INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Antibiotic resistance research has been and still is confined primarily to the study of cultivable bacterial isolates of mostly clinical origin. However, the cultivable isolates may represent only a fraction of the actual microbiota (1) where the antibiotic resistance genes reside. For example, no bacteria can be grown from more than 80% of all clinical samples sent to clinical microbiology laboratories (4), and certainly no antibiotic resistance profile can be determined if cultivation fails. Another important issue with antibiotic resistance is the fact that the wide use of antibiotics not only selects for drug-resistant pathogenic bacteria but also exerts selective pressure on the normal commensal microbiota. In light of the ubiquitously demonstrated phenomenon of horizontal antibiotic resistance gene transfer in the microbial world, the presence of such reservoirs may explain the rapid dissemination of antibiotic resistance from commensal organisms to the pathogenic microbiota. However, information regarding the antibiotic resistance pool in commensal microbiotas is very scarce, and the data are mostly phenotypical. Therefore, development of genotyping tools for detection and tracking of antibiotic resistance genes in a variety of commensal and pathogenic bacteria, as well as in the environment, is essential for understanding the ecology of antibiotic resistance.

One of the attractive models for studying the ecology of antibiotic resistance could be the genes conferring resistance to tetracyclines. Tetracyclines belong to a family of broad-spectrum antibiotics that includes tetracycline, chlortetracycline, doxycycline, and minocycline. These antibiotics inhibit protein synthesis in gram-positive and gram-negative bacteria by preventing the binding of aminoacyl-tRNA molecules to the 30S ribosomal subunit (36). Bacterial resistance to tetracycline is mediated mainly by two mechanisms, protection of ribosomes by large cytoplasmic proteins (5, 6, 23, 33, 43) and energy-dependent efflux of tetracycline (18, 33, 36). A third mechanism, enzymatic inactivation of tetracycline, is relatively uncommon and has been described in only one species (41). The first nomenclature for tetracycline resistance determinants was proposed in 1989 (17), and a recent update appeared in 1999 (19). The ribosomal protection mechanisms identified so far fall into six classes: Tet M, Tet O, TetB P, Tet Q, Tet S, and otrA (43). Almost all representatives of these classes have been sequenced and have been shown to encode proteins with N-terminal amino acid sequence similarity to translation elongation factors EF-Tu and EF-G (6, 22, 35, 43).

Since their introduction in the 1950s, tetracyclines have been widely used in human and veterinary medicine, as growth promoters in animal industry, and for prophylaxis in plant agriculture and aquaculture. At present, resistance to tetracyclines has spread to almost all bacterial genera, and this situation perhaps is the consequence of previous overuse. Among the ribosomal protection determinants, Tet M was described originally in streptococci (5, 24) and subsequently in a broad variety of gram-positive and gram-negative bacteria (32). The Tet S determinant was encountered first on a plasmid in the food pathogen Listeria monocytogenes (7), later in a number of Enterococcus faecalis strains (8), and recently on a plasmid of Lactococcus lactis isolated from raw milk (31). Tet O-related sequences were determined first in plasmids from campylobacteria (40, 42), then in streptococci (16, 45), and recently in a rumen bacterium, Butyrivibrio fibrisolvens (2). Finding nearly identical tet(Q) sequences in Prevotella ruminicola (a typical inhabitant of the rumen) and in Bacteroides (a typical inhabitant of the human gastrointestinal tract) (29) suggested that bacteria normally found in the guts of different species can exchange DNA, presumably during transient colonization of the animal intestine by human-associated bacteria or vice versa. This scenario is supported by recent work which described the occurrence of the tet(W) gene in the rumen and in human and pig intestinal microbiotas (37). Sequence analysis of this gene from taxonomically divergent rumen and human bacterial isolates showed that there was no or just one nucleotide substitution in a 1.25-kb amplified internal fragment. Misincorporation errors during amplification could not be ruled out, suggesting that the sequences may actually be identical in ruminal B. fibrisolvens, Selenomonas ruminantium, and Mitsuokella multiacidus isolates and in human isolates of Fusobacterium prausnitzii and Bifidobacterium longum (37). These findings demand that there be further research targeted at development of genotyping tools for tracking the movement of antibiotic resistance genes in the environment.

