*For correspondence at the Institute of Animal Science and Health. E-mail firstname.lastname@example.org; Tel. (+31) 320 238 161; Fax (+31) 320 238 153.
Molecular epidemiology studies suggest that horizontal genetic exchange is a major cause of pathogen biodiversity. We tested this concept for the bacterial enteropathogen Campylobacter jejuni by seeking direct in vivo evidence for the exchange of genetic material among Campylobacter strains. For this purpose, two antibiotic resistance markers were inserted into the hipO or htrA gene of genetically distinct and naturally transformable C. jejuni strains. Genetic exchange of the resistance markers was analysed after co-cultivation of homologous and heterologous strains in vitro and in vivo during experimental infection of chickens. Double-resistant recombinants were obtained both in vitro and from the chicken intestine for all combinations of strains tested. Bidirectional genetic exchange of DNA between homologous and heterologous strains was confirmed by Southern blotting in combination with flaA polymerase chain reaction–restriction fragment length polymorphism (PCR–RFLP), amplified fragment length polymorphism (AFLP) and pulsed field gel electrophoresis (PFGE). Extensive PFGE analyses of isolated recombinants indicated the frequent occurrence of genetic rearrangements during the experimental infection, in addition to the homologous recombination of the antibiotic resistance genes. Together, the data indicate unequivocally that interstrain genetic exchange as well as intragenomic alterations do occur in vivo during C. jejuni infection. These events probably explain the genome plasticity observed for this pathogen.
Genetic diversity is considered a major trait of many pathogen populations. This variation in genotype provides a bacterial population with a genome plasticity that may enlarge the adaptation potential and thus the survival of the pathogen in hostile environments. Advanced molecular epidemiological analyses suggest that genotype diversity is being generated continuously. A number of genetic mechanisms have been proposed to contribute to the diversity including the horizontal transfer of genes within and between bacterial populations and intragenomic events such as rearrangements, point mutations, deletions, duplications and inversions (reviewed by Feil et al., 2001). One pathogen that has a largely non-clonal population structure is the enteropathogen Campylobacter jejuni. This bacterium is the major cause of food-borne bacterial gastroenteritis worldwide (Tauxe, 1992) and has been associated with the development of the Guillain–Barré syndrome, an acute inflammatory polyneuropathy (Nachamkin et al., 1998; Wassenaar and Blaser, 1999). C. jejuni is widespread in the environment and appears to exhibit typical commensal behaviour in livestock. C. jejuni genotype diversity has been demonstrated via the analysis of a large number of isolates and with a variety of molecular epidemiological techniques including multilocus sequence typing (MLST) (Duim et al., 1999; Wassenaar and Newell, 2000; Dingle et al., 2001; Suerbaum et al., 2001). On the basis of the molecular typing, it has been suggested that C. jejuni generates extensive genetic diversity through intra- and interspecies recombination (Dingle et al., 2001; Suerbaum et al., 2001).
Despite the wide acceptance and theoretical considerations, direct in vivo experimental evidence that horizontal transfer of DNA generates genetic diversity among bacteria in their natural habitat is sparse. The event requires the simultaneous presence of multiple strains at a distinct niche and active mechanisms that allow DNA transfer and integration into the chromosome. C. jejuni appears to fulfil these criteria, as multiple strains are frequently isolated from the same host, and several C. jejuni strains have been demonstrated to be naturally competent for DNA uptake. Furthermore, homologous recombination has been described for the virulence-associated flagellin genes (Wassenaar et al., 1995; Harrington et al., 1997; Nuijten et al., 2000). In order to demonstrate unequivocally that genetic exchange contributes to the generation of genetic diversity in C. jejuni in vivo in the absence of any apparent selective (immunological) pressure, we have investigated the bidirectional transfer of two non-essential genes between C. jejuni strains during colonization of chickens. Our data indicate that genetic exchange does occur in vivo between both homologous and heterologous strains. In addition, evidence was found for in vivo intragenomic events such as duplications and/or point mutations contributing to genetic diversity.
