Present address: Sigrid Rita Andersen, University College Øresund, Faculty of Laboratory Sciences, Copenhagen, Denmark
Editor: Julian Marchesi
Correspondence: Andrea Wilcks, Department of Microbiology and Risk Assessment, Danish Institute for Food and Veterinary Research, Mørkhøj Bygade 19, DK-2860 Søborg, Denmark. Tel.: +45 7234 7185; fax: +45 7234 7698; e-mail: email@example.com
Two wild-type strains of Lactobacillus plantarum previously isolated from fermented dry sausages were analysed for their ability to transfer antibiotic resistance plasmids in the gastrointestinal tract. For this purpose, we used gnotobiotic rats as an in vivo model. Rats were initially inoculated with the recipient Enterococcus faecalis JH2-2 at a concentration of 1010 CFU mL−1. After a week, either of the two donors L. plantarum DG 522 (harbouring a tet(M)-containing plasmid of c. 40 kb) or L. plantarum DG 507 [harbouring a tet(M)-containing plasmid of c. 10 kb and an erm(B)-containing plasmid of c. 8.5 kb] was introduced at concentrations in the range of 108–1010 CFU mL−1. Two days after donor introduction, the first transconjugants (TCs) were detected in faecal samples. The detected numbers of tet(M)-TCs were comparable for the two donors. In both cases, this number increased to c. 5 × 102 CFU g−1 faeces towards the end of the experiment. For erm(B)-TCs, the number was significantly higher and increased to c. 103 CFU g−1 faeces. To our knowledge, this is the first study showing in vivo transfer of wild-type antibiotic resistance plasmids from L. plantarum to E. faecalis.
The major financial and societal costs caused by the emergence and evolution of antibiotic resistance in pathogenic bacteria represent a well-known problem. The attenuation of this problem is complicated by commensal bacteria that may act as reservoirs for antibiotic resistance determinants found in pathogens (Salyers & Shoemaker, 1996; Levy & Marshall, 2004). This statement is supported by the fact that the same type of genes encoding resistance to, for example, tetracycline, erythromycin, chloramphenicol, streptomycin and streptogramin have been found in commensal lactococci and lactobacilli as well as in potentially pathogenic enterococci and pathogenic streptococci (Teuber et al., 1999). A remarkable similarity in resistance genes has also been observed for tetracycline-resistant Lactobacillus plantarum, Lactobacillus sakei ssp. carnosus, L. sakei ssp. sakei, Lactobacillus curvatus and Lactobacillus alimentarius, all of which have been previously isolated from Belgian fermented dry sausages (Gevers et al., 2000, 2003a). Partial sequencing of the tetracycline resistance genes detected in all these species revealed two sequence homology groups (SHGs) with ≥99.6% identity to tet(M) genes of Enterococcus faecalis and Neisseria meningitidis (SHG I) and of Staphylococcus aureus (SHG II) (Gevers et al., 2003a; Huys et al., 2004).
Lactobacilli are generally recognized as safe, and are industrially important food-grade organisms used as probiotics and starter cultures in fermented foods. They are present in ready-to-eat foods and are also indigenous members of the human intestinal microbiota. Data on antibiotic resistance in lactobacilli are relatively limited. Only recently, a number of studies have recorded antibiotic susceptibility profiles for various Lactobacillus species in order to facilitate differentiation of intrinsic resistance from acquired (and thus potentially transferable) resistance (Charteris et al., 1998; Danielsen & Wind, 2003; Cataloluk & Gogebakan, 2004; Delgado et al., 2005; Florez et al., 2005). Although reports on the presence of antibiotic resistance genes associated with mobile genetic elements are scarce among lactobacilli, the safety implications connected to their presence in organisms for human consumption should be considered (Saarela et al., 2000). Lactobacilli that harbour antibiotic resistance determinants have been found in a broad selection of food products such as fermented drinks and yoghurts (Temmerman et al., 2003), cheese (Herrero et al., 1996; Florez et al., 2005), and meat products (Gevers et al., 2000). Owing to their wide environmental distribution, it is possible that these commensal bacteria act as vectors for the dissemination of antibiotic resistance determinants via the food chain to the consumer, a risk that has so far been poorly addressed.