In this study, we initiated research to examine the molecular ecology of antibiotic resistance. As a model, we used tetracycline resistance genes encoding the ribosomal protection proteins (RPPs). Phylogenetic analysis revealed the monophyletic origin of these genes, which allowed us to design a set of PCR primers suitable for detection of RPP genes in general, as well as different classes. After validation, this set was used to detect the corresponding genes in the total DNA of swine fecal and rumen samples and in swine feed, as well as in fecal streptococcal isolates from pigs. The primers were also used in a PCR-denaturing gradient gel electrophoresis (DGGE) analysis to demonstrate the overall diversity and similarity of RPP genes in different ecosystems. The methods used in this work can be applied to study other phylogenetically coherent antibiotic resistance gene families.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Phylogenetic analysis and primer design. All currently available nucleotide sequences encoding RPPs, as well as the phylogenetically most closely related elongation factors, EF-Gs, were downloaded from the GenBank database (3). These included the sequences of the following RPP genes (the numbers in parentheses are GenBank accession numbers): E. faecalis DS16 tet(M) (M85225), E. faecalis Tn916 tet(M) (X56353), E. faecalis Tn1545 tet(M) (X04388), Neisseria meningitidis tet(M) (X75073), Ureaplasma urealyticum tet(M) (U08812), Gardnerella vaginalis tet(M) (U58986), Neisseria gonorrhoeae 6418 tet(M) (L12241), N. gonorrhoeae 2903 tet(M) (L12242), Staphylococcus aureus tet(M) (M21136), Streptococcus pneumoniae Tn5251 tet(M) (X90939), L. monocytogenes BM4210/pIP811 tet(S) (L09756), L. lactis K214/pK214 tet(S) (X92946), S. pneumoniae tet(O) (Y07780), Streptococcus mutans DL5 tet(O) (M20925), Campylobacter jejuni tet(O) (M18896), B. fibrisolvens tet(W) (AJ222769), Bacteroides fragilis tet(Q) (Z21523), Prevotella intermedia PDRC-11 tet(Q) (U73497), P. ruminicola tet(Q) (L33696), B. fragilis BF-2 tet(Q) (Y08615), Bacteroides thetaiotaomicron tet(Q) (X58717), Streptococcus pyogenes A498 tet(T) (L42544), Clostridium perfringens CW92 tetB(P) (L20800), Streptomyces lividans 1326 tet (M74049), and Streptomyces rimosus otrA (X53401). Elongation factor EF-G-encoding genes were obtained from Bacillus subtilis (D64127), E. faecalis (retrieved from www.tigr.org), Escherichia coli (X00415), Helicobacter pylori (AE001539), Thermus thermophilus (X16278), and Aquifex aeolicus (AE000669).

Environmental samples and DNA extraction. Samples of whole rumen contents were obtained from eight fistulated steers maintained at the Research Farm of the University of Illinois at Urbana-Champaign. The animals used are treated with antibiotics only in case of disease, and no antibiotics are given for prophylaxis or growth promotion. No tetracyclines had been used for disease treatment in this group of animals, and the animals were considered free of tetracycline selective pressure. Fecal samples from six sows were collected at the Swine Research Farm of the University of Illinois at Urbana-Champaign. In this facility, in addition to therapeutic use, antibiotics are added to feed for prophylactic and growth-promoting purposes. The sows were routinely fed Tylan (Elanco Animal Health, Indianapolis, Ind.) at a concentration of 40 mg per kg of feed. The antibiotic was switched to chlortetracycline (400 mg per kg of feed) for 2 weeks and then back to Tylan. Fecal samples were collected 3 weeks after the switch back to Tylan. Rumen and fecal samples were frozen at 20°C for future isolation of total DNA. Samples of pig feed were stored at room temperature before DNA isolation. Total DNA was isolated from the fecal, rumen, and feed samples by using a Soil DNA Purification Kit (Mo Bio, Solana Beach, Calif.) according to the manufacturer's protocol.

