Skip to main content

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

Campylobacter jejuni genotypes are associated with post-infection irritable bowel syndrome in humans

Abstract

Campylobacter enterocolitis may lead to post-infection irritable bowel syndrome (PI-IBS) and while some C. jejuni strains are more likely than others to cause human disease, genomic and virulence characteristics promoting PI-IBS development remain uncharacterized. We combined pangenome-wide association studies and phenotypic assays to compare C. jejuni isolates from patients who developed PI-IBS with those who did not. We show that variation in bacterial stress response (Cj0145_phoX), adhesion protein (Cj0628_CapA), and core biosynthetic pathway genes (biotin: Cj0308_bioD; purine: Cj0514_purQ; isoprenoid: Cj0894c_ispH) were associated with PI-IBS development. In vitro assays demonstrated greater adhesion, invasion, IL-8 and TNFα secretion on colonocytes with PI-IBS compared to PI-no-IBS strains. A risk-score for PI-IBS development was generated using 22 genomic markers, four of which were from Cj1631c, a putative heme oxidase gene linked to virulence. Our finding that specific Campylobacter genotypes confer greater in vitro virulence and increased risk of PI-IBS has potential to improve understanding of the complex host-pathogen interactions underlying this condition.

Introduction

Irritable bowel syndrome (IBS) is a chronic, disabling gastrointestinal disorder that affects up to 15% of people worldwide1. Patients suffer from chronic symptoms of abdominal pain and alteration in frequency and/or form of the stool. Acute infectious gastroenteritis is a major risk factor for development of this condition with the onset of chronic symptoms often emerging post-infection (PI-IBS)2. Most PI-IBS patients have abdominal pain with either diarrhea or alternating episodes of diarrhea and constipation3. Campylobacter species are among the most common causes of bacterial enterocolitis worldwide and estimates of PI-IBS rates associated with this enteric pathogen are greater than 20%3,4. It is known that the severity of Campylobacter enterocolitis is linked to the likelihood of developing PI-IBS. For example, bloody stools, hospitalization, and duration of acute illness are known to be associated with increased risk of PI-IBS development3,5,6. However, while Campylobacter species and strains vary in importance as gastrointestinal pathogens, the lineages associated with PI-IBS and underlying genetic drivers are poorly understood.

Over 90% of human Campylobacter infections are caused by C. jejuni or C. coli, mostly acquired through consumption of contaminated meat and poultry7. Acute intestinal injury from Campylobacter species as well as the host response play a role in the pathogenesis of PI-IBS. In vitro studies have shown that C. jejuni can invade epithelial cells8,9, disrupt tight junctions8 and affect intestinal fluid transport10. In severe cases, acute infection can lead to impairment of the colonic barrier and absorptive function11. It is clear that not all C. jejuni strains are equally important as human pathogens, but the extent to which this reflects different infection dynamics or variable pathogenicity among strains is not well known12,13.

There is evidence that some strains are carried asymptomatically14 while others are associated with sequelae such as reactive arthritis or Guillain-Barre syndrome15. Furthermore, in vitro assays have shown variation in the propensity of strains to invade epithelial cells and invoke pro-inflammatory cytokine (IL-8) response16,17. Key to explaining the differential pathogenicity among strains is gaining an understanding of the genetic basis of phenotypic variation, specifically the genes that underlie traits conferring pathogenic potential. Outer membrane protein, toxin production, and flagella biosynthesis genes have all been implicated as virulence determinants17,18,19,20,21. However, to date, no single gene, plasmid, phage or pathogenicity island has been causally linked to acute illness in humans.

In this study, using Campylobacter isolates from a well-characterized cohort of patients who developed Campylobacter PI-IBS and those who did not, we identify sequence types and lineages associated with development of PI-IBS. Applying a genome-wide association study (GWAS) approach linked to clinically relevant phenotypes tested in vitro, we identify genes and genetic elements associated with pathogenicity and prediction of PI-IBS development. This approach has the potential to improve understanding of the genetic determinants of virulence and allows the calculation of a risk score for individual genotypes that, with further validation, could be a basis for medical interventions aimed at preventing PI-IBS following Campylobacter infection.

Results

Campylobacter isolates associated with PI-IBS cases and controls

As a part of statewide surveillance, the Minnesota Department of Health collects data on symptoms and exposures within 7–10 days upon notification of culture-positive Campylobacter cases and collects fecal samples. For this study, from 2011 through 2019, we prospectively sent surveys (Rome III and IBS symptom severity) to patients 6–9 months after Campylobacter infection. Patients with IBS prior to the infection or history of inflammatory bowel disease, celiac disease, and microscopic colitis were excluded from the prospective survey. The Rome III criteria requires presence of abdominal pain or discomfort at least 2–3 times a month and alterations in bowel function (stool consistency or frequency) for ≥6 months to demonstrate chronicity of symptoms3,22. Ninety-four isolates were collected from the Minnesota Department of Health and sequenced, of which 88 were C. jejuni and 6 C. coli (Supplementary Data 1). Infection with 49 of the C. jejuni isolates led to PI-IBS, whereas 30 did not (defined as control or PI-no-IBS isolates). Nine C. jejuni isolates lacked complete follow up data. The mean (SD) IBS symptom severity score (0-500 scale) for PI-IBS was 182.6 (117.6) suggesting moderate IBS symptom severity (defined by a score range 175–300)23. Demographic and clinical details of PI-IBS patients and controls are provided in Table 1.

Table 1 Demographic and clinical characteristics of patients during acute C. jejuni gastroenteritis.

Acute Campylobacter infection isolates are genetically diverse

A maximum likelihood phylogeny constructed from a concatenated gene-by-gene core genome alignment revealed a highly structured population (n = 94; Fig. 1a, Supplementary Data 2). Isolates shared 1225 core genes (loci present in >95% of genomes; equivalent to 75.7% of the NCTC11168 reference genome). C. jejuni (n = 88) and C. coli (n = 6) isolates clustered into 59 common sequence types (ST), as defined by seven MLST genes24 (Fig. 1a; Supplementary Data 3). Acute infection was predominantly caused by host generalist lineages, previously associated with more than one source reservoir (n = 25 for ST21 and ST45, most common clonal complex isolates); and poultry specialist (n = 18 for ST353, ST354 most common clonal complexes) (Fig. 1b)25. Fewer acute infection isolates were from ruminant-associated lineages (Supplementary Fig. 1A, B).

Fig. 1: Population structure of Campylobacter isolates collected in this study.
figure 1

a A maximum-likelihood phylogeny was constructed with IQ-TREE, using a GTR model and ultrafast bootstrapping (1000 bootstraps; version 1.6.8)99,100 from an alignment of all isolates (n = 94). Scale bar represents a genetic distance of 0.01. Leaves from isolates that lead to an onset of PI-IBS are colored red and those that did not (control) are blue; isolates for which follow up date was incomplete are gray. The tree is annotated with MLST, LOS, and Penner (capsule) types, presence (colored block) and absence (no block) of common Campylobacter virulence factors (VFdB), antimicrobial resistance genes (ARG-ANNOT) and the presence of two previously characterized plasmids (pTet: NC008790 and pVir: NC008770; presence indicated if 75% of more plasmid genes present). Interactive visualization is available on Microreact101: https://microreact.org/project/CampyIBS. b Ratio of how many isolates from each MLST clonal complexes (CC) lead to the onset of PI-IBS. Only CCs represented by three or more isolates are shown. Frequency of c clonal complexes, d LOS classes and e Penner types identified. p-value < 0.05 indicated with (*); Fisher’s LSD test. Raw data for (be) available in Supplementary Data 4.

PI-IBS emerges from multiple C. jejuni lineages

Risk of PI-IBS following infection by isolates from different lineages varied (36 STs for PI-IBS and 21 STs for controls), with 18 different STs representing both PI-IBS and control C. jejuni. The ST922 clonal complex isolates were consistently associated with PI-IBS (3 of 3 cases; Fig. 1b, c), as were three quarters of the ST22 clonal complex (3 of 4) infection cases (Fig. 1b, c). The globally disseminated host generalist sequence types, ST48 and ST206 clonal complexes, also lead to the onset of PI-IBS in more than three quarters of cases (4 of 5) (Fig. 1b, c). Conversely, neither acute infection with cattle associated ST42 clonal complex isolates (0 of 4), nor two thirds of cases involving the poultry-associated ST460 clonal complex, lead to the development of PI-IBS (1 of 3 led to PI-IBS; Fig. 1b, c). Despite ST21 clonal complex being the most common (n = 18) lineage in acute infection, PI-IBS developed in just over half (55%) of cases. No significant difference in genome size was noted between C. jejuni isolates from PI-IBS cases and controls (Supplementary Fig. 1C). Despite differences between lineages, no source reservoir dominated; rather isolates from all ecological backgrounds were equally capable of leading to PI-IBS (Supplementary Fig. 1D).

Association between lipooligosaccharide and capsule types and PI-IBS

The lipooligosaccharide (LOS) and capsular polysaccharide (CPS) loci are highly variable in gene content26,27,28,29 and this variability is reflected in the diversity of LOS and capsule types (Penner typing) for the PI-IBS and PI-no-IBS isolates (n = 10 LOS types; n = 16 capsule/Penner types; Fig. 1d, Supplementary Data 4). The most common LOS class locus was class B (n = 10 class B1; n = 18 class B2), which was evenly distributed between isolates that did and did not develop IBS. Most LOS class B isolates that developed IBS were of class B2 (n = 11/18). LOS class C was identified in 19 isolates, of which 9 (47%) developed PI-IBS. Three LOS classes showed some over-representation in PI-IBS isolates. LOS class EHOP was present in 20 isolates and 13/20 (65%) of these were from PI-IBS cases. LOS class F was present in six isolates and 4/6 (66%) of these were from PI-IBS cases. There were ten isolates with LOS class A and 80% developed PI-IBS symptoms (5 of 7 class A1; 3 of 3 class A2).

The most common CPS Penner type was HS:4-complex (n = 18) and exactly two thirds of these isolates were from PI-IBS cases (n = 12/18) and all were of LOS classes A and B (Fig. 1e, Supplementary Data 4). Additionally, Penner type HS:3,17 was present in ten isolates and 7/10 (70 %) of these were from PI-IBS cases (both Penner types statistically significantly over-represented in PI-IBS). Penner HS:19 was present in four isolates, with three quarters from PI-IBS cases (n = 3/4). Some of the isolates that lacked complete follow up data (1 of 4 ST-22CC, 2 of 7 LOS class A1, and 1 of 4 Penner type HS:19) were also from these high risk lineages raising the question if they might be able to better predict PI-IBS development in larger datasets.

