Genetic diversity in Campylobacter jejuni is associated with differential colonization of broiler chickens and C57BL/6J IL10-deficient mice.

Previous studies have demonstrated that Campylobacter jejuni, the leading causative agent of bacterial food-borne disease in the USA, exhibits high-frequency genetic variation that is associated with changes in cell-surface antigens and ability to colonize chickens. To expand our understanding of the role of genetic diversity in the disease process, we analysed the ability of three C. jejuni human disease isolates (strains 11168, 33292 and 81-176) and genetically marked derivatives to colonize Ross 308 broilers and C57BL/6J IL10-deficient mice. C. jejuni colonized broilers at much higher efficiency (all three strains, 23 of 24 broilers) than mice (11168 only, 8 of 24 mice). C. jejuni 11168 genetically marked strains colonized mice at very low efficiency (2 of 42 mice); however, C. jejuni reisolated from mice colonized both mice and broilers at high efficiency, suggesting that this pathogen can adapt genetically in the mouse. We compared the genome composition in the three wild-type C. jejuni strains and derivatives by microarray DNA/DNA hybridization analysis; the data demonstrated a high degree of genetic diversity in three gene clusters associated with synthesis and modification of the cell-surface structures capsule, flagella and lipo-oligosaccharide. Finally, we analysed the frequency of mutation in homopolymeric tracts associated with the contingency genes wlaN (GC tract) and flgR (AT tracts) in culture and after passage through broilers and mice. C. jejuni adapted genetically in culture at high frequency and the degree of genetic diversity was increased by passage through broilers but was nearly eliminated in the gastrointestinal tract of mice. The data suggest that the broiler gastrointestinal tract provides an environment which promotes outgrowth and genetic variation in C. jejuni; the enhancement of genetic diversity at this location may contribute to its importance as a human disease reservoir.

INTRODUCTION

Campylobacter jejuni is a small, motile, ubiquitous, Gram-negative bacterium with a small (1.7 Mbp), AT-rich genome (Parkhill ; Fouts ). This important foodborne disease agent is associated with 2.4–3 million human disease cases in the USA each year at an estimated cost of $1.6–6.2 billion (reviewed by Snelling ; Humphrey ; Young ). In immunocompetent individuals, C. jeuni infection results in a self-limiting enteritis with an average duration of 3–5 days. More rarely, auto-immune disorders such as Guillain–Barré (GBS) and Miller–Fisher syndromes occur after C. jejuni infection (1/1000 infections) (Humphrey ; Yuki, 2010); these disorders have been linked to certain serotypes that carry epitopes similar to human gangliosides, a process known as molecular mimicry (Linton ; Guerry ; Prendergast ; Godschalk ; van Doorn ; Vucic ).

C. jejuni establishes a commensal relationship in the gastrointestinal (GI) tract of migratory waterfowl and commercial broiler flocks, where it can persist in these animal reservoirs; the pathogen then gains access to the human food chain predominantly through the slaughter and processing of broiler chickens (Lee & Newell, 2006; Knudsen ). Poultry and poultry products therefore represent a major reservoir for human disease (reviewed by Suzuki & Yamamoto, 2009; Horrocks ). The prevalence of C. jejuni in commercial broiler flocks has been well documented in Europe, Australia and the USA (reviewed by Humphrey ; Horrocks ; Suzuki & Yamamoto, 2009).

Despite its high profile as the leading causative agent of foodborne bacterial gastroenteritis, the mechanisms by which C. jejuni causes disease have proven difficult to define and relatively few virulence factors [e.g. flagella, cytolethal distending toxin, FspA (Poly )] have been identified. However, recent studies demonstrate that homopolymeric tracts associated with several loci in the C. jejuni genome (Parkhill ) are subject to high-frequency mutation (as high as 1 in every 2 cells in a population) and that these genetic changes likely arise by slip-strand mutation (Bernatchez ; Linton ; Guerry ; Prendergast ; Hendrixson, 2006). Many of these loci are associated with the synthesis and modification of key surface antigens, including capsule, flagella and lipo-oligosaccharide (LOS); their expression appears to be regulated by phase variation, prompting their characterization as ‘contingency genes’ (Guerry ; Prendergast ; Moxon ; Hendrixson, 2008; van Alphen ). The ability to control the conformation and antigenicity of surface structures via mutation strongly suggests that this genetic diversity could affect host colonization and disease. In support of this idea, others have observed that mutations in specific contingency genes affect the ability of C. jejuni to invade epithelial cells (in vitro) (Guerry ), the antigenicity of specific cell-surface molecules (glycoproteins) associated with molecular mimicry both in vitro and in experimental human infection (Bernatchez ; Linton ; Guerry ; Prendergast ), and the colonization of chickens (Ashgar ; Hendrixson, 2008). Several groups have also demonstrated that the C. jejuni genotype can affect colonization of the GI tract of poultry (Ahmed ; Hook ; Coward ; Ridley ) and that passage through poultry can affect both the genotype and colonization of poultry (Cawthraw ; Wassenaar ; Ringoir & Korolik, 2003; Jones ; Kakuda & DiRita, 2006).

In order to understand the ability of C. jejuni to cause disease in a human host, we believe that it is important to understand the mechanisms by which C. jejuni establishes a reservoir in poultry, how the organism is transferred from the reservoir to the human host, and how growth in the reservoir affects subsequent ability to colonize and cause disease in the host. Based on the discussion above, we hypothesized that rapid mutation in contingency genes generates genetic diversity in C. jejuni subpopulations that enhances the ability to colonize poultry and to subsequently colonize and cause disease in humans. To begin to test this hypothesis, we analysed the ability of three C. jejuni human disease isolates (11168, 33292 and 81-176) and derivatives to colonize broiler chickens (Ross 308) and our mouse model for human disease (C57Bl/6J IL10-deficient mice) (Mansfield ). We then compared the genetic diversity in the genomes of these three isolates and derivatives (microarray DNA/DNA hybridization analysis) and analysed genetic changes that specifically occurred in the homopolymeric tracts in two contingency genes (DNA sequence analysis) in culture, in the poultry reservoir, and after passage through mice.