Organisms, plasmids, and culture techniques. The organisms and plasmids used in this study for validation and control are listed in Table 2. C. jejuni subsp. jejuni ATCC 43503 was grown at 37°C under microaerophilic conditions on ATCC medium 1115. C. perfringens JIR4202 was grown anaerobically at 37°C on BHIB medium (Difco Laboratories, Detroit, Mich.). S. pyogenes CIP105079 was grown aerobically on BHIB medium (Difco) at 37°C. E. coli strains were grown on Luria-Bertani medium at 37°C with aeration. Media were solidified when necessary with 1.8% (wt/vol) agar (Difco). Tetracycline (Sigma Chemical Co., St. Louis, Mo.) was added at a concentration of 10 µg/ml to cultures of C. jejuni, C. perfringens, and S. pyogenes. The cloned tet genes were maintained in E. coli by selection of plasmid antibiotic markers (ampicillin or kanamycin at a concentration of 50 µg/ml or chloramphenicol at a concentration of 20 µg/ml).

PCR and DGGE. A typical PCR mixture (total volume, 20 µl) contained 25 pmol of each primer (except for the degenerate universal primers, which were used at a concentration of 100 pmol per mixture), 1× ExTaq reaction buffer (Takara Shuzo, Orsu, Japan), each deoxynucleoside triphosphate at a concentration of 100 µM, and 1.0 U of ExTaq DNA polymerase (Takara Shuzo). A 200-ng portion of purified DNA or one-half of the biomass of a 1- to 2-mm-diameter colony was used as a template. PCR amplification (25 cycles) was performed with a GeneAmp 2400 PCR system (Perkin-Elmer, Norwalk, Conn.) as follows: initial denaturation at 94°C for 5 min, followed by 25 cycles of 94°C for 30 s, 30 s of annealing at the annealing temperatures shown in Table 1, and 30 s of extension at 72°C, and a final extension step at 72°C for 7 min. A touchdown PCR with the degenerate Ribo2 primers (Table 1) was performed as follows: initial denaturation at 94°C for 5 min; 22 cycles of denaturation at 94°C for 30 s, annealing for 30 s with 1°C decrements at temperatures of 72 to 50°C, and extension at 72°C for 30 s; 20 cycles of 94°C for 30 s, 50°C for 30 s, and 72°C for 30 s; and final extension at 72°C for 7 min. If unidentified substances in a DNA preparation inhibited the PCR, 1 µl of the reaction mixture was used as a template for further amplification. Aliquots (5 µl) were analyzed by electrophoresis on a 2.5% (wt/vol) agarose gel (NuSieve; FMC Bioproducts, Rockland, Maine) containing the fluorescent dye GelStar (FMC Bioproducts). Gels with amplicons generated by the Ribo2 primers were stained with ethidium bromide.

Cloning and sequencing of PCR amplicons. PCR products were cloned by using a TA Cloning kit (Invitrogen, Carlsbad, Calif.). White colonies of ampicillin-resistant transformants were screened for the presence of tet fragments by PCR by using the same primer set that was used for amplification. DNA sequence analysis of recombinant plasmids was performed for both strands (primers M13F and M13R) at the University of Illinois Biotechnology Center. On-line similarity searching was performed by using the BLAST (Basic Local Alignment Search Tool) family of programs in GenBank (21).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Phylogenetic analysis. Phylogenetic analysis was performed with 25 complete nucleotide sequences encoding RPPs and with six complete sequences encoding phylogenetically closely related translation elongation factors belonging to family G (EF-G). With 100% bootstrap support, this analysis confirmed the monophyletic origin of RPP genes and the early branching from the other group of elongation factors, EF-G (Fig. 1). The number of substitutions per base pair was approximately 2.4 times higher in the RPP cluster than in the EF-G cluster. Within the RPP supercluster, there are eight clusters corresponding to the recently revised classes Tet M, Tet S, Tet O, Tet W, Tet Q, Tet T, TetB P, and otrA. Three of these clusters, Tet W, Tet T, and TetB P, are represented by only a single sequence at present.

View larger version (29K): [in this window] [in a new window]   FIG. 1.   Phylogenetic placement of tetracycline resistance genes encoding RPPs. The sequence of the A. aeolicus fusA gene for translation elongation factor EF-G was used as the outgroup to root the tree. The number at each node is the number of times that that tree configuration occurred in 1,000 bootstrap trials. The scale bar indicates 0.1 fixed nucleotide substitution per sequence position. The sets of PCR primers (Table 1) targeting various classes of RPP genes are shown on the right.

