Investigation of the Enteric Pathogenic Potential of Oral Campylobacter concisus Strains Isolated from Patients with Inflammatory Bowel Disease

Background Campylobacter concisus, a bacterium colonizing the human oral cavity, has been shown to be associated with inflammatory bowel disease (IBD). This study investigated if patients with IBD are colonized with specific oral C. concisus strains that have potential to cause enteric diseases. Methodology Seventy oral and enteric C. concisus isolates obtained from eight patients with IBD and six controls were examined for housekeeping genes by multilocus sequence typing (MLST), Caco2 cell invasion by gentamicin-protection-assay, protein analysis by mass spectrometry and SDS-PAGE, and morphology by scanning electron microscopy. The whole genome sequenced C. concisus strain 13826 which was isolated from an individual with bloody diarrhea was included in MLST analysis. Principal Findings MLST analysis showed that 87.5% of individuals whose C. concisus belonged to Cluster I had inflammatory enteric diseases (six IBD and one with bloody diarrhea), which was significantly higher than that in the remaining individuals (28.6%) (P<0.05). Enteric invasive C. concisus (EICC) oral strain was detected in 50% of patients with IBD and none of the controls. All EICC strains were in Cluster 1. The C. concisus strain colonizing intestinal tissues of patient No. 1 was closely related to the oral C. concisus strain from patient No. 6 and had gene recombination with the patient’s own oral C. concisus. The oral and intestinal C. concisus strains of patient No. 3 were the same strain. Some individuals were colonized with multiple oral C. concisus strains that have undergone natural recombination. Conclusions This study provides the first evidence that patients with IBD are colonized with specific oral C. concisus strains, with some being EICC strains. C. concisus colonizing intestinal tissues of patients with IBD at least in some instances results from an endogenous colonization of the patient’s oral C. concisus and that C. concisus strains undergo natural recombination.


Introduction
Campylobacter oncisus is a Gram-negative bacterium with a curved shape and a polar flagellum, which was first isolated from human gingival plaques in 1981 [1]. C. concisus is a fastidious bacterium, requiring hydrogen enriched microaerobic conditions for growth [2,3].
Recently, C. concisus has been shown to be associated with inflammatory bowel disease (IBD). IBD is a chronic inflammatory disorder of the gastrointestinal tract; the two major forms of IBD are Crohn's disease (CD) and ulcerative colitis (UC) [4,5]. The aetiology of IBD is unknown. Studies have shown that multiple factors including genetic factors, environmental factors and intestinal microflora are involved in the development of IBD [4,5]. Despite strong evidence showing that the intestinal microbiota plays a key role in the pathogenesis of IBD, the exact causative or triggering agent still remains unknown [6,7,8].
A significantly higher prevalence of C. concisus in intestinal biopsies and fecal samples of patients with IBD as compared with controls were reported by a number of research groups [9,10,11,12]. Using in vitro cell culture models, C. concisus was shown to increase intestinal epithelial permeability and induce intestinal epithelial production of IL-8 and apoptosis [13,14,15]. Some C. concisus strains cultured from intestinal biopsies of patients with IBD and diarrheal stool samples were shown to be invasive to Caco2 cells [14]. The presence of bacterial virulence factors such as phospholipase A2 and a cytolethal distending toxin (CDT)-like toxin in some C. concisus strains has been reported [16,17].
C. concisus is a commensal bacterium of the human oral cavity. Zhang et al isolated C. concisus from 75% of saliva samples obtained from healthy individuals aged 3 to 60 years old and detected C. concisus by PCR in 95% of these samples [18]. The prevalence of C. concisus in the oral cavity of patients with IBD and healthy controls was not statistically different [18]. Furthermore, this study noted some bacterial protein banding similarities between a C. concisus strain colonizing the oral cavity and the C. concisus strain colonizing the intestinal tissues of a patient with IBD and proposed that specific oral C. concisus strains are involved in human IBD [18].
Currently, whether patients with IBD are colonized with specific oral C. concisus strains is not known. It is also not clear whether oral C. concisus strains have enteric pathogenic potential and whether C. concisus colonizing intestinal tissues of a given patient with IBD results from an endogenous colonization of the patient's own oral C. concisus. To investigate these issues, we compared the housekeeping genes and protein profiles of oral C. concisus isolated from patients with IBD and controls, as well as C. concisus isolated from intestinal biopsies of patients with IBD. In addition, we examined the invasiveness of oral C. concisus isolates to Caco2 cells and identified a number of bacterial proteins that may be important to C. concisus invasion of Caco2 cells.