METHODS

C. jejuni culture and growth conditions.

Glycerol stocks of C. jejuni 11168, 81-176 and 33292, and genetically marked derivatives, were streaked onto Trypticase Soy Blood Agar (TSBA) and incubated at 37 °C under microaerophilic conditions (10 % CO2/10 % H2/80 % N2) in BBL anaerobic jars (Becton Dickinson). The cultures were incubated for 16–24 h, harvested with a sterile Dacron or cotton swab (Fisher Scientific), and resuspended to a concentration of 1×109 c.f.u. ml−1 in Trypticase Soy Broth (TSB). Then 100 μl aliquots were plated onto TSBA and the plates incubated at 37 °C under a microaerophilic atmosphere (10 % CO2/10 % H2/80 % N2). The culture was incubated for approximately 12 h, harvested with a sterile Falcon cell scraper (Becton Dickinson) and suspended in TSB as either a single culture or a co-culture of two strains to a concentration of 5×1010 c.f.u. ml−1 for mouse inoculation or 5×108 c.f.u. ml−1 for chicken inoculation. DNase I (Roche) was added to a concentration of 1 U μl−1 in co-cultures to prevent DNA transfer prior to inoculation. This inoculum culture was immediately placed on ice prior to oral gavage.

On the day of necropsy, caecal tissue was excised, placed in sterile PBS, and its contents gently removed. Caecal tissue was weighed and separated into halves. One half of the tissue was dried at 75 °C for 2 days to determine tissue dry weight. The second half was homogenized in 150–300 μl TSB, serially diluted, and plated onto TSBA supplemented with 20 μg cefoperazone ml−1, 10 μg vancomycin ml−1 and 2 μg amphotericin B ml−1 (CVA). Chloramphenicol (20 μg ml−1) or kanamycin (30 μg ml−1) was added to TSBA-CVA when appropriate. Culturing for C. jejuni colonization prior to necropsy of mice was performed by collecting faecal pellets directly from the mice. Faecal pellets were homogenized in TSB, a portion dried as above, and a sample serially diluted and plated onto TSBA-CVA. Cultures recovered from animals were incubated for 72 h at 37 °C under standard microaerophilic conditions as described above.

Genetically marked C. jejuni strains.

Genetically marked C. jejuni strains were derived from 11168 through homologous recombination with transposed chromosomal DNA or with a pUC-based suicide vector. DNA was introduced into competent C. jejuni by electroporation as described by Wilson . Recombination occurred by double-crossover of a kanamycin or chloramphenicol resistance marker (Wilson ) with one integration per mutant as demonstrated by Southern blotting (data not shown).

Integration of the kanamycin marker into the C. jejuni 11168 chromosome was accomplished via the pJB23SK suicide vector described by Wilson . The suicide vector pDWhipOK, produced by methods similar to those utilized by de Boer , was used to introduce the kanamycin marker into the C. jejuni 11168 hipO chromosomal locus. Briefly, primers JL1100 (5′-AGAGCTAATGATACCCTACC-3′) and JL1101 (5′-ATGCTATTGTCTGGAGGAGC-3′) were used to amplify the 11168 hipO locus (1.9 kb), which was subsequently cloned into pCR2.1-Topo (Invitrogen) and electroporated into EC TOP10 (Invitrogen) according to the manufacturer's specifications. A 1.9 kb EcoRI fragment containing the hipO gene was excised from the above engineered plasmid, ligated into the EcoRI site of pBluescript II SK(+) (Stratagene), and transformed into Escherichia coli DH5α (Life Technologies) to produce plasmid pDWhipO. The Campylobacter aphA-3 kanamycin marker (Trieu-Cuot ) was excised from pILL600 by SmaI and blunt-end ligated into the SphI site of pDWhipO (polished with T4 DNA polymerase), generating pDWhipOK. Both pDWhipOK and pJB23SK were amplified in E. coli DH5α before electroporation into C. jejuni.

Mice and chickens.

C57BL/6J IL10-deficient (IL10−/−) mice were obtained from Jackson Laboratories and a breeding colony established in a Campylobacter/Helicobacter-free facility with autoclaved food, bedding and water at Michigan State University (MSU) (Mansfield ).

Ross 308 broiler eggs were acquired from a commercial hatchery that supplies fresh poultry to the Chicago and Detroit markets via regional grow-out farms and a centralized processing plant. Eggs were incubated for 18 days at the MSU Poultry Farm before transfer to a portable Brower incubator at the University Research Containment Facility (URCF). Eggs were sanitized in a dilute bleach solution prior to setting at URCF. Hatching occurred at about 3 days post-setting and chicks were housed in a single isolator for about 1 day before oral gavage. Chickens were provided a commercial high-protein chicken feed (Broiler Starter and Grower W/Maxiban 72 – medicated for prevention of coccidiosis), and sterile water ad libitum. Chicken feed was irradiated at a target dose of 1 mrad (0.01 mGy). Chickens, like mice, were housed in separate cages to maintain isolation of each individual C. jejuni population.

Animals were inoculated with 200 μl of culture through a 5.0 French feeding tube/catheter (Sovereign) and a 1 ml Luer-Lok syringe (Becton Dickinson). All animal protocols were reviewed and approved by the MSU All University Committee on Animal Use and Care.

Microarray DNA–DNA hybridization analysis.