Design and validation of PCR primers targeting RPP genes. Evidence of the monophyletic origin of the RPP genes opened the possibility of designing primers that target all genes in the cluster. However, early branching and further independent diversification, together with the high G+C contents of the tet and otrA genes, precluded incorporation of these genes into the alignment analysis. Thus, the design of the universal primer pair was based on the sequences belonging to seven classes of RPP genes. As mentioned above, the rate of nucleotide substitution in the RPP cluster is higher than that in other elongation factors, and therefore, the overall sequence structure is less conserved. Because of this, the design of the universal primer pair involved a substantial level of degeneracy (Table 1). Primer pairs specific for the individual classes, together with the expected amplicon sizes, are shown in Table 1.

Detection of RPP genes in the rumen. Total DNA preparations from the rumen samples were subjected to PCR amplification with the universal primer set, the Ribo2 set, followed by analysis with class-specific primers. Although the Ribo2 primer set performed well with pure cultures (Table 3), it was not sufficiently selective during amplification from total community DNA. According to the analysis with class-specific primers, the rumen microbiota appeared to bear the tet(O), tet(Q), and tet(W) genes (data not shown). The coamplified V3 region of 16S rDNA (data not shown) allowed a rough estimate of the proportion of the rumen bacterial microbiota carrying the corresponding resistance genes. Comparative analysis of densitograms by using biomass of C. jejuni as a standard suggested that up to 5% of ruminal bacteria may carry tet(O). The rumen samples were also subjected to PCR-DGGE analysis with the TetO, TetQ, and TetW primer sets, in which GC clamps were attached to the reverse primers. This analysis revealed the uniformity of the tet(W) and tet(O) genes circulating in the rumen microbiota of the cows (Fig. 2, lanes 2 through 9, and Fig. 3, lanes 13 through 20). DGGE bands from tet(W) samples were excised, reamplified, sequenced, and found to be identical to the tet(W) sequence (2). However, the DGGE band from this control template migrated farther than our samples (Fig. 2, lane 13), and careful inspection of the nucleotide sequence of the tet(W) control revealed a nucleotide substitution (TC), which may have been incorporated during amplification from B. fibrisolvens (2). A DGGE analysis with the TetO primers produced similar results, with no variations among animals but with two bands (Fig. 3, lanes 13 through 20). Sequence analysis revealed that these two bands, which were in the region of tet(O), were actually identical, but for unknown reasons, the upper band also contained the reverse DGGE primer misincorporated at the 5' end of the forward primer, thus increasing the melting temperature and creating artificial heterogeneity. A DGGE analysis with the TetQ primers revealed diversity among tet(Q) in the rumen, as well as animal-to-animal variation (Fig. 4). There were at least five bands, and sequence analysis confirmed the heterogeneity at the sequence level.

View larger version (100K): [in this window] [in a new window]   FIG. 2.   DGGE analysis of tet(W) amplicons from steer rumen and pig fecal samples. Lanes 1, 10, and 20, synthetic marker composed of known 16S rDNA sequences with various G+C contents; lanes 2 through 9, rumen samples from steers 277, 279, 280, 281, J277, J279, J280, and J281, respectively; lanes 11 and 12, negative controls pBT-1 [tet(Q)] and pJIR667 [tetB(P)], respectively; lane 13, positive control pGEM-tetW [tet(W)]; lanes 14 through 19, fecal samples from pigs 1 through 6, respectively.

View larger version (74K): [in this window] [in a new window]   FIG. 3.   DGGE analysis of tet(O) amplicons from pig fecal and steer rumen samples. Lanes 1, 12, and 21, synthetic marker composed of known 16S rDNA sequences with various G+C contents; lanes 2 through 7, fecal samples from pigs 1 through 6, respectively; lanes 8 and 9, S. alactolyticus O19 and O31, respectively; lanes 10 and 11, negative controls pBT-1 [tet(Q)] and pJIR667 [tetB(P)], respectively; lanes 13 through 20, rumen samples from steers 277, 279, 280, 281, J277, J279, J280, and J281, respectively.

