Molecular Evidence for the Thriving of Campylobacter jejuni ST-4526 in Japan

Campylobacter jejuni is a leading cause of human gastroenteritis worldwide. This study aimed at a better understanding of the genetic diversity of this pathogen disseminated in Japan. We performed multilocus sequence typing (MLST) of Campylobacter jejuni isolated from different sources (100 human, 61 poultry, and 51 cattle isolates) in Japan between 2005 and 2006. This approach identified 62 sequence types (STs) and 19 clonal complexes (CCs), including 11 novel STs. These 62 STs were phylogenetically divided into 6 clusters, partially exhibiting host association. We identified a novel ST (ST-4526) that has never been reported in other countries; a phylogenetic analysis showed that ST-4526 and related STs showed distant lineage from the founder ST, ST-21 within CC-21. Comparative genome analysis was performed to investigate which properties could be responsible for the successful dissemination of ST-4526 in Japan. Results revealed that three representative ST-4526 isolates contained a putative island comprising the region from Cj0737 to Cj0744, which differed between the ST-4526 isolates and the reference strain NCTC11168 (ST-43/CC-21). Amino acid sequence alignment analyses showed that two of three ST-4526 isolates expressed 693aa- filamentous hemagglutination domain protein (FHA), while most of other C. jejuni strains whose genome were sequenced exhibited its truncation. Correspondingly, host cell binding of FHA-positive C. jejuni was greater than that of FHA-truncated strains, and exogenous administration of rFHA protein reduced cell adhesion of FHA-positive bacteria. Biochemical assays showed that this putative protein exhibited a dose-dependent binding affinity to heparan sulfate, indicating its adhesin activity. Moreover, ST-4526 showed increased antibiotic-resistance (nalidixic acid and fluoroquinolones) and a reduced ability for DNA uptake. Taken together, our data suggested that these combined features contributed to the clonal thriving of ST-4526 in Japan.


Introduction
Campylobacter jejuni is one of the leading bacterial causes of foodborne gastroenteritis in humans and is responsible for an estimated 5-14% of the occurrence of human diarrhea worldwide, which translates into 400-500 million cases annually [1], [2]. This pathogen is typically located in the feces and gastrointestinal tracts of warm-blooded animals [3], [4], [5], [6], which are considered reservoirs for human infection. Human campylobacteriosis is most often sporadic; although poorly prepared food has often been implicated, it is generally difficult to determine the source since C. jejuni easily loses cultivability in food specimens under aerobic conditions. Therefore, a substantial effort has been made to isolate C. jejuni from a wide variety of reservoirs and type these strains using a number of methods, including amplified fragment length polymorphism (AFLP), flaA short variable region (SVR) sequencing, pulsed-field gel electrophoresis (PFGE), and multilocus sequence typing (MLST) [7]. MLST involves the phylotyping of bacterial isolates based on the DNA sequence comparison of a standardized panel of housekeeping genes [8]. As this method has advantages for accuracy and reproducibility of the data between different laboratories, MLST is one of the best approaches to understand the population genetics, host association, and seasonal or geographical variation of this pathogen [9], [10], [11], [12], [13], [14], [15], [16]. To date, 5,957 sequence types (STs) from 18,562 isolates have been obtained from MLST data, which have been deposited in the Campylobacter MLST database (http://www. plst.org/campylobacter) as for August 17, 2012. The clonal diversity and evolutional linkage of the STs could be visually analysed with eBURST [17], facilitating the evolutional analysis.
C. jejuni is a public health threat as one of the most common foodborne pathogens in Japan [18]. Results of a previous epidemiological study suggest that chicken and feedlot cattle are the main reservoirs for human campylobacteriosis [19]. Several molecular approaches, including PFGE, serotyping, and flaA genotyping, have provided a possible explanation for the widespread of this pathogen, which coincides with partial host association in Japan [20], [21], [22]. For example, serotypes O:2 and O:4 were common in human, poultry, and bovine isolates, whereas serotypes O:23, O:36, O:53 were common in human and bovine isolates in Japan [20]. The PFGE analysis also showed a partial similarity between poultry and human isolates [20]. In a more recent study, Yabe et al. [23] conducted a small-scale MLST analysis, revealing the presence of novel STs in C. jejuni isolated from poultry and humans in Japan. However, it remains unclear how and whether such host associations might be linked to human campylobacteriosis and whether there are geographic specifications that contribute to the thriving of this pathogen in certain populations within Japan.
In the present study, we used an MLST approach to identify the population structure of C. jejuni isolated from human, poultry, and cattle across Japan for the first time. We identified a unique genotype, ST-4526, that thrived in poultry and human populations in Japan. A phylogeographical linkage analysis, in combination with comparative genome analyses, provided a possible explanation for this thriving.

