INTRODUCTION

Top Abstract Introduction Materials and Methods Results and Discussion References

Members of the genus Aeromonas are ubiquitous in most aquatic environments (19). Several species have been implicated in fish disease (e.g., Aeromonas hydrophila, A. sobria, A. allosaccharophila, A. salmonicida, and A. veronii; 6), and pathogenicity in humans has been demonstrated by A. hydrophila, A. veronii, A. jandaei, A. trota, and A. schubertii (23). Treatment or prevention of disease in both humans and fish has been undertaken extensively using antimicrobial agents. It has been estimated that the total amount of antibiotics used in aquaculture and agricultural practices is approximately equal to that employed in the therapy of human disease (46), as typified by Norway in 1989 (14). This has resulted in an increase in the frequency of bacteria resistant to these agents to the extent that they may affect the treatment of human and fish diseases, in addition to impacting the aquaculture environment directly (40). It was concluded that there was inadequate information to allow a quantitative or even a qualitative assessment of the risk to human health posed by antimicrobial agent usage (28, 40). Smith et al. (40) suggested that the greatest potential risk was presented by the transfer of plasmid-encoded resistance genes between the aquatic compartment and the human compartment of the environment. Furthermore, they identified our lack of understanding of the survival of bacterium-plasmid complexes in the environment as the major limitation in our knowledge (40).

Tetracyclines (TCs) have been used exclusively in aquaculture, particularly to control furunculosis in salmonids (40). Inevitably, resistance to oxytetracycline (OT) and TC emerged and was found to be plasmid encoded (1, 37, 38). The frequency of resistance to TCs in A. salmonicida has increased with greater usage from 4% of isolates from 1979 to 1981 (4) to greater than 50% of isolates examined from Scottish fish farms in the early 1990s (34). Assessing the extent of antibiotic resistance in bacterial isolates presents many difficulties (25, 26). Adams and coworkers (1) recognized that there was a need to assess the true nature of OT^r at the molecular level. Their strain set contained 66% of isolates with transferable OT^r which was subsequently shown to be encoded by large plasmids (the pASOT group) that carried a 5.4-kb EcoRI restriction fragment containing the Tet A determinant (1). The detection of OT^r bacteria in the aquaculture industry showed that this environment, like hospitals, was facing a threat from the use of antibiotics (1). Their study also highlighted the need to extend the investigation to isolates from other geographical locations and to other fish pathogens (1). Furthermore, they suggested that the ability of plasmids to spread among the bacterial population present in, and beyond, fish farm environments should be investigated (1).

In this study, we have assessed whether there is sufficient evidence to suggest that the aquaculture and human environments exist as two separate entities with distinct transfer events or whether they comprise a single interactive compartment of the environment where free exchange of genetic information occurs. To test these hypotheses, we investigated the dissemination of conjugative OT^r plasmids between isolates in the human and aquaculture compartments of the environments using OT^r mesophilic aeromonads recovered from untreated hospital effluent and from a fish farm facility as model organisms and environments.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results and Discussion References

Bacterial strains, plasmids, and sampling. The bacterial strains and plasmids used in this study are described in Table 1. Escherichia coli strains and Pseudomonas putida PaW340 (a derivative of P. putida nt-2) were maintained on Iso-Sensitest agar (ISA; Oxoid, Basingstoke, United Kingdom). All OT^r control strains were maintained on ISA supplemented with OT-HCl (25 µg ml¹) obtained from Vetrapharm (Hampshire, United Kingdom).

1

1

Antibiotic susceptibility testing. Susceptibility of aeromonad field isolates and transconjugants to the antibiotics AP (25 µg), OT (30 µg), TC (30 µg), streptomycin (SM; 10 and 30 µg), nalidixic acid (NA; 30 µg), rifampin (RD; 25 µg), kanamycin (KM; 30 µg), and trimethoprim (TM; 5 µg) was assessed on ISA in accordance with the established Kirby-Bauer procedure (7) using disks supplied by Oxoid.