Gene-specific primers were designed using Primer 3 software and used to amplify 1685 predicted ORFs in the C. jejuni 11168 sequence (ftp://ftp.sanger.ac.uk/pub/pathogens/cj/) carried in the Sanger Centre and TIGR database. C. jejuni 11168 chromosomal DNA was used as template for PCR to produce a clone library which was subsequently reamplified; the PCR products were ethanol precipitated and resuspended in 3× SSC print buffer. DNA representing each clone was spotted in triplicate on UltraGAPS slides (Corning) using a GeneMachines Omnigrid 100 robot and Telechem Chipmaker. Test and reference sample DNA was sonically sheared to produce a majority of fragments between 500 bp and 2000 bp and labelled with Cy3 and Cy5 fluorescent dyes (Amersham).

Each sample consisted of a combined statistical analysis of four chips that included two chips for biological replicates and two chips for dye reversals of test and reference chromosomal DNA. Each chip contained triplicate representation of the C. jejuni chromosome. Competitive hybridizations were performed in a SlideHyb (Ambion) at 54 °C followed by post-hybridization washes at 42 °C. Chips were scanned for fluorescent signal with an Affymetrix 428 scanner. Hybridization data were analysed as the distribution of the two-colour signal ratios by using GACK (Genomotyping Analysis by Charlie Kim) (Kim ). The GACK program uses the shape of the log2 fluorescent signal distributions to assign estimated probability of presence (EPP) values to genes. For each array, analysis of the log2(test strain/reference strain signals) distribution of good probes was performed.

DNA sequence analysis.

The chromosomal DNA of C. jejuni cultures recovered from broilers and mice was extracted using Easy-DNA (Invitrogen) according to the manufacturer's recommendations. DNA concentration was determined with a Nanodrop ND-1000 spectrophotometer, and 200–300 ng was amplified with the high-fidelity Pfx50 DNA polymerase (Invitrogen). Primers JL 1273 (5′-TGC TGG GTA TAC AAA GGT TGT G-3′) and JL 1274 (5′-GTG CTA AAG TAG CAA CTT CAC C-3′) were used to amplify a 415 bp fragment of wlaN. Primers JL 1225 (5′-GAG CGT TTA GAA TGG GTG TG-3′) and JL 1226 (5′-GCC AGG AAT TGA TGG CAT AG-3′) were used to amplify a 390 bp fragment of flgR. PCR concentrations were as follows: 1× Pfx50 PCR buffer (Invitrogen), 0.2 mM (each) dNTP, 0.5 pmol each primer μl−1, 0.1 unit polymerase μl−1. The PCR thermocycling consisted of an initial denaturation at 94 °C for 2 min, followed by 35 cycles of DNA annealing at 60 °C for 30 s, polymerization at 68 °C for 30 s, and denaturation at 94 °C for 30 s. A final polymerization at 68 °C for 5 min was performed. In preparation for TA cloning, each PCR was incubated with 0.5 units Taq DNA Polymerase (Invitrogen) for an additional 10 min after thermocycling, and purified using the QIAquick PCR Purification kit (Qiagen).

PCR products were cloned into the pCR 4-TOPO vector using the TOPO TA Cloning Kit for Sequencing (Invitrogen) and transformed into One Shot TOP10 competent E. coli. For each PCR and cloning reaction, plasmids from 20 clones were prepared using standard alkaline lysis (Ausubel ) and purified with Wizard Plus SV Miniprep Columns (Promega). Plasmid insert DNA was sequenced in both directions using T3 and T7 primers and either an ABI 3730 genetic analyser or an ABI Prism 3700 DNA analyser (Applied Biosystems) through the Research Technology Support Facility at Michigan State University.

RESULTS

C. jejuni colonization of C57BL/6J IL10-deficient (IL10−/−) mice

In initial pilot experiments, C. jejuni failed to colonize C57BL/6J IL10-deficient mice consistently. We hypothesized that single-colony isolation during preparation of the mouse inoculum may have introduced a bottleneck in the genetic diversity of the C. jejuni population, thereby reducing colonization potential and generating inconsistencies in the colonization data. We eliminated the single-colony isolation step during inoculum preparation for all strains and achieved consistent colonization data with strain 11168; this change in protocol did not result in colonization by strain 33292 or 81-176.

Three sets of eight C57BL/6J IL10-deficient mice were inoculated with C. jejuni strains 11168, 33292 and 81-176. Each set (4 male and 4 female) was inoculated with one strain and each mouse received an inoculum of 1010 c.f.u. Two control mice received TSB only. Colonization was monitored by faecal culture on days 1, 4, 7, 10 and 13. Mice were necropsied on day 14 and colonization assayed by caecal culture. Control mice were culture negative at all time points. On day 1 (24 h), all 24 mice inoculated with C. jejuni were culture positive. On average, mice carried 11168, 33292 and 81-176 at 29 000±8400, 1300±540 and 1500±790 c.f.u. per mg dry faecal mass, respectively (mean±se). Mice inoculated with C. jejuni 33292 and 81-176 were culture negative on days 4, 7, 10, 13, and 14, with the exception of a single mouse inoculated with 33292, which carried 4 c.f.u. per mg dry faecal mass on day 7. The limit of detection during culture ranged from 160 to 400 c.f.u. per g dry faecal or caecal mass based on the negative culture data from control mice.