Bacterial Isolates and Media
From 2005-2006, C. jejuni human clinical isolates (n = 100) were obtained from two prefectures, Tokyo and Osaka, in Japan. Poultry (n = 61) and cattle isolates (n = 51) were obtained from prefectures in Miyagi, Gunma, Akita, Kumamoto, Aichi, Hyogo, Yamaguchi (for poultry) or Oita and Hyogo (for cattle) during the same periods (Table S1, Fig. S1). The C. jejuni strain NCTC11168 [24] was used as a reference strain for comparative genomic and phenotypic studies. The C. jejuni peb4 mutant [25] was used as a donor strain for the natural transformation efficiency assay. Bacterial isolates were grown on Mueller-Hinton (MH) agar or MH broth (Becton Dickinson, Franklin lakes, NJ, USA) at 37uC in a humidified CO 2 AnaeroPack-Microaero gas system (Mitsubishi Gas Chemicals, Tokyo, Japan).

Multi-locus Sequence Typing (MLST) Analysis
Bacterial genomic DNA was extracted using a DNeasy Genomic Extraction kit (QIAGEN, Hilden, Germany) and stored at -20uC until further use. The PCR and cycle sequencing reactions were conducted according to the guidelines on the Campylobacter MLST database (http://pubmlst.org/ campylobacter/). We confirmed the absence of the non-specific PCR amplicon using a 1% agarose gel. ExoSAP-IT was subsequently used to purify the PCR products according to the manufacturer's instructions (Life Technologies, Carlsbad, CA, USA). For the sequencing reactions, we used both DNA strands in each allele with the BigDye terminator v. 3.1 Ready Reaction Cycle Sequencing kit on an ABI3730x DNA analyzer (Life technologies). The obtained sequences were assembled using CLC DNA Workbench software equipped with an MLST module (CLC bio, Aarhus, Denmark). The consensus sequences for each allele were assigned an allele number, a 7-locus (3,309 bp) sequence type (STs), and a clonal complex (CCs) through interrogation of the Campylobacter MLST database. Unassigned sequences were deposited in the database to obtain new allele or ST numbers according to the guidelines described on the website. The sequence data of each isolate were deposited in the database (each isolate ID is shown in Table S1).

Phylogenetic Analyses
The 7-locus allelic sequences were used to estimate the clonal genealogy of the STs using a model-based approach to determine the bacterial microevolution using ClonalFrame [26]. The output data were visualized using the iTOL (interactive Tree of Life) program (http://itol.embl.de/index.shtml) [27]. To assess clonal diversity and evolutional linkage, 7 allelic numbers for a total of 547 STs belonging to CC-21, were subjected to the Global Optimal eBURST (goeBURST, a java implementation of the eBURST algorithm proposed by Feil et al. [17]) [28].

Pulsed-Field Gel Electrophoresis (PFGE)
All ST-4526 isolates (n = 10, Tables 1 and 2) were subjected to PFGE with SmaI endonuclease (New England BioLabs, Ipswich, MA, USA) using the CHEF Mapper system (Bio-Rad Laboratories, Hercules, CA, USA) as previously described [29]. The gel images were obtained using ethidium bromide stain. The electrophoretic patterns from PFGE were compared based on band position using FingerPrinting II software (Bio-Rad Laboratories) and derived using the Dice coefficient with a maximum position tolerance of 1%. The strains were clustered using the unweighted pair group (UPGMA) method with arithmetic averages according to the manufacturer's instructions.