Identification of isolates. All isolates were initially confirmed as Aeromonas spp. at the genus level by PCR amplification of the aroA gene as described previously (42). Further characterization was carried out by gas-liquid chromatographic analysis of cellular fatty acid methyl esters using the Microbial Identification System (Microbial ID Inc.) as previously described (22). The resulting fatty acid methyl ester patterns were identified automatically and compared with the database AER48C (20) for the identification of mesophilic aeromonads. Isolates that could not be unambiguously classified into one of the known Aeromonas DNA hybridization groups (HGs) or genomic species by comparison with the AER48C database were subjected to whole-genome analysis with the fluorescent amplified fragment length polymorphism analysis technique as described previously (21). Digitized fluorescent amplified fragment length polymorphism fingerprints were further processed using the GelCompar software package (Applied Maths) and compared for identification with the laboratory database AEROLIB (21).

Conjugation experiments. Filter matings were performed by separately resuspending a loopful of freshly cultured donor and recipient cells in 300 µl of 1× phosphate-buffered saline (pH 7.4) (36), followed by overlaying 10 µl of each suspension onto a 0.22-µm-pore-size membrane filter (Supor-200; Gelman) and then incubation at 28 ± 0.5°C for 24 h. Controls (unmixed donors and recipient cells) were treated in the same manner. After incubation, cells and controls were resuspended in 450 µl of phosphate-buffered saline and transconjugants were selected by spreading onto either ISA supplemented with OT (25 µg ml¹) and RD (50 µg ml¹) at 37°C for 24 h (when E. coli J53-1 was used as the recipient) or ISA supplemented with OT (25 µg ml¹) and SM (500 µg ml¹) for 24 h at 30°C (when P. putida was the recipient). E. coli J53-1 transconjugants were confirmed by demonstration of OT and RD resistance phenotypes and auxotrophy for proline and methionine. P. putida transconjugants were confirmed by OT and SM resistance and auxotrophy for tryptophan.

DNA manipulations. Restriction endonuclease digestion and agarose gel electrophoresis were carried out using established techniques (36). Plasmid DNA was extracted from control strains and E. coli J53-1 transconjugants after growth in Luria-Bertani medium supplemented with OT (25 µg ml¹) at 37°C with shaking at 150 rpm for 18 h using Qiagen mini and midi columns or using the sucrose gradient plasmid extraction technique (47).

Construction and labeling of DNA probes. DNA probes for the determinants Tet B and Tet D were constructed as previously described (31) and were purified using GenElute agarose gel purification spin columns (Supelco). Probes for the determinants Tet A, Tet C, Tet E, and Tet G were created via PCR using primers specific to these determinants in accordance with an established protocol (15). PCR probes were purified using the Nucleon QC PCR/oligo clean up kit (Amersham-Pharmacia Biotech). Probes specific to Tn1722 and the methyl-accepting chemotaxis protein (MCP)-encoding gene of Tn1721 were constructed from pMT1286 (45). The entire Tn1722 region was produced by digesting pMT1286 with EcoRI and excision and purification of the 5.6-kb fragment. The MCP probe was prepared by digestion of pMT1286 with EcoRI and HindIII to produce a fragment of 1.2 kb. Excised DNA was purified using the Qiaex II DNA purification kit (Qiagen). DNA probes were labeled with [-³²P]dCTP as specified in the random-primed hexanucleotide labeling kit (Roche-Molecular Biochemicals).

DNA blotting and hybridizations. In all cases, DNA was blotted onto positively charged nylon membrane (Roche-Molecular Biology). Southern transfers (43) were carried out in accordance with standard procedures (36) using 0.4 M NaOH as the transfer, denaturation, and fixation buffer. Dot blots were performed with QIAGEN plasmid DNA (denatured at 95°C for 10 min), and the DNA was fixed with UV light (302 nm) for 3 min. Prehybridization was carried out in prewarmed (68°C) 5× SSPE (1× SSPE is 0.18 M NaCl, 10 mM NaH₂PO₄, and 1 mM EDTA [pH 7.7]) containing 5× Denhardt's solution, 0.5% (wt/vol) sodium dodecyl sulfate (SDS), and 0.25% (wt/vol) N-lauryl sarcosine for 5 to 6 h. Hybridization was performed in fresh prewarmed solution (prehybridization solution without the addition of Denhardt's solution) at 68°C for 18 to 20 h. Unbound radioactive probe DNA was removed by washing membranes twice for 10 min (each time) in 2× SSPE-0.1% (wt/vol) SDS at room temperature (20 to 25°C), followed by 15 min at 68°C in 1× SSPE-0.1% (wt/vol) SDS and two washes of 15 min (each) in 0.1× SSPE-0.1% (wt/vol) SDS at 68°C. The membranes were then wrapped in cling film and exposed to X-ray film (Hyperfilm-MP; Amersham-Pharmacia) at 70°C for up to 3 days.