Pathogenic Campylobacter have an open pangenome

Construction of the pangenome using PIRATE on 88 C. jejuni and 6 C. coli isolates in our collection identified an open pangenome30, meaning that the number of genes in the pangenome continues to increase with each additionally sequenced isolate (Supplementary Data 5). Accessory genes represented nearly three-quarters of the pangenome (3410 genes; 74% of pangenome; Supplementary Fig. 2) with a quarter of the genes identified (1225, 26%) considered core genes present in 95% or more of the isolates. A large proportion of the accessory genome, 2571/3,410 accessory genes (75.4%), was present in fewer than 15% of isolates. Differences in the distribution of pairwise genomic distances are indicative of multiple genetically distinct clusters that are diverging in both core sequences and accessory gene content. Visualization of this clustering revealed distinct C. jejuni and C. coli populations (Supplementary Fig. 1D). C. jejuni isolates clustered into three groups. One was composed primarily of isolates from poultry-associated clonal complexes (ST257, ST353, ST354, ST460, and ST607). Another was composed primarily of host generalist clonal complex isolates (ST21, ST52, and ST48). The third group consisted of isolates from the other large host generalist clonal complex ST21, the Guillain-Barre syndrome associated ST22 clonal complex15, wild bird lineages (ST179 and ST692 clonal complexes) and ruminant-associated clonal complexes (ST61 and ST42; Supplementary Fig. 1D).

Known virulence factors do not fully explain the onset of PI-IBS

Virulence is often associated with variation in accessory genome content and antimicrobial resistance genes7. In our isolate collection, over half the isolates harbored resistance elements for tetracycline (63.8% tetO) and/or beta-lactamase (61.7% blaOXA-61); over a third (35.1%) harbored the fluoroquinolone resistance single nucleotide polymorphism (SNP), gyrA C257T; small numbers of isolates harbored aminoglycoside resistance elements [12.8% aph(3’)-III; 5.3% aph7; 3.2% sat4A; 1.1% aadE]; and two isolates harbored a SNP in the 23S rRNA (A2075G) linked to erythromycin resistance (Fig. 1a; Supplementary Data 4). Tetracycline resistance in more than a third of the isolates (36.2%) was associated with a previously identified C. jejuni plasmid, pTet (accession: NC008790).

The presence of known virulence genes associated with Campylobacter adhesion, invasion, motility, toxin, and secretion system was also identified. Outer membrane proteins (cadF, pebA, porA), invasion proteins (ciaBC), glycosylation systems (N-linked, O-linked), motility (flagella genes), cytolethal distending toxin (cdtABC), lipooligosaccharide locus (LOS), and capsule production locus (CPS) were present in most isolates (74.5–100%), but only 4 isolates encoded a Type IV secretion system associated often found on the C. jejuni plasmid pVir (accession: NC008770). We did not observe significant differences in the frequency of virulence factors and antibiotic resistance genes between isolates from the PI-IBS cases and controls (Fig. 1a, Supplementary Data 4).

Pangenome-wide association study of PI-IBS-associated genes

A series of customized bacterial pGWAS were performed to investigate variation in the core and accessory genome of all C. jejuni isolates with PI-IBS follow-up data (n = 79). C. coli genomes (n = 6) were not included in the GWAS in order to minimize the confounding effects of bacterial lineages on identification of significant associations. Genomes from isolates that elicited PI-IBS (n = 49) were compared to control isolates (n = 30) where follow-up data indicated no disease sequelae. Input matrices of core SNPs, gene presence, allelic variation, and gene fusions and duplication were prepared following removal of low frequency alleles/genes (present in <0.05% of isolates) and SNPs (present in <0.1% of isolates).

Three correlation scores were employed (terminal, simultaneous, and subsequent) in order to differentiate between within- and between-lineage associations (Supplementary Data 6). In total, we identified variation in 24,528 genetic elements associated with PI-IBS (Table 2; Bonferroni corrected p-value ≤ 0.05): 22,686 SNPs, 1392 alleles and 211 genes and 239 gene duplication, fissions or fusions. The subsequent score incorporates a clonal frame phylogeny that accounts for the impact of recombination (Fig. 2a; Supplementary Data 7), reducing the effect of population structure and maximizing the chance of identifying elements associated with the onset of PI-IBS. In total, 6311 subsequent associations were identified with a Bonferroni corrected p-value below 0.05: 5,896 SNPs; 335 alleles, 35 genes and 45 gene duplication, fissions or fusions, representing less than 4% of the total variants investigated (Fig. 2b). Associated variants were mapped back to the PIRATE pangenome file, which incorporated the NCTC 11168 reference genome to allow comparison with established gene nomenclature and help identify putative gene function. The variants that were most significantly associated with PI-IBS (p-value <7.5 × 10−5) were in genes linked with bacterial stress response (Cj0145_phoX) and several genes involved in core biosynthetic pathways (biotin: Cj0308_bioD; purine: Cj0514_purQ [x2]; and isoprenoid: Cj0894c_ispH; Fig. 2a).

Table 2 Summary of all associated GWAS variants (p-value < 0.05), including terminal, simultaneous and subsequent scores from treeWAS110.
Fig. 2: Genome-wide association of genetic variants with onset of PI-IBS.
figure 2

a ClonalFrame-ML phylogeny of PI-IBS isolates, with recombination masked114. Red leaves are isolates from PI-IBS patients and leaves with blue circles are from control patients. Common disease-associated clonal complexes (CCs) are highlighted. Interactive visualization is available on Microreact101: https://microreact.org/project/CampyIBS-CF. Matrix shows presence of variants most associated with IBS, which are labeled below (terminal, n = 5; simultaneous, n = 9; subsequent, n = 6). b Pangenome position of associated GWAS results (subsequent p-value < 0.05; n = 6,311; Table S4). Each circle represents a variant mapped to a position in the C. jejuni reference strain NCTC1116856 or inferred pangenome from this study (Supplementary Data 5). Variants with the strongest association (p-value < 0.000075; n = 6) from four genes with homologs in NCTC11168 are annotated (Cj0145:phoX, Cj0308:bioD, Cj0514:purQ, and Cj0894c:ispH). c Hot spots of associated variants. The number of associated variants per gene in NCTC11168 and the pangenome (any association p-value < 0.05). Genes with more than 100 associated variants are annotated, including four from the O-linked glycosylation locus. d Breakdown of associated variant types (core genome SNPs, core and accessory alleles, gene fission/fusions, accessory gene presence, and gene duplications; any association p-value < 0.05, raw data are available in Supplementary Data 6).

In addition to identifying variants with the strongest association with PI-IBS, we identified hot spots of associated variants (Fig. 2c). Large numbers associated variants (multiplied terminal, simultaneous and subsequent p-value below 0.000125; Fig. 2c) were found in Cj0208 (DNA modification methylase), Cj0261c (transducer-like, methyl-accepting chemotaxis proteins), Cj0533 (succinyl-CoA ligase; biosynthesis of secondary metabolites), Cj0720_flaC (flagellar formation), the O-linked glycosylation locus (Cj1293, Cj1300, Cj1334_maf3, Cj1351_pldA; flagellar glycosylation), Cj1414c_kpsC (capsule biosynthesis), Cj1729c_flgE2 (flagellar secretion) and accessory genome gene cluster g00183 (hypothetical protein). Associated elements were predominantly core genome SNPs (Fig. 2d) with the most stringently associated elements found in multiple lineages (Fig. 2a).

Risk-score provided varying predictive accuracy for different lineages

The 1000 most associated GWAS elements (subsequent p-value below 0.0075) were tested for their ability to predict the onset of PI-IBS and their suitability as markers of infection risk. Twenty-two markers were able to accurately predict the onset of PI-IBS in more than 60% of cases (Fig. 3a). Four of the most accurate predictor variants were found in a single gene, Cj1631c a putative heme oxidase linked to virulence31 (Table 3). Two predictors were identified in genes that contained many associated variants (flagellin C; Cj0720c_flaC and phospholipase A; Cj1351_pldA). By weighting each marker by its subsequent association score, which incorporates the direction of association, presence in multiple lineages, and strength of association, we used these 22 markers to calculate a cumulative score for each isolate and estimate the risk of developing PI-IBS (Fig. 3b).

Fig. 3: Predicting risk of PI-IBS using risk markers derived from the primary GWAS.
figure 3

a Ability of top 1,000 (of 6,311) variants from the GWAS to predict the onset of PI-IBS (variants with subsequent p-value below 0.0075). b ClonalFrame-ML phylogeny of IBS isolates, with recombination masked. IBS onset is denoted by red (PI-IBS) and blue (control) squares. Common disease-associated clonal complexes (CCs) are highlighted. Predicted risk score indicated by the red or blue bars (color denoting predicted outcome). High risk isolates scored above 1500 and small tick () or cross (✗) indicated whether the prediction observed disease status. c Repartition of risk scores on 83 isolates used in the GWAS from acute C. jejuni infection, according to the onset of PI-IBS within 9 months. Each point corresponds to the risk score associated with a single isolate. Risk score was calculated by weighting the association score of the top 22 PI-IBS predictive variants (Table 2). PI-IBS was predicted with a score above 1500, which showed variation between lineages (disease outcome was correctly predicted in 65 of 83 isolates). d Repartition of risk-score prediction by LOS class and e Penner types. *All box plots show the 25th and 75th percentile, whiskers extending to the min and max values with all data points shown. Raw data for  (be) are available in Supplementary Data 9.

Table 3 Details of GWAS-informed markers (n = 22) used to predict risk of subsequent PI-IBS.

Reflecting our empirical clinical evidence, we observed variation in risk between clonal complexes represented by 3 or more isolates (Fig. 1c, Supplementary Data 8). The highest risk was posed by isolates from ST48 (n = 5), ST45 (n = 5), ST206 (n = 5), ST22 (n = 3), and ST922 (n = 3) clonal complexes (mean risk score 1593). The ruminant associated ST42 clonal complex (n = 4) was predicted to be the least likely to lead to PI-IBS, with the lowest mean risk score (−1344). Risk prediction was also influenced by lineage, with accuracy varying from 67% in ST460 and ST257 clonal complexes to 100% accuracy in clonal complexes ST922, ST42, ST22, and ST52 (Fig. 3b, c). Repartition of the risk scores according to LOS class and Penner types also identified LOS classes A1 (1593) and EHOP (1584); and Penner types HS:19 (1593), HS:3,17 (1575) as high-risk lineages (Fig. 3d, e). Our clinical data suggested that Penner type HS:4-complex was also high-risk (Fig. 1e), but variability in isolate scores for this well-represented group (n = 18) lowered group’s mean risk score and predictive accuracy (1415).

