Comparative studies of Campylobacter jejuni genomic diversity reveal the importance of core and dispensable genes in the biology of this enigmatic food-borne pathogen

Summary of recent events

MLST, DNA microarrays, and genome sequencing has allowed for a greater understanding of the metabolic capacity and epidemiology of Campylobacter jejuni. While strain-specific genes may provide an isolate a selective advantage in environments and contribute to the organism's pathogenicity, recent work indicates that C. jejuni pathogenicity is dictated by variations in the nucleotide sequence of core genes. Challenges facing C. jejuni researchers include determining: a) the degree to which genomic diversity enables this bacterium to persist in particular environments; b) if C. jejuni virulence and disease severity can be predicted based on genotype; c) the set of core and variable genes whose products contribute to virulence; and d) the genes in which nucleotide changes can affect a strain's pathogenicity.

Introduction

Campylobacter jejuni, a Gram-negative, spiral shaped bacterium [1], is one of the leading bacterial causes of food-borne human gastroenteritis. C. jejuni is currently estimated to cause 5 to 14% of diarrhea worldwide, which translates into 400−500 million cases per year [2]. Most cases of C. jejuni mediated gastroenteritis (campylobacteriosis) are characterized by nausea, abdominal cramps, diarrhea and fatigue. Although campylobacteriosis is most often self-limiting, certain strains of C. jejuni have been implicated as an antecedent to the development of Guillain-Barré Syndrome (GBS), an acute autoimmune mediated polyneuropathy characterized by ascending paralysis [3,4].

While outbreaks of campylobacteriosis occur, predominantly through consumption of contaminated milk and untreated water [5], most Campylobacter infections are sporadic in nature and linked to the improper handling and consumption of poultry. The linkage between human infection and the handling of raw poultry is not unexpected, as C. jejuni is a common commensal organism of chickens. In fact, C. jejuni colonize the intestinal tract of a variety of animals, including common livestock (cattle, sheep, pigs), domestic animals (dogs, cats), poultry, and wildlife (rabbits, pheasant, quail) [6-9]. A number of methods [e.g., Penner serotyping, Lior serotyping, fla-short variable region (SVR) sequencing, pulsed-field gel electrophoresis (PFGE), multilocus sequence typing (MLST)] are useful for the discrimination of C. jejuni isolates in epidemiological investigations. These methods have enabled investigators to identify the strain responsible for an outbreak [10,11]. The use of MLST, in particular, has provided researchers with the benefit of a defined molecular fingerprint to compare strains. The recent explosion of genome sequences and comparative genomic data has increased our understanding of the epidemiology and metabolic capacity of this organism.

The number of Campylobacter species and strain-specific genome sequences is increasing

The importance of Campylobacter in human gastrointestinal illness has resulted in at least eighteen isolates from eight different Campylobacter species having been or being sequenced (Table 1). The genomes of Campylobacter organisms are characterized by a low mol% (G+C) content (between 29.5% and 44.5%), small size (ranging from 1.5 Mb for Campylobacter lari RM2100 to 2.1 Mb for Campylobacter concisus 13826), and relatively few open reading frames (between 1425 and 1931 ORFs). The availability of sequenced genomes has aided in the development of methodologies to address questions concerning the relationship between genomic content and C. jejuni biology.

Table 1.