Cloning. Plasmid DNA was cloned into pUC18 (Amersham-Pharmacia-Biotech) in accordance with standard procedures (36). Transformation was carried out using E. coli DH5 competent cells in accordance with the methodology of the supplier (Clontech).

DNA sequencing and analyses. Cloned DNA was partially sequenced using the universal forward and reverse primers homologous with pUC18 using an ABI Prism 373 sequencer. FASTA and BLAST DNA homology searches were performed using the programs within the DNA Data Bank of Japan (DDBJ) at the following internet address: http://www.ddbj.nig.ac.jp/E-mail/homology.html.

	RESULTS AND DISCUSSION

Top Abstract Introduction Materials and Methods Results and Discussion References

Isolation of OT^r aeromonads and conjugation experiments. A total of 163 presumptive OT^r mesophilic aeromonads were isolated from the hospital (72 isolates) and fish farm (91 isolates) sample sites. All isolates were assessed for susceptibility to the antibiotics TC, OT, RD, SM, NA, and KM using sensitivity test disks. All isolates were confirmed as resistant to TC and OT. The most common resistance observed in both environmental groups was to NA (94% of isolates from the hospital and 52% of isolates from the fish farm), while 14% of hospital isolates and 40% of the fish farm isolates were resistant to SM. Resistance to both NA and SM was observed in isolates from both environments. A total of 90% of those isolates that were SM resistant, and 13% of the NA-resistant isolates from the hospital were resistant to both antibiotics, while 75% of SM-resistant and 57% of NA-resistant isolates from the fish farm were resistant to both. No isolates were RD or KM resistant.

Identification of host strains. The identification of host strains that carried the 17 transferable OT^r plasmids is shown in Table 2. Eight isolates were identified as members of A. hydrophila DNA HG3 and contained plasmids pFBAOT3 to -5, -7, -9 to -11 and -15. Six others were identified as members of A. veronii bv. sobria HG8 (plasmids pFBAOT1, -8, -12, -13, -14, and -17). The isolate that originally carried pFBAOT16 could not be identified beyond the genus level (Table 2).

Host range assessments. In an attempt to determine whether the pFBAOT plasmids are capable of transfer to recipients other than E. coli, conjugal transfer was carried out between each of the 17 original isolates and P. putida PaW340. Transfer of OT^r to P. putida was demonstrated for plasmids pFBAOT3 to -7, -9, and -11. It was also shown that these plasmids could be retransferred from P. putida transconjugants back into OT^s plasmid-free E. coli J53-1 (Table 2). These results suggested that pFBAOT3 to -7, -9, and -11 were potentially broad-host-range plasmids.

Molecular characterization of pFBAOT plasmids. We assessed which tet determinants and replicons were carried by the pFBAOT plasmids. Minipreparations of plasmids pFBAOT1 to -17 were screened by dot hybridization using the Tet A to E and Tet G probes and those specific to broad-host-range replicons N, P, Q, W, and U. It was shown that the Tet A determinant was carried by 10 plasmids (pFBAOT1, -3 to -7, -9, and -11 to -13) (Table 2), and no other characterized determinants were detected. Therefore, the remaining seven plasmids (pFBAOT2, -8, -10, -14, -15, -16, and -17) may possess previously undescribed tet determinants. Seven plasmids hybridized with the IncU probe (Table 2), and these were the same plasmids that displayed broad-host-range characteristics (plasmids pFBAOT3 to -7, -9, and -11). In addition, all of these plasmids carried Tet A (Table 2). The remaining 10 plasmids did not hybridize with the inc-rep probe set.