The aim of this study was to examine the ability of two wild-type L. plantarum strains from food origin horizontally to transfer tetracycline and erythromycin resistance genes in vivo to the well-documented recipient strain Enterococcus faecalis JH2-2. Conjugative dissemination of resistance genes between these strains has been shown previously using an in vitro filter mating approach (Gevers et al., 2003b). In order to better represent the natural situation, gnotobiotic rats were used to investigate the potential of the Lactobacillus strains to function as vehicles of transferable antibiotic resistance in the gastrointestinal tract.
Materials and methods
Bacterial strains and growth conditions
The two bacterial donors used in this study were L. plantarum DG 522 (LMG 21687) and L. plantarum DG 507 (LMG 21684) isolated from fermented dry sausages (Gevers et al., 2003a). Strain L. plantarum DG 522 contains a tet(M) tetracycline resistance gene located on a c. 40-kb plasmid (ptet(M)DG 522). Strain L. plantarum DG 507 contains a tet(M) gene located on a c. 10-kb plasmid (ptet(M)DG 507) and an erm(B) erythromycin resistance gene located on a c. 8.5-kb plasmid (perm(B)DG 507) (Gevers et al., 2003b). The donor strains were grown at 30°C in De Man, Rogosa and Sharpe (MRS) medium (Oxoid, Hampshire, UK) containing only tetracycline (L. plantarum DG 522) or both tetracycline and erythromycin (L. plantarum DG 507) for 24–48 h. The same growth conditions for donor strains were used both for monocultures and for isolation from faecal samples.
Strain Enterococcus faecalis JH2-2, showing resistance to rifampicin and fusidic acid (Rifr, Fusr) (Jacob & Hobbs, 1974) (LMG 19456), was used as plasmid-free recipient, and was grown in brain heart infusion (BHI) medium (Oxoid) containing rifampicin and fusidic acid for 24 h. The recipient was incubated at 37°C when grown from monoculture and at 42°C when isolated from faecal samples.
Transconjugants (TCs) were selected from faecal samples on both BHI and Slanetz & Bartley agar (Oxoid), resulting in similar counts for the two media. For selection of TCs E. faecalis JH2-2,ptet(M)DG 522 and E. faecalis JH2-2,ptet(M)DG 507 plates were supplemented with rifampicin, fusidic acid and tetracycline, whereas for selecting TCs E. faecalis JH2-2,perm(B)DG 507 plates were supplemented with rifampicin, fusidic acid and erythromycin. Plates were incubated at 42°C for 48 h. In succeeding analyses, TCs were subcultured in BHI medium with the appropriate antibiotics at 37°C.
Antibiotics (Sigma, Bornem, Belgium) were used at the following concentrations: rifampicin, 50 μg mL−1; fusidic acid, 25 μg mL−1; tetracycline, 10 μg mL−1; erythromycin, 5 μg mL−1; streptomycin, 500 μg mL−1; and spectinomycin, 500 μg mL−1.
Animal management and experimental design
Six male and six female germ-free Sprague–Dawley rats, originally supplied by Taconic (Germantown, NY) were bred at the Danish Institute for Food and Veterinary Research. The rats had an age of c. 3 months at the beginning of the experiment and were housed and fed as previously described (Wilcks et al., 2004). The germ-free status of the rats was verified prior to bacterial dosing by analysing faeces for aerobic and anaerobic growth of bacteria and yeasts. The rats were divided into three groups: (A) five rats receiving L. plantarum DG 522 and E. faecalis JH2-2, (B) five rats receiving L. plantarum DG 507 and E. faecalis JH2-2, and (C) two rats receiving only E. faecalis JH2-2 as a control.
At day 0, all rats received 1 mL of 1010 CFU mL−1 of the recipient strain E. faecalis JH2-2. The recipient strain was allowed to establish in the rats for 1 week, after which the donor strains were introduced. Each day from day 7 to 10 and 13 to 16 all rats in groups A and B received 1 mL of 108–1010 CFU mL−1L. plantarum DG 522 or L. plantarum DG 507, respectively. At days 13–16, 1 μg tetracycline or 1 μg tetracycline +0.5 μg erythromycin was added to the dosing cultures of L. plantarum DG 522 and L. plantarum DG 507, respectively. During the same period, one rat in control group C was given 1 mL of 1 μg mL−1 tetracycline whereas the other received 1 mL of 1 μg mL−1 tetracycline +0.5 μg mL−1 erythromycin. All bacterial cultures were grown overnight and washed in phosphate-buffered saline (PBS) (Oxoid) before they were given peros by gavage after the collection of faecal samples.