View larger version (135K): [in this window] [in a new window]   FIG. 4.   DGGE analysis of tet(Q) amplicons from steer rumen samples. Lanes 1 and 11, synthetic marker composed of known 16S rDNA sequences with various G+C contents; lanes 2 through 9, rumen samples from steers 277, 279, 280, 281, J277, J279, J280, and J281, respectively; lane 10, positive control pBT-1 [tet(Q)].

Detection of RPP genes in swine feces. Total DNA preparations from swine fecal samples were subjected to PCR amplification with the universal primer set, the Ribo2 set, and also with class-specific primers. Tet M, Tet O, and Tet W determinants were detected in the swine intestinal microbiota (data not shown). Further PCR-DGGE analysis (Fig. 5) and sequencing demonstrated that swine tet(M) is identical to the control template, tet(M) cloned from Streptococcus agalactiae. Interestingly, the tet(O) and tet(W) genes circulating in the pig herd had the same mobility on DGGE gels as the corresponding genes from the rumina of steers (Fig. 2 and 3). Sequence analysis of the excised and cloned major DGGE bands confirmed that these two classes of genes were identical in the two types of animals. The difference was that swine fecal samples produced an additional minor tet(W) band migrating farther than the major band (Fig. 2, lanes 14 through 19). Several attempts to clone this minor band were unsuccessful, and sequence information for this band is not available.

View larger version (35K): [in this window] [in a new window]   FIG. 5.   DGGE analysis of tet(M) amplicons from pig fecal samples and streptococcal isolates. Lanes 1, 11, and 19, synthetic marker composed of known 16S rDNA sequences with various G+C contents; lanes 2 through 7, fecal samples from pigs 1 through 6, respectively; lane 8, positive control pFD310 [tet(M)]; lanes 9 and 10, negative controls pBT-1 [tet (Q)] and PCR mixture without a template, respectively; lanes 12 through 18, S. alactolyticus M15, M113, M118, M33, M35, M30, and M32, respectively.

Detection of RPP genes in fecal streptococci from swine. Fecal streptococcal isolates from swine (n = 150; 25 isolates from each of six animals) were characterized by RFLP and 16S rDNA sequence analyses. As determined by the RFLP analysis, these strains could be divided into at least three groups (Fig. 6). Sequence analysis allowed identification (sequence similarity, >99%) as strains of Streptococcus alactolyticus (Table 4). A majority (94.7%) of the isolates were resistant to tetracycline at a concentration of 10 µg/ml (Table 5). PCR analysis with our set of primers revealed that all of the resistant isolates carried the tet(O) gene (Table 5).Approximately 22% of the strains carried tet(M) in addition to tet(O). No other tetracycline resistance determinants conferring ribosomal protection were detected in these isolates (Table 5). Thus, the tetracycline-resistant swine S. alactolyticus populations were characterized by the invariable presence of tet(O), and 22% of the strains carried both tet(O) and tet(M).

View larger version (165K): [in this window] [in a new window]   FIG. 6.   RFLP analysis of swine S. alactolyticus isolates. Lanes 1 and 12, 1-kb ladder (Gibco BRL); lanes 2 through 11, isolates M15, M19, M113, M118, M33, M35, M310, M312, M321, and O31, respectively. The first group includes only M15; the second group includes M19, M113, M33, M35, M310, and M321; and the third group consists of M118, M312, and O31.

Detection of RPP genes in swine feed. Because of the presence of unidentified inhibitory substances, a second round of PCR was necessary in the experiments performed with swine feed, and therefore, the detection limit of this assay was lower than that of the assay performed with the fecal and rumen samples. The presence of bacterial DNA in all premix and mixed samples was confirmed by amplification of the V3 region of bacterial 16S rDNA (Table 6). The presence of RPP genes in these samples was confirmed first with the Ribo2 primer set and then with class-specific primers (Table 6). First, the feed components were sampled before the corresponding diet mixes were prepared for three different age groups. These groups were the starters (ages, 3 to 6 weeks), growers (6 weeks to 6 months), and finishers (antibiotics were withdrawn before slaughtering). The corn component used to prepare the mixes for all age groups contained tet(W), tet(O), tet(Q), and tet(M), while the soybean component also contained the tet(S) gene (Table 6). The resistance gene profiles of the commercial whey preparation and the protein plasma product were similar to that of the soybean component. Interestingly, the commercial preparation of Tylan (a macrolide which was used in the grower diet) also contained tetracycline resistance genes with a profile similar to that of the corn component (Table 6). The antibiotic mixture used for the starter group (chlortetracycline, sulfonamide, and penicillin) contained DNA of tetracycline resistance genes, particularly that of tet(W), tet(O), tet(Q), tet(M), and tet(S), and had a profile similar to those of the soybean, whey, and plasma product components (Table 6).