Analysis of Tet A-containing plasmids. The restriction fragment length polymorphism (RFLP) patterns of plasmid DNA from Tet A-carrying pFBAOT plasmids were compared after digestion with EcoRI. Plasmids from hospital isolates were shown to be closely related to each other, with pFBAOT7 and pFBAOT11 being identical (Fig. 1A). Plasmid extractions from fish farm E. coli transconjugants were repeatedly of poor quality, and so RFLP patterns for plasmids pFBAOT12 to -17 were difficult to interpret. For this reason, the profiles of these plasmids are absent from Fig. 1. However, it was clear that they did not share close homology and were larger than the hospital-derived plasmids (pFBAOT1 to -11) (data not shown). Two EcoRI restriction fragments of approximately 5.2 and 5.5 kb were common to seven of the eight Tet A-carrying hospital-derived plasmids (pFBAOT3 to -7, -9, and -11). pFBAOT1, which was not IncU (and therefore was not included in Fig. 1) and was originally found in A. veronii bv. sobria HG8, carried only the 5.5-kb fragment. In addition, common fragments of approximately 5.7 and 6.5 kb were observed in pFBAOT3 to -5, -7, -9, and -11 (all originally found in A. hydrophila HG3) but were absent from pFBAOT1 and -6 (from A. caviae HG5A) (Fig. 1A; see Fig. 4A). It was difficult to discern from RFLP data alone whether any of the fish farm-derived plasmids carried these common fragments.

View larger version (46K): [in this window] [in a new window]   FIG. 1.   (A) RFLP profiles (generated by EcoRI digestion) of IncU OTr plasmids from aeromonads and corresponding Southern hybridizations with determinant Tet A and pASOT plasmid. Lanes: 1 and 11,  DNA markers digested with HindIII and a 1-kb ladder, respectively; 2, pIE420; 3, pASOT; 4, pFBAOT3; 5, pFBAOT4; 6, pFBAOT5; 7, pFBAOT6; 8, pFBAOT7; 9, pFBAOT9; 10, pFBAOT11. (B) Hybridization of the Tet A determinant probe to plasmids. All plasmids harbored Tet A on a 5.5-kb fragment. (C) Hybridization of plasmid pASOT to IncU plasmids.

View larger version (24K): [in this window] [in a new window]   FIG. 2.   Comparison of plasmids pRAS1 and pIE420 by hybridization of radiolabeled plasmid pRAS1 to both plasmids. Lanes: 1, pRAS1; 2, pIE420.

Cloning and partial sequencing of the 5.5-kb EcoRI Tet A-carrying fragment. The results presented here show that the 5.5-kb EcoRI Tet A-carrying fragment is more widespread than has been previously suggested (1) and that it is not exclusively associated with IncU plasmids. This fragment was shotgun cloned from plasmid pFBAOT6 into pUC18, and sequencing was carried out from each end of the cloned insert such that partial sequences of 647 and 620 bp were obtained. The first 647 bp of one end of the insert shared 95% homology with Tn1721 (encompassing the first right inverted repeat (IRR-I), the tccA gene (27), and part of the tetR gene, respectively). The first 620 bp of the opposite end of the insert shared 95% homology with IRR-II and part of the tnpA' region of Tn1721 (data not shown). This demonstrated that the 5.5-kb EcoRI fragment from pFBAOT6 was highly homologous to the 5.5-kb EcoRI fragment found in Tn1721. In order to assess whether the complete transposon Tn1721 was present on these plasmids, a DNA probe was constructed from plasmid pMT1286 (45) that was specific to the gene (orfI) which encodes MCP, situated at the left-hand end of Tn1721 (2) (Fig. 3). This probe was hybridized to plasmid pASOT (pASOT2 and pASOT3 were not assessed), pIE420, and the IncU pFBAOT plasmids. The results showed that all of the IncU pFBAOT plasmids and pASOT carry this gene, unlike plasmid pIE420 (Fig. 4C). Plasmids pFBAOT3, -4, -5, -7, -9, and -11 carry the MCP-encoding gene on a 5.6- to 5.8-kb EcoRI fragment (slightly larger than the region carrying the Tet A determinant). Plasmid pFBAOT5 also carried a second copy of the gene on a fragment of approximately 12 to 13 kb, suggesting that this plasmid contains two copies of a Tn1721-like element. Plasmid pASOT carries the region on a 7-kb EcoRI restriction fragment, while a 9-kb fragment was detected in pFBAOT6 (Fig. 4). The presence of the MCP region on a 5.6- to 5.8-kb EcoRI fragment, in addition to a 5.5-kb Tet A-carrying EcoRI fragment, is indicative of possession of the complete Tn1721 transposon (Fig. 3). Therefore, plasmids pFBAOT3 to -5, -7, -9, and -11 appear to possess complete transposon sequences (indicating that the MCP-encoding gene-containing region is 5,610 bp in length; Fig. 3). The differences between this group and pASOT and pFBAOT6 might be explained in light of recent findings by Schnabel and Jones (39). They demonstrated that the Tet A-mediated TC resistance observed in phylloplane Pseudomonas sp. was due to a truncated form of Tn1721 (designated Tn1720). Tn1720 lacks the MCP-encoding region (orfI) and has minor variations in the sequence of the three inverted repeats on Tn1721. This variation within the left inverted repeat (IRL) resulted in an alteration at the EcoRI site (39). Plasmids pASOT and pFBAOT6 possessed the MCP-encoding region and therefore did not harbor Tn1720. However, the location of the MCP-encoding region on fragments larger than 5.6 kb along with a 5.5-kb Tet A-carrying EcoRI fragment is indicative of the loss of the EcoRI site within the IRL. The absence of the MCP-encoding region from pIE420 suggested that it may contain a Tn1720-like transposon. If this were the case, then the transposition genes tnpR and tnpA would still be present on the plasmid. To test this hypothesis, the plasmid was probed with the complete Tn1722 transposon and no hybridization signal was obtained (data not shown). This result suggested that the 5.5-kb Tet A-carrying EcoRI fragment was deposited by Tn1721 but with subsequent loss of minitransposon Tn1722 (Fig. 3). This scenario might also be applicable to the unrelated plasmids pFBAOT1, -12, and -13, which also carry Tet A on the 5.5-kb EcoRI fragment.