Genome Sequencing and Bioinformatics Analysis
Genomic DNA was extracted from three representative ST-4526 isolates (H_0002, H_0089, P_0053, which were selected randomly from the three different PFGE groups a, b, and c, respectively), using a Genomic tip-500 kit (QIAGEN, Hilden, Germany) and subjected to 8-kb paired-end pyrosequencing in a GS FLX+ system (Roche Diagnostics, Burgdorf, Switzerland) according to the manufacturer's instructions. The sequence data (151,327, 173,181, and 167,955 reads for H_0002, H_0089, and P_0053, respectively) were assembled using the GS De Novo Assembler version 2.6 (454 Life Sciences, Branford, CT, USA). 77, 63, and 54 contigs were obtained (.500 bases) for the isolates H_0002, H_0089, and P_0053 at 41, 50, and 48-fold coverages, respectively. The draft genome sequences reported in this paper have been deposited in the GenBank database with accession number DRA000568.
The detection of open-reading frames and annotation of all draft genomes were automatically performed using the RAST service pipeline [30]. The Artemis Comparison Tool (ACT; http://www.sanger.ac.uk/resources/software/act/) was used for genome-wide comparisons of the nucleotide sequences [31]. Genome comparisons were also performed using a protein sequence-based bidirectional BLAST (Bi-BLAST) approach. The Island Viewer, which is a computational tool that integrates three different genomic island prediction methods, i.e., IslandPick, IslandPath-DIMOB, and SIGI-HMM [32], was used to identify the genomic islands. The results of all Bi-BLAST comparisons are shown in Table S6.

LOS Isolation, SDS-PAGE, and Western Blot
Lipooligosaccharides (LOS) were prepared from C. jejuni grown on MH agar plates for 20 h at 37uC as described previously [33]. The samples were loaded onto 15% polyacrylamide gels and visualized using silver stain. The detection of LOS sialylation was performed using western blot analysis with cholera toxin B-subunit (Sigma Aldrich, St. Louis, MO, USA) as previously described [34].

Cloning, Expression, and Purification of Recombinant FHA Protein in E. coli
The gene encoding the putative filamentous hemagglutination domain protein (FHA) was PCR-amplified from ST-4526 isolates H_0002 and H_0089 using primers (forward for H_0002: CAC. CATGAAAAAGA TGAGTAAACATATAG, forward for H_0089: CACCATGGGAGTTTTGATT. GGCAAAACAG, reverse: TAATTCTTCTTTAAAAC-TAGCGAAAT), directly cloned into the pBAD202 D-Topo vector, and transformed into E. coli LMG194 (Life Technologies). The expression of the His-tagged protein was induced in the presence of 0.1% L-arabinose according to the manufacturer's instructions. The rFHA protein was purified using Ni-NTA resin (QIAGEN) according to the manufacturer's instructions.

Binding Affinity of rFHA to Heparan Sulfate
A total of 5 mg of heparan sulfate from porcine intestine (Sigma Aldrich) in sodium carbonate buffer (pH 9.6) was captured onto a 48-well tissue culture plate (TPP Techno Plastic Products AG, Trasadingen, Switzerland) at 4uC overnight. After washing with phosphate buffered saline (PBS, pH7.4) (Life technologies), 100 mg and serial dilutions of rFHA protein were added, and the plates were incubated at room temperature for 2 h. Following three washes with PBS, the bound proteins were solubilized in 20 mM Tris-HCl (pH 7.4) containing 0.1% SDS, and the protein concentrations were measured chronometrically using a Bradford assay.