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

This work was the first attempt to use the molecular ecology approach to study antibiotic resistance and, in particular, to estimate the gene pool and flux of antibiotic resistance genes in production animals. With this approach, the first step is elucidation of the evolutionary history of the genes of interest. From the phylogenetic analysis, it is evident that the elongation factors conferring resistance to tetracycline form, with a high degree of confidence, a phylogenetically coherent group separated from other elongation factors. Within this group, there are eight clusters, which correspond to the eight currently defined classes of RPPs (Tet M, Tet S, Tet O, Tet W, Tet Q, Tet T, TetB P, and otrA).

The most deeply branching class, exemplified by tet and otrA, is the class obtained from the antibiotic-producing organisms S. lividans and S. rimosus. Based on the sequence information available, there is no evidence of recent horizontal transfer of RPP genes from antibiotic-producing strains to commensal or pathogenic microbiotas. Hybridization data indicate, however, that some mycobacteria may actually carry the resistance genes originally described in streptomycetes (30). Additional sequence information concerning the mycobacterial RPP genes is required to decide whether there was a potential horizontal transfer event from antibiotic-producing strains. Another interesting aspect of the two available gene sequences of antibiotic-producing streptomycetes is that they are quite divergent. The length of the branch between the two genes is actually comparable to the length of the branch separating the Tet M, Tet S, and Tet O classes (Fig. 1). If more sequence data from this class of genes were available, perhaps definition of at least two new classes would be necessary.

The available sequence data support the scenario that early branching and lengthy independent diversification of eight (or more) clusters of RPPs occurred well before the "antibiotic era." While the functional role of these proteins in antibiotic-producing bacteria is evident (they provide protection against the synthesized antibiotics), it is more challenging to explain their presence and function in bacteria from other ecological niches that have no or limited contact with the soil microbiota (e.g., the gastrointestinal tract). The long evolutionary history of RPP genes supports the hypothesis that these genes might have served some metabolic functions other than providing antibiotic resistance. Protein synthesis is a vital cell process, and there should be mechanisms that support proper functioning of the translation machinery and buffer possible undesirable effects of low-molecular-weight metabolites of the cell. Thus, the alternative elongation factors may have been selected in this way in some bacteria and may have assumed a role in protecting ribosomes against tetracyclines only recently.

At the same time, the rapid movement of the tetracycline-resistant elongation factors to taxonomically divergent commensal and pathogenic bacteria is a very recent evolutionary event on the phylogenetic time scale and can most probably be attributed to horizontal transfers within the clusters in the antibiotic era. Some of the genes are located on plasmids (e.g., pOZ101, pIP811, or pK214) or conjugative transposons (e.g., Tn916, Tn5251, or Tn1545), thus facilitating transfer between species and genus boundaries.

Proof of the monophyletic origin of the RPP genes opened the possibility of designing primer sets targeting all classes, as well as class-specific primers. However, it appeared that the early branching and further independent diversification of the otrA genes, together with a high G+C content, precluded incorporation of these genes into the alignment. Also, the number of substitutions per base pair appeared to be higher in the RPP genes than in other elongation factors, and therefore, the overall sequence structure is less conserved. Thus, the overall design of the universal primer pair involves a substantial level of degeneracy and does not include the genes from the antibiotic-producing streptomycetes. Primers were validated in PCR that included crude bacterial biomass and fecal material, which is notorious for the presence of PCR-inhibiting substances. This fact could be useful for rapid screening for the presence of the RPP genes in bacteria without a DNA isolation step. The primers also were designed to amplify short sequences, thus allowing use in PCR-DGGE analysis. Therefore, such an analysis could be performed with total DNA preparations of environmental origin, thus allowing for the first time access to the pool and diversity of RPP genes in a given ecosystem.