View larger version (11K): [in this window] [in a new window]   FIG. 3.   Physical and genetic structure of transposon Tn1721 (length, 11,139 bp). The 38-bp inverted repeats IRL, IRR-I, and IRR-II are highlighted. The designation, position, and direction of transcription of genes are shown by the arrowed boxes. The region that comprises the minor transposon Tn1722 is highlighted, along with the location of EcoRI restriction sites. This diagram was adapted from Schnabel and Jones (39) and Tsuda et al. (45).

View larger version (31K): [in this window] [in a new window]   FIG. 4.   (A) EcoRI-generated RFLP patterns for IncU OTr plasmids. The gel was electrophoresed for an extended period in order to separate the fragments in the 5- to 7-kb range. Lanes: 1 and 12,  DNA markers digested with HindIII and a 1-kb ladder, respectively; 2, pMT1286; 3, pASOT; 4, pIE420; 5, pFBAOT3; 6, pFBAOT4; 7, pFBAOT5; 8, pFBAOT6; 9, pFBAOT7; 10, pFBAOT9; 11, pFBAOT11; 12, 1-kb marker DNA. (B) Southern hybridization of the MCP probe to the DNA in panel A.

Conclusions. Adams et al. (1) highlighted the need for research on OT usage in aquaculture and the resistance associated with relevant bacteria to extend to isolates from other geographical locations and to other fish pathogens. In addition, these authors highlighted the need to determine whether these resistance plasmids extend beyond fish farm environments (1). The main objectives of this study were to carry out these assessments and to investigate whether the aquaculture and human compartments of the environment should be considered separate entities with distinct transfer events or a single interactive compartment of the environment. In the course of these analyses, we showed that closely related IncU R plasmids previously associated only with fish farm environments were common to those impacting humans (i.e., pASOT plasmids and pRAS1 from fish farms and pFBAOT plasmids and pIE420 from hospital sewage and a German hospital patient, respectively). In addition, this dissemination occurred among at least four separate countries (Norway, Scotland, England, and Germany). Furthermore, the occurrence of pFBAOT plasmids in A. hydrophila and A. caviae demonstrated that they could disseminate to other related bacteria under natural conditions. This study also showed for the first time that plasmids pRAS1 and pIE420 are probably identical. The original carriage of pIE420 in E. coli (44) and pRAS1 in A. salmonicida (37) demonstrated interspecies transfer from (or to) a human commensal (and potentially pathogenic) organism under natural conditions. Collectively, these findings provide evidence that suggests that we should consider the two environments (fish farm and hospital) one interactive compartment and are therefore contrary to the hypothesis of Smith and coworkers (40). The involvement of Tn1721 and Tn1721-like elements in the dissemination of the Tet A determinant, as demonstrated here and by other workers (13, 39), is extremely relevant to this interaction and, globally, to the potential dissemination of these and similar plasmids.