PI-IBS isolates are more adhesive and invasive on colonocytes

A random set of PI-IBS and control isolates (n = 28/group; each isolate tested in biological and technical replicates; indicated in Supplementary Data 9) was used for determining in vitro virulence. Adhesion was found to be significantly higher with PI-IBS (2.19 × 10−3 ± 1.71 × 10−3) compared to control strains (1.75 × 10−3 ± 2.68 × 10−3), Mann–Whitney, p value = 0.0025 (Fig. 4a). Invasion was also found to be significantly higher with PI-IBS strains (5.65 × 10−4 ± 1.77 × 10−3) compared to control strains (1.55 × 10−4 ± 2.33 × 10−4), p value = 0.005 (Fig. 4b). Results are expressed as (mean ± SD). Adhesion and invasion showed strong positive correlation (Spearman correlation, r = 0.87, p < 0.0001) (Fig. 4c). TER drop (% from baseline) was not significantly different between PI-IBS and control strains (PI-IBS: 90.07% ± 7.98 vs. control: 83.79% ± 16.05, p = 0.058) at 24 h post-infection (Fig. 4d).

Fig. 4: In vitro pathogenicity phenotypes of PI-IBS and control C. jejuni isolates.
figure 4

a PI-IBS causing C. jejuni isolates demonstrate greater adhesion to T84 epithelial cells. b Using gentamycin protection assay, PI-IBS isolates demonstrate greater intracellular invasion compared to the control isolates. c Positive correlation between adhered and invaded bacteria on T84 colonocytes (non-parametric Spearman correlation, r = 0.87; p-value < 0.0001). d T84 colonocytes were grown as epithelial monolayers with transepithelial resistance (TER) > 750 Ω cm2. No differences in % drop in TER in monolayers exposed to PI-IBS causing C. jejuni isolates compared to control isolates. (n = 28/group, Mann–Whitney U test). *All box plots show the 25th and 75th percentile, whiskers extending to the min and max values with all data points shown. Raw data are available in Supplementary Data 9.

Variation in accessory genes is associated with in vitro virulence

Endonuclease (Cj0139), Campylobacter adhesion protein capA (Cj0628), O-linked glycosylation (Cj1293-Cj1342) and glycan transferase (Cj1438) represented the greatest number of unique alleles per isolate (Fig. 5a). Prevalence of core gene variants acetyltransferase (Cj0295), and the genes with uncharacterized/unknown function (Cj0567, Cj0569, Cj0570) were higher in PI-IBS while Cj1392 was higher in control isolates (Fig. 5b). Adhesion was associated with capA (Cj0628), periplasmic protein (Cj0967), β-1,3 galactosyltransferase (Cj1139) and O-linked glycosylation (Cj1293-Cj1342) (Fig. 5c). Invasion was also associated with capA (Cj0628) as well as lipo-oligosaccharide locus (Cj1145) genes (Fig. 5d). One variant (Cj0139) was associated with TER drop (Fig. 5e). Increased variation in accessory genes was found to be significantly associated with adhesion (Fig. 5f, Supplementary Data 10). However, there were no overall differences in core, accessory, GWAS hit and O-linked glycosylation genes for invasion and barrier disruption (Fig. 5f).

Fig. 5: Effect of genome-wide variation on in vitro pathogenicity-related phenotypes.
figure 5

a Allelic diversity of NCTC 11168 homologs in 83 isolates used in the GWAS. b Gene prevalence of all NCTC 11168 genes. Genes over-represented in PI-IBS or control isolates appear further away from the mid-point. c Correlation of gene presence with pathogenicity-related phenotypes. Average number of adhered bacteria as a ratio to starting inoculum when NCTC 11168 genes are present in an isolate. d Average number of invaded bacteria as a ratio to starting inoculum when NCTC 11168 genes are present in an isolate. e Average drop in baseline TER when NCTC 11168 genes are present in an isolate. f Summary of variation in in vitro pathogenicity phenotypes grouped by core (present in ≥95% of isolates) and accessory genes. Significantly more variation (t-test, Mann–Whitley, p-value = 0.0012) was attributed to the accessory genome in adhesion assays (blue), than invasion assays (green) or barrier integrity assays (red). Raw data are available in Supplementary Data 10. *All box plots show the 25th and 75th percentile, whiskers extending to the min and max values with all data points shown.

Chemokine and cytokine induction by Campylobacter isolates

Apical media fractions from T84 cells exposed to PI-IBS associated C. jejuni strains had significantly higher IL-8 as compared to media fractions from control strains (PI-IBS: 117.7 (27.38) pg/mL vs. control: 97.8 (27.85) pg/mL, p = 0.01) (Fig. 6a). Basolateral concentrations of IL-8 were not different between the strains (Supplementary Fig. 3A). Apical media fractions from T84 cells exposed to PI-IBS associated C. jejuni strains showed lower chemokine (C–C motif) ligand 2 (CCL2) or monocyte chemoattractant protein-1 concentrations compared to the media fractions from cells exposed to control strains (PI-IBS: 1.81 (0.96) pg/mL vs. control: 2.84 (1.49), p = 0.004) (Fig. 6b), while basolateral concentrations were not different (Supplementary Fig. 3B). Basolateral TNFα concentrations were higher after exposure to PI-IBS strains compared to control strains (PI-IBS: 0.30 (0.37) pg/mL vs. control: 0 (0) pg/mL, p = 0.0001) (Fig. 6c), as well as C-X-C motif chemokine ligand 10 (CXCL10) or IP-10 (PI-IBS: 6.81 (7.01) pg/mL, control: 2.72 (4.64) pg/mL, p = 0.01) (Fig. 6d), while apical concentrations of either were not different (Supplementary Fig. 3C, D). The IL-1α (apical), IL-6 (apical), MIG (apical or basolateral), and I-TAC (apical or basolateral), levels were unchanged in response to the two bacterial groups (p > 0.05) (Supplementary Fig. 3E–J). IFNγ, IL-10, IL-1β, and IL-4 levels (apical or basolateral) and IL-1α, IL-6 (basolateral) were not within the readable range upon exposure to any C. jejuni strain.

Fig. 6: Cytokine responses in colonocytes following exposure to PI-IBS and control C. jejuni isolates.
figure 6

T84 epithelial monolayers were exposed to C. jejuni isolates for 24 h followed by collection of media from apical and basolateral compartments. a Apical IL-8 concentration higher in PI-IBS C. jejuni exposed colonocytes than exposed to control isolates. b Apical hemokine (C–C motif) ligand 2 (CCL2) or monocyte chemoattractant protein 1 (MCP-1) concentration lower in PI-IBS C. jejuni than control isolates. c Basolateral TNF-alpha concentration higher in PI-IBS C. jejuni than control isolates. d Basolateral C-X-C motif chemokine ligand 10 (CXCL10) or interferon gamma-induced protein 10 (IP-10) concentration higher in PI-IBS C. jejuni than control isolates. e Differences in apical IL-8 concentration, apical MCP-1 concentration, basolateral TNF-alpha concentration and basolateral IP-10 concentration in response to C. jejuni infection grouped by clonal complexes, LOS class and Penner type. Red and blue bars highlight high and low risk lineages, respectively. (n = 28/group, Mann–Whitney U test). *All box plots show the 25th and 75th percentile, whiskers extending to the min and max values with all data points shown. Raw data are available in Supplementary Data 9.

High risk lineages elicit greater apical IL-8 and basolateral TNFa induction

Comparison of cytokine responses within and between lineages highlighted that the response of each individual isolate was not similar across all cytokines (Fig. 6e). Variation was greatest in response to apical IL-8 and basolateral TNFα and significant differences (ANOVAs with Tukey correction for multiple comparisons) were observed in response to several of our clinically observed/GWAS predicted high risk lineages. Apical IL-8 responses for isolates from ST-206 and ST-22 clonal complexes, LOS class A1 and Penner type HS:19 were significantly higher than other tested lineages. Similarly, the same lineages (apart from ST-206 CC) elicited increased basolateral TNFα production (Fig. 6e). However for most cytokines, PI-IBS status was the more significant variable in the analysis despite variation in cytokine induction between lineages (Supplementary Fig. S3).

Discussion

The epidemiology of PI-IBS is well established but understanding the pathophysiology remains an obstacle to effective interventions6. Intestinal barrier dysfunction32,33,34,35,36, microbial dysbiosis33,37,38,39, and immune dysregulation32,40 have all been described in PI-IBS patients, and the intensity of acute illness is linked to the development of PI-IBS6. Host factors are undoubtedly important as evidenced by increased risk observed in younger people and women after enteric infection. However, the possibility that some strains are more virulent, and, therefore, more likely to cause the onset of PI-IBS has remained unexplored.

Using comparative genomics techniques and in vitro studies, we identified pathogen factors associated with the onset of PI-IBS at multiple levels. First, we examined variation in the strains associated with the development of PI-IBS to identify high and low risk lineages. Second, we recorded the difference in presence/absence patterns of known Campylobacter virulence and pathogenicity genes among patients who develop PI-IBS and those who do not. Third, we conducted a GWAS approach that was agnostic to existing assumptions about gene function and virulence to identify genetic variation in genes that was associated with the development of PI-IBS. Finally, we investigated clinically relevant C. jejuni phenotypes in vitro to characterize variation in adhesion, invasion, barrier function disruption, and induction of pro-inflammatory cytokine production among PI-IBS and control C. jejuni isolates.

The most common isolates obtained from patients with PI-IBS belonged to the ST21 clonal complex. This lineage is among the most common causes of acute human infection41,42 and has been associated with extra-intestinal spread and liver infection43. However, the prevalence of ST21 clonal complex isolates was approximately equal among PI-IBS cases and controls. A more striking finding was the differential association observed in other strains. Specifically, isolates belonging to the ST22 clonal complex were over-represented in isolates that lead to the development of PI-IBS, while ST42 clonal complex were very low risk. Interestingly, while ST22 complex isolates are a relatively infrequent cause of gastroenteritis, they account for up to a third of infections among patients who developed Guillain-Barre syndrome following campylobacteriosis15,44. It is known that the structure of C. jejuni lipooligosaccharide (LOS) has a role in the outcome of infection. Specifically, isolates that can produce sialylated ganglioside mimics45, LOS (class A, B, or C), have greater potential for invasion and translocation across epithelial cell monolayers46,47,48 and are more frequently associated with bloody diarrhea and a longer duration of symptoms49,50. All class A isolates in our study lead to development of PI-IBS, although class B and C isolates were found among control and PI-IBS isolates. While this provides a possible explanation for the PI-IBS association of ST22 clonal complex isolates51; there was no clear association of sialylated LOS classes with PI-IBS, as exhibited in Guillain-Barre syndrome and Miller Fisher Syndrome. Lastly, Penner (capsule) HS:3,17 and HS:4 complex were significantly more common in PI-IBS isolates.

The most obvious place to start is to consider differential prevalence of putative virulence genes in Campylobacter. However, while many have been described46,47,48,49, there is limited data on the contribution of these virulence genes to variability in acute disease severity or the risk of post-infection sequelae. Furthermore, as virulence genes are typically those that are common in isolates from infected patients, it is no surprise that the majority of Campylobacter from human infection studies possess these common genes52. Consistent with this, we found no significant difference among PI-IBS cases and controls in the prevalence of common virulence genes including outer membrane proteins (cadF, peb1, porA); invasion proteins (ciaB); motility (flagella genes); cytolethal distending toxin (cdtA-C. It is, therefore, necessary to take other approaches to investigate genomic differences between the strains.