Cell Adhesion/Invasion Assay
Caco-2 cell adhesive/invasive assays were performed as described [35]. Briefly, the cells were infected with C. jejuni at an multiplicity of infection (MOI) of 100, and incubated for 6 h. The wells were gently washed three times with sterile PBS (pH 7.4) (Life technologies), and the cells were detached using 0.2% saponin in PBS for 20 min. The suspensions were subsequently plated onto MH agar to count the colony-forming units (CFU). To enumerate the numbers of intracellular bacteria, a gentamicin protection assay was conducted after the same incubation period as previously described [35]. To minimize the effect of motility on cell adhesion, the culture plate was spun down (5006g, 3 min) immediately after the bacterial inoculation to synchronize the bacteria onto the cell surface. To examine the blocking effect of rFHA, the cells were pre-treated for 30min with 0, 5, or 20 mg of rFHA derived from the H_0002 isolate prior to C. jejuni infection.  The numbers of cell-associated bacteria were counted after gentle washing with PBS.

DNA Uptake and Natural Transformation Efficiency Assays
The DNA uptake assay was performed essentially as described [36]. Briefly, 0.1 mg of plasmid pRY108 DNA [37] labeled with [a-32 P]dCTP was co-incubated with approximately 4.5-5.8610 7 CFU of C. jejuni MH-broth suspension for 30 min. After washing and DNase I treatment, the intrabacterial uptake of radioactivity was measured using a scintillation counter. The transformation efficiency was also assayed in parallel according to the method of Wang and Taylor [38]. Briefly, equal amounts of C. jejuni MH-broth suspension as prepared above were incubated with 1 mg of chromosomal DNA from the C. jejuni peb4 mutant (carrying Cm-resistant cassette) [25] on MH agar supplemented with 5% horse blood for 6 h. The bacterial cells were subsequently resuspended in 1 ml of MH broth, and 50 ml of the bacterial suspension were spread onto MH agar plates supplemented with or without Cm (20 mg/ml). The numbers of colonies grown on each plate were comparatively measured to determine the transformation efficiency.

Statistics
The START ver.2 program [39] was used to determine the number of informative and parsimony informative sites, MLST alleles for each gene, and polymorphic sites according to the website guidelines (http://pubmlst.org/software/analysis/start2/). The genetic distances and rates of synonymous (dS) and nonsynonymous substitutions (dN) were calculated using the MEGA ver.5 (Molecular Evolutionary Genetic Analysis) program [40]. The MEGA software was also used to draw amino acid alignment for the FHA protein. The data from all phenotypic assays represent the means 6 standard deviations (SD) from at least three independent experiments.  (Table S2), and uncA retained the greatest synonymous means (123.8) and lowest dN/dS value (0.0106). The total site variation within each gene ranged from 3.0% (gltA) to      (Table S1).

Population Structure of the MLST Dataset
To reveal the population structure of the C. jejuni isolates, a ClonalFrame analysis was conducted. This phylogenetic program grouped the 62 STs into 6 clusters (I to VI) ( Fig. 2A); one of the predominant CCs, CC-21, was grouped into Cluster I. No cattle isolates were included in Clusters II and IV, which frequently included CC-460, CC-257, CC-353, and CC-354 from human and poultry populations. ST-58 was only identified in the cattle population (n = 8) and showed a distinct lineage from the other STs. Clusters III, V, and VI were generated in three (human, poultry, and cattle) populations. Thus, these data revealed a partial host association of distinct CCs/STs.

Evolution Linkage of ST-4526 within CC-21
The MLST analysis revealed that ST-4526 was represented in 7 human and 3 poultry isolates, although their isolation areas and dates were diverse (Tables 1 and 2). Therefore, a search for this ST in the Campylobacter MLST database led to the identification of a single poultry isolate that was also isolated in Japan (Isolate ID 10249). For a more comprehensive analysis of the possible patterns of evolutionary descent, the goeBURST (global optimal eBURST) analysis [28] was carried out using a total of 547 STs belonging to CC-21 deposited on the Campylobacter pubMLST database (as of 2, August, 2012). This approach resulted that the ST-4526 showed one of the most distant lineage from the original (founding) ST-21 genotype, together with the genotypes originated only from human and poultry (ST-883 and the derivatives, Fig. 2B). Other 13 STs reported only from Japan (i.e. ST-4253, ST-4387, ST-5233) were classified into different clusters (Fig. S2). Together, these data support the idea that ST-4526 might be distinctly evolved from the ancestor.