All eight mice inoculated with C. jejuni 11168 were culture positive at each time point analysed (Fig. 1). Differences in colonization level between individual mice inoculated with 11168 fluctuated over the 14 day period. At day 1, colonization level in individual mice ranged from 8.3×102 to 7.1×104 c.f.u. per mg dry faecal mass, a difference of approximately 100-fold. The difference in colonization level peaked at day 7, with a range from 4.7×100 to 1.3×105 c.f.u. per mg dry faecal mass (a difference of approximately 10 000-fold). The diversity of colonization levels between individual mice inoculated with 11168 was again demonstrated during examination of caecal tissue at necropsy, with a range of colonization from 8.9×101 to 3.5×105 c.f.u. per mg dry caecal mass. C. jejuni isolates 11168-728 and 11168-732 were obtained from individual colonized mice and used for DNA analysis because of their extreme differences in colonization level. On day 1, 11168-728 and 11168-732 were each present at 2.0×104 c.f.u. per mg dry faecal mass (C. jejuni). On day 14, 11168-728 (caecal sample) was present at 3.1×105 c.f.u. per mg dry caecal mass (C. jejuni) and 11168-732 (caecal sample) at 8.9×101 c.f.u. per mg dry caecal mass, a greater than 3000-fold difference in population level.

Fig. 1.

Time-course of C. jejuni 11168 colonization of individual C57BL/6J IL10-deficient mice. Mice were orally gavaged with 1010 c.f.u. of C. jejuni 11168 at day 0. Faecal samples were taken at days 1, 4, 7, 10 and 13 and plated onto TSBA-CVA selective medium. Mice were sacrificed on day 14 and caecal tissue was homogenized and plated onto TSBA-CVA. (a) Plot of the mean and standard error of colonization data for eight C. jejuni 11168-inoculated mice. (b) Each plot represents colonization data from a single, separately caged mouse. The chromosomal DNA from 11168-728 and 11168-732 was harvested for DNA/DNA microarray analysis to determine the stability of the C. jejuni chromosome during colonization.

C. jejuni colonization of Ross 308 broilers

The lack of colonization of mice by C. jejuni 33292 and 81-176 prompted us to determine if these strains were able to efficiently colonize broiler chickens, an important human disease reservoir. Three sets of eight Ross 308 broiler chickens were inoculated with C. jejuni strains 11168, 33292 and 81-176. Each set was inoculated with one strain and each chicken received an inoculum of 108 c.f.u. The chicken experiments were performed after the mouse studies, and chicken doses were reduced to 108 c.f.u. after it was demonstrated by Mansfield that there was not a significant difference in colonization between C57BL/6J IL10-deficient mice inoculated with 108 and 1010 c.f.u. of C. jejuni. Six control chickens received TSB only. Colonization was monitored by cloacal swab on days 1, 4, 7 and 10. Chickens were necropsied on day 14 and colonization assayed by caecal culture. Chickens, unlike mice, produce a watery faeces which is difficult to collect from the bird; thus cloacal swabs were used to assay colonization. Cloacal swab data were reported as colonization positive or negative and they were not normalized to faecal mass.

Control chickens were culture negative at all time points. Ross 308 broilers inoculated with C. jejuni were all culture positive on day 1 post-inoculation. Twenty-one of twenty-four broilers inoculated with C. jejuni were culture positive at all time points analysed. A single Ross 308 broiler (17) inoculated with strain 81-176 was culture negative (cloacal swab) at days 7 and 10; however, this bird carried a low concentration of 81-176 (5×10−1 c.f.u. per mg dry caecal tissue) on day 14. Two chickens (24 and 25) inoculated with strain 33292 were culture negative on day 4, and a single chicken (27) inoculated with this strain was culture negative on days 7 and 10, and 33292 was not detected in the caecum sample at day 14. Chickens 24 and 25 on day 14 carried C. jejuni at concentrations of 1.4×103 and 3.5×103 c.f.u. per mg dry caecal tissue, respectively. The limit of detection for caecal culture of chickens was similar to that of mice.

At day 14 post-inoculation all birds inoculated with C. jejuni strain 11168 were culture positive (caecal samples; Fig. 2). The range of 11168 colonization levels in chickens was 5.2×104 to 1.6×106 c.f.u. per mg dry caecal tissue, while the 11168 colonzation levels in mice varied by approximately 1000-fold (see above); the range in colonization levels in chickens was 100-fold greater than that in mice. While strains 81-176 and 33292 failed to colonize mice they successfully colonized chickens; colonization ranged from 6.7×102 to 9.2×103 c.f.u. per mg dry caecal tissue in 14 of 16 chickens.

Fig. 2.

C. jejuni colonization of C57BL/6J IL10-deficient mice and Ross 308 broiler caecal tissue at 14 days post-inoculation. Eight mice and eight chickens were sacrificed on day 14 and caecal tissue was homogenized, serially diluted, and plated onto TSBA-CVA. Mice inoculated with strains 33292 and 81-176 were culture negative at 14 days post-inoculation, with the limit of detection in these experiments ranging from 150 to 300 c.f.u. per g dry caecal tissue. Chicken 17 had a relatively low C. jejuni concentration of 5×10−1 c.f.u. per mg dry caecal tissue while chicken 27 did not have a detectable level of C. jejuni. Each animal was housed separately in an individual cage.

Colonization of Ross 308 broilers and C57BL/6J IL10-deficient mice by genetically marked C. jejuni 11168

A long-term goal of our research is to understand the process of natural transformation and its role in the generation of genetic diversity in C. jejuni strains. To this end, we examined the ability of populations of genetically marked progeny derived from the same parent to colonize chickens and mice. Genetically marked strains (chloramphenicol-resistant and kanamycin-resistant populations) derived from C. jejuni 11168 were co-inoculated into 42 C57BL/6J IL10-deficient mice (Table 1). At approximately 1 day post-inoculation all mice were culture positive for C. jejuni, but by day 4, 39 of 42 (93 %) mice were culture negative, and by day 7, 40 of 42 (95 %) mice were culture negative for C. jejuni. Two mice remained culture positive on day 14 (necropsy). Mouse 712 carried a chloramphenicol-resistant population (MATn5CamR6) at a concentration of 4.9×105 c.f.u. per mg dry caecal tissue and mouse 682 maintained a kanamycin-resistant population (MA23SKanR4) at a concentration of 2.4×105 c.f.u. per mg dry caecal tissue.

