Evidence for natural horizontal transfer of tetO gene between Campylobacter jejuni strains in chickens
Kempf Isabelle, Unité de Mycoplasmologie – Bactériologie, Agence française de Sécurité Sanitaire des Aliments, BP 53, F-22440 Ploufragan, France (e-mail: email@example.com).
Aims: The transfer of tetO gene conferring resistance to tetracycline was studied between Campylobacter jejuni strains, in the digestive tract of chickens.
Methods and Results: In vitro conjugation experiments were first performed in order to select donor/recipient couples for further in vivo assay. Then, chickens were inoculated with a donor/recipient couple of C. jejuni strains displaying spontaneous in vitro tetracycline resistance gene transfer. The donor was a tetracycline-resistant ampicillin-susceptible strain, and the recipient was a tetracycline-susceptible ampicillin-resistant strain. Chicken droppings were streaked on antimicrobial selective media and bi-resistant Campylobacter isolates were further characterized according to their donor or recipient flaA gene RFLP profile. The acquisition of tetracycline-resistance gene by the recipient C. jejuni strain from the donor C. jejuni strain was confirmed by tetO PCR.
Conclusions: The study showed that transfer of tetO gene occurs rapidly and without antimicrobial selection pressure between C. jejuni strains in the digestive tract of chickens.
Significance and Impact of the Study: The rapid and spontaneous transfer of tetO gene may explain the high prevalence of tetracycline resistance in chicken Campylobacter strains.
Campylobacter are commonly encountered bacteria in chicken digestive tract (Jeffrey et al. 2001; Refrégier-Petton et al. 2001; Zhao et al. 2001). They frequently display resistance to antibiotics like tetracycline and quinolones (Saenz et al. 2000; Avrain et al. 2001; Van Looveren et al. 2001). In avian production, tetracyclines are the most widely used antibiotics because of their low cost but their antimicrobial spectrum decreases (Speer et al. 1992; Roberts 1997).
The tetO gene, responsible for tetracycline resistance in Campylobacter, seems plasmid mediated (Taylor et al. 1983; Tenover et al. 1985). It may be present in other bacteria such as Enterococcus, Streptococcus, Mobiluuncus, Staphylococcus or Peptostreptococcus and shares 75% identity with tetM, present in Enterococcus and Streptococcus strains (Zilhao et al. 1988; Chopra et al. 1992; Roberts et al. 1993). TetO and TetM proteins confer ribosomal protection. TetM determinant is associated with conjugative chromosomal elements which code for their own transfer, whereas TetO determinant can generally be found on conjugative plasmids or chromosomal elements, a feature which may explain its wide distribution (Roberts 1996). The purpose of this study was to document on in vivotetO gene transfer between two Campylobacter strains. Therefore, in vitro mating experiments were first realized in order to select a couple of donor/recipient strains for which tetO gene transfer could be demonstrated. This couple of donor and recipient C. jejuni strains was then simultaneously inoculated to 20 chickens. Molecular analysis by restriction fragment length polymorphism (RFLP) of flaA gene and tetO gene amplification from colonies collected from chickens, were performed to evidence the transfer of tetracycline resistance gene from the donor to the recipient C. jejuni strain.
Campylobacter jejuni and C. coli strains, previously isolated from broilers in French slaughterhouses in 1999, were used (Avrain et al. 2001). Among these strains, 10 donor strains, D1–D10, were selected for their resistance to tetracycline and susceptibility to ampicillin (TER/AMS) or enrofloxacin (TER/ENRS) and six recipient strains, R1–R6, for their susceptibility to tetracycline and resistance to ampicillin (TES/AMR) or enrofloxacin (TES/ENRR). The previously determined (Avrain et al. 2001) resistances of the strains for ampicillin, erythromycin, nalidixic acid, enrofloxacin, tetracycline or gentamicin, are indicated in Table 1. Donors D1–D10 and recipient R1–R6 were used in in vitro mating experiments and donor D7 and recipient R5 C. jejuni strains were used for in vivo assay.