Summary of Genomic Features of C. jejuni ST-4526 Strains
Given the potent thriving and unique evolutional lineage of ST-4526, all ST-4526 isolates from human and poultry (n = 10) were macrogenotyped using pulsed-field gel electrophoresis (PFGE) analysis to determine their genomic similarities. As shown in Fig. 3, these isolates were divided into 4 groups (a-d), in which 3 poultry isolates (P_0016, P_0017, P_0053; group c) were closely related to the human isolates H_0089 and H_0092 that belonged to group b. To further clarify the genomic similarities and characteristics that might account for the thriving of these bacteria, three representative isolates (H_0002, group a; H_0089, group b; P_0053, group c) were genome sequenced, resulting in contigs comprising 1,625,252 bp (H_0002), 1,627,144 bp (H_0089), and 1,626,331 bp (P_0053). The sequences were subsequently processed with the RAST annotation pipeline [30], which annotated 1,836 to 1,865 features (Tables S3-S5). The three ST-4526 genomes were compared with the genomes of NCTC11168 (ST-43/CC-21, accession No. AL11168) and IA3902 (ST-8/CC-21, accession No. CP001876). DNA Plotter depicted the comparative genome circular maps for the genomes discussed above (Fig. 4).

ST-4526 Expresses Asialylated forms of Lipooligosaccharides (LOS)
The genome alignments using ACT (Artemis Comparison Tool) combined with a Bi-BLAST comparison of C. jejuni ST-4526 strains with strains NCTC11168/IA3902 clearly showed significant overall synteny between the five genomes. However, the LOS modification locus was not conserved (Cj1137c to Cj1144c) (Fig. 5A, Table S6). In contrast to strain NCTC11168, the LOS of the 3 ST-4526 isolates did not react with the cholera toxin, a ligand for GM1-oligosaccharide structures (Fig. 5B). Sialylated LOS structure is known to contribute to cell invasion [41] and gastrointestinal colonization in poultry [42]. The Caco-2 cell adhesion/invasion assays indicated, however, that the H_0002 and P_0053 isolates exhibited greater cell adhesion and invasion properties than the H_0089 and NCTC11168 isolates (Fig. 5C), providing an idea that the LOS differences did not essentially contribute to cell adhesion/invasion, hence certain determinant(s) might preferentially promote cell adhesion and invasion in H_0002 and P_0053.