Features of Sequenced Campylobacter Genomes

Species/Strain	Size (Mb)	%GC	ORFs	Origin	Diseasea	GenBankb	Ref.
Campylobacter jejuni subsp jejuni NCTC 11168	1.64	30.5	1643	Clinical	Food Poisoning	AL111168	[36]
Campylobacter jejuni subsp jejuni RM1221	1.78	30.3	1835	Chicken		CP000025	[22]
Campylobacter jejuni subsp jejuni 81−176	1.6	30.6	1653	Clinical	Food Poisoning	CP000538	[24]
Campylobacter jejuni subsp jejuni 81116	1.63	30.5	1626	Clinical	Food Poisoning	CP000814	[43]
Campylobacter jejuni subsp jejuni CG8421	1.6	30.4	1512	Clinical	Food Poisoning	ABGQ00000000
Campylobacter jejuni subsp jejuni HB93−13	1.7	30.6	1710	Clinical	GBSc	AANQ00000000
Campylobacter jejuni subsp jejuni CG8486	1.65	30.4	1425	Clinical	Food Poisoning	AASY00000000	[44]
Campylobacter jejuni subsp jejuni CF93−6	1.67	30.5	1757	Clinical	MFSd	AANJ00000000
Campylobacter jejuni subsp jejuni 84−25	1.67	30.4	1748	Clinical	Meningitis	AANT00000000
Campylobacter jejuni subsp jejuni 260.94	1.65	30.5	1716	Clinical	GBS	AANK00000000
Campylobacter jejuni subsp doylei 269.97	1.8	30.6	1731	Clinical	Bacteremia	CP000768
Campylobacter coli RM2228	1.68	31.4	1765	Chicken			[22]
Campylobacter concisus 13826	2.1	39.4	1929	Clinical	Food Poisoning	CP000792
Campylobacter curvus 525.92	2.0	44.5	1931	Clinical	Periodontitis	CP000767
Campylobacter fetus subsp. fetus 82−40	1.8	33.3	1719	Clinical	Septicemia	CP000487
Campylobacter hominis ATCC BAA-381	1.7	31.7	1682	Clinical	Non-pathogenic	CP000776
Campylobacter lari RM2100	1.5	29.6	1554	Clinical	Food Poisoning		[22]
Campylobacter upsaliensis RM3195	1.66	34.5	1782	Clinical	GBS		[22]

^aDisease associated with origin of isolate

^bGenBank Accession Number

^cGuillain-Barre Syndrome

^dMiller-Fisher Syndrome

MLST provides greater insight into genetic diversity and population structure

MLST allows researchers to differentiate strains based on alleles at seven “unlinked” housekeeping loci [12]. Each unique allele is assigned a number based on its order of discovery and the combination of allelic numbers is the basis for its sequence type (ST). The ST is indicative of the isolate's genotype. This method is advantageous because it yields data that are accurate, reproducible, unaffected by changes in gene order, and readily comparable among laboratories. In addition to its usefulness in investigating C. jejuni outbreaks, MLST has demonstrated that C. jejuni are genetically diverse. Dingle et al. [12] found 194 C. jejuni isolates to be genetically diverse with a weakly clonal population structure. The genetic diversity was evident in that of the 155 STs observed, 104 STs could be grouped into 11 major genetic lineages or clonal complexes whereas the remaining 51 STs were unique. Manning et al. [13] performed MLST on 266 C. jejuni veterinary and human isolates and found that the populations overlapped among the 19 clonal complexes identified. Dingle et al. [14] applied the MLST scheme used for C. jejuni to Campylobacter coli and found the two species to share approximately 86% nucleotide sequence identity at the housekeeping loci and found some evidence for horizontal gene transfer. The investigators found additional evidence for horizontal gene transfer by sequencing the short variable region (SVR) of the flaA flagellin gene [14]. Evidence for horizontal gene transfer was also reported by Meinersmann et al. [15] using flaA SVR sequence anlaysis. These studies demonstrate that recombination between C. jejuni and C. coli occurs, and supports a hypothesis that these two species are continuing to evolve [16]. Taken together, these studies demonstrate the genetic diversity of Campylobacter strains. While it may never be possible to predict an isolate's pathogenicity or an individual's clinical symptoms or disease severity based on the ST alone, the use of MLST has already contributed to a greater understanding of C. jejuni population structures and their relationships with a variety of hosts.