GWAS analysis revealed variation in gene presence, core and accessory gene alleles, core SNPs, gene fission/fusions, and gene duplications. Five variants were most significantly associated with PI-IBS (Bonferroni corrected p-value < 7.5 × 10−5). These included genes associated with bacterial stress response (Cj0145_phoA), core biosynthetic pathways (biotin: Cj0308_bioD; two variants in purine: Cj0514_purQ; and isoprenoid: Cj0894c_ispH). Iron, though important for several physiological processes can be toxic. Thus, several Gram-negative bacteria control cytoplasmic iron levels through the transcriptional regulator Fur53. Iron and Fur genes alter production of transducer-like proteins (tlp) that allow C. jejuni to sense environmental signals and aid bacterial invasion54. The Cj0145 (phoA) gene is located immediately downstream of tlp2 and induced in the presence of iron, likely regulating cytoplasmic iron levels53. It also shows homology with Vibrio cholerae alkaline phosphatase and plays a role in phosphate regulated two-component system. Inactivation of Cj0145 resulted in reduced alkaline phosphatase activity, which regulates PhosS/PhosR mediated regulation of intracellular phosphate and survival in less favorable habitats55. ATP-dependent dethiobiotin synthetase Cj0308 (bioD) catalyzes the first step in biotin biosynthesis from 7,8-diaminononanoate56. However, not much is known about the biological importance of biotin synthesis in Campylobacter pathogenesis. Cj0514 (purQ) is part of the pur operon required for purine metabolism. The product (phosphoribosylformylglycinamidine synthase subunit I) catalysis ATP and glutamine dependent formation of formyl glycinamidine ribonucleotide, ADP, P (i), and glutamate in the de novo process of purine biosynthesis57. Cj0894 (ispH) is involved in step 6 of the sub pathway that synthesizes isopentenyl diphosphate from 1-deoxy-D-xylulose 5-phosphate56. It is plausible that survival in a stressed environment provides the isolates greater opportunity for mediating epithelial injury and immune responses. However, the mechanistic interaction between these core biosynthetic genes and PI-IBS development would need further studies.

Eleven genes (Cj0208_fokIM, Cj0261c_ubiG, Cj0533_sucC, Cj0720c_flaC, Cj1293_pseB, Cj1300_ubiG, Cj1334, Cj1351_pldA, Cj1414c_kpsC, Cj1729c_flgE, g00183) had the greatest number of variants associated with PI-IBS. Three of these are involved in energy metabolism; Cj0208, a DNA modification methylase; Cj0261c, a ubiquinone biosynthesis O-methyltransferase; and Cj0533, an ADP-forming Succinate–CoA ligase subunit beta. Cj0720 (flaC) encodes for non-structural flagellin which is secreted through the flagella and has been implicated in bacterial invasion58. Interestingly, the flaC showed significant sequence identity with TLR5-activating flagellins of other bacteria, like Salmonella, but not with other Campylobacter flagellins. In vitro studies showed direct interaction of flaC with TLR5 and p38 activation59. The immunomodulatory property of this protein may play a role in pathogenesis of PI-IBS. UDP-N-acetylglucosamine 5-inverting 4,6-dehydratase (Cj1293_PseB) catalyzes the first step in the biosynthesis of pseudaminic acid, which mediates flagellin glycosylation60. This has been shown to facilitate bacterial motility61. Cj1334 is an uncharacterized gene but shares 90% homology with bacterial motility associated factor glycosyltransferase family protein. Cj1351 (pldA) cleaves phospholipids resulting in lysophospholipids that contain only acyl chain instead of two62. These lysophospholipids accumulate in conditions of environmental stress and mediates survival in stressed conditions63. KpsC (Cj1414c), KpsM, and KpsS mutants lacked the ability to produce O antigen or the capsular polysaccharide64. kpsM mutant of strain 81–176 showed a 10-fold reduction in intestinal epithelial cells invasion in vitro65. flgE encodes for flagellar hook, which is a joint between the motor and flagellar filament, and is found across a number of motile bacterial species66,67. Thus, in addition to stress response genes, variants in flagellar- and motility-associated genes are significantly associated with PI-IBS.

We found that both adhesion and invasion associated with CapA, a surface-exposed autotransporter protein that mediates invasion into the intestinal epithelial cells68. It was found to be the only conserved putative adhesins in a study of 97 C. jejuni isolates using dot blot hybridization and in vitro adhesion assays69. Adhesion was also associated with periplasmic protein Cj0967, which been shown to play a role in polysaccharide production, and biofilm formation, which promotes virulence and stress survival70. In NCTC 11168, WlaN or CgtB (Cj1139) demonstrated β-1,3 galactosyltransferase activity which mediates biosynthesis of the LOS core oligosaccharide71,72. We found Cj1139 to be associated with adhesion. Another closely related LOS locus and homopolymeric G/C tract containing gene (Cj1145) is associated with invasion in our study which has been reported in the past72. Cj0139 is a putative endonuclease whose role was investigated in heat shock mediated conjugation efficiency, an important process to horizontal gene transfer73. However, its inactivation did not result in conjugation efficiency in heat shocked or untreated cells. In our study, its variants were associated with disruption of epithelial barrier in vitro. Overall, we found that PI-IBS causing isolates displayed elevated in vitro virulence. Adhesion to epithelial cells, intracellular invasion and survival, elevated pro-inflammatory cytokine response and disruption to barrier function are all known to be associated with virulence in humans74. Invasion as well as CDT production plays a role in IL-8 secretion75,76. In addition to providing a route for its own translocation into the sub-epithelial space, invasion is known to facilitate translocation of commensal E. coli77,78. The PI-IBS causing isolates in our study were twice as invasive as the control isolates. In a recent study based on transcriptomics, genes associated with oxidative stress, N-, and O- glycosylation systems, translocases and the membrane-integrated component of the flagellar apparatus associated with bacterial internalization79. Bacterial motility induced by the flagella is critical for Campylobacter adhesion and invasion80, and flagellar export apparatus also mediate intracellular secretion of other Campylobacter invasive antigen (Cia) proteins19,81, which have been associated with facilitating intracellular survival82,83. C. concisus strains from patients with chronic inflammatory bowel disease had greater invasion potential and caused induction of IFN-γ84. However, PI-IBS differs in pathophysiology and patients do not demonstrate chronic colonization of Campylobacter species in the gastrointestinal tract.

Predicting the development of disease sequel not only increases understanding of pathogenesis but also has considerable potential to inform treatment and preventative interventions. Similar to other studies that have aimed to predict disease severity following infection by different bacterial genotypes, we used a GWAS-informed approach to select genetic markers for the onset of PI-IBS85,86,87,88,89. By weighting the presence of 22 PI-IBS associated markers according to their disease-association score, we were able to correctly predict the long-term disease outcome in up to 78% of cases. Particularly, ST22 and 922 were universally associated with PI-IBS and ST42 with the control state. Our results show promise that an artificially enhanced disease outcome prediction model could be applied with large reference datasets, incorporating the breadth of bacterial species (and host) population structure. Challenges for risk prediction based upon infection associated markers include accurate diagnosis of disease sequelae and confounding variability of the host immune system. A more sophisticated approach may balance the association of disease markers identified through bacterial GWAS correlated with differences in the host genome90,91. Our prior study showed that longer time before starting antibiotic therapy may increase the risk of PI-IBS3. Additionally, PI-IBS development is associated with reduced diversity of commensal microbiota which can be enhanced through a targeted set of commensals or fecal microbiota transplantation33. In future, a validated risk-score may allow targeted interventions to prevent establishment of chronic phenotype of PI-IBS.

In conclusion, acute epithelial injury to cellular and paracellular compartments can signal downstream changes in immune and neuromuscular apparatus that may lead to changes in motility and sensation, which may contribute to the development of PI-IBS. Additionally, a subset of patients may be at risk for long-term changes in luminal (microbiota, bile acids, proteases, etc.) and epithelial compartment (epigenetic changes in secretory, barrier and other signaling mechanisms), which may make them vulnerable for PI-IBS development. Different strains of Campylobacter have also been shown to have unique toxigenic effects in vitro. This study provides the systematic evaluation of bacterial genomics and in vitro virulence using isolates acquired from a prospectively identified cohort of patients developing Campylobacter PI-IBS. Identification of key genes associated with bacterial metabolism, stress response, motility and ganglioside mimics, in conjunction with greater adhesion and invasion, demonstrate a clear contribution of bacterial factors in the long-term sequel of Campylobacter enterocolitis. In combination with host epidemiological factors, these findings pave the way for identification of high-risk infections that can be amenable for preventative strategies. Furthermore, experimental studies testing these genetic elements in animal models using sensory-motor assessments will advance our understanding of PI-IBS pathophysiology.

Methods

Study participants and bacterial strains

Patients with stool culture positive Campylobacter enteritis were surveyed 6–9 months post infection to determine the presence of PI-IBS using Rome III criteria22. Details on survey strategy and clinical metadata collected are published3. Demographic and clinical symptom data during acute campylobacteriosis was collected for each patient. The study was approved by the institutional review boards at the Minnesota Department of Health and Mayo Clinic. All participants provided informed consent for participation in the survey and for sharing of biospecimens.

C. jejuni isolates were cultured on Campylobacter sheep blood agar plates (Hardy Diagnostics, #A40) along with reference isolates NCTC 1116856 and 81–17692 that were chosen as control strains. C. jejuni were routinely grown in Müeller-Hinton (MH) broth (BD BBL™) or on agar at 37 °C in a tri-gas incubator (Thermo Scientific, Heracell™ 150i) under microaerobic conditions (85% N2, 10% CO2 and 5% O2).

DNA extraction and sequencing

Genomic DNA of all isolates was extracted using Wizard genomic DNA purification kits (Promega), following the manufacturer’s instructions. DNA quality was measured using a Nanodrop (Thermo Scientific, USA) and Qubit fluorometer (dsDNA HS) kit (Thermo Scientific, USA). Indexed Illumina sequencing libraries were prepared using Nextera XT DNA Library Preparation Kits (Illumina Inc., San Diego, CA) according to the manufacturer’s protocols. Briefly, 1 ng genomic DNA of each sample was tagmented by Nextera transposome, and then libraries were amplified by barcoded primers. Library cleanup and size selection were performed to remove amplicons <300-bp using AMPure XP beads (NEB, USA). Libraries were normalized, pooled, and sequenced on Illumina HiSeq 3000 platform with 150-bp paired-end reads for all the isolates. Raw sequencing reads were quality trimmed and filtered using Trimmomatic (version 0.33)93 and assembled using the de novo assembly algorithm, Velvet (version 1.2.08). The VelvetOptimiser script (version 2.2.4) was run for all odd k-mer values from 21 to 9994,95. Read data were deposited in the Sequence Read Archive under BioProject ID PRJNA675124.