ST-4526 Expresses a Putative Adhesin Protein with a Filamentous Hemagglutination Domain (FHA)
The Bi-BLAST analyses further revealed that the region from Cj0738 to Cj0752 was highly diverse between ST-4526 and NCTC11168 (Table S6). The Island Viewer [32] subsequently predicted that the three ST-4526 isolates contained a putative island comprising the region from Cj0737 to Cj0744, which harbored genes which differed between ST-4526 and NCTC11168/IA3902 (Fig. 6A). The amino acid alignment analyses showed that the Cj0742-homologue filamentous hemagglutination domain protein (FHA) was expressed in isolates H_0002, P_0053 and strain CF93-3 (accession No. CJJCF936_0827), which encoded a 693 aa protein with 100% sequence similarity; notably, this protein was truncated in H_0089 and NCTC11168/IA3902, with 585 aa (lacking Nterminus) and 358 aa (lacking C-terminus), respectively (Fig. S3). Other strains also exhibited distinct aa sequences, ranging from 358aa (strains 84-25 and DFVF1099) to 633aa (strain 1213) (Fig. S3).
FHA functions as an adhesin in Bordetella pertussis, showing binding affinity to carbohydrates, such as heparan sulfate, distributed on epithelial cell surfaces [43]; therefore, we produced recombinant FHA (rFHA) protein in E. coli (Fig. S4) and measured its affinity to heparan sulfate. The results showed that the rFHA from the H_0002 isolate showed greater binding  locus among the NCTC11168, IA3902, H_0002, H_0089, and P_0053 strains, which was imaged using the Artemis Comparison Tool (ACT). (B) SDS-PAGE and detection of sialylation in C. jejuni LOS. The samples were extracted from NCTC11168, H_0002, H_0055, H_0089, and P_0053 strains and visualized using silver stain after loading onto 15% acrylamide gels (left panel). In parallel, the same samples on the acrylamide gels were subjected to western blot analysis using the GM1-reactive, HRP-conjugated cholera toxin B-subunit to detect sialylated LOS (right panel). (C) Cell adhesion and invasion properties of NCTC11168 and C. jejuni ST-4526 representatives. No adhesive or invasive bacteria were shown in either blue bar (means corresponding to the left Y-axis) or red lines (means corresponding to the right Y-axis), respectively. doi:10.1371/journal.pone.0048394.g005 affinity to heparan than the rFHA from the H_0089 isolate (Fig. 6B), indicating that the N-terminus region of FHA protein was required for this binding affinity. To further support the idea that FHA mediated bacterial adhesion on Caco-2 cells (Fig. 5C), the cells were pre-treated with rFHA prior to infection with H_0002 and P_0053 isolates, and a significant reduction in their cell adhesion levels was observed (Fig. 6C). Thus, these data indicated that the ST-4526 isolates H_0002 and P_0053 expressed functional FHA, which functions as an adhesin.

ST-4526 Increases Antibiotic Resistance and Reduces DNA Uptake/Recombination
The single nucleotide polymorphism (SNP) analyses with NCTC11168 additionally revealed unique genetic variations in the genes for antibiotic resistance, DNA uptake/recombination, and flagellar motility in the ST-4526 (Table 3, Table S7).
(ii) DNA uptake and transformation efficiency. C. jejuni takes up exogenous DNA via the type II secretion system (T2SS), which is likely to affect genomic plasticity in this pathogen [36]. The Bi-BLAST analyses revealed a reduced amino acid similarity in the Cj1474c gene product (CtsD, a putative type II secretion system D protein) compared with that in NCTC11168 (% similarity at 65.5-66.3%) (Table S6). In addition, many variations were observed in the restriction-modification (R-M) system genes (Cj0690c, 39 SNPs, Cj0031, and 17 SNPs) ( Table 3, Table S7), which are also involved in a barrier against foreign DNA and bacteriophages [46], [47]. We thus reasoned that ST-4526 might lose the ability for uptake of exogenous nucleotides, thereby reducing recombination. The ST-4526 isolates showed a more reduced (2.0-3.0%) uptake of exogenous 32 P-labeled plasmid DNA than the NCTC11168 strain (Fig. 7A). Similarly, ST-4526 also exhibited reduced transformation efficiency compared with NCTC11168 (5.0-6.1% relative efficiency, Fig. 7B). Taken together, our data indicated that ST-4526 exhibited a loss-offunction for the DNA uptake and recombination.
(iii) Flagellar-mediated motility. A total of 4 genes for flagellar-mediated motility, which is prerequisite for establishing host colonization during an early course of infection [48], [49], [50], also showed sequence variations between ST-4526 and NCTC11168, with 22 SNPs in Cj1729c (flgE2), 20 SNPs in Cj1313 (pseH), 18 SNPs in Cj1312 (pseG), and 9 SNPs in Cj1317 (pseI) ( Table 3, Table S7). The soft agar motility assay, however, showed that the ST-4526 isolates were motile, with no significant difference with NCTC11168 (Fig. S5). Thus, we showed that the The FHA protein shows affinity to heparan sulfate in vitro. The tissue culture plates were pre-coated with heparan sulfate (5 mg/well). After washing with PBS, the plates were co-incubated with 100 mg and serial dilutions of recombinant FHA (rFHA) from isolates H_0002 or H_0089. The bound rFHA were solubilized with 0.1% SDS, and the protein concentrations were measured chronometrically using a Bradford assay. (C) The rFHA blocks the adhesion of FHA-positive C. jejuni on Caco-2 cells. The cells were pre-treated with 0, 5, or 20 mg of rFHA derived from H_0002 for 30 min prior to C. jejuni infection. The numbers of cell-associated bacteria were counted after gentle washing with PBS. doi:10.1371/journal.pone.0048394.g006