Introduction of genetic markers in C. jejuni strains 2412 and 2535
In order to assess whether genetic exchange between C. jejuni strains contributes to genetic diversity in vivo, Cmr and Kmr antibiotic resistance markers were introduced into C. jejuni strains 2412 and 2535. For this purpose, the hipO gene, encoding the non-essential enzyme benzoylglycine amidohydrolase (hippuricase) (Hani and Chan, 1995) was cloned into pBSK− and disrupted by insertion of the cat gene. Sequence analysis confirmed the correct amplification of hipO (data not shown). In a similar approach, the aphA-3 gene was inserted into the htrA gene, encoding the heat shock protein HtrA, a serine protease (Henderson, 1996). Both genes were selected because they were genetically conserved among strains, not essential for colonization of chickens (see below) and supposedly not subject to selective pressure in chickens.
Electrotransformation of strains 2412 and 2535 with pHipCat (pHipO::cat) or pHtrA2T (pHtrA::aphA-3) yielded the desired four different Cmr and Kmr mutants verified by Southern blotting using cat, aphA-3, hipO and htrA probes (data not shown). The mutants were designated 2412hipO::Cmr, 2412htrA::Kmr, 2535hipO::Cmr and 2535htrA::Kmr.
Exchange of DNA among C. jejuni strains under laboratory conditions
The potential of the constructed Cmr and Kmr strains to exchange genetic material was first determined under laboratory conditions. In these experiments, the Cmr and Kmr mutants were mixed in all four possible combinations (i.e. 2412hipO::Cmr + 2412htrA::Kmr, 2535hipO::Cmr + 2535htrA::Kmr, 2412hipO::Cmr + 2535htrA::Kmr and 2412htrA::Kmr + 2535hipO::Cmr). Mixed suspensions were added to HIS plates and to a biphasic medium known to favour natural competence (Wang and Taylor, 1990). After incubation for 24 h in biphasic medium and growth for 48 h on HIS plates, the bacteria were collected from both media and used to inoculate fresh media containing either none or various combinations of antibiotics. After growth, both procedures resulted in ≈ 104–105 double-resistant mutants per 107–109 bacteria for all combinations tested (Table 1). These data indicate that the strains used had the intrinsic ability to exchange DNA with both the parent strain and the heterologous strain and, thus, were suitable for use in in vivo recombination experiments.
Table 1. Results of in vitro recombination experiments on plates and in biphasic medium of isogenic and heterogenic combinations of C. jejuni mutants.
Cm + Km
Cfus are indicated as numbers counted on HIS plates containing the following antibiotics: Cm, chloramphenicol; Km, kanamycin; Cm + Km, chloramphenicol + kanamycin.
2412hipO::Cmr + 2412htrA::Kmr
2535hipO::Cmr + 2535htrA::Kmr
2412hipO::Cmr + 2535htrA::Kmr
2412htrA::Kmr + 2535hipO::Cmr
2412hipO::Cmr + 2412htrA::Kmr
2535hipO::Cmr + 2535htrA::Kmr
2412hipO::Cmr + 2535htrA::Kmr
2412htrA::Kmr + 2535hipO::Cmr
Exchange of DNA among C. jejuni strains during experimental infection of chicken
Campylobacter jejuni is naturally adapted to colonize the chicken intestine in large numbers, and different strains are able to colonize the intestine simultaneously (Jacobs-Reitsma et al., 1995). Therefore, a chicken infection model was used to study the genetic exchange between the constructed recombinant C. jejuni strains in vivo. Culturing of cloacal swabs taken before infection using both enrichment and direct sampling on selective plates confirmed the negative status of the animals for Campylobacter and Salmonella at the start of the experiment.