Table 1.

Co-inoculation and colonization of C57BL/6J IL10-deficient mice in caecal tissue 14 or 16 days post-inoculation by genetically marked C. jejuni 11168

Mouse caecal samples were homogenized in TSB, serially diluted, plated onto TSBA-CVA medium and TSBA-CVA supplemented with chloramphenicol and/or kanamycin, and incubated at 37 °C under under microaerophilic conditions for 72 h (see Methods). Culture-positive mice carried C. jejuni at necropsy.

Cj 11168 mutant progeny co-inoculum	Mice inoculated	Culture-positive mice, day 14–16 post-inoculation
Tn5CamR2+23SKanR4	5 male, 5 female	Detected in 1 of 10 mice (*Cj 11168 KanR) 2.4×105 c.f.u. (mg caecal tissue)−1
Tn5CamR2+hipOKanR1	5 male, 5 female	Not detected
Tn5CamR6+23SKanR4	5 male, 5 female	Detected in 1 of 10 mice (*Cj 11168 CamR) 4.9×105 c.f.u. (mg caecal tissue)−1
Tn5CamR6+hipOKanR1	5 male, 5 female	Not detected
hipOCamR1+23SKanR4	1 male, 1 female	Not detected
*Tn5CamR6+*23SKanR4	4 male, 4 female	Detected in 8 of 8 mice

*Represents a mouse-adapted strain.

In a subsequent experiment, the MATn5CamR6 and MA23SKanR4 cultures recovered from mouse caecal tissue were co-inoculated into eight C57BL/6J IL10-deficient mice (Table 1). All mice inoculated with these mouse-adapted C. jejuni cultures were Campylobacter positive at every time point examined and carried 2.1×101 to 1.9×105 c.f.u. C. jejuni per mg dry caecal tissue (necropsy), a range of approximately 10 000-fold (Fig. 3). Kanamycin-resistant C. jejuni cultures were not recovered from caecal tissue in these mice, and serial dilutions of recovered cultures gave similar C. jejuni densities on TSBA-CVA and TSBA-CVA plus chloramphenicol, indicating that a majority if not all of the population recovered was chloramphenicol resistant. Six control mice (TSB only) were culture negative for C. jejuni at all time points. MATn5CamR6, MA23SKanR4 and 11168-Tn5CamR2 were also inoculated as individual cultures into four C57BL/6J IL10-deficient mice per strain. At day 14 post-inoculation, all eight mice inoculated with the mouse-adapted cultures were culture positive for C. jejuni and carried 1.5×102 to 1.5×106 c.f.u. per mg dry caecal tissue (Fig. 3). Three of the four 11168-Tn5CamR2-inoculated mice were culture negative for C. jejuni on day 14 (caecal culture); the fourth mouse carried 3.0×104 c.f.u. per mg dry caecal tissue (Fig. 3). Two control mice (TSB only) were culture negative for C. jejuni at all time points.

Fig. 3.

Colonization of C57BL/6J IL10-deficient mice and Ross 308 broiler caecal tissue 14 or 16 day post-inoculation by genetically marked C. jejuni 11168. Colonization data for wild-type C. jejuni 11168 in mouse and chicken caecal tissue are presented (as in Fig. 2), for comparison with 11168 mutant progeny. For inoculation of mice, the majority of genetically marked 11678 strains were inoculated as co-cultures as defined in Table 1. Of 46 mice inoculated with non-adapted mutant progeny, 43 were culture negative for C. jejuni 2 weeks post-inoculation, with 42 of the mice receiving co-inocula. The mouse-adapted cultures were 11168-Tn5CamR6 (indicated by circles in the Adapted column) and 11168-23SKanR4 (indicated by triangles in the Adapted column), recovered from the caecal tissue on day 14 post-inoculation of mouse 712 and mouse 682 respectively. The limit of detection in these experiments ranged from 150 to 900 c.f.u. per g dry caecal tissue. Each animal was housed separately in an individual cage.

C. jejuni 11168-Tn5CamR2 and 11168-23SKan4 were inoculated as individual cultures into seven Ross 308 broilers each. Broilers were assayed for C. jejuni by cloacal swab culture on days 1, 4, 7 and 10, and by caecal culture on day 14. The chickens were culture positive on all days examined. Each chicken carried between 9.7×100 and 3.2×105 c.f.u. per mg dry caecal tissue (Fig. 3). Five control broilers (TSB only) were culture negative for C. jejuni at all time points.

Microarray DNA–DNA hybridization analysis

We conducted microarray analysis to explore whether genetic diversity might at least in part explain the observed differences in colonization levels between different C. jejuni strains inoculated into mice (comparison of 11168, 81-176, and 33292 for example) and between high- and low-level colonizing isolates of the same strain (11168-728 and 11168-732) obtained from mice. Chromosomal DNAs isolated from ‘test’ strains were fluorescently labelled and used as probes in a competitive hybridization against a labelled chromosomal DNA probe from 11168, the ‘reference’ strain. PCR products from 1681 ORFs representing the entire 11168 genome were printed in replicate on the microarray slide and these served as a reporter target for hybridization. Three adjacent gene clusters demonstrated a high degree of genetic diversity between 11168, 33292 and 81-176 (Fig. 4). The first of these clusters, a protein glycosylation (PG) gene cluster, lies adjacent to the second cluster, a LOS biosynthesis cluster. Together they form one ‘locus’ called PGLOS. This ‘locus’ comprises approximately 33 600 bp in the 11168 chromosome; the ends of the locus are delineated by the genes wlaM/pglG (gene position 1086) and waaD/hldD (gene position 1116). The capsule biosynthesis locus (CAP) is approximately 42 600 bp in length; its ends are defined by the genes kpsS (gene position 1373) at one end and kpsM (gene position 1408) at the other. The PGLOS locus has a G+C content of 29 mol%, compared to 30–31 mol% for the equivalent-sized regions flanking PGLOS, and the CAP locus has a G+C content of 28 mol% compared to 30 mol% in the flanking sequences. The G+C content of the entire 11168 chromosome is 31 mol% (Parkhill ).