Analysis of Housekeeping Genes of Oral and Enteric C. Concisus Isolated from Patients with IBD and Controls by Multilocus Sequence Typing
Six housekeeping genes amplified from 70 C. concisus isolates, which were obtained from eight patients with IBD and six controls (details of these C. concisus isolates were described in Materials and Methods section), were analysed by multilocus sequence typing (MLST). The six housekeeping genes amplified were aspA (aspartase A), glnA (glutamine synthetase), tkt (transkelotase), asd (aspartate semialdehyde dehydrogenase), atpA (ATP synthase alpha subunit) and pgi (glucose-6-isomerase). MLST analysis was based on the sequences of six housekeeping genes with a total of 2,561 bp from each isolate analysed. The sequence types (ST) and allelic profiles of C. concisus isolates analysed are shown in Table 1. The polymorphic nucleotides were submitted as supplementary data ( Figure S1). The criteria to define strains and variants were described in the Materials and Methods section.
The sequences of all six housekeeping genes of the 10 oral isolates (P3UCO1-O10) of patient No. 3 were identical, suggesting that this patient was colonized with a single oral C. concisus strain (ST12) ( Table 1).
One oral C. concisus isolate was available from each of the remaining four patients (patients No. 5 to No. 8). The four C. concisus isolates from these four patients differed at all six housekeeping genes, each representing a different strain.
The nine isolates (H1O1-O9) obtained from the healthy individual No. 1 had identical housekeeping genes, suggesting that this individual was colonized with a single oral C. concisus strain (ST15) ( Table 1). One oral C. concisus isolate was available from each of the remaining five healthy controls (H2-H6). Four of these healthy controls were colonized with a different C. concisus strain (differing at six housekeeping genes), except for H2O1 which had atpA and tkt genes identical to C. concisus ST12 (Table 1).

MLST Analysis of Enteric C. Concisus Isolated from Patients with IBD
The intestinal biopsy C. concisus isolate (P1CDB1(UNSWCD)) of patient No. 1 (designated as ST6) had five unique housekeeping genes (Table 1). However, its asd gene was identical to the patient's own oral ST1, ST3 and ST5, suggesting that ST6 has acquired its asd gene from the patient's own oral C. concisus isolates ( Table 1).
The 10 intestinal biopsy C. concisus isolates (P3UCB1-B10) obtained from patient No. 3 had six housekeeping genes identical to that of the patient's own oral C. concisus isolates, indicating that the oral and intestinal biopsy C. concisus isolates are the same strain (ST12) ( Table 1). Two luminal-washout isolates (P3UCLW1 and P3UCLW2) were obtained from patient No. 3. P3UCLW2 (ST13) had five housekeeping genes identical to the patient's own oral and intestinal biopsy isolates (ST12) and differed only by one base in pgi, suggesting that ST13 is a mutational variant of ST12 (Table 1). P3UCLW1 (ST14) had six housekeeping genes that were different from the patient's own oral and intestinal biopsy C. concisus isolates, representing a different C. concisus strain ( Table 1).