Table 1. flaA RFLP profiles of five randomly chosen bi-resistant colonies resulting from in vitroCampylobacter mating experiments
|C. jejuni||L||D1 (TE)||R1 (ENR)||ENR/TE||Five colonies with D1 profile|
|F||D7 (TE)||R5 (AM)||AM/TE||Three colonies with R5 profile, two NT¶|
|F||D8 (TE)||R5 (AM)||AM/TE||Two colonies with R5 profile, 3 NT|
|C. coli||L||D2 (TE, E)||R2 (AM, ENR)||ENR/TE||Two colonies with D2 profile, three with R2 profile|
|F||D4 (TE)||R2 (AM, ENR)||ENR/TE||Four colonies with D4 profile, one with R2 profile|
|F||D5 (TE, AM)||R3 (AM, ENR)||ENR/TE||Four colonies with D5 profile, one with R3 profile|
|F||D6 (TE)||R3 (AM, ENR)||ENR/TE||Five colonies with D6 profile|
|L||D9 (TE, AM)||R6 (EN)||ENR/TE||Four colonies with D9 profile, one NT|
|L||D10 (TE)||R6 (EN)||ENR/TE||Five colonies with D10 profile|
|F||D3 (TE, EN)||R3 (AM, ENR)||AM/TE||Four colonies with D3 profile, one with R3 profile|
|L||D3 (TE, EN)||R3 (AM, ENR)|| ||Five colonies with D3 profile|
In vitro mating experiments
Both liquid and filter-mating methods as described by Taylor et al. (1981) with slight modifications, were tested for 17 donor and recipient strains combinations. Cultures were adjusted at 2 McFarland in Brucella broth. For both methods, dilutions of initial suspensions were made in sodium phosphate buffer 0·05 m for titration. For the liquid-mating method, 0·5 ml of donor strain was added to 1 ml of recipient strain and 1 ml of fresh Brucella broth. Filter-mating method consisted in mixing 0·5 ml of donor strain with 1 ml of recipient strain on nitrocellulose filter (Amersham Biosciences, Orsay, France, 82 mm ø, pore size 0·45 μm) placed on Mueller–Hinton agar supplemented with 5% sheep blood (MHs). Tubes and plates containing mating mixtures were incubated 48 h at 37°C under microaerobic atmosphere. Colonies on nitrocellulose filters were removed with 1 ml of sodium phosphate buffer 0·05 m. Five colonies grown on MHs supplemented with tetracycline 4 mg l−1 and ampicillin 16 mg l−1 (TE/AM) or tetracycline 4 mg l−1 and enrofloxacin 1 mg l−1 (TE/ENR) after 48 h at 37°C under microaerobic atmosphere, were randomly selected and streaked on MHs, incubated 48 h at 37°C and frozen at −80°C in 25% glycerol Brucella broth for further characterization.
Experimental facility and inoculation
Seven-week-old specific pathogen-free (SPF), Campylobacter-free chickens, obtained from the experimental poultry unit of AFSSA, Ploufragan, France, were used for tetracycline-resistance gene transfer assay. They were placed in four isolators (A, five birds; B, five birds; C and D, 10 birds each). A feed free of antibiotics, coccidiostats and other chemical or biological growth promoters was given. Water, heated to 80°C before cooling at room temperature, was provided in one bell drinker per isolator. The five birds from isolator A were orally inoculated with a culture of the D7 donor strain (5·9 × 109 CFU per bird). In isolator B, the five chickens were inoculated with the R5 recipient strain (2 × 106 CFU per bird). The 20 chickens in isolators C and D were similarly inoculated with both donor and recipient strains. Four days postinoculation (PI) and then twice a week during 3 weeks, cloacal samples were collected from each bird.
Faecal materials were weighted and diluted 10−1 to 10−8 in Preston broth (Oxoid, Dardilly, France). Ten microlitres of each dilution were streaked on Butzler no. 2 selective media supplemented with tetracycline 4 mg l−1 or ampicillin 16 mg l−1 or both. Plates were incubated for 48 h at 42°C under microaerobic atmosphere. Samples in Preston broth, dilutions 1 : 10, were incubated 18 h at 42°C under microaerobic atmosphere and then streaked on Butzler no. 2 selective media supplemented with antibiotics as described before. Numbers of chickens from isolators C and D contaminated by donor, recipient and bi-resistant Campylobacter strains were calculated according to presence or absence of colonies on each antimicrobial supplemented media, the detection threshold being 9·102 CFU ml−1. From each sample, a maximum of five colonies grown on plates supplemented with both antibiotics were streaked on MHs, incubated 48 h at 37°C under microaerobic atmosphere for later molecular analysis.