Fig. 4.

Microarray DNA/DNA hybridization of C. jejuni chromosomal DNAs. A glass slide array of C. jejuni 11168 gene-specific PCR ORFs served as a microarray reporter. C. jejuni strains D0121 and D2600 are distantly related human isolates as demonstrated by MLST analysis (personal communication, Erin Stanley). These strains were not used in colonization studies but are included here to provide more robust analysis between chromosomes. Samples consisted of two fluorescently labelled chromosomal preparations (test and reference conditions) in a competitive hybridization against the reporter DNA. C. jejuni 11168 chromosomal DNA served as reference in all hybridizations. A GACK score of –0.5 represents a 100 % chance of sequence divergence and an estimated probability of presence (EPP) ≤0.5 %. A GACK score of 0.5 represents a 100 % chance of identical sequence and an EPP ≥99.5 %.

In contrast, over 99 % of the 1681 ORFs analysed in the isolate of the high-efficiency mouse colonizer 11168-728 (Fig. 1) exhibited high sequence similarity (≥99.5 % EPP) to ORFs in the 11168 reference strain (original inoculum). Similarly, over 97 % of ORFs in the isolate of the low-efficiency colonizer 11168-732 demonstrated high sequence similarity (≥99.5 % EPP) to the 11168 reference strain. Of particular significance, very few significant genetic differences were observed between 11168-728 and 11168-732 within the three gene clusters discussed above.

Analysis of homopolymeric tracts in contingency genes

To further explore the potential role of genetic diversity in the ability of C. jejuni to colonize broilers and mice, we analysed the frequency of slip-strand mutations in two genes, wlaN and flgR, shown to carry homopolymeric tracts (Parkhill ), as indicators of genetic diversity.

The C. jejuni gene wlaN encodes a β-1,3-galactosyltransferase involved in LOS biosynthesis (Linton ). If wlaN expression or enzyme activity is blocked, the LOS structure lacks a terminal galactose residue. Wild-type wlaN generates a LOS structure that carries a terminal galactose and mimics human ganglioside GM1, resulting in an increased risk for GBS in infected patients. Linton demonstrated that WlaN enzyme activity is subject to phase variation (a high-frequency genetic event that alters gene expression or protein activity) by slip-strand mutations (see below) that occur at very high frequency within a homopolymeric [poly(G)] tract in the wlaN open reading frame. We measured mutation frequency in wlaN as an indicator of the degree of genetic diversity in C. jejuni populations; this is particularly relevant because this genetic change appears to have a direct impact on human health. wlaN carries a homopolymeric tract of eight G residues.

flgR encodes one protein in a two-component regulatory pathway that controls flagellar synthesis and modification (Hendrixson, 2006). The homopolymeric sequence in flgR consists of four A/T-rich stretches (7 to 8 residues each) within a 50 bp fragment. Slip-strand mutations in either gene eliminate protein activity.

We purified genomic DNA from C. jeuni populations isolated from: (1) wild-type 11168 broiler inoculum (CI); (2) wild-type 11168 mouse inoculum (MI); (3) wild-type 11168-colonized broilers (four birds, C4, 5, 7 and 11); and (4) wild-type 11168-colonized mice (four mice, M51, 719, 728 and 732). DNA (400 bp) surrounding the wlaN homopolymeric tract was amplified (PCR), the PCR fragment ligated into plasmid PCR4-TOPO, and the recombinant plasmid transformed into E. coli. The wlaN region carried in 20 independent plasmids derived from each of the 10 PCR fragments (200 plasmids total) was subjected to DNA sequence analysis (Fig. 5). In ‘wild-type’ C. jejuni 11168 animal inocula, 30–40 % of the wlaN clones carried a single base insertion in the wlaN homopolymeric tract (+1) while 60–70 % of the clones carried wild-type wlaN. These data suggest that nearly one in every two bacterial cells in the so-called ‘wild-type’ inocula carried a mutation in the wlaN homopolymeric tract. In contrast, in wlaN clones obtained from four different broilers (80 plasmids total), nearly 80 % of the clones carried a single base insertion (+1), 5 % carried two base insertions (+2), 2 % carried single base deletions (−1), and 13 % were wild-type in wlaN; the (+2) and (−1) mutant categories were not observed in the original broiler C. jejuni inoculum (CI), suggesting that they either originated or became enriched in the broiler GI tract. Of particular interest, in wlaN clones obtained from four different colonized mice (80 plasmids total), the vast majority (>90 %) carried the wild-type wlaN homopolymeric tract. Only a small number of mouse wlaN clones (<10 %) carried single base (+1) insertions and these occurred in only one of four mice.

Fig. 5.

DNA sequence analysis of the homopolymeric GC tract of wlaN in C. jejuni 11168 cultures recovered from the caecum of C57BL/6J IL10-deficient mice and Ross 308 broilers. We prepared chromosomal DNA extracts from C. jejuni 11168 used to inoculate C57BL/6J IL10-deficient mice (MI) and Ross 308 broilers (CI) as well as from C. jejuni isolates recovered from caecal contents of four mice and four chickens. DNA extracts were amplified with Pfx50 high-fidelity DNA polymerase and wlaN primers. A specific 415 bp amplification product containing the wlaN homopolymeric GC tract was purified, ligated and transformed into E. coli using the PCR4-TOPO cloning system. Twenty individual wlaN clones were selected from each chromosomal DNA amplification and sequenced with either an ABI 3730 genetic analyser or an ABI Prism 3700 DNA analyser using T3 or T7 primers. The relative wlaN DNA sequence populations from each culture are reported (C, Ross 308 Broiler; M, C57BL/6J IL10-deficient mouse).