Isolated from Patients with IBD and Controls
To further illustrate the genetic relationship between oral and enteric C. concisus strains isolated from patients with IBD and controls, a phylogenetic tree was constructed based on the sequences of the six housekeeping genes analysed ( Figure 1). The housekeeping genes of the whole genome sequenced C. concisus strain 13826 were also included.
C. concisus isolates obtained from the majority of the patients with IBD (6/8) formed one cluster (Cluster 1). The C. concisus strain (P1CDB1(UNSWCD)) isolated from intestinal biopsies of patient No. 1 with CD, C. concisus strain 13826 (a strain isolated from fecal sample of a patient with bloody diarrhea) and the oral C. concisus isolate obtained from healthy individual No. 3 were also grouped into Cluster 1. All enteric invasive Campylobacter concisus (EICC) strains (see results in Table 2) were grouped into Cluster 1 ( Figure 1). Seven of the eight individuals (7/8, 87.5%) whose isolates belonged to Cluster 1 had inflammatory enteric diseases (six patients with IBD and one had bloody diarrhea), which was significantly higher than the remaining individuals (2/7, 28.6%) (P,0.05). Among the C. concisus isolates that did not belong to Cluster 1, six isolates were genetically closely related, forming a second cluster (Cluster 2). The oral C. concisus isolate obtained from a healthy control (H6O1) was genetically distinct from the other C. concisus isolates ( Figure 1).
The intestinal biopsy isolate of patient 1 (P1CDB1(UNSWCD)) was genetically most closely related to the oral C. concisus isolate of a patient with CD (P6CDO1) (Figure 1). The intestinal biopsy isolates of patient No. 3 (P3UCB1-B10) were identical to the patient's own oral isolates (P3UCO1-10). One luminal-washout isolate (P3UCLW2) of patient No. 3 was very closely related to the patient's own oral and intestinal biopsy isolates. The second luminal-washout C. concisus isolate (P3UCLW1) of this patient was different from the patient oral and intestinal C. concisus isolates, but was more closely related to the oral isolates of healthy control No. 1 ( Figure 1).

Comparison of Protein Profiles of Different C. Concisus Isolates
All 70 C. concisus isolates analyzed by MLST were also subjected to whole cell protein profile analysis. The protein profile types of all 70 C. concisus isolates were shown in Figure 1 and the representative protein profiles of C. concisus isolates from each individual are shown in Figure 2. The 21 oral C. concisus isolates (P1CDO1-O21) of patient 1 showed two different protein profiles (P1-1 and P1-2). Thus, the different strains showed different protein profiles, the variants did not generate new protein profiles (see MLST data). The intestinal biopsy isolate (P1CDB1(UNSWCD)) of patient No. 1 showed a protein profile (P1-3) that was different from the patient's oral strains. The seven oral C. concisus isolates (P2CDO1-O7) from patient No. 2 revealed two protein profiles (P2-1 and P2-2). Again, individual strains showed different protein profiles, the recombinant variants did not generate new protein profiles. The oral C. concisus isolates of patient No. 3 (P3UCO1-O10) showed an identical protein profile (P3-1), consistent with the finding that all oral C. concisus isolates of this patient had identical sequences of housekeeping genes and representing a single strain. The intestinal biopsy C. concisus isolates (P3UCB1-B10) and one luminal-washout isolate of patient 3 (P3UCLW2) showed the same protein profile (P3-2), which was identical to the protein pattern of the patient's oral C. concisus isolates except for the disappearance of a 210 kD band (band A in Figure 2). The second luminal-washout C. concisus isolate (P3UCLW1) showed a different protein profile (P3-3) (Figures 1  and 2).
Sequencing of the 210 kD band of P3UCO5 (oral isolate) identified 47 C. concisus proteins, the majority of which were ribosomal proteins and various proteins involved in metabolism. Interestingly, one protein was a bacterial virulence protein S-layer-RTX.
All oral C. concisus isolates of the healthy control No. 1 showed an identical C. concisus protein profile (H1-1) (Figure 2), consistent with the finding that all oral C. concisus isolates from this individual had identical housekeeping genes.
The C. concisus isolates obtained from the remaining patients and controls showed individual protein patterns ( Figure 2).
Disease associated protein profiles were not identified.