Identification of hypervariable plasticity regions

The availability of genome sequences has made it possible to construct whole genome DNA microarrays for comparative genomic hybridization (CGH) analysis. Analysis of 11 C. jejuni clinical isolates by CGH revealed extensive genetic diversity and enabled the researchers to identify the genetic core of this organism [17]. Approximately 84% of the 1654 genes analyzed were common to all strains tested and encoded proteins involved in housekeeping functions, including metabolic, biosynthetic, cellular and regulatory processes. Strain-specific gene differences were involved in the biosynthesis and modification of cell surface structures including flagella, lipooligosaccharide (LOS) and capsular polysaccharide. Pearson et al. [18] examined the genomic diversity of 18 C. jejuni strains from different sources and found that the variable genes were often present in clusters, suggesting they were acquired or lost in groups during evolution [18]. In particular, seven hypervariable plasticity regions (PR) (Figure 1) were identified (PR1 – PR7). PR1 contained genes important in the utilization of alternative electron acceptors during respiration, possibly conferring a selective advantage in restricted oxygen environments. PR2, PR3 and PR7 contained genes encoding outer membrane and periplasmic proteins, which may be linked to phenotypic variation and adaptation to different ecological niches. PR4, PR5 and PR6 contained genes involved the biosynthesis and modification of flagella, LOS, and capsule. Further studies are required to elucidate the contribution of genes within the hypervariable regions to a strain's phenotype.

Figure 1.

Whole genome alignments of C. jejuni NCTC 11168, RM1221, 81−176 and 81116 performed using Mauve [42]. Each genome is laid out horizontally and homologous segments are shown as colored blocks. The average sequence identity is proportional to the height of the colored region with in each horizontal block. C. jejuni integrated elements (CJIEs) in strain RM1221 are indicated by blue rectangles. The seven plasticity regions (PR1-PR7) are indicated by red rectangles.

Meta-analysis of CGH data from four separate data sets revealed that many of the variable genes were absent or divergent in only one of the 97 strains [19]. In a separate study, phylogenomic analysis of 111 isolates from human, animals, and environmental sources provided insight into the population structure of C. jejuni. The investigators identified livestock and non-livestock associated clades (Figure 2A) [20]. Although analysis of C. jejuni isolates recovered from individuals with different clinical symptoms did not cluster (Figure 2B), more than half of the clinical isolates were phylogenomically related to strains from non-livestock sources. The investigators concluded that environmental sources serve as an important reservoir for C. jejuni infectious isolates. In contrast, Wilson et al. [21] concluded that livestock is the principal source of C. jejuni infection based on modeling DNA sequence evolution and rates of zoonotic transmission [21]. They suggested that the frequency of recombination in C. jejuni makes a single phylogenetic tree an inappropriate representation of the relationship between C. jejuni genomes [21]. Based on these studies, it is apparent that additional work is needed to determine the relative importance of reservoirs for C. jejuni.

Figure 2.

Phylogenomic relationship of C. jejuni strains. Strains are designated at the end of branches and are colored according to the (A) ecological niche from which the C. jejuni strain was isolated or (B) clinical symptoms or livestock/environmental source. P = 1.0 represents 100% of all phylogenies showing a given topology. Adapted with permission from the authors [20].

C. jejuni genomes are syntenic and some contain integrated elements

Presumably, all C. jejuni strains sequenced to date are pathogenic. As such, it is not possible to compare the genomic sequence of pathogenic and non-pathogenic strains. However, sequencing and comparative genomic analysis of five Campylobacter genomes (C. jejuni NCTC 11168, C. jejuni RM1221, C. coli RM2228, C. lari RM2100, and Campylobacter upsaliensis RM3195) revealed major structural differences between the strains [22]. While the genome of C. jejuni RM1221 was syntenic with the previously sequenced C. jejuni NCTC 11168, it contained four inserted genomic islands termed Campylobacter jejuni-integrated elements (CJIEs) (Figure 1). The CJIEs from C. jejuni RM1221 were not present among the other three C. jejuni sequenced strains (NCTC 11168, 81116, and 81−176). CJIE1 was found to be a Campylobacter Mu-like phage while CJIE2 and CJIE4 contained genes predicted to encode phage related endonucleases, methylases and repressors. A total of 73% of the CJIE3 predicted proteins showed sequence similarity with those encoded on the C. coli RM2228 megaplasmid, suggesting that it may be an integrated plasmid. Comparative genomic analysis of 67 C. jejuni and 12 C. coli strains from various geographical, clinical and veterinary sources revealed the CJIEs widely distributed, and more than half of the strains tested contained at least one CJIE [23]. C. coli RM2228 was also highly syntenic with C. jejuni RM1221, while C. lari and C. upsaliensis were not [22]. Noteworthy is that the regions of sequence variability aligned with the previously identified PR in C. jejuni NCTC 11168, suggesting the PR are likely physical loci within C. jejuni genomes. Additionally, CJIEs 2 and 4 appeared at loci adjacent to PR 2 and 5, respectively. The relevance of the CJIEs in the biology of C. jejuni remains to be determined.