The primers were used to detect the occurrence of RPP genes in the rumina of cows, in swine feed and feces, and in swine fecal streptococci. The Tet O and Tet W determinants were found in the intestinal contents of both types of animals, while Tet M was confined to pigs and Tet Q was confined to the rumen. Approximate estimates suggest that up to 5% of the bacteria in the rumen and swine intestine may carry the tet(O) gene. Another interesting observation is that tet(W) and the majority of the tet(O) genes circulating in the two different animal herds, which had very different antibiotic use regimens, were actually identical. The identity of the tet(W) genes obtained from bovine and ovine rumen and human intestinal isolates was demonstrated in a recent study (37). Obviously, this finding could be extended to include yet another animal model (pig) and another gene [tet(O)]. The occurrence of identical tetracycline resistance genes in different hosts provides additional evidence that there are extensive pools of antibiotic resistance genes that are actively exchanged at least between domestic animals. However, genetic transfer itself is not a guarantee that the transferred antibiotic resistance gene will be maintained in another host. The second observation concerning the persistence of antibiotic resistance in the apparent absence of antibiotic selective pressure [cattle that have no antibiotic in their feed but carry intestinal bacteria with tet(O), tet(Q), and tet(W) and swine feed containing a diverse group of resistance genes] raises the question of how resistance persists. The possession of an antibiotic resistance gene by a bacterium is certainly advantageous in the presence of the corresponding antibiotic. In the absence of the antibiotic, however, the cost of carrying of the resistance gene should reduce the bacterial fitness and the resistant phenotype should be replaced by the sensitive phenotype. However, a recent reexamination of this topic suggested that bacteria may have been able to adapt to the burden of resistance with little or no cost to their fitness (25). In this scenario, the antibiotic-resistant microbiota would successfully compete with the sensitive counterpart even in the absence of selection. Such adaptations would preclude resistant lineages from reverting to sensitivity and make control of antibiotic resistance even more difficult.

The significant outcome of the sequence analysis of tetracycline resistance genes in cultivable streptococcal isolates is that the nucleotide sequences of tet(M) and the majority of tet(O) genes are identical to those of the corresponding genes acquired directly from fecal DNA. This is yet another validation of the in vitro analysis approach and suggests that the pool of resistance genes, initially discovered in total DNA, could be tracked to specific bacterial populations in the gut. In our case, S. alactolyticus could be considered one of the main reservoirs of the tet(M) and tet(O) genes in the swine intestinal microbiota. Based on RFLP and sequence analyses of 16S rDNA of S. alactolyticus isolates, this is not a clonal population but is represented by at least three subpopulations. Therefore, circulation of identical tet(M) and tet(O) genes in this genetically diverse group of bacteria suggests that there is horizontal exchange of tetracycline resistance genes rather than coexistence of several tetracycline-resistant clones.

Compared with the rumen and fecal samples, the components of the swine feed appeared to be contaminated with a more diverse group of RPPs, and only two classes (Tet T and TetB P) were absent. No attempt to isolate resistant bacteria was made, but the ubiquitous presence of these genes, together with the bacterial V3 markers, suggests that the feed components may have been contaminated by bacteria carrying the corresponding resistance genes. It is not clear whether these bacteria were dead or viable; regardless, the feed was genetically contaminated. The experiments were designed to detect possible cross-contamination of the swine feed by on-farm dust and fecal material, but it appeared that the components of swine feed already carried more diverse markers of tetracycline resistance, including that in the swine gut microbiota. This suggests that the actual source of antibiotic resistance gene contamination of swine feed was something else and requires further independent research. At this time, we hypothesize that at least for the corn and soybean components the source may have been manure from farms on which antibiotics were used, which was applied to the land. Whey, a by-product of cheese manufacturing, may contain a residual biomass of tetracycline-resistant lactic acid bacteria and propionobacteria. However, we have no information concerning the source of antibiotic resistance gene contamination in other components of the swine feed, such as the plasma protein and especially the antibiotic preparations, which are supposedly the products of sterile fermentation.

In this study molecular ecology tools were used to study the antibiotic resistance problem, and the results suggest that this approach has the potential of uncovering the reservoirs and determining the identities of antibiotic resistance genes in a variety of ecosystems. This approach could be easily extended to other classes of antibiotic resistance genes in order to understand the pathways leading to acquisition of drug resistance by human- and animal-pathogenic bacteria.