Core and accessory genome characterization

All unique genes present in at least one isolate (the pangenome) were identified by automated annotation using PROKKA (version 1.13) followed by PIRATE, a pangenomics tool that allows for orthologue gene clustering in bacteria. We defined genes in PIRATE using a wide range of amino acid percentage sequence identity thresholds for Markov Cluster algorithm (MCL) clustering (45, 50, 60, 70, 80, 90, 95, 98). Genes in the pangenome were ordered initially using the NCTC 11168 reference followed by the order defined in PIRATE based on gene synteny and frequency. Core and accessory genome variation analysis was performed as previously described14,96. Briefly, a matrix was produced summarizing the presence/absence and allelic diversity of every gene in the pangenome list. Core genes were defined as present in 95% of the genomes and accessory genes as present in at least one isolate (Supplementary Data 1). The pangenome was visualized using phandango, as a matrix of gene presence alongside a core genome phylogeny97.

An alignment of all 94 Campylobacter isolates was constructed from concatenated gene sequences of all core genes (found in ≥95% isolates) using MAFFT (version 7)98 on a gene-by-gene basis (size: 429,515 bp; Supplementary Data 2). Maximum-likelihood phylogenies were constructed for C. jejuni and C. coli using a GTR + I + G substitution model and ultra-fast bootstrapping (1000 bootstraps)99 implemented in IQ-TREE (version 1.6.8)100 and visualized on Microreact (https://microreact.org/project/CampyIBS)101. Pairwise core and accessory genome distances were compared using PopPunk (version 1.1.4), which uses pairwise nucleotide k-mer comparisons to distinguish shared sequence and gene content to identify divergence of the accessory genome in relation to the core genome. A two-component Gaussian mixture model was used to construct a network to define clusters (Components: 43; Density: 0.1059; Transitivity: 0.8716; Score: 0.7793)102.

Molecular typing and identification of virulence and antibiotic resistance genes

MLST analysis and examinations of known virulence associated genes and antibiotic resistance genes were carried out using the SRST2 pipeline, which uses raw Illumina reads as input103. Capsule polysaccharide (CPS) and lipooligosaccharide (LOS) locus types of each C. jejuni isolate were also characterized from their raw sequence data: short read sequences were mapped to known capsule and LOS locus types using BLAST as previously described27,29,104,105. Antimicrobial resistance genes and putative virulence genes were detected through comparison with reference nucleotide sequences in the VFDB106 and ARG-ANNOT107 databases, respectively. Point mutations related to antibiotic resistance genes were identified by PointFinder using the pointfinder_db database108.

Pangenome wide association studies (pGWAS)

GWAS analyses were used to compare PI-IBS cases confirmed by follow-up (n = 49) with isolates that did not elicit IBS symptoms 6–9 months following initial infection (n = 30). Variation in core and accessory genome elements were identified using PIRATE109 and incorporated into treeWAS110 to investigate associations in [i] gene presence, [ii] core and [iii] accessory allelic variants, [iv] core SNPs, [v] gene fission/fusions and [vi] gene duplications. We removed any alleles/genes with a frequency of <0.05% and any SNPs with a frequency <0.1%. Three scores were calculated in order to differentiate between within- and between-lineage associations: (1) the terminal score is phylogenetically naïve and measures sample-wide association across the leaves of the tree; (2) the simultaneous score identifies within-lineage association by measuring the parallel change in genotype and phenotype across the tree; and (3) the subsequent score incorporates the phylogeny to measure the proportion of the tree in which genotype and phenotype co-exist. The subsequent score is more sensitive and should be able to detect subtle patterns of association, while pooling the scores provides a comprehensive description of association in the dataset.

A C. jejuni specific phylogeny was constructed of the 79 C. jejuni isolates with complete IBS follow-up data by mapping assemblies against the NCTC 11168 reference genome [accession: AL11168.1; Supplementary Data 3]56,111,112. A maximum likelihood phylogeny was constructed using RAxML v8.2.11113 with the GTRGAMMA model (https://microreact.org/project/CampyIBS-CF). Recombination regions inferred by ClonalFrame-ML114 were masked from the core genome sequence alignment (i.e., replaced with gaps) using a custom script (https://github.com/kwongj/cfml-maskrc) to represent the clonal frame, against which associations were compared to correct for clonal descent and lineage effects115. Therefore, associations identified in multiple genetic backgrounds (lineages/STs) will be given greater importance and enriched in our analysis. Variants associated with PI-IBS among the three empirical association scores are deemed to be statistically significant, and candidates for biological association, pending subsequent confirmatory analyses. We prioritized IBS-associated variants that returned a subsequent p-value < 0.05 (which incorporates the phylogeny and corrected for multiple test using Bonferroni correction) or genes that contained over 100 associated variants (p > 0.05 in any of the three tests) for further investigation.

Risk score

A simple risk score was calculated based on the presence of our most significant pGWAS associations, similar to the score previously described to estimate risk of gastric cancer as a result of infection by different Helicobacter pylori genotypes85. Briefly, we screened 1000 variants with the most significant subsequent scores (Bonferroni corrected p-value ≤ 0.0075). By plotting the accuracy of each variant to predict the onset of PI-IBS in our dataset, we drew a threshold in upper tail of the distribution. This provided GWAS-informed markers for risk of developing PI-IBS. A cumulative score was then calculated by weighting the presence of each variant according to its subsequent score from treeWAS, which incorporates the direction of association, as scores with positive values representing an association for that variant with PI-IBS and a negative value representing an association with the control group.

Epithelial cell culture and polarization

T84 human colon cancer cells (from ATCC CCL-248) were cultured in DMEM/F12 medium (GIBCO, US) supplemented with 5% heat-inactivated fetal bovine serum (Life Technologies, US). No antibiotic supplementation was used. Cells were routinely passaged in 75 cm2 flasks (Corning, US) at 5% CO2, 37 °C. For the invasion assay, T84 cells were split into 96-well plates at a density of 105 cells/well. For polarization, T84 cells were seeded at density of 7.5 × 105 cells/well into the apical chamber of 12-well Transwells (12 mm diameter; 3 μm pore size, Corning, US). The culture medium, DMEM/F12 medium supplemented with 5% heat inactivated FBS, was replaced every 2–3 days. Transepithelial electric resistance (TER) was measured manually every 2–3 days using an epithelial voltohmmeter 2 connected to a STX2 chopstick electrode (World Precision Instruments, USA) until the reading reached above 750 Ω cm2. TER values for monolayers were calculated by subtracting the reading of an unseeded well from the seeded wells and by subsequent correction for surface area.

Bacterial invasion

Invasion assays (or gentamicin protection assay) were performed using established protocols116,117. Briefly, epithelial cells grown in 96-well plates, once reaching confluence, were infected with C. jejuni at concentration of 107 CFU/well (i.e., multiplicity of infection of 100). The plates were centrifuged at 600 × g for 5 min to promote contact of C. jejuni with host cells. After a 3-h incubation in a normal humidified incubator (5% CO2, 37 °C), wells were washed five times with 1× phosphate-buffered saline to remove non-adherent bacteria, and then split in two halves. One half of the wells (3 wells) were lysed with 0.25% (wt/vol) sodium deoxycholate (Sigma) for 15 min, and serial 10-fold dilutions of the lysates were plated on MH agar plates to determine the total number of adherent and invaded bacteria. Antibiotic-free cell culture medium with gentamicin (200 μg/mL, Sigma), proven to be effective in killing extracellular C. jejuni116,118, was then added to the other half of the wells (3 wells) and incubated for 3 h in a humidified incubator (37 °C and 5% CO2). Cells were then washed five times with phosphate-buffered saline and lysed with 0.25% sodium deoxycholate (15 min; Sigma). Serial 10-fold dilutions of the lysates were plated on MH agar plates to determine the number of invaded bacteria. During each invasion experiment, highly invasive C. jejuni strain 81–176117 and genome-sequenced low invasive strain NCTC 11168119 were used as positive and negative controls, respectively. For each C. jejuni isolate, independent experiments were performed in triplicate and each triplicate on three different days, for a total of nine observations per strain.

CellZScope system for C. jejuni-T84 interaction

CellZscope is an automated, parallel, real-time recording system based on impedance spectroscopy that can be used for cells grown on porous filters120. T84 epithelial cell lines can be used as in vitro model systems of functional epithelial barriers. Polarized T84 monolayers grown on transwells (3 μm pore size, Corning, US) were transferred into a 24-well cell module of a CellZscope system (NanoAnalytics, Germany) with DMEM/F12 medium without serum. The 24-well cell module of CellZScope was placed in a humidified incubator (37 °C and 5% CO2) overnight to allow equilibration. TER values were recorded every 40 min. The next day, C. jejuni cells were added to the apical chamber at the density of 2 × 108 CFU/well. This infection dose was used because a pilot experiment proved that lower than this dose did not result in significant TER change at 24 h post infection for most C. jejuni strains. At 24 h post infection, medium fractions from both apical and basolateral chambers were collected for cytokine analysis. The TER data was extracted, and TER change was reflected by the percentage of TER at 24 h post incubation versus the initial. For each experiment, 3 replicates were measured.

Cytokine analysis

Apical and basolateral post-infection media fractions at the end of the 24 h CellZScope experiment were pelleted and bacteria removed with a 0.22 µm filter (Spin-X, Corning). Chemokines and cytokines in the media were measured using custom multiplex ELISAs (EMD Millipore, Darmstadt, Germany) for the following cytokines: IL-1β, IL-4, IL-6, IL-8, IL-10, TNFα, IFNγ, MCP-1, IP-10, MIG, and I-TAC. Samples were run in duplicate and average values were calculated and used for comparisons.

Statistics and reproducibility

All the results of genomics and GWAS analyses can be reproduced with the parameters indicated. In vitro studies of C. jejuni adhesion and invasion were performed in triplicate and each triplicate on three different days, for a total of nine observations per strain. For effects on barrier function in CellZScope, each strain was studied in triplicate.

Mean and standard deviation are reported for continuous variables, whereas frequencies and percentages are reported for categorical variables. A two-sided Mann–Whitney U-test assuming non-Gaussian distribution was used to compare variables between groups. Spearman’s correlation (nonparametric) was used to analyze the strength of relationship between paired data. All analyses were done using GraphPad Prism 7 (GraphPad Software, San Diego, CA, USA). A p-value of <0.05 was considered significant.