Comparative genomic analyses have helped identify dispensable genes amongst C. jejuni isolates (i.e., genes absent or highly divergent in one or more of the isolates). The genes unique among the four C. jejuni sequenced strains (NCTC 11168, RM1221, 81−176 and 81116) are listed in Supplemental Table 1. The majority of these genes mapped to the variable loci identified previously and encode hypothetical proteins (Figure 3). Some of the unique genes were predicted to be involved in capsular polysaccharide biosynthesis, DNA modification, lipoprotein synthesis, or were phage-related. Additionally, the DNA sequences flanking the variable loci were conserved, suggesting that large regions of the C. jejuni chromosome are relatively stable. Sequencing of C. jejuni 81−176 revealed aerobic and anaerobic respiratory pathways that may confer advantages in low oxygen environments (i.e., gastrointestinal tract) [24]. Additionally, an ortholog of γ-glutamyltranspeptidase was identified, which is important for Helicobacter pylori colonization [25,26]. A mutation of this gene, resulting in loss of function, significantly reduced C. jejuni colonization of mice [24]. Together, these findings indicate that strain variable genes can provide an isolate an advantage in selective environments.

Figure 3.

Classification of genes unique to one of the four C. jejuni strains (NCTC 11168, RM1221, 81−176 and 81116). Unique genes are listed in Supplemental Table 1 and were classified based upon genome annotations

Nucleotide variations alter pathogenic behavior

The availability of C. jejuni genome sequences has provided the genetic basis for investigating the metabolism, gene regulation, and physiology of these organisms. These data indicate that many of the previously identified putative virulence determinants, including cytolethal distending toxin (CDT), various adhesins (e.g., CadF, JlpA, and PEB1), and the flagellar structural proteins, are conserved amongst strains [17]. Despite conservation of these genes, it is likely that the variations in the pathogenicity of C. jejuni strains are influenced by nucleotide differences in virulence genes. In this regard, none of the techniques discussed above are capable of determining whether point mutations, nucleotide insertions, nucleotide deletions, and gene rearrangements are present in one strain versus another. Below we provide specific examples of where nucleotide variations may affect the pathogenicity of C. jejuni strains.

Comparative analysis of CDT, a multi-subunit toxin [27,28], in C. jejuni, C. coli, and Campylobacter fetus revealed that the cdt gene cluster is widely distributed in a species-specific manner [29]. CDT has been shown to induce tissue damage and fluid accumulation in the colon of infected mice [30]. However, the contribution of CDT in campylobacteriosis is not clear, as C. jejuni CDT-negative strains have been isolated from individuals with clinical signs of diarrhea [31]. Detection of the cdtABC gene cluster does not necessarily result in the expression and synthesis of a functional product [31]. AbuOun et al. [31] identified two mutations that result in a CDT-negative phenotype. One mutation was characterized by a 667 bp deletion, encompassing a portion of both the cdtA and cdtB genes, and the second mutation was a nucleotide change that resulting in nonsynonymous residue at position 95 of CdtB [31]. This example illustrates the importance of performing experiments to test for a functional product even when a gene is detected by hybridization.

