet C, Personnic N, Cossart P (2007) Internalins: a complex family of leucine-rich repeat-containing proteins in Listeria monocytogenes. Microbes and Infection 9: 1156–1166.
- 53. Bruggemann H, Bauumer S, Fricke WF, Wiezer A, Liesegang H, et al. (2003) The genome sequence of Clostridium tetani, the causative agent of tetanus disease. Proceedings of the National Academy of Sciences of the United States of America 100: 1316–1321.
- 54. Sebaihia M, Peck MW, Minton NP, Thomson NR, Holden MTG, et al. (2007) Genome sequence of a proteolytic (Group I) Clostridium botulinum strain Hall A and comparative analysis of the clostridial genomes. Genome Res 17: 1082–1092.
- 55. Bella J, Hindle K, McEwan P, Lovell S (2008) The leucine-rich repeat structure. Cellular and Molecular Life Sciences 65: 2307–2333.
- 56. Rutenber E, Ready M, Robertus JD (1987) Structure and evolution of ricin B chain. Nature 326: 624–626.
- 57. Carpusca I, Jank T, Aktories K (2006) Bacillus sphaericus mosquitocidal toxin (MTX) and pierisin: the enigmatic offspring from the family of ADP-ribosyltransferases. Molecular Microbiology 62: 621–630.
- 58. Manasherob R, Itsko M, Sela-Baranes N, Ben-Dov E, Berry C, et al. (2006) Cyt1Ca from Bacillus thuringiensis subsp. israelensis: production in Escherichia coli and comparison of its biological activities with those of other Cyt-like proteins. Microbiology 152: 2651–2659.
- 59. Arndt JW, Gu J, Jaroszewski L, Schwarzenbacher R, Hanson MA, et al. (2005) The Structure of the Neurotoxin-associated Protein HA33/A from Clostridium botulinum Suggests a Reoccurring [beta]-Trefoil Fold in the Progenitor Toxin Complex. Journal of Molecular Biology 346: 1083–1093.
- 60. Nakamura T, Tonozuka T, Ide A, Yuzawa T, Oguma K, et al. (2008) Sugar-binding Sites of the HA1 Subcomponent of Clostridium botulinum Type C Progenitor Toxin. Journal of Molecular Biology 376: 854–867.
- 61. Morimoto K, Karita S, Kimura T, Sakka K, Ohmiya K (1997) Cloning, sequencing, and expression of the gene encoding Clostridium paraputrificum chitinase ChiB and analysis of the functions of novel cadherin-like domains and a chitin-binding domain. Journal of Bacteriology 179: 7306–7314.
- 62. Morimoto K, Karita S, Kimura T, Sakka K, Ohmiya K (1999) Sequencing, expression, and transcription analysis of the Clostridium paraputrificum chiA gene encoding chitinase ChiA. Applied Microbiology and Biotechnology 51: 340–347.
- 63. Jolles P, Muzzarelli R (1999) Chitin and Chitinases. Basel, Switzerland: Birkhauser Verlag.
- 64. Gupta KC, Kumar MNVR (2000) An Overview on Chitin and Chitosan Applications with an Emphasis on Controlled Drug Release Formulations. 40: 273–308.
- 65. Khambualai O, Yamauchi K, Tangtaweewipat S, Cheva-Isarakul B (2009) Growth performance and intestinal histology in broiler chickens fed with dietary chitosan. British Poultry Science 50: 592–597.
- 66. Gusils C, Oppezzo O, Pizarro R, Gonzalez S (2003) Adhesion of probiotic lactobacilli to chick intestinal mucus. Can J Microbiol 49: 472–478.
- 67. Kirn TJ, Jude BA, Taylor RK (2005) A colonization factor links Vibrio cholerae environmental survival and human infection. Nature 438: 863–866.
- 68. Bhowmick R, Ghosal A, Das B, Koley H, Saha DR, et al. (2008) Intestinal Adherence of Vibrio cholerae Involves a Coordinated Interaction between Colonization Factor GbpA and Mucin. Infect Immun 76: 4968–4977.
- 69. Jonas K, Edwards AN, Ahmad I, Romeo T, Römling U, et al. (2009) Complex regulatory network encompassing the Csr, c-di-GMP and motility systems of Salmonella Typhimurium. Environmental Microbiology 9999:
- 70. Brouillette E, Hyodo M, Hayakawa Y, Karaolis DKR, Malouin F (2005) 3′,5′-Cyclic Diguanylic Acid Reduces the Virulence of Biofilm-Forming Staphylococcus aureus Strains in a Mouse Model of Mastitis Infection. Antimicrob Agents Chemother 49: 3109–3113.
- 71. Sudarsan N, Lee ER, Weinberg Z, Moy RH, Kim JN, et al. (2008) Riboswitches in Eubacteria Sense the Second Messenger Cyclic Di-GMP. Science 321: 411–413.
- 72. Olkowski AA, Wojnarowicz C, Chirino-Trejo M, Laarveld B, Sawicki G (2008) Sub-clinical necrotic enteritis in broiler chickens: Novel etiological consideration based on ultra-structural and molecular changes in the intestinal tissue. Research in Veterinary Science 85: 543–553.
- 73. Larsen RA, Cusumano C, Fujioka A, Lim-Fong G, Patterson P, et al. (2007) Treadmilling of a prokaryotic tubulin-like protein, TubZ, required for plasmid stability in Bacillus thuringiensis. Genes Dev 21: 1340–1352.
- 74. Tam R, Saier MH Jr (1993) Structural, functional, and evolutionary relationships among extracellular solute-binding receptors of bacteria. Microbiol Mol Biol Rev 57: 320–346.
- 75. Sommer D, Delcher A, Salzberg S, Pop M (2007) Minimus: a fast, lightweight genome assembler. BMC Bioinformatics 8: 64.
- 76. Quinlan AR, Hall IM (2010) BEDTools: A flexible suite of utilities for comparing genomic features. Bioinformatics.
- 77. Rutherford K, Parkhill J, Crook J, Horsnell T, Rice P, et al. (2000) Artemis: sequence visualization and annotation. Bioinformatics 16: 944–945.
- 78. Krzywinski M, Schein J, Birol Än, Connors J, Gascoyne R, et al. (2009) Circos: An information aesthetic for comparative genomics. Genome Res 19: 1639–1645.
- 79. Cheung JK, Dupuy B, Deveson DS, Rood JI (2004) The Spatial Organization of the VirR Boxes Is Critical for VirR-Mediated Expression of the Perfringolysin O Gene, pfoA, from Clostridium perfringens. J Bacteriol 186: 3321–3330.
- 80. Myers GS, Rasko DA, Cheung JK, Ravel J, Seshadri R, et al. (2006) Skewed genomic variability in strains of the toxigenic bacterial pathogen, Clostridium perfringens. Genome Res 16: 1031–1040.