Detection of Enteric Invasive C. Concisus (EICC) Oral Strains in Patients with IBD
The invasiveness of all 70 C. concisus isolates to Caco2 cells was examined and expressed as invasive index. EICC strains, which have an invasive index $1, are shown in Table 2. EICC oral strains were detected in 50% (4/8) patients with IBD and none of the controls (0/6) ( Table 2). However, the prevalence of EICC oral strains in patients with IBD was not statistically different from that in controls (P.0.05). The remaining C. concisus isolates had an invasive index ,1. The positive control invasive C. concisus strain (P1CDB1(UNSWCD)) showed an invasion index of 1.3.

Identification of Bacterial Proteins that may be Associated with C. Concisus Enteric Epithelial Invasion
Given that patient 1 was colonized with both EICC and non-EICC oral C. concisus isolates, we sequenced the most abundantly expressed protein band (band B shown in Figure 2) of an EICC isolate (P1CDO3) and its corresponding band (band C shown in Figure 2) of a non-EICC oral isolate (P1CDO2), attempting to identify some bacterial proteins that may be associated with C. concisus invasion to intestinal epithelial cells.
Twenty-three and 21 C. concisus proteins were identified from the protein band of the EICC isolate and the non-EICC isolate respectively. Seventeen proteins were common proteins identified from both the EICC isolate and the non-EICC isolate, and these proteins are involved in protein transport, metabolism and protein synthesis. General glycosylation pathway protein and Type II secretion system protein E, which were previously shown to be associated with bacterial virulence, were identified only from the protein band of EICC isolate. The distinctive proteins identified from the EICC isolate and the non-EICC isolate are listed in Table 3.

Bacterial Morphology of EICC and Non-EICC C. Concisus Isolates
To observe whether EICC isolates and non-EICC isolates are morphologically different, an EICC oral isolate (P1CDO3), a non-EICC oral isolate (P1CDO2) and an EICC enteric isolate (P1CDB1(UNSWCD)) were examined using electron microscopy. Both EICC and non-EICC isolates showed a similar morphology. Flagellum was present in all isolates ( Figure 3).

Discussion
This study examined whether patients with IBD are colonized with specific oral C. concisus strains and whether oral C. concisus strains have enteric pathogenic potential. Furthermore, this study Figure 1. Neighbour-Joining tree based on the sequences of six housekeeping genes (asd, aspA, atpA, glnA, pgi and tkt) illustrating the phylogenetic relationships of oral and enteric Campylobacter concisus isolates analysed in this study. Isolates from patients with IBD are coloured red. Isolates from a healthy control are coloured blue. Clusters are indicated in roman numerals. Campylobacter curvus is used as an outgroup. Alongside the tree on the right are the MLST sequence types (ST) and protein profiles. Campylobacter concisus strain 13826 is the whole genome sequenced strain (Accession No.: CP000792.1). doi:10.1371/journal.pone.0038217.g001 investigated whether C. concisus colonizing intestinal tissues of a given patient with IBD results from the endogenous colonization of the patient's own oral C. concisus strain. MLST analysis revealed an association between Cluster I C. concisus strains and enteric inflammatory diseases including IBD (see results section). The oral C. concisus strains isolated from patients with IBD were predominantly grouped into Cluster I ( Figure 1). EICC oral C. concisus strains were detected in 50% of patients with IBD and none of the controls ( Table 2). All EICC oral C. concisus strains were in Cluster I. C. concisus colonizing intestinal tissues of patients with IBD, at least in some instances, originated from the colonization of the patients own oral C. concisus strain ( Figure 1 and Table 1). Taken together, these results suggest that patients with IBD are colonized with a specific group of oral C. concisus strains that have enteric pathogenic potential if colonizing the intestinal tract.
Patient No. 1 was colonized with both EICC strain and non-EICC oral strains, which were morphologically indistinguishable ( Figure 3). We attempted to identify proteins that may be associated with C. concisus intestinal epithelial invasion by sequencing the most abundantly expressed protein band of an EICC isolate and its corresponding protein band of a non-EICC isolate (Table 3 and Figure 2). Use of EICC and non-EICC C. concisus strains isolated from the oral cavity of the same individual would minimize the influence of environmental factors on bacterial protein expressions. We found that two of the proteins that were identified from the EICC isolate, including general glycosylation pathway protein and type II secretion system protein E, are particularly interesting in relation to bacterial virulence.
General glycosylation system has been reported to be important for C. jejuni to attach to and invade human epithelial cells [19]. The  Table 3. Distinctive proteins identified from the most abundantly expressed protein band # of an oral EICC isolate and the corresponding band of a non-EICC oral C. concisus isolate obtained from patient 1.