Reporting summary

Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Data availability

Short read data are available on the NCBI Sequence Read Archive, associated with BioProject PRJNA675124 http://www.ncbi.nlm.nih.gov/bioproject/675124. Assembled genomes, GWAS summary files, and supplementary material are available from FigShare: https://doi.org/10.6084/m9.figshare.12493106. Raw data availability: Fig. 1b–e (Supplementary Data 4), Fig. 2d (Supplementary Data 6), Fig. 3b–e, Fig. 4 (Supplementary Data 9), Fig. 5f (Supplementary Data 10), Fig. 6 (Supplementary Data 9). Phylogenetic trees can be visualized and manipulated on Microreact for the whole dataset: https://microreact.org/project/CampyIBS and the recombination-free phylogeny used in the GWAS at https://microreact.org/project/CampyIBS-CF.

Code availability

Scripts for manipulation of additional PIRATE outputs and treeWAS helper scripts are available on GitHub: https://github.com/SionBayliss.

References

  1. Mearin, F. et al. Bowel disorders. Gastroenterology https://doi.org/10.1053/j.gastro.2016.02.031 (2016).

  2. Barbara, G. et al. Rome foundation working team report on post-infection irritable bowel syndrome. Gastroenterology 156, 46–58.e47 (2019).

    PubMed  Article  Google Scholar 

  3. Berumen, A. et al. Characteristics and risk factors of post-infection irritable bowel syndrome following campylobacter enteritis. Clin Gastroenterol Hepatol. https://doi.org/10.1016/j.cgh.2020.07.033 (2020).

  4. Marshall, J. K. et al. Incidence and epidemiology of irritable bowel syndrome after a large waterborne outbreak of bacterial dysentery. Gastroenterology 131, 445–450; quiz 660 (2006).

    PubMed  Article  Google Scholar 

  5. Thabane, M., Simunovic, M., Akhtar-Danesh, N. & Marshall, J. K. Development and validation of a risk score for post-infectious irritable bowel syndrome. Am. J. Gastroenterol. 104, 2267–2274 (2009).

    PubMed  Article  Google Scholar 

  6. Klem, F. et al. Prevalence, risk factors, and outcomes of irritable bowel syndrome after infectious enteritis: a systematic review and meta-analysis. Gastroenterology 152, 1042–1054.e1041 (2017).

    PubMed  Article  Google Scholar 

  7. Dai, L., Sahin, O., Grover, M. & Zhang, Q. New and alternative strategies for the prevention, control, and treatment of antibiotic-resistant Campylobacter. Transl. Res. https://doi.org/10.1016/j.trsl.2020.04.009 (2020).

  8. Beltinger, J. et al. Disruption of colonic barrier function and induction of mediator release by strains of Campylobacter jejuni that invade epithelial cells. World J. Gastroenterol. 14, 7345–7352 (2008).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  9. van Spreeuwel, J. P. et al. Campylobacter colitis: histological immunohistochemical and ultrastructural findings. Gut 26, 945–951 (1985).

    PubMed  PubMed Central  Article  Google Scholar 

  10. MacCallum, A., Hardy, S. P. & Everest, P. H. Campylobacter jejuni inhibits the absorptive transport functions of Caco-2 cells and disrupts cellular tight junctions. Microbiology 151, 2451–2458 (2005).

    CAS  PubMed  Article  Google Scholar 

  11. Bucker, R. et al. Campylobacter jejuni impairs sodium transport and epithelial barrier function via cytokine release in human colon. Mucosal Immunol. 11, 575–577 (2018).

    CAS  PubMed  Article  Google Scholar 

  12. Nichols, G. L., Richardson, J. F., Sheppard, S. K., Lane, C. & Sarran, C. Campylobacter epidemiology: a descriptive study reviewing 1 million cases in England and Wales between 1989 and 2011. BMJ Open https://doi.org/10.1136/bmjopen-2012-001179 (2012).

  13. Kirk, K. F. et al. Molecular epidemiology and comparative genomics of Campylobacter concisus strains from saliva, faeces and gut mucosal biopsies in inflammatory bowel disease. Sci. Rep. 8, 1902 (2018).

    PubMed  PubMed Central  Article  CAS  Google Scholar 

  14. Pascoe, B. et al. Genomic epidemiology of Campylobacter jejuni associated with asymptomatic pediatric infection in the Peruvian Amazon. PLoS Negl. Trop. Dis. 14, e0008533 (2020).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  15. Nielsen, L. N. et al. MLST clustering of Campylobacter jejuni isolates from patients with gastroenteritis, reactive arthritis and Guillain-Barre syndrome. J. Appl. Microbiol. 108, 591–599 (2010).

    CAS  PubMed  Article  Google Scholar 

  16. Harvey, P., Battle, T. & Leach, S. Different invasion phenotypes of Campylobacter isolates in Caco-2 cell monolayers. J. Med. Microbiol. 48, 461–469 (1999).

    PubMed  Article  Google Scholar 

  17. Thornley, J. P. et al. Relationship of Campylobacter toxigenicity in vitro to the development of postinfectious irritable bowel syndrome. J. Infect. Dis. 184, 606–609 (2001).

    CAS  PubMed  Article  Google Scholar 

  18. Konkel, M. E., Larson, C. L. & Flanagan, R. C. Campylobacter jejuni FlpA binds fibronectin and is required for maximal host cell adherence. J. Bacteriol. 192, 68–76 (2010).

    CAS  PubMed  Article  Google Scholar 

  19. Neal-McKinney, J. M. & Konkel, M. E. The Campylobacter jejuni CiaC virulence protein is secreted from the flagellum and delivered to the cytosol of host cells. Front. Cell Infect. Microbiol. 2, 31 (2012).

    PubMed  PubMed Central  Article  CAS  Google Scholar 

  20. Samuelson, D. R. et al. The Campylobacter jejuni CiaD effector protein activates MAP kinase signaling pathways and is required for the development of disease. Cell Commun. Signal. 11, 79 (2013).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  21. Ketley, J. M. The 16th C. L. Oakley Lecture. Virulence of Campylobacter species: a molecular genetic approach. J. Med. Microbiol. 42, 312–327 (1995).

    CAS  PubMed  Article  Google Scholar 

  22. Longstreth, G. F. et al. Functional bowel disorders. Gastroenterology 130, 1480–1491 (2006).

    PubMed  Article  Google Scholar 

  23. Francis, C. Y., Morris, J. & Whorwell, P. J. The irritable bowel severity scoring system: a simple method of monitoring irritable bowel syndrome and its progress. Aliment Pharm. Ther. 11, 395–402 (1997).

    CAS  Article  Google Scholar 

  24. Dingle, K. E. et al. Multilocus sequence typing system for Campylobacter jejuni. J. Clin. Microbiol. 39, 14–23 (2001).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  25. Sheppard, S. K. et al. Cryptic ecology among host generalist Campylobacter jejuni in domestic animals. Mol. Ecol. 23, 2442–2451 (2014).

    PubMed  PubMed Central  Article  Google Scholar 

  26. Karlyshev, A. V. et al. Analysis of Campylobacter jejuni capsular loci reveals multiple mechanisms for the generation of structural diversity and the ability to form complex heptoses. Mol. Microbiol. 55, 90–103 (2005).

    CAS  PubMed  Article  Google Scholar 

  27. Parker, C. T., Gilbert, M., Yuki, N., Endtz, H. P. & Mandrell, R. E. Characterization of lipooligosaccharide-biosynthetic loci of Campylobacter jejuni reveals new lipooligosaccharide classes: evidence of mosaic organizations. J. Bacteriol. 190, 5681–5689 (2008).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  28. Parker, C. T. et al. Comparison of Campylobacter jejuni lipooligosaccharide biosynthesis loci from a variety of sources. J. Clin. Microbiol 43, 2771–2781 (2005).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  29. Poly, F. et al. Discrimination of major capsular types of Campylobacter jejuni by multiplex PCR. J. Clin. Microbiol 49, 1750–1757 (2011).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  30. Meric, G. et al. A reference pan-genome approach to comparative bacterial genomics: identification of novel epidemiological markers in pathogenic Campylobacter. PLoS ONE 9, e92798 (2014).

    PubMed  PubMed Central  Article  CAS  Google Scholar 

  31. Ridley, K. A., Rock, J. D., Li, Y. & Ketley, J. M. Heme utilization in Campylobacter jejuni. J. Bacteriol. 188, 7862–7875 (2006).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  32. Spiller, R. C. et al. Increased rectal mucosal enteroendocrine cells, T lymphocytes, and increased gut permeability following acute Campylobacter enteritis and in post-dysenteric irritable bowel syndrome. Gut 47, 804–811 (2000).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  33. Edogawa, S. et al. Serine proteases as luminal mediators of intestinal barrier dysfunction and symptom severity in IBS. Gut 69, 62–73 (2020).

    CAS  PubMed  Article  Google Scholar 

  34. Swan, C. et al. Identifying and testing candidate genetic polymorphisms in the irritable bowel syndrome (IBS): association with TNFSF15 and TNFalpha. Gut 62, 985–994 (2013).

    CAS  PubMed  Article  Google Scholar 

  35. Marshall, J. K. et al. Intestinal permeability in patients with irritable bowel syndrome after a waterborne outbreak of acute gastroenteritis in Walkerton, Ontario. Aliment Pharm. Ther. 20, 1317–1322 (2004).

    CAS  Article  Google Scholar 

  36. Dunlop, S. P. et al. Abnormal intestinal permeability in subgroups of diarrhea-predominant irritable bowel syndromes. Am. J. Gastroenterol. 101, 1288–1294 (2006).

    PubMed  Article  Google Scholar 

  37. Jalanka-Tuovinen, J. et al. Faecal microbiota composition and host-microbe cross-talk following gastroenteritis and in postinfectious irritable bowel syndrome. Gut 63, 1737–1745 (2014).

    PubMed  Article  Google Scholar 

  38. Sundin, J. et al. Altered faecal and mucosal microbial composition in post-infectious irritable bowel syndrome patients correlates with mucosal lymphocyte phenotypes and psychological distress. Aliment Pharm. Ther. 41, 342–351 (2015).

    CAS  Article  Google Scholar 

  39. Jeffery, I. B. et al. An irritable bowel syndrome subtype defined by species-specific alterations in faecal microbiota. Gut 61, 997–1006 (2012).

    PubMed  Article  Google Scholar 

  40. Dunlop, S. P., Jenkins, D., Neal, K. R. & Spiller, R. C. Relative importance of enterochromaffin cell hyperplasia, anxiety, and depression in postinfectious IBS. Gastroenterology 125, 1651–1659 (2003).

    PubMed  Article  Google Scholar 

  41. Sheppard, S. K. et al. Campylobacter genotyping to determine the source of human infection. Clin. Infect. Dis. 48, 1072–1078 (2009).

    PubMed  Article  Google Scholar 

  42. Sheppard, S. K. et al. Campylobacter genotypes from food animals, environmental sources and clinical disease in Scotland 2005/6. Int. J. Food Microbiol. 134, 96–103 (2009).