Campylobacter colonization of chicken was established via oral administration of four combinations of Cmr and Kmr mutants (105 cfu of each mutant) to four groups of 30 7-day-old broilers. The animals in groups 1 and 2 received the homologous combinations of strains 2412hipO::Cmr + 2412htrA::Kmr and 2535hipO::Cmr + 2535htrA::Kmr respectively, whereas the other groups received the heterologous combinations of 2412hipO::Cmr + 2535htrA::Kmr (group 3) and 2412htrA::Kmr + 2535hipO::Cmr (group 4). Quantitative sampling of the caeca from five chickens from each group at days 2, 7, 10, 14, 21 and 29 after inoculation showed that all chickens were colonized with 105–109 cfu per g of caecum content throughout the entire period and that all recombinant strains showed comparable colonization characteristics, i.e. co-colonization was achieved for each of the combinations of strains (data not shown). The co-colonization numbers of the mutants were comparable with the colonization numbers found for the wild types (P. de Boer et al., unpublished), confirming the assumption that the hipO and htrA mutations had no effect on the (co)colonization potential of the mutants.
In search for in vivo genetic transfer events, chicken intestinal flora was tested for the presence of double-resistant isolates using selective media. Double-resistant C. jejuni were obtained from all groups of chickens. The first double-resistant mutants were found at 2 days (group 1) to 10 days (groups 2–4) after inoculation, and these recombinants remained present during the entire infection period. Chickens inoculated with the homologous combination 2412hipO::Cmr + 2412htrA::Kmr yielded double-resistant mutants in considerably more animals than the comparable combination of 2535 mutants and the combinations of heterologous strains (Table 2). Recombination frequencies were not calculated, as it cannot be deciphered whether the recovered double-resistant mutants resulted from separate recombination events or were mainly the progeny of a few mutants.
Table 2. Number of chickens from which double-resistant mutants were successfully isolated at various periods post inoculation (p.i.).
The number of double-resistant mutants ranged from 1.0*102 to 8.9*103 g−1 caecal content. In general, the total number of Campylobacters isolated ranged from 105 to 109 g−1 caecal content.
Infection with 2412hipO::Cmr + 2412htrA::Kmr
Infection with 2535hipO::Cmr + 2535htrA::Kmr
Infection with 2412hipO::Cmr + 2535htrA::Kmr
Infection with 2412htrA::Kmr + 2535hipO::Cmr
Southern blot analysis of double-resistant strains
In order to verify that the double-resistant phenotype was caused by horizontal transfer of DNA, the chromosomal DNA of two randomly selected double-resistant mutants derived from each group (mutants 1, 9, 14, 18, 22, 35, 38 and 52) was subjected to Southern blotting with cat, aphA-3, hipO and htrA as probes (Fig. 1). All eight double-resistant mutants showed hybridizing bands of similar size to the parental strains 2412 or 2535 except 35, in which an ≈ 10 kb-sized fragment appeared to have shifted into a 7 kb fragment (Fig. 1A and C, indicated by an asterisk). Together, the results indicate that both resistance genes were present in the mutants and were located on the same fragment as the hipO and htrA genes and, thus, that genetic exchange of DNA had occurred between the various strains in the chicken.
Bidirectional transfer of genetic material
In order to investigate which of the two co-colonizing strains had acted as donor and recipient of the foreign DNA in the chickens, the eight double-resistant C. jejuni strains that had been analysed by Southern blotting were genotyped by flaA polymerase chain reaction– restriction fragment length polymorphism (PCR–RFLP) typing, amplified fragment length polymorphism (AFLP) analysis and pulsed field gel electrophoresis (PFGE). The patterns obtained were compared with those from the four parental strains. Double-resistant strains 1, 9, 22, 38 and 52 showed a genotype similar to the parental strain 2412hipO::Cmr with both flaA PCR– RFLP and AFLP (Table 3), indicating that these strains must have acted as recipients of the aphA-3 gene from 2535htrA::Kmr. Similarly, strains 14, 18 and 35 showed a 2535htrA::Kmr genetic background, suggesting that, in these cases, strain 2412hipO::Cmr had served as the DNA donor. Corresponding results were found with PFGE typing, except for strain 52, which showed a genotype dissimilar from both 2412 and 2535 (Fig. 2, Table 3). Thus, in vivo, both strains served as donor as well as acceptor strain in the transfer of genetic material.
Table 3. Summary of the DNA typing data of all analysed double-resistant mutants and the initially used mutants.
The original typing patterns of strains 2412 and 2535 are indicated by A and B respectively. A1 to A7 and B1 to B9 represent PFGE patterns that closely resembled but were not identical to those of A or B respectively. PFGE patterns C and D indicate new PFGE patterns that could not be correlated to the original PFGE patterns.
Recombinants that have been analysed by MLST are indicated in bold.
Chickens 1–5 were sacrificed 2 days p.i., chickens 6–10 at 7 days, chickens 11–15 at 10 days, chickens16–20 at 14 days, chickens 21–25 at 21 days and chickens 26–30 at 29 days p.i.
Analysis of an additional seven (group 1) and nine (group 2) double-resistant mutants yielded the same patterns as A and B, respectively, with all the typing methods used.
Not possible to determine because of a homologous genetic background.
The identification of a novel PFGE type for the double-resistant mutant 52 suggested that perhaps the C. jejuni genotype may not be stable in vivo. To investigate this further, we analysed a total of 47 double-resistant mutants by PFGE. This analysis revealed several novel PFGE types that closely resembled patterns from 2412htrA::Kmr and 2535htrA::Kmr (e.g. strain 29 closely resembled 2535htrA::Kmr; Fig. 2, Table 3) as well as completely novel patterns that were totally dissimilar from the parental genotypes (e.g. strain 52; Fig. 2, Table 3). Novel PFGE types were not only identified for double-resistant mutants from different animals colonized with heterologous strains (PFGE type of strain 32 matched strain 37 from group 3 and strains 47 and 49 matched strain 50 from group 4; Table 3), but also for different double-resistant mutants isolated from a single chicken (e.g. chicken 12 from group 3 yielded four different PFGE types; Table 3). Interestingly, novel genotypes were also found for isolates derived from animals that had been colonized with strains that had an identical genetic background except for the resistance marker (e.g. strain 9; Table 3). MLST analysis of 10 double-resistant mutants showed identical sequences for all seven sequenced housekeeping genes (Table 3), indicating that the altered PFGE patterns were not caused by changes within the loci used for MLST.
In order to investigate whether the changes in PFGE types were caused by specific in vivo conditions, 48 double-resistant recombinants obtained in the in vitro recombination experiments were analysed. Three mutants showed a PFGE pattern that was different from the parental strains. Two of the mutants were isolated from the combination of 2535 hipO::Cmr + 2535htrA::Kmr, and one was derived from the heterologous combination of 2412htrA::Kmr + 2535hipO::Cmr (Fig. 2). Remarkably, the novel PFGE patterns in these mutants were all identical and appeared to be similar to the patterns of double-resistant mutants 24 and 46 that had been isolated from chickens. Analysis of individual strains that underwent >300 passages in vitro yielded no novel PFGE patterns (data not shown). Together, these data strongly suggest that, in addition to allelic exchange, additional (intra)genomic alterations (i.e. rearrangements, mutations, inversions, deletions) do occur in C. jejuni that may alter the bacterial genotype and contribute to pathogen diversity.
The population of C. jejuni consists of genetically diverse strains and a limited number of seemingly clonal lineages. The results presented in this study provide direct experimental evidence for horizontal DNA transfer among C. jejuni strains in their natural in vivo habitat leading to genetic diversity. In addition, intragenomic alterations were observed, leading to even more diversity.
Horizontal gene transfer among C. jejuni strains during infection of chicken was established with strains that were naturally competent for DNA uptake and contained anti-biotic resistance markers inserted into the hipO or htrA genes. The mutant strains efficiently colonized chickens, indicating that the affected genes were not essential for colonization. Direct evidence for in vivo DNA transfer between C. jejuni strains was obtained by the recovery and genetic analysis of double-resistant recombinants from chickens co-colonized with strains carrying the different antibiotic resistance markers. Double-resistant mutants were obtained already at 2 days after inoculation and were isolated throughout the entire study period, indicating that the double-resistant phenotype maintained its colonization potential. Homologous recombination between heterologous strains (2412 and 2535) in vivo appeared to occur less frequently than during co-colonization of strain 2412 hipO::Cmr + 2412htrA::Kmr. This difference was not found in vitro, which suggests that it is caused by environmental differences rather than by the activity of restriction–modification systems (R–M systems).
The exact nature of the genetic mechanism(s) that drive the alterations in PFGE genotype has yet to be investigated. Our results indicate that the formation of new PFGE types is not caused solely by the exchange of DNA between heterologous strains, as novel PFGE patterns were also observed for mutants derived from strains with a virtually identical genetic background (i.e. 2412htrA::Kmr + 2412hipO::Cmr). It is possible that the formation of novel PFGE types is limited to strains that undergo allelic recombination, as prolonged propagation (300 passages) of single strains on agar plates did not result in novel PFGE patterns. A striking finding was that several of the in vivo- and in vitro-generated double-resistant mutants appeared to have acquired a similar novel PFGE pattern characterized by a gain of ≈ 80 kb to the largest DNA fragment. This may point to a duplication of a distinct region. Measurement of the minimal inhibitory concentration (MIC) of antibiotics in the various mutants did not reveal an association between the 80 kb fragment and the level of resistance (data not shown), suggesting that the possible duplication was not driven by antibiotic pressure. The finding that the number of novel PFGE patterns was higher among the transformants derived from the heterologous combinations than from homol-ogous combinations of strains suggests that intragenomic events are not a major cause of the altered PFGE types. This is in line with recent MLST data from Dingle et al. (2001), which indicate that the contribution of recombinations to the genetic diversity of C. jejuni may be six times more important at the locus level, and 41 times more important at the nucleotide level, than the contribution of point mutations.
Irrespective of the nature of the underlying mechanism(s), the data provide evidence that the genome of a C. jejuni strain is potentially unstable and that allelic exchange as well as other intragenomic events contribute to the population diversity in vivo. This in vivo evidence of the generation of genetic diversity in C. jejuni impacts directly on the phylogenetic structure of this species. Although the significance in terms of the actual rates of recombination is difficult to assess, our results show that simple co-colonization of C. jejuni strains, which occurs frequently in chickens (Jacobs-Reitsma et al., 1995), results in C. jejuni diversity. Our data support the assumption that, as suggested for the closely related gastrointestinal pathogen Helicobacter pylori, recombination in C. jejuni occurs frequently enough to create many different combinations of alleles (Dingle et al., 2001; Feil et al., 2001; Suerbaum et al., 2001).
The results presented have direct implications for the application of genotyping techniques for C. jejuni. FlaA PCR–RFLP, AFLP and MLST were all useful for identifying the corresponding parental strains of the analysed recombinants. However, the impression with these tech-niques that the analysed mutants carried a stable genotype similar to the parental strains was clearly wrong, as PFGE analysis indicated many differences compared with the parental patterns. Although the majority of these differences were small, some were so extensive that the original PFGE pattern could not be deduced. PFGE is widely applied in studying the molecular epidemiology of C. jejuni but, as new PFGE genotypes occur from two strains and even from the same strain, this method is too sensitive for the determination of genetic relatedness of strains. Our data suggest that, because of the genome diversity of C. jejuni, the use of a combination of typing methods, e.g. PFGE combined with MLST (Dingle et al., 2001; Suerbaum et al., 2001), AFLP or flaA PCR–RFLP, is needed for reliable determination of interstrain relationships and the evolutionary history of C. jejuni.
The present in vivo findings on the genome (in)stability of C. jejuni clearly indicate that, under natural conditions, the generation of genetic diversity can be very rapid even in the absence of selective pressure. Furthermore, interstrain genetic exchange as well as intragenomic alterations contribute to the population diversity observed for this pathogen. These observations may explain the genome plasticity of this pathogen and probably conceal existing phylogenetic relationships between strains and lineages.
Bacterial strains and plasmids
Bacterial strains and plasmids used in this study are shown in Table 4. Escherichia coli was grown onto LB agar or in LB broth (Sambrook et al., 1989) at 37°C under aerobic conditions. When necessary, the medium was supplemented with ampicillin (50 μg ml−1), chloramphenicol (12.5 μg ml−1) or kanamycin (50 μg ml−1). Bacterial stocks were stored at –80°C in LB broth containing 15% glycerol. C. jejuni strains 2412 and 2535 were originally isolated from poultry and grown onto heart infusion agar plates supplemented with 5% sheep blood (HIS plates) for 48 h under microaerobic conditions created with an Anoxomat system at 42°C. For reisolation from caecal contents, charcoal cefaperazone desoxychelate agar (CCDA; Oxoid) plates were used. When necessary, media were supplemented with chloramphenicol (12.5 μg ml−1) and/or kanamycin (50 μg ml−1). Campylobacter strains were stored at –80°C in heart infusion broth containing 15% glycerol.
Table 4. Bacterial strains and plasmids used in this study.
Chromosomal DNA for AFLP and flaA typing was isolated from 48 h cultures using a Puregene chromosomal DNA isolation kit (Biozym). For Southern blot analysis of double-resistant mutants, DNA was isolated according to the Puregene chromosomal DNA isolation kit except that, after the protein precipitation step, phenol–chloroform–isoamyl alcohol (25:24:1) extraction was performed on the supernatant. After a final chloroform–isoamyl alcohol (24:1) extraction, the DNA was ethanol precipitated and dissolved in distilled water.
PCR mixtures contained 50 pmol of forward primer, 50 pmol of reverse primer, 50 mM KCl, 10 mM Tris-HCl (pH 9.0), 0.01% (w/v) gelatine, 2 mM MgCl2, 0.2 μM dNTPs, 50 pmol of template and 2.5 units of AmpliTaq DNA polymerase (Perkin-Elmer) with a total reaction volume of 50 μl. Reaction conditions were 60 s at 94°C followed by 30 cycles of 45 s at 94°C, 45 s at 55°C, 60 s (120 s for htrA) at 72°C and ended with 5 min at 72°C.
Construction of suicide replacement plasmids
The hipO gene from strain 2412 was amplified by PCR using forward primer HipO-F (5′-TTCCAGAAATACTAGACTTACA-3′) and reverse primer HipO-R (5′-AAAAATCCAAAATC CTCA-3′) and cloned into the pCRII-TOPO vector (Invitrogen) yielding pHipO. This cloned hipO gene was sub-sequently subcloned into the EcoRI site of pBluescript SK– (pBSK−), resulting in pHipO-1. The cat gene of Campylo-bacter coli was amplified from plasmid pUOA23 using forward primer Cat-F (5′-CACAACGCCGGAAACAAG-3′) and reverse primer Cat-R (5′-CCGCAGGACGCACTACTC-3′) and blunt end ligated into the SphI site of the cloned hipO gene of pHipO-1, resulting in pHipCat.
Plasmid pHtrA2T was constructed from C. jejuni 81116-2T, a mutant strain that contained a kanamycin resistance cassette inserted into the htrA gene. The disrupted htrA gene was PCR amplified using forward primer HtrA-F (5′-AATC GACTGCAACGGCTAATC-3′) and reverse primer HtrA-R (5′-ATAATTCACCCTCTTGGAAACC-3′). The PCR amplicon was cloned into the pCRII-TOPO vector and subsequently subcloned as an EcoRI fragment into the EcoRI site of pBSKII− yielding pHtrA2T. The nature of the cloned PCR products was confirmed by sequence analysis.
Electrotransformation of C. jejuni 2412 and 2535
Electrocompetent cells were prepared according to the method of Wassenaar et al. (1993). After preparation, the cells were immediately used for electrotransformation with 1 μg of DNA (pHipCat or pHtrA2T) added to 50 μl of electrocompetent cells (Wassenaar et al., 1993). After the transformation, the cells were plated on HIS plates without antibiotics for viability counts and onto selective HIS plates containing chloramphenicol or kanamycin to select for mutants expressing the cat or aphA-3 gene respectively. The plates were then incubated for 2 days at 42°C under microaerobic conditions.
Southern blot analysis
For Southern blot analysis, ClaI-digested chromosomal DNAs were transferred to nitrocellulose membranes and hybridized with alkaline phosphatase-labelled (Alk Phos Direct labelling kit; Amersham) pBSK−, cat and aphA-3 probes. Detection was performed using ECF substrate on a Storm 840 (Amersham Pharmacia Biotech).
Exchange of DNA among C. jejuni strains under laboratory conditions
Allelic exchange between C. jejuni strains was measured by suspending the individual mutants carrying the resistance markers into HI broth (OD600 of 0.12) and mixing of the suspensions in four combinations (2412hipO::Cmr + 2412htrA::Kmr, 2535hipO::Cmr + 2535htrA::Kmr, 2412hipO::Cmr + 2535htrA::Kmr, 2412htrA::Kmr + 2535hipO::Cmr) at ratios of 1:1. The mixed suspensions were plated onto HIS plates containing either chloramphenicol or kanamycin and onto HIS plates containing both antibiotics. After 2 days, cfus were counted.
For assessing recombination in biphasic medium, strains were grown separately onto HIS plates, suspended into HI broth (OD600 of 0.12), mixed in the same combinations as for the plate experiment described above and loaded (400 μl) onto HI agar in a polypropylene tube (12 × 75 mm). After incubation (24 h at 37°C under microaerobic conditions), the cultures were plated onto HIS plates containing either chloramphenicol or kanamycin and onto HIS plates containing both antibiotics. After 2 days, the cfus were counted.
Exchange of DNA among C. jejuni strains during experimental infection of chickens
Chicken experiments were performed with Ross 308 broiler chickens housed in isolators in groups of 30 animals each. All feed was irradiated, and water was filter sterilized before ad libitum administration to the animals. Before infection, cloacal swabs taken from all chickens were examined for the presence of Salmonella and Campylobacter by streaking directly on selective plates as well as after enrichment. At day 7 after hatching, the animals were inoculated with mixtures of two C. jejuni mutants (105 cfu of each mutant). The animals in group 1 received 2412hipO::Cmr + 2412htrA::Kmr, group 2 received 2535hipO::Cmr + 2535htrA::Kmr, group 3 received 2412hipO::Cmr + 2535htrA::Kmr and group 4 received 2412htrA::Kmr + 2535hipO::Cmr. At days 2, 7, 10, 14, 21 and 29 after inoculation, five animals from each group were sacrificed, and C. jejuni was reisolated from caecal contents. The caecal contents were serially diluted and plated onto CCDA plates supplemented with chloramphenicol or kanamycin and onto CCDA plates containing both antibiotics to estimate colonization and the formation of double-resistant mutants respectively.
Analysis of double-resistant strains
Initially, two double-resistant mutants were selected randomly from each in vivo group, and together with the four initial single-resistant mutants analysed by Southern blot analysis. ClaI-digested chromosomal DNA was Southern blotted onto nitrocellulose membranes and hybridized with alkaline phosphatase-labelled cat, aphA-3, hipO and htrA probes (Alk Phos Direct labelling kit; Amersham). At a later stage, a total of 47 double-resistant mutants from the in vivo experiments (four mutants per chicken when available) were typed by flaA PCR–RFLP, AFLP analysis and PFGE typing, as described previously (de Boer et al., 2000). Furthermore, a total of 48 double-resistant mutants from the in vitro plate experiments were also analysed by PFGE typing. Finally, a total of 10 double-resistant mutants from the in vivo experiments were analysed by MLST at the National Institute for Public Health and the Environment (RIVM), Bilthoven, The Netherlands, as described previously (Dingle et al., 2001) (Table 3).
Stability of PFGE patterns in vitro
In order to assess the stability of strains in vitro, 2412, 2535 and their Cmr mutants were subcultured 300 times on HIS plates, and the PFGE profiles were determined.
We thank the Experimental Animal and Laboratory Services for assistance with the animal experiments, and Dr Jan van Embden, Dr Rob Willems and Sanne van Reulen (National Institute for Public Health and the Environment, Bilthoven, The Netherlands) for MLST typing. Dr Julian Ketley is gratefully acknowledged for advice on the use of htrA and providing pHtrA-2T, Dr Diane Taylor for providing the plasmids encoding the antibiotic resistance genes, and Dr Trudy Wassenaar for fruitful discussions. This work was partly funded by the Product Boards for Livestock, Meat and Eggs, Rijswijk, The Netherlands.