Collection and processing of faecal samples
Faecal samples were collected directly from the rats each working day by careful squeezing of the abdomen. Intestinal samples from duodenum, ileum, caecum and colon were taken at sacrifice. The samples were homogenized by whirly mixing in PBS. Ten-fold dilution series were prepared in PBS and incubated on the appropriate selective agar-plates for enumeration of donors, recipients and TCs as described above. The detection limit for TCs was determined as 2 × 101 CFU g−1 faeces.
Verification of TCs by PCR
TCs isolated from the rats were selected on the basis of their phenotypic resistance profile, i.e. Rifr, Fusr and Tetr or Rifr, Fusr and Ermr, by testing their capability of growth on BHI agar supplemented with these antibiotics. In order to verify that these isolates were true TCs and not mutants, PCR assays with primers specific for E. faecalis or targetting tet(M) and erm(B) resistance genes, respectively, were performed. At least five isolates of each TC type (E. faecalis JH2-2,ptet(M)DG 522, E. faecalis JH2-2,ptet(M)DG 507 and E. faecalis JH2-2,perm(B)DG 507) were selected from faecal samples of each of the five rats. The isolates were selected from different days during the experiment representative for the period from the first appearance of the TCs until euthanasia. Donor and recipient strains were included as positive and negative controls, respectively.
Bacterial DNA template was prepared by boiling of one colony in 200 μL TAE (20 mM Tris, 10 mM acetate, 0.5 mM EDTA, pH 7.4) for 10 min. After cooling, excess cell material was settled by centrifugation at 5000 g for 2 min and the supernatant was transferred to new vials and stored at −20°C until use. PCR reaction mixtures contained 5 μL DNA template, one PuReTaq Ready-To-Go PCR bead (Amersham Biosciences, Buckinghamshire, UK) and 10 pmol of each primer in a total volume of 25 μL. The primers, which were used to check for the presence of tet(M) and erm(B) genes, were tetM-F and tetM-R (Warsa et al., 1996; Wilcks et al., 2004) (406 bp) and ermB-F and ermB-R (Jensen et al., 1999) (424 bp), respectively. Enterococcus faecalis species-specific primers were ddl E. faecalis E1 and ddl E. faecalis E2 (Dutka-Malen et al., 1995) (941 bp).
All PCR amplifications were performed in a PTC-225 Peltier Thermal Cycler (MJ Research, Bio-Rad, Waltham, MA) using the PCR programs as previously described for detection of tet(M) and erm(B) (Gevers et al., 2003b) and for E. faecalis (Dutka-Malen et al., 1995). The PCR products were run on a 1% agarose gel and visualized by ethidium bromide staining.
Determination of minimum inhibitory concentration values
The minimum inhibitory concentrations (MICs) of tetracycline and erythromycin were evaluated by application of an E-test strip (AB Biodisk, Solna, Sweden) on Mueller-Hinton (BBL, Sparks, USA) agar. E-tests were read after overnight incubation at 37°C for recipients and TCs and at 30°C for donors.
Plasmid profiles and Southern hybridization
Plasmid DNA was isolated from TCs, donor and recipient strains using the QIAprep Spin Miniprep Kit (Qiagen, Hilden, Germany) according to the manufacturer's protocol with slight modifications. Exponentially growing cultures were used instead of overnight cultures, and a lysozyme step (2 mg mL−1 lysozyme in P1 buffer for 25 min at 37°C) was included. The plasmid extractions were run on 0.7% agarose gels and the gels were vacuum blotted onto Hybond-N+ membrane filters (Amersham Biosciences, Little Chalfont, UK). Labelling of DNA probes for tet(M) and erm(B) with alkaline phosphatase and chemiluminescent detection with CDP-star were carried out using the AlkPhos Direct labelling kit RPN 3690 (Amersham Biosciences) as described by the manufacturer. Amplification products obtained with primer pairs tetM-F and tetM-R, and ermB-F and ermB-R were used as probes.