Discussion
Genetic attribution of bacterial genotypes has become a major tool in the investigation of the epidemiology of campylobacteriosis and has implicated retail chicken meat as the major source of human infection in several countries [51], [52]. In this study, we analyzed the population genetics of C. jejuni from different sources and areas in Japan using a MLST approach. We also attempted to understand the molecular basis underlying the unique and widespread predominance of ST-4526 in Japan using comparative genomic analyses. We identified several genomic traits beneficial for host adaptation, which seemingly accounted for the thriving of this genotype.
Many campylobacters are commensals that pass through a wide range of warm-blooded hosts and insects. Within farm animals, certain MLST lineages were identified in chickens and cattle, with different frequencies, whereas several CCs (i.e. CC-21) were identified at high frequencies in both sources [53]. CC-21 and -45 were the most frequent CCs in humans [54], [55], whereas CC-21, CC-45, and CC-574 were predominant in the poultry population in Great Britain from 2003 to 2006 [56]. In addition, Kwan et al. [57] reported the predominance of CC-61, CC-21, and CC-42 among cattle isolates in the United Kingdom. Our data are consistent with the finding for host assignment discussed above, regardless of geographic distance. In a recent large-scale MLST data analysis, Sheppard et al. [58] demonstrated that host association is more robust than geographic associations.
The results of the ClonalFrame analysis also revealed the presence of the rare genotype ST-58. In this study, however, this genotype was detected in 8 cattle isolates derived from the same location (Oita prefecture), although they were collected at different periods. Therefore, we could not rule out that this genotype might be adapted to the cattle population at a particular geographical location (farms). The additional collection of cattle isolates from different locations, including this area, would help to ascertain whether ST-58 also is a cattle-adaptive genotype.
There is an increasing evidence for the local and international transmission of host-associated lineages between food animal species [58]. In particular, CC-21, a widely distributed group, showed high genomic similarity regardless of origin [59]. We observed the frequent distribution of ST-4526 within CC-21 between the human and poultry populations of our collection. This genotype has not been previously reported from other countries; therefore, we examined the evolutionary linkage of this genotype within CC-21, which revealed a distinct lineage of this genotype from frequent genotypes worldwide. It remains unknown whether such phylogenetic distances represent bacterial traits for adaptation or host colonization. Lang et al. [60] showed that host species was a significant factor in explaining the genetic variation and macrogeography between Western Europe and the USA, and geographic variation of C. jejuni genotypes have been previously reported in Great Britain [56] and the USA [61]. The evolutionary linkage data herein thus suggested that ST-4526 genotype might be evolved in the Asian geography for adaptation. Comparative genomics between the closely related STs in goeBURST analysis (i.e. ST-4526 vs ST-883) would clarify whether the housekeeping genes-based clonal analysis might represent bacterial fitness for environmental survival or host infection, host specificity.
The availability of genome sequences facilitates a qualitative advance in our understanding of the epidemiology, ecology, and molecular biology of various organisms. To characterize the genomic features that account for the thriving of ST-4526, a comparative genomic analysis was performed for three ST-4526 isolates (H_0002, H_0089, and P_0053). This genotype showed (asialylated) LOS forms distinct from those of NCTC11168, although the latter showed forms similar to the widespread genotypes within CC-21. Habib et al. (2008) [62] showed a strong (85%) correlation between CC-21 and the sialylated class C LOS, indicating a potential role for class C LOS in the evolution of CC-21 that might be of particular importance in the poultry meatrelated transmission of C. jejuni to humans.