EICC isolate (P1CDO3)
Non-EICC isolate (P1CDO2) General glycosylation pathway protein* Pyridoxal phosphate-dependent enzyme Outer membrane protein Peptide chain release factor 2 Hypothetical protein CCC13826_1624 Type II secretion system protein E* 3-isopropylmalate dehydratase large subunit Glutamate dehydrogenase Signal recognition particle protein Threonine dehydratase # Protein bands were shown in Figure 3. *Proteins related to bacterial virulence. doi:10.1371/journal.pone.0038217.t003 Type II secretion system (T2SS), consisting of at least 12 core components including T2SS protein E, is one of the five protein secretion systems that Gram-negative bacteria use to export proteins from within the bacterial cell to the extracellular environment or into host cells [20,21]. T2SS has been shown to be important in bacterial pathogenesis [22]. A study by Iwobi et al showed that a novel type II secretion gene cluster is present only in high-pathogenicity Yersinia enterocolitica strains [23]. In our study, the fact that General glycosylation system protein and T2SS protein E were identified only from the protein band of the EICC isolate, not from the corresponding protein band of the non-EICC isolate indicates a difference of these proteins in EICC and non-EICC isolates. Future studies examining genes coding for these proteins in EICC and non-EICC isolates will further illustrate whether General glycosylation system and T2SS play a role in C. concisus invasion to intestinal epithelial cells. A recent study by Kaakoush et al detected exotoxin 9 gene which is located in the plasmid in four enteric C. concisus strains that were invasive to Caco2 cells including the UNSWCD strain [24]. The presence of plasmid and the exotoxin 9 gene in oral C. concisus strains isolated from patients with IBD and controls should be examined in future studies. There are other evidence supporting that specific oral C. concisus strains may be enteric pathogenic. For example, Nielsen et al showed that both oral and fecal C. concisus strains induced epithelial barrier dysfunction [13].
The oral cavity is the primary colonization site of C. concisus in humans. Using a filtration method, we previously isolated C. concisus from 75% of saliva samples collected from healthy individuals and 85% of saliva collected from patients with IBD [18]. In comparison to the human oral cavity, the human intestine represents a less favourable environment for C. concisus growth. Using the same filtration method, Engberg et al isolated C. concisus from 2.8% of fecal samples collected from healthy individuals [25]. C. concisus requires H 2 -enriched microaerobic condition for growth [3]. H 2 in the human intestine is generated by bacterial fermentation of unabsorbed carbohydrates; the amount of H 2 available in an individual's intestine is influenced by the ingested food and the composition of the local intestinal microbial community [26,27]. Thus, whether oral C. concisus strains are able to establish intestinal colonization and multiply to a sufficient number to cause enteric disorders will be determined by both the properties of C. concisus strains and the local intestinal environ-ment, the latter may fluctuate due to the change of ingested food and the composition of an individual's intestinal microbial community.
Humans were previously considered as the only host of C. concisus, however, recently C. concisus was detected in fecal and saliva samples of domestic dogs and cats [28,29]. Whether domestic pets are an additional source of human intestinal C. concisus infection needs to be investigated. A study by Lynch et al isolated C. concisus from chicken and beef meat [30]. While chicken and beef meat may serve as a source of human infection, it is unable to conclude whether chicken and cattle are a natural host of C. concisus.
Oral and intestinal biopsy C. concisus isolates of patient 3 revealed identical protein profiles, except for the disappearance of a 210 kD protein band in the intestinal biopsy C. concisus isolates ( Figure 2). Interestingly, S-layer-RTX protein was identified from this 210 kD protein band of the oral isolate. S-layer is a cell surface protein found in a range of bacterial species. RTX proteins are a family of proteins secreted by a variety of Gram-negative bacteria, which exhibit various biological functions including formation of the S-layer protein in some bacterial species [31]. S-layer contributes to bacterial pathogenesis by adhering to host cells and immune evasion [32]. Campylobacter fetus and Campylobacter rectus possess S-layer [33]. High frequency antigenic variation of S-layer in C. fetus, resulting from DNA inversion, has been reported [33,34]. Future studies are required to examine whether the disappearance of S-layer-RTX protein in intestinal biopsy C. concisus isolates in this patient is due to antigenic variation. S-layer RTX protein was previously detected in the P1CDB1(UNSWCD) strain [35]. Kalischuk et al reported that the gene coding for S-layer RTX was detected in two C. concisus strains isolated from fecal samples of healthy individuals, but not in C. concisus strains isolated from fecal samples of patients with diarrhea [15].
This study showed that the genome of C. concisus species is highly diverse. All individuals included in this study were colonized with different C. concisus strains, demonstrated by the findings that these C. concisus isolates differed at all six housekeeping genes (except for H2O1 strain, which had atpA and tkt genes identical to that of ST12) and showed unique protein profiles. This is consistent with the finding by Aabenhus et al [36]. Aabenhus et al examined 62 C. concisus strains using amplified fragment length polymorphism (AFLP) and found that all C. concisus strains gave unique AFLP profiles [36]. Our finding that C. concisus strains undergo natural recombination (Table 1) offer an explanation for the high genetic diversity of C. concisus species, as observed by other research groups [36,37,38,39].
From the luminal-washout of patient No. 3, two C. concisus isolates were isolated. While one isolate (P3UCLW2) was the variant of the C. concisus strain colonizing the intestinal tissues of this patient, the other isolate (P3UCLW1) was a different strain which was closely related to the oral C. concisus strain of healthy control No. 1 (Figure 1 and Table 1). This suggests that C. concisus detected in fecal samples may contain both C. concisus strains colonizing the intestinal tissues and C. concisus strains transiently colonizing the fecal materials.
In summary, this study provides the first evidence that patients with IBD are colonized with a group of specific oral C. concisus strains that may have enteric pathogenic potential. In addition, this study showed that C. concisus colonizing intestinal tissues of patients with IBD, at least in some instances, results from an endogenous colonization of the patient's own oral C. concisus and that C. concisus strains undergo natural recombination.

C. Concisus Isolates and Cultivation
A total of 70 C. concisus isolates obtained from eight patients with IBD and six controls were included in this study. Enteric C. concisus refers to isolates cultured from intestinal biopsies or luminal-washout fluid. Oral C. concisus refers to isolates cultured from saliva. Luminal-washout fluid was the fluid collected from luminal fluid draining tube prior to the start of the colonoscopy, which contains fecal bacteria and the mucosa associated bacteria flushed out from the intestinal mucus due to the severe diarrhoea induced during the preparation for colonoscopy. The enteric C. concisus isolate of patient 1 (P1CDB1) was isolated by Zhang et al [9], which was named as UNSWCD in a following study by Man et al [14]. To maintain the consistency with the previous publications and the naming system in this study, we used P1CDB1 (UNSWCD) to label this strain in this study. The enteric C. concisus isolates of patient 3 were isolated by Mahendran et al [11]. The oral C. concisus isolates were either isolated in our previous studies or in this study [18]. The identities of these C. concisus isolates were confirmed by microscopic examination of bacterial morphology and C. concisus specific PCR [12]. Details of the C. concisus isolates used in this study are listed in Table 4.
C. concisus isolates were cultured on Horse blood agar (HBA), prepared using Blood Agar Base No. 2 supplemented with 6% (v/ v) defribinated horse blood and 10 mg/ml vancomycin (Oxoid Australia Pty Limited, South Australia). Culture plates were incubated at 37uC for 48 hours under microaerobic conditions generated using a Campylobacter gas generating kit (Oxoid).
The clinical information of patients included in this study is shown in Table 5.
Bacterial DNA Extraction C. concisus DNA was extracted using the Puregene DNA Extraction kit (Gentra, Minneapolis, USA) following the manufacturer's instructions.

Comparison of the Sequences of Housekeeping Genes of Oral and Enteric C. Concisus Isolates by Multilocus Sequence Typing
Choice of housekeeping genes and primer design. Six housekeeping genes (aspA, glnA, tkt, asd, atpA and pgi) were selected for MLST analysis. These housekeeping genes have previously been used for MLST analysis of C. jejuni [40,41,42].
The polymerase chain reaction (PCR) primers used to amplify each of the above genes were designed using the software Primer 3 plus, based on the genome sequence of C. concisus strain 13826 (Accession No. CP000792.1). The sequences of the PCR primers used are shown in Table 6.
Amplification and sequencing of MLST genes. To amplify the MLST genes, hot start PCR reactions were performed in a 25 ml reaction mixture containing 16 PCR buffer, 200 nM of deoxynucleotide triphosphate, 2.5 mM MgCl 2 , 5.5 U of Taq polymerase (Fisher Biotech, Subiaco, Australia), 10 pmol of each primer and 10 ng of bacterial DNA extracted from each C. concisus isolates. The thermal cycling conditions consist of denaturing at 96uC for 2 minutes, followed by 40 cycles of 94uC for 10 seconds, annealing for 10 seconds and 72uC for 45 seconds. The annealing temperatures were 55uC for aspA and atpA, 51uC for tkt and asd. The annealing temperatures for gln and pgi were 55uC-57uC, depending on individual isolates. Both strands of all PCR products were sequenced using BigDye TM terminator chemistry (Applied Biosystems, Foster City, CA) and separated on an ABI Capillary DNA Sequencer ABI3730 (Applied Biosystems).
MLST analysis. Sequences of housekeeping genes of C. concisus isolates were aligned and compared using software programs of MEGA 4 [43] and an in-house script MULTICOMP [44]. PHYLIP was used to generate neighbour-joining trees [45]. Campylobacter curvus (Accession No. CP000767.1) was used as an outgroup.
For each housekeeping gene, the different sequences present in different isolates were assigned distinct allele numbers. The allele numbers at each of the housekeeping genes defined the allelic profiles. Each isolate with a distinct allelic profile was referred as an individual sequence type (ST). Isolates with identical sequences at all six housekeeping genes were defined as the same strain. Isolates with different sequences at all six housekeeping genes were defined as different strains. In the case that a given individual was colonized with multiple C. concisus strains (more than one C. concisus strain), if evidence suggesting that some C. concisus isolates were generated due to gene recombination or mutations of the C. concisus strains colonizing the same individual, the generated C. concisus isolates were defined as recombinant or mutational variants.

GenBank Sequence Submission
Sequences of housekeeping genes amplified by PCR were submitted to GenBank. The accession numbers of the sequences of the PCR products of housekeeping genes submitted to GenBank were JQ683402-JQ683505.

Comparison of Whole Cell Protein Profiles of Oral and Enteric C. Concisus Isolates
Whole cell proteins of C. concisus isolates were prepared. Briefly, C. concisus was harvested from HBA plates following cultivation for two days. After washing three times with PBS, the pellet was frozen-thawed three times using liquid nitrogen, then suspended in 600 ml of PBS. The bacterial mixtures were sonicated on ice for 3 minutes with 0.5 seconds intervals (40% amplitudes). The protein concentrations were determined using the BCA TM protein assay kit (Pierce, Rockford, USA). 15 mg whole cell proteins of each isolate were loaded on to 12% sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS-PAGE) to examine the whole cell protein profiles as described previously [46].

Investigation of Enteric Invasiveness of C. Concisus Isolates
Invasive abilities of C. concisus isolates to Caco2 cells were examined using previously described gentamicin protection assay with modifications [14,47,48].
Briefly, Minimum Essential Medium (MEM) supplemented with 10% heat inactivated fetal bovine serum (HI-FBS), 1 mM Sodium  Following washing of the Caco2 monolayer for 5 times using PBS, the cells were lysed with 1% Triton X-100. Serial dilutions of cell lysates were inoculated onto HBA plates and incubated for two days. CFU of C. concisus were recorded, which were regarded as the numbers of C. concisus that have been internalised into Caco2. Gentamicin assay was repeated three times. Invasive index, which is the number of internalized C. concisus/ number of adherent C. concisus 6 100, was calculated using the formula described by Larson et al [49]. A C. concisus strain that has an invasive index $1 was defined as an enteric-invasive C. concisus (EICC) strain. This definition was based on previous studies showing that the established human enteric pathogen Campylobacter jejuni with an invasive index $1 causes symptoms in infected pigs similar to human symptoms [49] . The UNSWCD strain, which was previously shown to invasive to Caco2 cells [14), was used as the positive control.

Mass Spectrometry Analysis
A number of C. concisus protein bands separated on SDS-PAGE were subjected to mass spectrometry analysis to identify proteins that may be important for C. concisus invasion to Caco2 cells and for C. concisus intestinal colonization. Briefly, protein bands of interest were excised from Commassie Blue stained polyacrylamide gel and were digested with trypsin. Digested peptides were separated by nano-liquid chromatography (LC) using an Ultimate 3000 HPLC and autosampler system (Dionex, Amsterdam, The Netherlands) and then subjected to analysis using a LTQ-FT Ultra mass spectrometer (Thermo Electron, Bremen, Germany). All MS/MS spectra were searched against NCBI database using MASCOT (version 2.3). Mass spectrometric analysis was carried out at the Bioanalytical Mass Spectrometry Facility, University of New South Wales, Australia.

Scanning Electron Microscopy
Scanning electron microscopy was used to examine the morphology of two different oral C. concisus isolates, one shown to have an invasive index .1 (defined as EICC in this study) and another shown to have an invasive index ,1 (defined as non-EICC in this study). C. concisus suspension prepared using MEM containing 10% HI-FBS was placed on glass cover slips. The glass cover slips were incubated at 37uC with 5% CO 2 for 2 hours, the bacteria were fixed with 2% glutaraldehyde and 2.5% paraformadehyde in 0.1 mol/L phosphate buffer (pH 7.4). Following dehydration in ethanol and critical point drying, C. concisus was mounted on carbon tabs, gold-coated then observed using Nova TM NanoSEM 230 high resolution scanning electron microscope (FEI, Oregon, USA). Scanning electron microscopic examination was performed at the Electron Microscope Unit at the University of New South Wales, Australia.

Statistic Analysis
Fisher's exact test (two tailed) was used in this study. Statistical analysis was performed using GraphPad Prism 5 software (San Diego, CA). Figure S1 The polymorphic nucleotides of MLST genes and Sequence types (ST) of C. concisus isolates. Six housekeeping genes (asd, aspA, atpA, glnA, pgi and tkt) of 70 oral and enteric C. concisus isolates cultured from eight patients with IBD and six healthy control were analysed. Dot indicates that the base is the same as the consensus. *The whole genome sequenced Campylobacter concisus strain 13826 (Accession No.: CP000792.1). (DOCX)