In vitro mating
The ability of the TCs to function as new donors of the tet(M) and the erm(B) resistance genes was assessed using a filter mating approach. Five isolates of TCs E. faecalis JH2-2,ptet(M)DG 522, five isolates of E. faecalis JH2-2,ptet(M)DG 507 and five isolates of E. faecalis JH2-2,perm(B)DG 507 obtained from faecal samples and representing different plasmid profiles were evaluated as potential donors. Enterococcus faecalis JH2SS resistant to streptomycin and spectinomycin (Strepr, Specr) (Tomich et al., 1980) isogenic with E. faecalis JH2-2 and E. faecalis OG1SS (strepr, specr) (Franke & Clewell, 1981) were used as recipients. As a positive control of the mating conditions, transfer of pAMβ1 from E. faecalis JH2-2 [which had received pAMβ1 from Lactococcus lactis SH 4174 (Gasson & Davies, 1980)] to E. faecalis JH2SS was applied. The mating procedure was as described earlier (Gevers et al., 2003b). In short, exponentially growing donor and recipient cultures were mixed and poured onto a sterile filter (HAWP04700, Millipore, Bedford, MA), and the filters were incubated on non-selective BHI agar plates at 37°C for 18–20 h. The bacteria were washed off the filters with PBS and appropriate dilutions were spread onto donor-, recipient- and TC-selective agar plates.
Reported values are the means from five-fold repetitions. In order to facilitate log transformation of the data for CFU counts, measurements of zero CFU on a plate were set to 1 CFU g−1 faeces. Comparison of the erm(B)- and tet(M)-TC numbers were performed using the Wilcoxon test.
CFU counts of animal samples
In vivo transfer of wild-type plasmids encoding tetracycline and erythromycin resistance was assessed from the donor strains L. plantarum DG 522 and L. plantarum DG 507 to the recipient E. faecalis JH2-2 using germ-free rats. The recipient given to the animals as a single-dose at day 0 colonized readily to attain a stable and high population size of c. 5 × 109 CFU g−1 faeces throughout the experiment (Figs 1 and 2). The donors were introduced one week after the recipient and were administered daily for two dosing periods each of 4 days (interrupted by a two-day break). The number of donors was lower and less stable than the number of recipients, but remained within the range of 105–107 CFU g−1 faeces (Figs 1 and 2). The fact that the number of donors in the inoculation dose varied might explain the variation in the numbers observed in the animals, because these two numbers seemed to correlate (data not shown). After the first dosing period, the donor strains persisted in the animals at relatively unchanged numbers for 2 days without re-inoculation. Subsequently, the number increased again during the second dosing period, in which sub-therapeutic levels of antibiotics were also added. The sub-therapeutic treatment with antibiotics did not have any detectable effect on the number of recipients, either in the control rats inoculated only with recipients (data not shown) or in the rats inoculated with both recipients and donors (Figs 1 and 2).
The first TCs were detected in faecal samples 2 days after the introduction of either of the donors and increased slightly in number towards the end of the experiment (Figs 1 and 2). During the first 48 h, no bacterial growth was observed on TC-selective agar plates. Throughout the entire transfer experiment, no growth was observed from control rats on plates selective for TCs (data not shown). TCs were detected in all but one rat before the sub-therapeutic treatment with antibiotics was initiated. The development in numbers of TCs selected for resistance to tetracycline E. faecalis JH2-2,ptet(M)DG 522 and E. faecalis JH2-2,ptet(M)DG 507 was comparable for the two mating pairs and reached a level of c. 5 × 102 CFU g−1 faeces at the end of the experiment (Figs 1 and 2). For rats inoculated with donor L. plantarum DG 507, TCs were additionally selected for erythromycin resistance. The number of E. faecalis JH2-2,perm(B)DG 522 TCs was slightly but significantly (P<0.005) higher than that of the tetracycline-resistant TCs, and reached a level of c. 103 CFU g−1 faeces (Fig. 2).