In order to detect, in chicken's digestive flora, other possibly bearing tetO bacteria, droppings samples were obtained, 14 days PI, from birds of isolators A and B. They were diluted 1 : 10 in Mueller–Hinton broth and 10 μl were streaked on m-Enterococcus agar (Becton Dickinson SA, Pont de Claix, France) supplemented with tetracycline 16 mg l−1. Plates were incubated 48 h at 42°C. A maximum of five colonies were then streaked on m-Enterococcus agar, incubated 48 h at 37°C and processed for tetO gene PCR.
Strain characterization by PCR-RFLP
DNA lysates from D1 to D10 and R1 to R6 strains, from five randomly chosen bi-resistant colonies obtained after each successful in vitro mating assay, and from Campylobacter colonies isolated from chicken faecal materials were obtained by heating 10 colonies, at 95°C for 10 min in 200 μl of TE 10·1 buffer (Tris–HCl, 10 mm and EDTA, 1 mm) and, after centrifugation at 2000 g for 2 min, supernatant was diluted 1 : 10. A PCR was used to amplify a 1·45 kbp DNA fragment representing the Campylobacter flagellin gene (flaA) (Nachamkin et al. 1993). A quantity of 22 μl of the PCR product were digested with three units of the restriction enzyme DdeI (Invitrogen, Cergy Pontoise, France) in 2·5 μl of 10X REact® buffer (Fitzgerald et al. 2001). After 4 h at 37°C, digested products were analysed by agarose gel electrophoresis using 3% (w/v) agarose low melting/gelling temperature (Amersham Biosciences) in 1X Tris–Borate–EDTA (TBE) buffer. The DNA fragments were stained with ethidium bromide and visualized under u.v.
Construction of an internal positive control to check for the presence of tetO PCR inhibitors
The internal positive control (IPC) was synthesized by PCR according to Sachadyn and Kur (1998) method, with the forward M6-MD16S1 5′-GTTTATCACGGAAGYGCWAATCTAATGGCTTAACCATTAAAC-3′ and the reverse M4-MD16S2 5′-GGAGCCCAGAAAGGATTYGGGGACGGTAACTAGTTTAGTATT-3′ primers (Genome express, Meylan, France) possessing, respectively, 5′ over-handing ends identical to the primers M6 5′-GTTTATCACGGAAGYGCWA-3′ and M4 5′-GGAGCCCAGAAAGGATTYGG-3′ used in the tetO PCR (Roberts et al. 1993), and 3′ ends complementary to a predetermined 16S rRNA sequence of C. jejuni and C. coli. Thus, the IPC was obtained by amplification of a 16S rRNA fragment of 896 bp according to Denis et al. 1999, from a tetracycline susceptible C. jejuni strain distinct from any of the recipients used in this study. The IPC amplicon was purified by PCR purification kit (Qiagen S.A., Courtaboeuf, France) and adjusted to 1 pg μl−1.
Detection of tetO gene by PCR with IPC
The presence of tetO gene was analysed in all donor and recipient strains, as well as in bi-resistant colonies obtained after in vitro mating experiments and colonies isolated from faecal materials. Amplification of a fragment of tetO gene was performed using the method described by Roberts et al. (1993) with slight modifications. Bacterial cultures were suspended in 200 μl of TE 10·1 buffer. DNA was obtained by heating the cells at 95°C for 10 min. After centrifugation at 2000 g for 2 min, the supernatants were diluted 1 : 10 in TE 10·1 buffer. The PCR reaction mix (total 50 μl) contained 100 ng of each primer (M4 and M6), 5 μm of dNTPs solution (Eurobio, Les Ulis, France), 1X PCR buffer with MgCl2 (Roche Diagnostics, Meylan, France), 1 U of Taq polymerase (Roche Diagnostics), 1 pg of IPC and 5 μl of DNA lysate. Amplification reactions were performed in a Perkin Elmer 9600 (Applied Biosystems, Courtaboeuf, France) thermocycler: 5 min at 95°C, 35 cycles each consisting of 30 s at 94°C, 1 min at 60°C, 1 min at 72°C, a final extension 14 min at 72°C before a cooling down to 4°C. DNA fragments were analysed by agarose gel electrophoresis using 1·5% (w/v) agarose (Eurogentec, Angers, France) in 1X TBE buffer, stained with ethidium bromide and visualized under u.v. The detection limit of this tetO PCR performed with IPC was tested with the tetracycline-resistant D7 C. jejuni strain and evaluated at 55 CFU per assay.
Antimicrobial susceptibility testing
Antimicrobial susceptibility was tested according to Aarestrup et al. (1997). Plates containing MHs were supplemented with tetracycline (0·125–128 mg l−1) or ampicillin (0·25–64 mg l−1). One microlitre of each bacterial suspension containing about 104 CFU per spot was inoculated on each supplemented plate. Cultures were incubated 48 h at 37°C under microaerobic atmosphere. MIC of (TER/AMR) colonies were compared with previously determined MICs of donor and recipient strains. The results were validated by reference strains (Escherichia coli (ATCC 25922), Staphylococcus aureus (ATCC 29213), Pseudomonas aeruginosa (ATCC 27853) and Enterococcus faecalis (ATCC 29212) according to the National Committee of Clinical Laboratory Standards document M7-A4. Susceptibility categorization was carried out according to the statement 1999 of the Antibiogram Committee of the French Society for Microbiology (1999).
In vitro mating experiments
The RFLP analysis of flaA gene could be performed for all donor and recipient strains except for R6 and D8 strains for which enough flaA gene fragment PCR product could not be obtained. Generated patterns contained five to eight fragments with size varying from 50 to 780 bp (data not shown). flaA RFLP with DdeI restriction enzyme enabled differentiation of donor from recipient strains according to one, two or three different fragment sizes. Among the 14 strains for which enough flaA gene product could be obtained, 10 had a singular pattern and two couples of strains (R4 and R5, and D3 and D4) exhibited the same RFLP pattern.
Among the 17 mating experiments tested, bi-resistant colonies could be obtained on media supplemented with tetracycline and enrofloxacin or tetracycline and ampicillin from three C. jejuni couples and seven C. coli couples (Table 1). For each of these mating assays, five randomly chosen bi-resistant colonies were analysed by flaA RFLP and tetO PCR. Patterns of bi-resistant colonies were compared with those of donor and recipient strains. Colonies resulting from C. jejuni or C. coli mating experiments on enrofloxacin and tetracycline media had most often the donor RFLP profile, indicating spontaneous mutations of donor strains (initial MIC of 0·25 mg l−1) becoming resistant to enrofloxacin (MIC > 2 mg l−1). When the selective antimicrobials used for C. jejuni mating experiments were ampicillin and tetracycline, bi-resistant colonies displayed the recipient pattern or could not be typed (because the quantity of flaA gene fragment PCR was too low) whereas most C. coli bi-resistant colonies had the donor profile, indicating probable spontaneous mutation leading to ampicillin resistance (initial MIC of 2 mg l−1, final MIC > 16 mg l−1).
The six recipient strains did not contain tetO gene whereas it was detected in the 10 donor strains and in all colonies selected on (TE/AM) or (TE/ENR) plates after mating experiments. Thus the tetO gene was transferred in vitro between C. jejuni strains (from D7 to R5 strain and from D8 to R5) and between C. coli strains (from D4 to R2, D3 to R3, D5 to R3 and D2 to R2 strains). The five former transfers were obtained by the filter-mating method and the latter by the broth-mating method.
Chicken colonization by donor D7 and recipient R5 strains
All chickens inoculated with D7 strain (isolator A) were colonized up to 21 days PI. The numbers of colonies collected on tetracycline-supplemented Butzler no. 2 medium fluctuated, depending on birds (2·1 × 105–2·4 × 109 CFU per gram of droppings) and time PI. On day 4 PI, the mean was 1·3 × 109Campylobacter CFU per gram whereas on day 21 PI, the rate was reduced to 6·4 × 106 CFU g−1. For birds inoculated with R5 recipient strain only (isolator B), on day 4 PI, the mean was 7·2 × 108 CFU g−1 whereas on day 21 PI, only one chicken still yielded positive culture. All analysed colonies from isolators A or B exhibited the D7 or R5 profile respectively. No bi-resistant colony could be isolated from samples collected from D7 or R5 inoculated birds.
In the D7/R5 inoculated group (isolators C and D), on day 4 PI, the mean number of tetracycline resistant colonies was 8·4 × 106 CFU g−1, significantly lower than results obtained from birds inoculated with D7 strain alone (isolator A).
In vivo tetracycline-resistance gene transfer
The numbers of chickens from isolators C and D contaminated by donor, recipient or bi-resistant Campylobacter are shown in Table 2. Among the 20 chickens inoculated with both strains, TER/AMR colonies could be obtained after direct isolation from 11 chickens; colony numbers varied from 9 × 103 to 2·7 × 106 CFU g−1 depending on chickens. For six birds (two in isolator C and four in isolator D), bi-resistant colonies were isolated as soon as day 4 PI (first sampling day PI); three other chickens yielded bi-resistant colonies on day 7 PI whereas bi-resistant colonies were not isolated from two chickens until day 21 PI.
Table 2. Numbers of chickens from isolators C and D from which donor, recipient and bi-resistant colonies could be isolated
|4 days PI||19/20||20/20||6/20|
|7 days PI||20/20||20/20||8/20|
|19 days PI||20/20||3/20||1/20|
|21 days PI||19/20||3/20||3/20|
The PCR-RFLP flaA method revealed that all five colonies randomly picked from tetracycline- and ampicillin-supplemented media from each faecal sample, exhibited the R5 recipient strain profile and the presence of tetO gene was confirmed by PCR for all these bi-resistant colonies.
Detection of tetO gene in SPF chicken digestive flora
The tetO PCR was performed on chicken faecal samples in order to detect the possible presence of tetO-positive bacteria (other than Campylobacter spp.) in the digestive tract of SPF chickens. The results showed that faecal material contained PCR inhibitors that hampered direct detection of tetO. However, tetracycline-resistant Gram-positive colonies were isolated on m-Enterococcus agar and shown to contain tetO gene according to PCR results.
Antimicrobial susceptibility testing
From the fourth day PI and on all sampling occasion thereafter, (TER/AMR) Campylobacterjejuni colonies collected from chickens exhibited ampicillin MICs of 32 mg l−1 and tetracycline MIC of 128 mgl−1 whereas D7 strain was resistant to tetracycline (MIC = 128 mg l−1) and susceptible to ampicillin (MIC = 4 mg l−1), and R5 strain was susceptible to tetracycline (MIC = 0·25 mg l−1) and resistant to ampicillin (MIC = 32 mgl −1).
Spontaneous tetO gene transfer in C. jejuni and C. coli strains was first demonstrated in vitro, and then in the digestive tract of chickens inoculated with a pair of donor and recipient C. jejuni strains. Enrofloxacin- and ampicillin-selective antimicrobials were chosen for in vitro experiments after preliminary unsuccessful mating experiments with erythromycin or nalidixic acid associated with tetracycline. However, when enrofloxacin was the selective antimicrobial, many Campylobacter colonies isolated on MHs supplemented with tetracycline and enrofloxacin exhibited a donor RFLP profile, indicating mutation of donor strain rather than acquisition of foreign DNA. Quinolones are usually the selective antibiotics in tetracycline gene transfer experiments (Taylor et al. 1981, 1983). However, Taylor et al. (1985) demonstrated that nalidixic acid resistant mutants of C. jejuni could be selected in vitro by spontaneous mutations. High-level resistance to quinolones in Campylobacterjejuni is often associated with amino acid substitutions due to mutations in gyrA (Zirnstein et al. 1999). Such frequent mutations were not observed with ampicillin as the majority of C. jejuni colonies selected on MHs with tetracycline and ampicillin had a recipient RFLP pattern. Ampicillin-resistance mechanism in Campylobacter may be due to β-lactam enzymes genes situated on the bacterial chromosome (Taylor and Courvalin 1988). Only C. coli D3 strain (ampicillin-susceptible strain, MIC = 2 mg l−1) seemed able to become resistant to ampicillin by mutation; nine of 10 colonies isolated from D3/R3 mating experiment presented D3 pattern and only one colony had R3 profile.
Natural tetO gene transfer was observed with both filter and liquid methods, but for each tested donor and recipient couple, transconjugants selected on (TE/AM) or (TE/ENR) plates were isolated with one or the other method but never with both, as previously observed by Taylor et al. 1981.
In our experimental conditions, tetO gene was evidenced in Campylobacter bi-resistant colonies obtained from R5, R2 and R3 recipient strains. However, only combinations between, D7 and R5, or D8 and R5 gave only colonies with recipient profile, the other mating experiments yielding colonies with either donor or recipient profiles. Because not enough flaA gene fragment PCR product could be obtained for D8 strain, the couple D7/R5 was chosen for the in vivo assay.
In in vivo experiment, donor and recipient strains colonized the digestive tract of chickens of isolators A and B respectively as shown by cultures obtained from these birds. R5 strain recovery from birds in isolator B decreased over time, whereas all chickens from isolator A remained heavily colonized up to 21 days PI. This difference may be due to different inoculum titres, to a better fitness of D7 strain or to a strain-specific barrier effect of the intestinal flora.
Bi-resistant colonies isolated from birds of isolators C and D, presented the R5 recipient flaA RFLP profile and harboured the tetO gene according to PCR results. Because natural transformation may occur in Campylobacter (Taylor 1992), the presence of tetO gene was investigated on plasmid and chromosomal preparations; results showed that tetO was borne on a plasmid in D7 strain and on a similar plasmid in bi-resistant colonies isolated from chickens of isolators C and D. This tetO-bearing plasmid was not detected in R5 strain (data not shown). And although tetO gene was present in bacteria of the intestinal flora of SPF chickens, no bi-resistant colony could be isolated from birds of isolator B, indicating that R5 did not acquire the tetO gene (or other tetracycline resistance gene) by transformation or conjugation from these bacteria in our experimental conditions. These data strongly suggest that the bi-resistant colonies resulted from conjugation between D7 and R5 strains. Moreover this conjugation event occurred in the digestive tract of at least two chickens as bi-resistant Campylobacter could be observed in isolators C and D. After these initial transfer events, bi-resistant Campylobacter could originate either from other transfer events in individual birds or from transmission of bi-resistant clones between birds within each isolator. It is noteworthy that transconjugants were isolated as soon as 4 days PI and persisted until the end of the assay. Such spontaneous and rapid gene transfer associated with persistence of bi-resistant population at a high level may partly explain the high percentage of tetracycline-resistant Campylobacter strains in avian productions (Avrain et al. 2001).
In vivo gene transfer experimentations between enterobacteria and E. coli, or Enterococcus and E. coli were previously documented. Thus, R-plasmid transfer was demonstrated by Duval-Iflah et al. (1980) from Serratia liquefaciens to E. coli in gnotobiotic mice associated with human faecal flora and Guillot and Boucaud (1988) used chickens as animal model to evidence resistance gene transfer between E. coli strains. They report that transconjugants rapidly appeared after inoculation and remained in chicken intestinal flora. Doucet-Populaire et al. (1992) showed conjugal transfer of plasmid pAT191, which confers resistance to kanamycin, from Ent. faecalis to E. coli. In our study, although tetO gene was naturally present in the digestive tract of SPF birds, in the absence of tetracycline selective pressure, the R5 recipient Campylobacter strain did not acquire tetracycline-resistance gene. However, transfer frequency of mobilisable elements, like tetracycline-inducible transposon Tn916 first described in Ent. faecalis strain DS16, is increased in the presence of tetracycline (Duval-Iflah et al. 1980; Celli and Trieu-Cuot 1998) and Speer et al. (1992) reported that tetM transfer from Ent. faecalis Tn 925 conjugative transposons was enhanced 10-fold after pre-exposure to tetracycline. Wu et al. (1999) reported that in Ent. faecalis a subinhibitory concentration of tetracycline gave rise to the secretion of the peptide sex pheromone cAD1 in recipient strains that favoured an aggregation response and consequently plasmid transfer. To our knowledge, such protein is not described in the C. jejuni sequenced genome.
To summarize, although in vitro gene transfer between Campylobacter strains was reported a long time ago, these observations are the first demonstration of spontaneous in vivo conjugation between C. jejuni strains.
The authors thank F. Doucet-Populaire (Centre Hospitalier de Versailles, France) for judicious comments on protocol, C. Marois (AFSSA Ploufragan, France) for useful advice and Y. Morin (AFSSA Ploufragan, France) for skilled technical assistance.