In wlaN clones prepared from genomic DNA from C. jejuni 11168 Tn5CamR2 animal inocula (which colonized only 1 in 24 mice), the vast majority of C. jejuni isolates (93 %) carried the wild-type wlaN hompolymeric tract. Since the identification of marked strains during strain construction (such as Tn5CamR2) requires single-colony isolation, these data provide strong supporting evidence that single-colony isolation of C. jejuni reduces the overall genetic diversity in the population and reduces the ability of C. jejuni to colonize mice.

A similar analysis was conducted on the flgR homopolymeric sequence in genomic DNA prepared individually from mouse inoculum (MI), chicken inoculum (CI), two mice (M728 and M732) and one broiler (C4). Genetic variation in flgR was observed at lower frequency overall than in wlaN. A single base deletion (−1) was observed in 5 % of the flgR clones in MI and C4. No mutations were observed among 40 flgR clones prepared from C. jejuni populations in the two mice or in the animal inoculum. The data generated from this second contingency gene tend to support our observations based on wlaN. Controls demonstrated that genetic variation in wlaN and flgR occurred within C. jejuni and not during cloning in E. coli.

DISCUSSION

Genetic diversity and host colonization

In the current work, we generated several independent lines of evidence that support our research hypothesis; that is that genetic diversity in subpopulations of C. jejuni enhances the ability to colonize the host animal. Previous studies demonstrated a high mutation frequency in specific contingency genes and support a role for the resulting genetic diversity in the ability to invade INT407 cells, to modify the host immune response (molecular mimicry) in experimental human infection (Guerry ; Prendergast ), and to colonize chickens (Ashgar ; Hendrixson, 2008). Our work expands on these previous studies by demonstrating a strong correlation between the degree of genetic diversity in C. jejuni populations and the differential ability to colonize broilers and mice.

Genetic diversity within gene clusters required for synthesis/modification of cell surface structures.

The three C. jejuni human disease isolates chosen for these analyses (strains 11168, 33292 and 81-176) colonized Ross 308 broilers at higher frequency (all birds) and efficiency (higher c.f.u. per g caecal material) than our mouse model for disease (C57 BL/6J IL10-deficient mice). In general, 11168 colonized all mice and broilers, whereas 33292 and 81-176 colonized all broilers, but failed to colonize mice. Microarray DNA–DNA hybridization analysis demonstrated genetic diversity between each of the three C. jejuni strains within gene clusters associated with the synthesis and modification of the cell-surface structures flagella, capsule and LOS. Although these analyses did not tell us the exact nature of the genetic changes that occurred within these gene clusters at the DNA sequence level, the data do strongly suggest that (1) the degree of DNA sequence divergence between these three strains within these three gene clusters is large, as evidenced by the fact that the genetic differences can be detected reproducibly by a relatively insensitive tool (the probability of presence of a particular gene sequence as measured by DNA/DNA hybridization); and (2) the high degree of genetic divergence occurred predominantly within these three gene clusters. Because a high level of DNA sequence divergence can be demonstrated over a span of several genes within each of the three gene clusters, the data also suggest that the observed genetic diversity likely accumulated over a relatively long time span (years as opposed to hours). Presumably, genetic diversity within the gene clusters generated changes in flagella, LOS and capsule that modulated interaction between the pathogen and the host via effects on host cell attachment and invasion, and on the ability of the host to mount an effective immune response. Future work will analyse surface-structure changes that accompany accumulation of genetic diversity and the specific mechanisms by which these changes affect colonization and disease.

Genetic diversity within contingency genes.

We examined genetic variation in two well-characterized genes that carry homopolymeric tracts susceptible to slip-strand mutation (Linton ; Hendrixson, 2006). The wlaN gene in C. jejuni 11168 encodes a β-1,3-galactosyltransferase; the presence or absence of a terminal galactose on LOS impacts attachment of sialic acid residues via the activity of sialic acid transferases. Sialic acid modification of LOS contributes to its antigenic specificity and to the ability of specific LOS structures to mimic GM1a, GD3, GD1a, GT1a and/or GD3 gangliosides associated with GBS (Linton ; Gilbert ; Godschalk ). An 8 nt poly(G) homopolymeric tract in the coding sequence of wlaN generates a full-length active enzyme (wild-type); one or two base insertions or deletions in this tract generate frameshift mutations that terminate translation and eliminate enzyme activity.

The homopolymeric tract in wlaN exhibited a higher frequency (>80 %) and diversity (one and two base insertions, one base deletions) of mutations in C. jejuni populations isolated from broilers as compared to the C. jejuni animal inoculum (40 %; predominantly one base insertions). In contrast, C. jejuni populations after passage in mice carried predominantly wild-type homopolymeric tract (>80 %). We observed a similar frequency of mutation as detected previously by others in C. jejuni populations (Parkhill ; Linton ; Gilbert ; Karlyshev ). In contrast to these studies, Wassenaar observed that the wlaN homopolymeric tract in C. jejuni 11168 remained stable under all conditions examined, including chicken colonization. Because Wassenaar detected high mutation frequencies in other contingency genes, the absences of mutation in wlaN in their study may reflect differences in selective pressure experienced by C. jejuni during growth in vivo.