The intestinal distribution of recipients, donors and TCs was relatively similar within the different rats, and also between those inoculated with different donor strains (Table 1). In all four intestinal segments, the recipient concentration was higher than the concentration of the donors, and these were in turn higher than the concentration of TCs (as also observed for the faecal samples). In all rats, bacterial concentrations were lower in the upper (duodenum and ileum) than in the lower (caecum and colon) intestinal segments (Table 1). In the duodenum, TCs were detected in only one of the rats (detection limit 2 × 101 CFU g−1 faeces). In the ileum, TCs were detected in half of the rats, whereas in caecum and colon TCs were observed in all but one rat. The number of Tetr TCs reached an average of c. 2 log CFU g−1 in the two latter segments and slightly higher for the Ermr TCs (Table 1).
Table 1. Intestinal distribution of recipients, donors and TCs from rats inoculated with (A) Enterococcus faecalis JH2-2 and Lactobacillus plantarum DG 522, and (B) E. faecalis JH2-2 and L. plantarum DG 507
The numbers represent average log CFU g−1 segment of five rats. The standard deviations are given in parentheses. Several rats contained TCs at numbers below the limit of detection (LD) of 1.30 log CFU g−1 intestinal segment. In those cases where positive samples could not be obtained by all five rats, n indicates the number of samples used for calculation of the average value and standard deviation.
E. faecalis JH2-2
L. plantarum DG 522
E. faecalis JH2-2,ptet(M)DG 522
E. faecalis JH2-2
L. plantarum DG 507
E. faecalis JH2-2,ptet(M)DG 507
E. faecalis JH2-2,perm(B)DG 507
Verification of TCs by PCR
At least 25 isolates of each of the three types of TCs from the faecal samples were selected for verification. All the verified isolates were confirmed to be true TCs. They were demonstrated to be of the same species as the recipient as witnessed by a positive PCR reaction with primers specific for E. faecalis. In addition, it was confirmed by PCR that all tetracycline-resistant isolates [E. faecalis JH2-2,ptet(M)DG 522 and E. faecalis JH2-2,ptet(M)DG 507] had received the tet(M) gene, and that all erythromycin-resistant isolates [E. faecalis JH2-2,perm(B)DG 507] had received the erm(B) gene (representative data are shown in Fig. 3). The recipient and donor strains were used as controls for the E. faecalis-specific PCR and for the resistance-gene-specific PCR. The controls gave the expected negative or positive results (Fig. 3).
Determination of cross-resistance
In order to determine whether the verified TCs E. faecalis JH2-2,ptet(M)DG 507 and E. faecalis JH2-2,perm(B)DG 507 had received the tet(M) gene, the erm(B) gene or both genes from donor L. plantarum DG 507, parallel PCR reactions with the two gene-specific primer sets were performed. Among 54 examined isolates (27 of each TC type), one of each type had received both genes (data not shown), meaning that c. 4% of all TCs were cross-resistant to both tetracycline and erythromycin. This low ratio was supported by plating of faecal samples on BHI agar (supplemented with rifampicin, fusidic acid, tetracycline and erythromycin) selective for TCs with both resistance genes. These results showed that, in six positive duplicate samples taken from two rats during four individual days, the number of cross-resistant isolates accounted for c. 4% (±2% SD) and 17% (±7% SD) of the erythromycin- and the tetracycline-resistant isolates, respectively.
Determination of MIC values
The MIC of tetracycline and erythromycin was determined for the recipient, the donors and all the verified TCs. The MIC of tetracycline was 64–96 μg mL−1 in all examined tetracycline-resistant TCs. This value was lower than observed for the donors, which were resistant to >256 μgmL−1. The MIC for erythromycin was >256 μg mL−1 in all the erythromycin-resistant TCs and similar to the value measured for the donor L. plantarum DG 507. The recipient E. faecalis JH2-2 was susceptible to both antibiotics and the MIC was <1 μg mL−1.