    PubMed  PubMed Central  Article  Google Scholar 

  43. Humphrey, S. et al. Heterogeneity in the infection biology of Campylobacter jejuni isolates in three infection models reveals an Invasive and virulent phenotype in a ST21 isolate from poultry. PLoS ONE 10, e0141182 (2015).

    PubMed  PubMed Central  Article  CAS  Google Scholar 

  44. Islam, Z. et al. Comparative genotyping of Campylobacter jejuni strains from patients with Guillain-Barre syndrome in Bangladesh. PLoS ONE 4, e7257 (2009).

    PubMed  PubMed Central  Article  CAS  Google Scholar 

  45. Godschalk, P. C. et al. The crucial role of Campylobacter jejuni genes in anti-ganglioside antibody induction in Guillain-Barre syndrome. J. Clin. Invest. 114, 1659–1665 (2004).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  46. Louwen, R. et al. Campylobacter jejuni translocation across intestinal epithelial cells is facilitated by ganglioside-like lipooligosaccharide structures. Infect. Immun. 80, 3307–3318 (2012).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  47. Louwen, R. et al. The sialylated lipooligosaccharide outer core in Campylobacter jejuni is an important determinant for epithelial cell invasion. Infect. Immun. 76, 4431–4438 (2008).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  48. Ellstrom, P., Feodoroff, B., Hanninen, M. L. & Rautelin, H. Lipooligosaccharide locus class of Campylobacter jejuni: sialylation is not needed for invasive infection. Clin. Microbiol. Infect. 20, 524–529 (2014).

    CAS  PubMed  Article  Google Scholar 

  49. Mortensen, N. P. et al. Sialylation of Campylobacter jejuni lipo-oligosaccharides is associated with severe gastro-enteritis and reactive arthritis. Microbes Infect. 11, 988–994 (2009).

    CAS  PubMed  Article  Google Scholar 

  50. Huizinga, R. et al. Sialylation of Campylobacter jejuni endotoxin promotes dendritic cell-mediated B cell responses through CD14-dependent production of IFN-beta and TNF-alpha. J. Immunol. 191, 5636–5645 (2013).

    CAS  PubMed  Article  Google Scholar 

  51. Revez, J. et al. Finnish Campylobacter jejuni strains of multilocus sequence type ST-22 complex have two lineages with different characteristics. PLoS ONE 6, e26880 (2011).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  52. Redondo, N., Carroll, A. & McNamara, E. Molecular characterization of Campylobacter causing human clinical infection using whole-genome sequencing: Virulence, antimicrobial resistance and phylogeny in Ireland. PLoS ONE 14, e0219088 (2019).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  53. Butcher, J., Sarvan, S., Brunzelle, J. S., Couture, J. F. & Stintzi, A. Structure and regulon of Campylobacter jejuni ferric uptake regulator Fur define apo-Fur regulation. Proc. Natl Acad. Sci. USA 109, 10047–10052 (2012).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  54. Chandrashekhar, K. et al. Transducer-like protein in Campylobacter jejuni with a role in mediating chemotaxis to Iron and phosphate. Front. Microbiol. 9, 2674 (2018).

    PubMed  PubMed Central  Article  Google Scholar 

  55. Wosten, M. M. et al. The Campylobacter jejuni PhosS/PhosR operon represents a non-classical phosphate-sensitive two-component system. Mol. Microbiol. 62, 278–291 (2006).

    PubMed  Article  CAS  Google Scholar 

  56. Parkhill, J. et al. The genome sequence of the food-borne pathogen Campylobacter jejuni reveals hypervariable sequences. Nature 403, 665–668 (2000).

    CAS  PubMed  Article  Google Scholar 

  57. Hoskins, A. A., Anand, R., Ealick, S. E. & Stubbe, J. The formylglycinamide ribonucleotide amidotransferase complex from Bacillus subtilis: metabolite-mediated complex formation. Biochemistry 43, 10314–10327 (2004).

    CAS  PubMed  Article  Google Scholar 

  58. Song, Y. C. et al. FlaC, a protein of Campylobacter jejuni TGH9011 (ATCC43431) secreted through the flagellar apparatus, binds epithelial cells and influences cell invasion. Mol. Microbiol. 53, 541–553 (2004).

    CAS  PubMed  Article  Google Scholar 

  59. Faber, E. et al. Novel Immunomodulatory Flagellin-Like Protein FlaC in Campylobacter jejuni and Other Campylobacterales. mSphere https://doi.org/10.1128/mSphere.00028-15 (2016).

  60. Morrison, J. P., Schoenhofen, I. C. & Tanner, M. E. Mechanistic studies on PseB of pseudaminic acid biosynthesis: a UDP-N-acetylglucosamine 5-inverting 4,6-dehydratase. Bioorg. Chem. 36, 312–320 (2008).

    CAS  PubMed  Article  Google Scholar 

  61. Salah Ud-Din, A. I. M. & Roujeinikova, A. Flagellin glycosylation with pseudaminic acid in Campylobacter and Helicobacter: prospects for development of novel therapeutics. Cell Mol. Life Sci. 75, 1163–1178 (2018).

    CAS  PubMed  Article  Google Scholar 

  62. Cao, X. et al. The unique phospholipidome of the enteric pathogen Campylobacter jejuni: lysophosholipids are required for motility at low oxygen availability. J. Mol. Biol. https://doi.org/10.1016/j.jmb.2020.07.012 (2020).

  63. Zheng, L., Lin, Y., Lu, S., Zhang, J. & Bogdanov, M. Biogenesis, transport and remodeling of lysophospholipids in Gram-negative bacteria. Biochim Biophys. Acta Mol. Cell Biol. Lipids 1862, 1404–1413 (2017).

    CAS  PubMed  Article  Google Scholar 

  64. Karlyshev, A. V., Linton, D., Gregson, N. A., Lastovica, A. J. & Wren, B. W. Genetic and biochemical evidence of a Campylobacter jejuni capsular polysaccharide that accounts for Penner serotype specificity. Mol. Microbiol. 35, 529–541 (2000).

    CAS  PubMed  Article  Google Scholar 

  65. Bacon, D. J. et al. A phase-variable capsule is involved in virulence of Campylobacter jejuni 81-176. Mol. Microbiol. 40, 769–777 (2001).

    CAS  PubMed  Article  Google Scholar 

  66. Barker, C. S., Meshcheryakova, I. V., Kostyukova, A. S., Freddolino, P. L. & Samatey, F. A. An intrinsically disordered linker controlling the formation and the stability of the bacterial flagellar hook. BMC Biol. 15, 97 (2017).

    PubMed  PubMed Central  Article  CAS  Google Scholar 

  67. Kido, Y., Yoon, Y. H. & Samatey, F. A. Crystallization of a 79 kDa fragment of the hook protein FlgE from Campylobacter jejuni. Acta Crystallogr Sect. F. Struct. Biol. Cryst. Commun. 67, 1653–1657 (2011).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  68. Ashgar, S. S. et al. CapA, an autotransporter protein of Campylobacter jejuni, mediates association with human epithelial cells and colonization of the chicken gut. J. Bacteriol. 189, 1856–1865 (2007).

    CAS  PubMed  Article  Google Scholar 

  69. Flanagan, R. C., Neal-McKinney, J. M., Dhillon, A. S., Miller, W. G. & Konkel, M. E. Examination of Campylobacter jejuni putative adhesins leads to the identification of a new protein, designated FlpA, required for chicken colonization. Infect. Immun. 77, 2399–2407 (2009).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  70. McLennan, M. K. et al. Campylobacter jejuni biofilms up-regulated in the absence of the stringent response utilize a calcofluor white-reactive polysaccharide. J. Bacteriol. 190, 1097–1107 (2008).

    CAS  PubMed  Article  Google Scholar 

  71. Linton, D. et al. Phase variation of a beta-1,3 galactosyltransferase involved in generation of the ganglioside GM1-like lipo-oligosaccharide of Campylobacter jejuni. Mol. Microbiol. 37, 501–514 (2000).

    CAS  PubMed  Article  Google Scholar 

  72. Karlyshev, A. V., Ketley, J. M. & Wren, B. W. The Campylobacter jejuni glycome. FEMS Microbiol. Rev. 29, 377–390 (2005).

    CAS  PubMed  Google Scholar 

  73. Zeng, X., Ardeshna, D. & Lin, J. Heat shock-enhanced conjugation efficiency in standard Campylobacter jejuni strains. Appl Environ. Microbiol 81, 4546–4552 (2015).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  74. Aguilar, C., Jimenez-Marin, A., Martins, R. P. & Garrido, J. J. Interaction between Campylobacter and intestinal epithelial cells leads to a different proinflammatory response in human and porcine host. Vet. Immunol. Immunopathol. 162, 14–23 (2014).

    CAS  PubMed  Article  Google Scholar 

  75. Hickey, T. E., Baqar, S., Bourgeois, A. L., Ewing, C. P. & Guerry, P. Campylobacter jejuni-stimulated secretion of interleukin-8 by INT407 cells. Infect. Immun. 67, 88–93 (1999).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  76. Hickey, T. E. et al. Campylobacter jejuni cytolethal distending toxin mediates release of interleukin-8 from intestinal epithelial cells. Infect. Immun. 68, 6535–6541 (2000).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  77. Campana, R. & Baffone, W. Intracellular survival and translocation ability of human and avian Campylobacter jejuni and Campylobacter coli strains. Adv. Exp. Med Biol. 1282, 115–125 (2020).

    CAS  PubMed  Article  Google Scholar 

  78. Lamb-Rosteski, J. M., Kalischuk, L. D., Inglis, G. D. & Buret, A. G. Epidermal growth factor inhibits Campylobacter jejuni-induced claudin-4 disruption, loss of epithelial barrier function, and Escherichia coli translocation. Infect. Immun. 76, 3390–3398 (2008).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  79. Kovacs, J. K. et al. Virulence Traits of Inpatient Campylobacter jejuni Isolates, and a transcriptomic approach to identify potential genes maintaining intracellular survival. Microorganisms https://doi.org/10.3390/microorganisms8040531 (2020).

  80. Guerry, P. Campylobacter flagella: not just for motility. Trends Microbiol 15, 456–461 (2007).

    CAS  PubMed  Article  Google Scholar 

  81. Konkel, M. E. et al. Secretion of virulence proteins from Campylobacter jejuni is dependent on a functional flagellar export apparatus. J. Bacteriol. 186, 3296–3303 (2004).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  82. Buelow, D. R., Christensen, J. E., Neal-McKinney, J. M. & Konkel, M. E. Campylobacter jejuni survival within human epithelial cells is enhanced by the secreted protein CiaI. Mol. Microbiol. 80, 1296–1312 (2011).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  83. Christensen, J. E., Pacheco, S. A. & Konkel, M. E. Identification of a Campylobacter jejuni-secreted protein required for maximal invasion of host cells. Mol. Microbiol. 73, 650–662 (2009).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  84. Kaakoush, N. O. et al. The pathogenic potential of Campylobacter concisus strains associated with chronic intestinal diseases. PLoS ONE 6, e29045 (2011).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  85. Berthenet, E. et al. A GWAS on Helicobacter pylori strains points to genetic variants associated with gastric cancer risk. BMC Biol. 16, 84 (2018).