While not well understood, single nucleotide variations may contribute to molecular mimicry of gangliosides concentrated in peripheral nervous tissue by the LOS of C. jejuni [32,33]. Although CGH analyses of C. jejuni strains from individuals with GBS and uncomplicated gastroenteritis have failed to identify GBS associated genes [34], direct sequence analysis has revealed information regarding phase variation in the LOS biosynthesis genes [35]. Analysis of the genome sequence of C. jejuni 81−176 identified a homopolymeric G tract in cgtA, which encodes N-acetylgalactosaminyltransferase. Variation in the number G residues resulted in differences in the LOS core structure, affecting ganglioside mimicry and the ability of this strain to invade epithelial cells in vitro [35]. This example illustrates that slipped-strand mispairing occurs during DNA replication at homopolymeric tracts consisting of consecutive guanine residues. The resulting differences in homopolymeric tract length can affect translation and result in phase variation. Given that these regions are frequently associated with genes that synthesize surface structures [22,24,36] their presence may influence a strain's pathogenicity and immunogenicity. Lastly, it is worth noting that although homopolymeric G tracts are common in the C. jejuni strains sequenced, their number varies among strains.

Comparison of the C. jejuni NCTC 11168 (11168-GS) isolate used for generating the genomic sequence and the original clinical isolate (11168-O) revealed differences in their virulence-associated phenotypes including colonization, invasion, translocation and motility [37]. Gene expression profiling of these two strains revealed dramatic differences in the expression of flagellar and motility related genes. Sequence analysis of three sigma factors (i.e., RpoD, RpoN, and FliA) in the 11168-GS and 11168-O strains identified single-nucleotide polymorphisms (SNPs) in each sigma factor gene, resulting in at least one amino acid substitution in each sigma factor. The investigators proposed that the differences in gene expression were due to changes in global gene regulation. The possibility that a nucleotide change in a sigma factor could influence gene expression on a global level was subsequently proven by the identification a defective rpoN gene in C. jejuni strain S2B and its complementation [38].

Malik-Kale et al. [38] assessed the virulence potential of two C. jejuni strains (Turkey and CS) that were indistinguishable by PFGE, MLST, and CGH. Interestingly, these two C. jejuni strains showed dramatically different virulence potential in both in vitro cell culture and piglet models. Gene expression analysis revealed dramatic differences in gene expression profile. In particular flagellar Class II genes were found to have lower expression in the C. jejuni CS strain (i.e., the less virulent isolate), suggesting expression of these genes are important to pathogenesis by C. jejuni. Additionally, DNA sequence analysis of genes that regulate flagellar synthesis revealed a point mutation in flgR may be responsible for the loss of virulence. Previous work indicated that flagellar synthesis is modulated via phase variation [39]. In particular, phase variation of FlgR is due to the loss or gain of a nucleotide in homopolymeric adenine or thymine tracts within flgR. The identified point mutation in flgR did not occur in a region previously identified as being phase variable. Based on these data, we conclude that allelic variation may also dictate a strain's pathogenic potential.

Conclusions and future perspectives

The identification of genetic markers predictive of ecological source and virulence potential are important to detecting and preventing the dissemination of C. jejuni via food sources. As we have reviewed, MLST, DNA microarrays, and genome sequencing of C. jejuni strains have demonstrated the genetic diversity of this important food-borne pathogen. Comparative genomic studies have demonstrated C. jejuni population structure relates to ecological source (livestock versus non-livestock sources) and identified genes (e.g., Cj1321−1326 in C. jejuni NCTC 11168) that are associated with livestock sources. Additionally, DNA sequence analyses implicate phase and allelic variation as possible mechanisms for altered gene expression and protein synthesis. In spite of recent advances, significant gaps exist in our knowledge of C. jejuni biology. First, researchers have yet to uncover a correlation between genomic diversity and disease severity. Second, C. jejuni virulence and disease pathology are not yet predictable based on genotype. Third, the core genes necessary for disease and the variable (i.e., dispensable) genes whose products contribute to C. jejuni disease are not known. Fourth, based on the observation that nucleotide changes in certain genes alter a strain's pathogenicity, studies are needed to identify additional genes/proteins whose expression/function is influenced by nucleotide/residue variations. To address these questions, a small infectious disease animal model is needed to test the pathogenic potential of C. jejuni isolates. However, few animal models are currently available to assess C. jejuni virulence [40]. One possibility is the use of the interleukin-10-deficient murine model to determine the relationship between C. jejuni genetic variation and disease spectrum [41]. Continued work focusing on the relationship of genotype to phenotype is important in understanding this enigmatic organism.