Plasmid profiles and Southern blotting
Among the verified TCs, ten isolates of each type (two from each of the five rats) were selected for further genotypic characterization of the transferred plasmids. In plasmid profiles of most of the E. faecalis JH2-2,ptet(M)DG 522 isolates, the presence of a plasmid band corresponding to c. 40 kb was visible (Fig. 4A1). This band was not present in the plasmid-free recipient strain E. faecalis JH2-2 (data not shown), but matched the size of a band present in the plasmid profile of donor strain L. plantarum DG 522 (Fig. 4A1). Using Southern hybridization, the tet(M) probe showed a positive signal with this band in both the donor and the TCs (Fig. 4A2). In addition to the c. 40-kb band, hybridization to a second band with lower electrophoretic mobility was observed in the TC isolates and in the donor (Fig. 4A2). This band could represent a dimer of the same plasmid, but this possibility was not investigated further. In five of the E. faecalis JH2-2,ptet(M)DG 522 isolates, two additional smaller bands were observed apart from the larger band corresponding to the resistance plasmid. These two bands appeared together in all cases, and hence two distinct plasmid profiles were observed for the examined isolates. Within most of the rats, both types of profiles were present (Fig. 4A1).
Characterization of the plasmid profiles was also performed for 10 tetracycline- and 10 erythromycin-resistant isolates selected from mating pair DG 507/JH2-2. In the E. faecalis JH2-2,ptet(M)DG 507 isolates, a bright plasmid band of c. 10 kb was evident in all isolates as well in the donor L. plantarum DG 507 (Fig. 4B1). Likewise, in the E. faecalis JH2-2,perm(B)DG 507 TCs a band of c. 8.5 kb was evident in all isolates, which corresponded to a similar size band in the donor (Fig. 4C1). These two bands showed positive hybridization with probes for tet(M) and erm(B), respectively (Fig. 4B2–C2). Again, hybridization to more than one band was observed with both the tet(M) and the erm(B) probe, possibly owing to different forms of the plasmids. The bands that hybridized with the probes were the same in both the donor and all the TCs. As for E. faecalis JH2-2,ptet(M)DG 522 TCs, co-transfer of plasmids other than the resistance plasmids was detected in several of the TC isolates derived from donor L. plantarum DG 507, and different plasmid profiles appeared within the same rat (Fig. 4B1–C1).
In vitro mating
In vitro transferability of the tet(M) and the erm(B) resistance genes from 15 distinct TC isolates was investigated using a filter mating procedure. No transfer was observed to recipients E. faecalis JH2SS and E. faecalis OG1SS (the detection limit was <10−8 TCs/recipient). The positive control (transfer of pAMβ1 from E. faecalis JH2-2 to E. faecalis JH2SS) was performed simultaneously with mating between the TCs and E. faecalis JH2SS. The control showed high transfer rates of 1.7 × 10−3 TCs/recipient.
To our knowledge, this is the first study to demonstrate in vivo transfer of wild-type antibiotic resistance plasmids from L. plantarum to E. faecalis. Only a limited number of studies have investigated antibiotic resistance transfer from Lactobacillus and these studies have concentrated on the introduced broad-host-range conjugative plasmid pAMβ1 encoding resistance to macrolide lincosamide and streptogramin B antibiotics. As such, it has been shown that L. plantarum, L. reuteri, L. fermentum and L. murinus can function as donors of pAMβ1 to other lactic acid bacteria in vitro (West & Warner, 1985; Shrago et al., 1986; Tannock, 1987) and in the gastrointestinal tract of mice (Morelli et al., 1988; McConnell et al., 1991). Likewise, it has been shown that a mobile native plasmid encoding chloramphenicol resistance can be transferred from L. plantarum to Carnobacterium piscicola by means of co-mobilization with pAMβ1 (Ahn et al., 1992).
The implementation of this study was brought about by the finding that wild-type L. plantarum strains isolated from Belgian fermented dry sausages were able to transfer tetracycline and erythromycin resistance genes to E. faecalis JH2-2 in filter mating experiments (Gevers et al., 2003b). However, the gastrointestinal tract is a very hostile environment to allochthonous bacteria, and in vitro transfer of plasmids is therefore not necessarily synonymous with transfer in vivo. A very important barrier that the ingested bacteria meet in passage through the gastrointestinal tract is the low pH and bile as encountered in the stomach and the upper intestine. Earlier studies have shown that strains of L. plantarum can be rather resistant to these conditions, and thus have a relatively good survival rate during transit of the gastrointestinal tract (Johansson et al., 1998; Vesa et al., 2000; Bron et al., 2004; Goossens et al., 2005). In contrast, it has been shown that several food-associated Lactobacillus species, such as L. curvatus and L. Sakei, were detectable in faeces only by molecular methods and not by culturing, perhaps because they were dead or in a non-culturable state (Walter et al., 2001).
In the present study, the two donor strains were introduced to the rats through multiple doses, thus simulating a worst-case scenario of daily intake of food products containing the resistant bacteria. The intestinal establishment of the donors was not investigated in this study, but the colonization potential would probably be different from that in a normal intestine, where the indigenous microbiota would represent a much greater colonization resistance. However, whereas the physiological state of the donor bacteria is critical for conjugation of resistance genes, the ability to colonize is not a prerequisite. Hence, several studies have shown transfer from donors, that were quickly eliminated from the intestine (Gruzza et al., 1993; Gruzza et al., 1994; Schlundt et al., 1994; Walter et al., 2001; Licht et al., 2002).
In this study, TCs were generated exhibiting resistance to either tetracycline or erythromycin or both. The TCs made up c. 10−7 of the recipient population at the end of the experiment, which is a low fraction taking the in vitro transfer rates of 10−4–10−5 TCs/recipient into consideration. However, one should keep in mind that the gastrointestinal tract is a dynamic system and that mating in vivo is a very complex scenario, both in time and in space, compared with filter mating. Consequently, a direct comparison with in vitro transfer ratios is not feasible. In the in vivo situation, three events may have participated in the development of TCs: (1) primary transfer of resistance genes from the Lactobacillus donors to the recipients; (2) secondary transfer from TCs to recipients; and (3) growth of TCs. Clearly, the contribution of each of these events is difficult to assess and was outside the scope of this study. However, it is obvious that primary transfer of resistance genes from the Lactobacillus donors is the prerequisite for establishing a TC population. The ability of the TCs to facilitate secondary transfer of the resistance plasmids was evaluated by in vitro matings, but no such transfer could be detected. However, these results do not rule out the possibility that the TCs can indeed function as donors under different conditions from the one tested here.
It was thought that the sub-therapeutic administration of antibiotics to the rats in this study would promote the spread of antibiotic resistance genes, because this effect had been reported previously (Salyers & Shoemaker, 1996; Licht et al., 2003). However, this effect was not observed. The population size of the sensitive recipient strain E. faecalis JH2-2 did not decrease in response to the antibiotics, but the expected increase in donors and TCs was not evident either. Possibly, the sub-therapeutic concentration of tetracycline was too low. This assumption is supported by the findings of Bahl et al. (2004), who showed that the concentration of bioavailable tetracycline within the bacterial growth habitat of the intestine represented only c. 0.4% of the intake concentration of the antibiotic.
Ultimately, the true risk of antibiotic resistance plasmids circulating in food and in the intestinal environment is the eventual transfer of resistance to human pathogens. The findings of our study show that in vivo resistance gene transfer can take place from wild-type L. plantarum isolated from food to E. faecalis, which represent a natural inhabitant of the human gut and a potentially pathogenic species. However, as the present experiment was performed in conditions with artificially high numbers of recipients and in the absence of competition, the extent of this potential safety hazard should be further investigated. For instance, animal models with a conventional biota or human-biota-associated animals could be used to investigate the magnitude of resistance gene transfer in the presence of a colonization barrier or during competition from a bacterial population imitating that of a human. The number of transfer events in more complex models will most likely be very restricted compared with that in the gnotobiotic gut. On the other hand, the persistence and growth advantage of indigenous recipients that receive a resistance gene under antibiotic selective conditions may be much improved. Hence, the maintenance and potential growth of TC populations during antibiotic treatment is an additional issue that needs to be addressed in future studies.
We would like to thank Anne Ørngren and her department, especially Kenneth Worm, for handling of the animals. Furthermore, we are grateful for the excellent technical assistance given by Bodil Madsen, Kate Vibefeldt and Rikke Kubert. This study was financed by the European Commission grant CT-2003-506214 (ACE-ART) under the 6th framework programme. G.H. and D.G. are indebted to the Fund for Scientific Research – Flanders (Belgium) for a postdoctoral fellowship.