We also observed variations in island-like structures, which included putative adhesin, in the genome of C. jejuni; i.e., fig|6666666.15583.peg.771 to fig|6666666.15583.peg.779 in H_0002 (Table S3), which showed less similarity to NCTC11168 (corresponded to the locus Cj0736-Cj743) ( Table S6). The amino acid alignment revealed the truncation of the putative filamentous hemagglutination domain protein (FHA) among the C. jejuni strains. The recombinant FHA protein showed dose-dependent binding affinity to heparan sulfate, which competitively inhibited bacterial adhesion, suggesting a putative role for this gene product as an adhesin. Future studies will be focused on the in-depth biochemical characterization of this putative adhesin protein (i.e., The bacterial cells were co-incubated with P 32 -labeled pRY108 DNA, and the intrabacterial radioactivity was measured using a scintillation counter. The means in the NCTC11168 strain were set to 100% for the comparative analysis. (B) The natural transformation efficiency was assayed. The appearance of the Cm-resistant clones was normalized to the total number of colonies on the plates without antibiotics, and the test strains were further normalized to the means in the NCTC11168 strain, which were set to 100%. doi:10.1371/journal.pone.0048394.g007 determination of the active domain for cell adhesion, with screening of the host cell surface carbohydrates as interactive partners), which will be helpful for clarifying the biological role of this putative molecule during infection.
The SNP analyses in combination with a Bi-BLAST search revealed several genetic variations unique to the ST-4526 isolates; the polymorphisms in the topoisomerase genes (especially gyrA), could explain the increased resistance to NA and CPFX/LVFX, which is likely to confer an advantage for the spread and stability of this pathogen in poultry hosts as previously reported [63], [64]. This SNP was also identified in the genes for DNA uptake and recombination, and phenotypic or biochemical assays confirmed these reduced abilities. The genetic exchange of this pathogen was previously demonstrated in the chicken intestinal environment [65]. Nevertheless, the observed phenotypes in ST-4526 could thus explain the inability of this genotype to cause frequent genomic alterations, i.e., expanding the ability for clonal spread in human and poultry populations in Japan. It remains unknown how ST-4526 acquired such phenotypes; one possible explanation is that ST-4526 might evolve during the colonization of the ancestor in the poultry host, as passage in the chicken gut could alter the genetic and phenotypic traits of this pathogen [66]. With consideration for unique phylogeny, future studies will attempt to clarify the potential role of farm animals in the genomic evolution in this pathogen.
The  [22]. The predominance of these genotypes in both the human and poultry populations reported herein reinforced the need for continued research on the potential importance of poultry as reservoirs for these human pathogens. We are currently collecting human isolates at different periods to determine whether the thriving of the ST-4526 is temporal or continuous with geographical significance.
In summary, this study is the first to show the population structure of C. jejuni from different sources in Japan, which deciphered the potent thriving of the ST-4526 among human and poultry populations. The results of genomic and phenotypic analyses provided possible reasons for the wide spread of this genotype. Our continuous examination of the molecular basis underlying the genotypic alterations and phylogeography of this pathogen will provide further insight into the link between the evolution and ecological features of this pathogen. Table S1 Campylobacter jejuni isolates used in this study. * Novel STs found in this study were shown in bold. ** UA, unassigned to any CCs. *** Isolate ID, deposited onto the Campylobacter MLST database (http://pubmlst.org/perl/bigsdb/ bigsdb.pl?db = pubmlst_campylobacter_isolates&page = query).

(XLS)
Table S2 Summary of the MLST alleles statistics for the C. jejuni dataset in Japan. Seven alleles obtained from a total of 212 C. jejuni isolates ( = 62STs) were statistically analyzed by START2 program, providing the allele frequency, and polymorphisms. * dN/dS: Mean non-synomous substitutions per nonsinoumous site (dN)/mean synonymous substitutions per suynomous site (dS). ** Sites are categorized by the degeneracy of the codon they belong to, rather than that of the individual site. (XLSX )   Table S3 Genomic annotation of C. jejuni H_0002 (Table S3), P_0053 (Table S4), and H_0089 (Table S5).