We also analysed flgR, which carries a series of four poly(A) tracts, each seven or more nucleotides in length, within the 11168 coding region. FlgR is a σ54-dependent NtrC-like protein that regulates flagellar expression in C. jejuni. It is the response regulator in a two-component regulatory system but is also subject to phase variation through slip-strand mutagenesis in its poly(A/T) tracts (Hendrixson, 2006; Joslin & Hendrixson, 2008). Our data suggest that mutations may occur more frequently in flgR in the chicken host than in the mouse host, implying that motility is essential in the mouse GI tract but is less so in chickens. However, more extensive experiments must be conducted to expand on these observations.

DNA sequence analysis of homopolymeric tracts embedded in wlaN and flgR demonstrated that subtle genetic differences (one or two base insertions or deletions) can arise at extremely high frequency within populations derived from a single strain during growth in culture and this mutation process is enhanced during colonization of broilers. Although high mutation frequencies have been observed previously in specific C. jejuni contingency genes (including wlaN) (Linton ; Guerry ; Prendergast ) we demonstrated a strong correlation between the degree of genetic diversity observed in the homopolymeric tract in wlaN and the differential ability to colonize mice and broilers. This is an extremely important observation because it suggests that C. jejuni can control the chemical composition and conformation of cell-surface structures via phase variation at high frequency and in real time.

It is important to note that wlaN and several other contingency genes are located within the three adjacent gene clusters that display a high degree of genetic diversity among the three C. jejuni human disease strains. This observation reinforces the notion that genetic diversity is important in colonization but also implies that this particular region of the genome is subject not only to high rates of mutation over very short time periods, but also to more stable accumulation of a large number of mutations that occurs over a much longer time interval. The data also demonstrate a strong ability of C. jejuni to adapt genetically in chickens. Future work will focus on the role of genetic adaptation in the ability to colonize and to cause disease in the mouse model.

Reduced genetic diversity (a genetic bottleneck) reduces colonization.

In initial studies, we constructed genetically marked strains of C. jejuni 11168 to trace the frequency of genetic exchange by natural transformation in the GI tract of mice. These marked strains were inoculated into C57BL/6J IL10-deficient mice and they either failed to colonize or colonized at low frequency (2 of 42 mice); a similar result was observed during initial attempts to colonize mice with wild-type 11168 that had been single-colony isolated during inoculum preparation. We hypothesized that single-colony isolation reduced genetic variation within the culture, resulting in reduced colonization efficiency. This supposition was supported by three observations: (1) when we eliminated single-colony isolation during inoculum preparation of wild-type 11168, mouse colonization increased markedly; (2) when we recovered a C. jejuni 11168 genetically marked derivative from a mouse and reinoculated it into a new set of mice, all mice were colonized; and (3) we detected only a low level of genetic diversity (>10 %) in wlaN in 11168 Tn5CamR2 (subjected to single-colony isolation during strain construction) that was used as an animal inoculum. Our C. jejuni culture protocol now avoids the bottleneck that results from single-colony isolation whenever practical.

These data also illustrate an extremely important principle; that is, that genetic adaptation occurs in the GI tract of mice and this directly affects the subsequent ability to colonize mice (although the mechanism and nature of this adaptation process are not completely understood). As discussed above, we also observed genetic adaptation in broilers. Of particular interest, the adaptation process in both host animals results in large fluctuations in genetic diversity within contingency genes. However, in chickens, adaptation enhances genetic diversity while mouse adaptation severely limits diversity. Our current work shows that mouse adaptation has a major impact on subsequent ability to colonize mice. Preliminary data generated in our laboratory also suggest that chicken adaptation enhances subsequent mouse colonization (unpublished). Future work will focus on understanding the mechanisms that underlie genetic adaptation in chickens and mice and the role of adaptation in regulating colonization and disease in the host animal.

Proposed role of the chicken reservoir in human disease

We now propose that the poultry reservoir plays at least two important roles in human disease caused by C. jejuni. The broiler GI tract provides a favourable environment for expansion of C. jejuni populations. This increases the potential for human exposure to this foodborne pathogen in improperly prepared poultry products. One novel contribution of the current study is that C. jejuni populations also experience a major expansion in genetic diversity during outgrowth in the broiler GI tract that appears to enhance colonization of our mouse model for human disease. Future work will explore the association between genetic diversity and mouse colonization in more detail.

The host immune response and genetic diversity

Based on these analyses, we propose a model that we believe, at least in part, explains the role of the host immune system in shaping genetic diversity in C. jejuni. According to this model, the host immune system provides strong selective pressure that acts as a filter for genetic diversity in C. jejuni populations. We hypothesize that the broiler immune system is much more tolerant in its ability to recognize and eliminate C. jejuni from the GI tract. As a result, most C. jejuni can colonize broilers at high population levels. In contrast, we hypothesize that the mouse immune system is more stringent. It recognizes C. jejuni cells that carry LOS lacking a terminal galactose (due to mutation and inactivation of wlaN) and successfully eliminates these from the GI tract. Cells expressing wild-type wlaN escape immune surveillance; this population becomes predominant in mice. We also propose that the host immune system plays an important role in the observed accumulation of a high level of genetic diversity within the PGLOS and CAP gene clusters (loci). Genetic variation within these clusters can generate subpopulations carrying cell-surface antigens that escape immune surveillance and these successful subpopulations expand. These subpopulations can then predominate in future rounds of colonization, resulting in a high level of diversity being ‘fixed’ over a long period of time. This analysis implies that selective pressure shapes the predominant genotypes observed in human clinical isolates of C. jejuni.

This analysis prompts us to hypothesize that the host immune system generates subpopulations with specific patterns of mutations within contingency genes and that understanding this pattern may allow us to predict the potential virulence of specific genotypes. Theoretically, one could analyse patterns of mutations in contingency genes to help understand the relative importance of individual genes and gene combinations in colonization and disease.