    PubMed  PubMed Central  Article  CAS  Google Scholar 

  86. Recker, M. et al. Clonal differences in Staphylococcus aureus bacteraemia-associated mortality. Nat. Microbiol. 2, 1381–1388 (2017).

    CAS  PubMed  Article  Google Scholar 

  87. Lees, J. A. et al. Improved prediction of bacterial genotype-phenotype associations using interpretable pangenome-spanning regressions. mBio https://doi.org/10.1128/mBio.01344-20 (2020).

  88. Wheeler, N. E., Gardner, P. P. & Barquist, L. Machine learning identifies signatures of host adaptation in the bacterial pathogen Salmonella enterica. PLoS Genet. 14, e1007333 (2018).

    PubMed  PubMed Central  Article  CAS  Google Scholar 

  89. Meric, G. et al. Disease-associated genotypes of the commensal skin bacterium Staphylococcus epidermidis. Nat. Commun. 9, 5034 (2018).

    PubMed  PubMed Central  Article  CAS  Google Scholar 

  90. Wilson, D. J. The harmonic mean p-value for combining dependent tests. Proc. Natl Acad. Sci. USA 116, 1195–1200 (2019).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  91. Lees, J. A. et al. Joint sequencing of human and pathogen genomes reveals the genetics of pneumococcal meningitis. Nat. Commun. 10, 2176 (2019).

    PubMed  PubMed Central  Article  CAS  Google Scholar 

  92. Hofreuter, D. et al. Unique features of a highly pathogenic Campylobacter jejuni strain. Infect. Immun. 74, 4694–4707 (2006).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  93. Bolger, A. M., Lohse, M. & Usadel, B. Trimmomatic: a flexible trimmer for Illumina sequence data. Bioinformatics 30, 2114–2120 (2014).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  94. Zerbino, D. R. & Birney, E. Velvet: algorithms for de novo short read assembly using de Bruijn graphs. Genome Res. 18, 821–829 (2008).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  95. Zerbino, D. R. Using the Velvet de novo assembler for short-read sequencing technologies. Curr Protoc Bioinformatics Chapter 11, Unit 11 15, https://doi.org/10.1002/0471250953.bi1105s31 (2010).

  96. Mourkas, E. et al. Agricultural intensification and the evolution of host specialism in the enteric pathogen Campylobacter jejuni. Proc. Natl Acad. Sci. USA 117, 11018–11028 (2020).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  97. Hadfield, J. et al. Phandango: an interactive viewer for bacterial population genomics. Bioinformatics 34, 292–293 (2018).

    CAS  PubMed  Article  Google Scholar 

  98. Katoh, K. & Standley, D. M. MAFFT multiple sequence alignment software version 7: improvements in performance and usability. Mol. Biol. Evol. 30, 772–780 (2013).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  99. Hoang, D. T., Chernomor, O., von Haeseler, A., Minh, B. Q. & Vinh, L. S. UFBoot2: Improving the Ultrafast Bootstrap Approximation. Mol. Biol. Evol. 35, 518–522 (2018).

    CAS  PubMed  Article  Google Scholar 

  100. Nguyen, L. T., Schmidt, H. A., von Haeseler, A. & Minh, B. Q. IQ-TREE: a fast and effective stochastic algorithm for estimating maximum-likelihood phylogenies. Mol. Biol. Evol. 32, 268–274 (2015).

    CAS  PubMed  Article  Google Scholar 

  101. Argimon, S. et al. Microreact: visualizing and sharing data for genomic epidemiology and phylogeography. Micro. Genom. 2, e000093 (2016).

    Google Scholar 

  102. Lees, J. A. et al. Fast and flexible bacterial genomic epidemiology with PopPUNK. Genome Res. 29, 304–316 (2019).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  103. Inouye, M. et al. SRST2: Rapid genomic surveillance for public health and hospital microbiology labs. Genome Med. 6, 90 (2014).

    PubMed  PubMed Central  Article  Google Scholar 

  104. Culebro, A. et al. Large sequence diversity within the biosynthesis locus and common biochemical features of Campylobacter coli lipooligosaccharides. J. Bacteriol. 198, 2829–2840 (2016).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  105. Poly, F. et al. Updated Campylobacter jejuni capsule PCR multiplex typing system and its application to clinical isolates from South and Southeast Asia. PLoS ONE 10, e0144349 (2015).

    PubMed  PubMed Central  Article  CAS  Google Scholar 

  106. Chen, L., Zheng, D., Liu, B., Yang, J. & Jin, Q. VFDB 2016: hierarchical and refined dataset for big data analysis-10 years on. Nucleic Acids Res. 44, D694–D697 (2016).

    CAS  PubMed  Article  Google Scholar 

  107. Gupta, S. K. et al. ARG-ANNOT, a new bioinformatic tool to discover antibiotic resistance genes in bacterial genomes. Antimicrob. Agents Chemother. 58, 212–220 (2014).

    PubMed  PubMed Central  Article  CAS  Google Scholar 

  108. Zankari, E. et al. PointFinder: a novel web tool for WGS-based detection of antimicrobial resistance associated with chromosomal point mutations in bacterial pathogens. J. Antimicrob. Chemother. 72, 2764–2768 (2017).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  109. Bayliss, S. C., Thorpe, H. A., Coyle, N. M., Sheppard, S. K. & Feil, E. J. PIRATE: A fast and scalable pangenomics toolbox for clustering diverged orthologues in bacteria. Gigascience https://doi.org/10.1093/gigascience/giz119 (2019).

  110. Collins, C. & Didelot, X. A phylogenetic method to perform genome-wide association studies in microbes that accounts for population structure and recombination. PLoS Comput. Biol. 14, e1005958 (2018).

    PubMed  PubMed Central  Article  CAS  Google Scholar 

  111. Pascoe, B. et al. Domestication of Campylobacter jejuni NCTC 11168. Microb. Genom. https://doi.org/10.1099/mgen.0.000279 (2019).

  112. Gundogdu, O. et al. Re-annotation and re-analysis of the Campylobacter jejuni NCTC11168 genome sequence. BMC Genomics 8, 162 (2007).

    PubMed  PubMed Central  Article  CAS  Google Scholar 

  113. Stamatakis, A. RAxML version 8: a tool for phylogenetic analysis and post-analysis of large phylogenies. Bioinformatics 30, 1312–1313 (2014).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  114. Didelot, X. & Wilson, D. J. ClonalFrameML: efficient inference of recombination in whole bacterial genomes. PLoS Comput. Biol. 11, e1004041 (2015).

    PubMed  PubMed Central  Article  CAS  Google Scholar 

  115. Earle, S. G. et al. Identifying lineage effects when controlling for population structure improves power in bacterial association studies. Nat. Microbiol 1, 16041 (2016).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  116. Wine, E., Chan, V. L. & Sherman, P. M. Campylobacter jejuni mediated disruption of polarized epithelial monolayers is cell-type specific, time dependent, and correlates with bacterial invasion. Pediatr. Res. 64, 599–604 (2008).

    PubMed  Article  Google Scholar 

  117. Monteville, M. R. & Konkel, M. E. Fibronectin-facilitated invasion of T84 eukaryotic cells by Campylobacter jejuni occurs preferentially at the basolateral cell surface. Infect. Immun. 70, 6665–6671 (2002).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  118. Watson, R. O. & Galan, J. E. Campylobacter jejuni survives within epithelial cells by avoiding delivery to lysosomes. PLoS Pathog. 4, e14 (2008).

    PubMed  PubMed Central  Article  CAS  Google Scholar 

  119. Gaynor, E. C. et al. The genome-sequenced variant of Campylobacter jejuni NCTC 11168 and the original clonal clinical isolate differ markedly in colonization, gene expression, and virulence-associated phenotypes. J. Bacteriol. 186, 503–517 (2004).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  120. Wegener, J., Abrams, D., Willenbrink, W., Galla, H. J. & Janshoff, A. Automated multi-well device to measure transepithelial electrical resistances under physiological conditions. Biotechniques 37, 590, 592–594, 596-597 (2004).

    Article  Google Scholar 

Download references

Acknowledgements

The study was funded by NIDDK K23 DK103911, R03 120745, and Department of Medicine K2R Program Award to M.G. Authors B.P., S.C.B. and S.K.S. are supported by the Medical Research Council (MRC) grant MR/L015080/1. C.T.P is supported by USDA CRIS Project 2030-42000-055-00D. Authors acknowledge Ms. Lori Anderson for administrative support.

Author information

Affiliations

Authors

Contributions

S.P.: acquisition, analysis, and interpretation of data; drafted manuscript. B.P.: acquisition, analysis, and interpretation of data; drafted manuscript. Z.W.: acquisition, analysis, and interpretation of data; revised manuscript. S.C.B.: acquisition, analysis, and interpretation of data; revised manuscript. X.Z.: acquisition and interpretation of data; revised manuscript. A.E.: acquisition and interpretation of data; revised manuscript. S.V.G.: interpretation of data; revised manuscript. S.J.: interpretation of data; revised manuscript. J.K.C.: interpretation of data; revised manuscript. E.M.: interpretation of data; revised manuscript. R.P.: designed work; revised manuscript. T.W.: acquisition and interpretation of data; revised manuscript. M.D.: acquisition and interpretation of data; revised manuscript. D.B.: designed work; acquisition of data; revised manuscript. K.S.: acquisition and interpretation of data; revised manuscript. C.T.P.: analysis, and interpretation of data; revised manuscript. G.F.: interpretation of data; revised manuscript. Q.Z.: designed work; acquisition, analysis, and interpretation of data; revised manuscript. S.K.S.: conceptualized and designed work; acquisition, analysis, and interpretation of data; drafted manuscript. M.G.: conceptualized and designed work; acquisition, analysis, and interpretation of data; drafted manuscript. All authors have approved the submitted version and are personally accountable for author’s own contributions.

Corresponding authors

Correspondence to Samuel K. Sheppard or Madhusudan Grover.

Ethics declarations

Competing interests

The authors declare no competing interests.

Additional information

Peer review information Communications Biology thanks Ozan Gundogdu and the other, anonymous, reviewers for their contribution to the peer review of this work. Primary Handling Editors: Nicolas Fanget and George Inglis. Peer reviewer reports are available.

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Peters, S., Pascoe, B., Wu, Z. et al. Campylobacter jejuni genotypes are associated with post-infection irritable bowel syndrome in humans. Commun Biol 4, 1015 (2021). https://doi.org/10.1038/s42003-021-02554-8

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1038/s42003-021-02554-8

Comments

By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.

Search

Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing