HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via CrossRef Citing Articles via Google Scholar Google Scholar Articles by Giraud, E. Articles by Donnay-Moreno, C. Search for Related Content PubMed PubMed Citation Articles by Giraud, E. Articles by Donnay-Moreno, C. Agricola Articles by Giraud, E. Articles by Donnay-Moreno, C.

Mechanisms of quinolone resistance and clonal relationship among Aeromonas salmonicida strains isolated from reared fish with furunculosis

¹Unité Mixte de Recherche INRA-ENVN Chimiothérapie Aquacole et Environnement, Ecole Nationale Vétérinaire, Atlanpôle, La Chantrerie, BP40706, 44307 Nantes, Cedex 03, France ²Unité de Biodiversité des Bactéries Pathogènes Emergentes, INSERM U 389, Institut Pasteur, 75724 Paris Cedex 15, France

Correspondence Etienne Giraud giraud{at}vet-nantes.fr

Received January 6, 2004
Accepted May 17, 2004

The mechanisms of resistance to quinolone and epidemiological relationships among A. salmonicida strains isolated from diseased fish in French marine farms from 1998 to 2000 were investigated. The quinolone resistance-determining regions of the gyrA and parC genes of 12 clinical A. salmonicida isolates with different levels of quinolone susceptibility were sequenced. MICs were determined in the presence of the efflux pump inhibitor (EPI) Phe-Arg ß-naphthylamide and E_max values (MIC without EPI/MIC in the presence of EPI) were calculated. Isolates fell into two classes: (i) those that had a wild-type gyrA gene with oxolinic acid MIC 0.5, flumequine MIC 1 and ciprofloxacin MIC 0.25 µg ml^–1; and (ii) those that had a single mutation in gyrA encoding Asp-87 Asn with oxolinic acid MIC 2, flumequine MIC 4 and ciprofloxacin MIC 0.125 µg ml^–1. No mutations were found in parC. High E_max values obtained for flumequine and oxolinic acid (up to 16 and 8, respectively, for the most resistant isolates of the two classes) indicated an important contribution of efflux to the resistance phenotype. Flumequine accumulation experiments confirmed that high E_max values were associated with a much lower level of accumulation. PCR/RFLP assays conducted on 34 additional isolates showed the presence of a mutation at codon 87 of gyrA in nearly all the quinolone-resistant isolates. This finding, together with PFGE typing results, strongly suggests a common clonal origin of these quinolone-resistant isolates.

The GenBank accession number for the partial sequence of the parC gene of A. salmonicida ATCC 14174 is AF473701.

	INTRODUCTION

TOP INTRODUCTION METHODS RESULTS DISCUSSION ACKNOWLEDGEMENTS REFERENCES

 
Aeromonas salmonicida is the causal agent of furunculosis, a debilitating and lethal disease affecting numerous farmed fish species including salmonids, turbot and carp (Noga, 2000). With daily mortality rates typically between 0.1 and 1 %, chronic furunculosis is responsible for massive economic losses for fish farms. Commercial vaccines and vaccination procedures have been developed for use in salmonid farming. However, they are still not widely used in France and offer poor protection in other fish species (Santos et al., 2003). Furunculosis is frequently treated orally with quinolone antibiotics, principally oxolinic acid and flumequine. In France, the recommended treatment durations are 5 and 6–7 days for flumequine and oxolinic acid, respectively, as reported in the specifications for products used to treat freshwater salmonids. Recommended dosages are 12 mg kg^–1 day^–1. Currently, extended treatment durations of up to 15 days and dosages up to 30 mg kg^–1 day^–1 have been observed in marine fish antibiotherapy for both quinolones. This use of quinolones has resulted in the emergence of A. salmonicida quinolone-resistant isolates, which have been isolated from diseased fish (Grant & Laidler, 1993; Hastings & McKay, 1987; Inglis & Richards, 1991; Oppegaard & Sorum, 1994; Tsoumas et al., 1989; Wood et al., 1986). In aquaculture, diagnoses are often presumptive and first-line treatments are administered in the absence of susceptibility data concerning the causative pathogen. The increasing prevalence of quinolone-resistant A. salmonicida strains has made the use of these antibiotics inappropriate in many cases. This misuse leads not only to therapeutic failures but also to the useless release of active ingredients into the aquatic environment, contaminating the surrounding water and sediments, either directly from uningested food or indirectly through excretion. Although the real selective pressure exerted by these antibiotics on the aquatic microflora is difficult to evaluate (Smith et al., 1994), many studies report an increase in the frequency of resistant bacteria isolated from water, sediments or wild animals in the vicinity of fish farms (Ervik et al., 1994; Guardabassi et al., 2000; Samuelsen et al., 1992; Schmidt et al., 2000; Spanggaard et al., 1993). Guidelines for a rational use of antibiotics such as those available for the treatment of bacterial infections in humans and land animals are still poorly developed in the case of fish diseases. In particular, susceptibility criteria, which should be taken into account when selecting an antimicrobial agent, are not clearly defined for fish-pathogenic bacteria. The definition of such criteria requires the acquisition and integration of clinical, microbiological, pharmacokinetic/pharmacodynamic and molecular data on the interactions between the fish, their pathogens and aquacultural antibiotics. Whereas the use of fluoroquinolones was recently banned in the USA in many food-animal productions, including aquaculture, it is still authorized in the European Union. Policies are needed particularly to regulate the recent but increasing use of veterinary fluoroquinolones in aquaculture. Indeed, sarafloxacin has been approved for oral use in fish, and other veterinary fluoroquinolones like enrofloxacin and marbofloxacin may be used ‘off-label’ under the prescription of veterinarians (Martinsen & Horsberg, 1995; FAO, 2002). The main concern is the risk of selecting bacterial human pathogens, which are commonly found in and near aquaculture systems, with a cross-resistance to the fluoroquinolones used in human medicine (Angulo, 2000; Lehane & Rawlin, 2000; Twiddy, 1995).

The aim of this study was to investigate the mechanisms of resistance to quinolones in strains of A. salmonicida isolated from cultured fish with furunculosis. We assessed: (i) the presence of mutations in the quinolone resistance-determining regions (QRDRs) of the gyrA and parC genes, encoding the A subunits of DNA gyrase and topoisomerase IV, respectively, the target enzymes for quinolones (Hooper, 2001); and (ii) the presence and contribution of an efflux mechanism (Poole, 2000). The results led us to investigate the epidemiological relationship among the isolates using PFGE.

	METHODS

TOP INTRODUCTION METHODS RESULTS DISCUSSION ACKNOWLEDGEMENTS REFERENCES

 
Bacterial strains and growth media.

A total of 46 A. salmonicida clinical strains and one reference strain, A. salmonicida ATCC 14174, were investigated. All clinical strains were isolated from diseased sea-water fish as part of an antibioresistance survey conducted between 1998 and 2000 in marine fish farms located on the west coast of France (Blanc et al., 2002). A first set of 12 clinical strains spanning various levels and patterns of resistance to quinolones was chosen for sequencing of the topoisomerase genes and efflux experiments. For PCR/RFLP experiments, 34 other strains were randomly sampled from two groups corresponding to two different clinical contexts. Some of these strains had been isolated in the same farm, during apparently independent furunculosis outbreaks that occurred at different times. Twenty of the strains had been isolated during furunculosis outbreaks that had been successfully treated with oxolinic acid or flumequine (defined as a decrease in daily mortality rates to stable values less than 0.1 %, obtained within a treatment time of less than 15 days) and were thus considered to be clinically susceptible. The other 14 strains had been isolated at the beginning of outbreaks for which the quinolone treatment had resulted in a therapeutic failure. They were thus considered to be clinically resistant. Considering the current treatment durations in France, the decrease in daily mortality rates within 15 days of treatment was retained as a preliminary indicator of a positive clinical outcome. All A. salmonicida strains were routinely grown at 22 °C on tryptic soy agar, on Mueller–Hinton agar or in Mueller–Hinton broth. Escherichia coli JM109 was used for cloning of the parC gene fragment of A. salmonicida ATCC 14174.

MIC determination.

For all strains, MICs of oxolinic acid, flumequine and ciprofloxacin were determined by the 2-fold agar dilution method according to the guidelines defined by the working group on the development of standard reference methods for antimicrobial agent susceptibility testing for bacterial fish pathogens (CEFAS, 1999). MICs of the three quinolones were also determined for all of the strains used in sequencing experiments, in the presence of serial 2-fold dilutions (1–256 µg ml^–1) of the efflux pump inhibitor (EPI) Phe-Arg ß-naphthylamide (Sigma). The plates were incubated for 48 h at 22 °C, after which E_max values were calculated. These were defined as the ratio between MIC without EPI and MIC in the presence of a maximal potentiating concentration of EPI (Lomovskaya et al., 2001). Analytical-grade oxolinic acid and flumequine were purchased from Sigma and ciprofloxacin from Bayer.

DNA isolation.

Chromosomal DNA was extracted from the strains by harvesting the cells from 1 ml of an overnight culture, resuspending the pellet in 100 µl sterile water and boiling for 3 min. After centrifugation for 3 min at 10 000 g, supernatants were collected and 1 : 100 dilutions in sterile water were used as a template for PCR.

Amplification and sequencing of the QRDR of the gyrA and parC genes.

The sequences of primers used in the PCRs are given in Table 1. Primers ASGYRA1 and ASGYRA2 were designed from the nucleotide sequence of the A. salmonicida gyrA gene (GenBank accession no. L47978). PCRs were carried out in a total volume of 25 µl containing 25 pmol of each primer, 200 µM dNTPs, 1.5 mM MgCl₂, 0.5 U Taq polymerase (Qiagen) and 5 µl of the template DNA dilution. Standard PCR conditions were used: 30 cycles of 94 °C for 1 min, 58 °C for 1 min and 72 °C for 1 min. The 663 bp amplicons were purified using a QIAquick Spin PCR Purification kit (Qiagen) and sequenced with the primers used for PCR by the fluorescent dideoxy chain-terminating method using a Perkin-Elmer 3700 capillary device (Genome Express).

An alignment of ParC amino acid sequences of various bacterial species, obtained using CLUSTAL W (http://www.infobiogen.fr/services), was used to design the degenerate primers ASPARC1 and ASPARC2, with the help of CODEHOP software (http://blocks.fhcrc.org/codehop.html). ParC sequences of the following species were included in the multi-alignment: E. coli (GenBank accession no. M58408), Salmonella enterica (NC_003197), Vibrio parahaemolyticus (AB023570) and Pseudomonas aeruginosa (AB003428). Amplification of a 447 bp amplicon overlapping the parC QRDR from A. salmonicida ATCC 14174 was possible using these degenerate primers. PCR conditions were the same as those described above, except that the annealing temperature was 68 °C. The PCR product was cloned into the pGEM-T Easy vector (Promega) and sequenced (Genome Express). Two specific primers, designated ASPARC3 and ASPARC4, were designed from this sequence to amplify the parC QRDR sequence of other A. salmonicida strains. PCR with primers ASPARC3 and ASPARC4 was carried out under the standard conditions described above, but with an annealing temperature of 60 °C. The 418 bp parC fragments were purified and sequenced directly, as for the gyrA sequence determination.

Sequence analysis.

The gyrA and parC QRDR sequences and the putative amino acid counterparts of the 12 clinical strains studied and of A. salmonicida ATCC 14174 were aligned and compared using CLUSTAL W. Percentage identities were calculated using LFASTA.

Detection of mutations at codon 87 by PCR/RFLP.

Primers ASGYRA3 and ASGYRA4 were used to amplify a 158 bp PCR product (Fig. 1). The forward primer, ASGYRA3, whose sequence is different by one base from the gyrA gene sequence, was designed to introduce an artificial Hpy188I cleavage site overlapping the first two bases of codon 87, according to the primer-specified restriction site modification method (Haliassos et al., 1989). PCR was performed as described for amplification of the sequenced fragments, with an annealing temperature of 58 °C. Three microlitres of the PCR product was digested with 1 U Hpy188I in a total volume of 8 µl. Digested PCR products were resolved in a 2.5 % agarose gel, alongside a lambda EcoRI/HindIII DNA marker. The presence of a mutation at codon 87 was detected by non-digestion of the 158 bp PCR product to two products of 122 and 32 bp (Fig. 2). This method makes it possible to detect all amino acid substitutions at position 87 with the exception of the Asp-87 Glu substitution (which is due to a mutation of the third base of codon 87, outside the Hpy188I restriction site).

View larger version (25K): [in this window] [in a new window]   Fig. 1. Schematic diagram of the PCR/RFLP assay used to detect mutations at codon 87 of the gyrA gene of A. salmonicida. The case of a wild-type strain is shown. The A C change directed by primer ASGYRA3 is shown in bold. The newly created Hpy188I restriction site in the PCR product is underlined.

View larger version (18K): [in this window] [in a new window]   Fig. 2. Results of the PCR/RFLP assay conducted on a reference strain and 12 clinical isolates of A. salmonicida. The 36 bp restriction product is not shown. Lane M, molecular mass marker; lane 1, strain ATCC 14174; lanes 2–13, strains 12, 15, 2, 5, 3, 56, 13, 47, 77, 74, 44 and 41, respectively. Fragment sizes (nt) are indicated.

Flumequine accumulation experiments.

Bacteria were grown to late exponential phase at 22 °C in Mueller–Hinton broth, harvested by centrifugation, washed in 50 mM sodium phosphate buffer (pH 7.0) and resuspended in the same buffer to an OD₆₀₀ of 4.0. The viable cells in this suspension were counted by plating the appropriate dilutions in triplicate on to tryptic soy agar plates after incubating overnight at 22 °C. The resuspended cells were equilibrated for 10 min at 22 °C in a water bath. Flumequine was then added to a final concentration of 10 µg ml^–1. Samples of 0.5 ml were taken 1, 2, 3 and 4 min after the addition of flumequine, diluted immediately in 1 ml ice-cold sodium phosphate buffer and centrifuged for 5 min at 5600 g. The pellet was washed once with 1 ml ice-cold buffer and resuspended in 1 ml 100 mM glycine hydrochloride (pH 3.0) for 15 h at room temperature. The samples were then centrifuged at 5600 g for 10 min at 4 °C. The concentration of flumequine in the supernatant was determined using an HPLC method adapted from Pouliquen & Pinault (1994). The results were expressed as ng flumequine incorporated (10⁹ c.f.u.)^–1.

PFGE.

Strains were grown on tryptic soy agar (Bio-Rad) at room temperature for 48 h. After a purity check, several colonies were resuspended in cell suspension buffer (100 mM Tris/HCl, 100 mM EDTA, pH 8) at an OD₆₁₀ of 1.35–1.4. A volume of 120 µl of the cell suspension containing 10 µg lysozyme (Sigma-Aldrich) was mixed with an equal volume of 1.6 % Incert agarose (BMA) and allowed to solidify in 100 µl moulds. Plugs were incubated in 1 ml lysis buffer A (6 mM Tris/HCl, pH 7.6, 1 M NaCl, 0.1 M EDTA, 0.5 % Brij 58, 0.2 % deoxycholate, 0.5 % Sarkosyl) with 1 mg lysozyme and 20 µg DNase-free RNase. After 2 h at 37 °C, the lysis buffer was removed and the plugs were washed with 1 ml TE buffer (10 mM Tris/HCl, 1 mM EDTA, pH 8.0) and incubated with 1 ml lysis buffer B (0.25 M EDTA, 20 mM ethylene glycol-bis(2-aminoethylether)-N,N,N',N'-tetraacetic acid, pH 9.0) supplemented with 500 µg proteinase K and 1 % Sarkosyl for 20 h at 50 °C. The plugs were washed three times with 1.5 ml TE buffer and three more times with distilled water and then digested with 30 U SpeI (Roche Diagnostics) overnight at 37 °C. Fragments of DNA were separated by PFGE in a 1 % agarose gel (Seakem Gold; BMA) in 0.5 x TBE buffer (0.045 M Tris/HCl, pH 8, 0.045 M boric acid, 0.01 M EDTA) using a CHEF-DR III (Bio-Rad). The running conditions were 6 V cm^–1 at 12 °C for 20 h, with pulse times ramped from 2.2 to 63.8 s. Lambda Ladder PFG Marker (New England BioLabs) was used as the molecular size marker.

The gel was stained in 1 µg ethidium bromide ml^–1 and the gel image was captured electronically using a video camera interfaced to a microcomputer (ImageMaster VDS; Amersham-Pharmacia Biotech).

Patterns were compared using the BIO 1D++ Software (Vilber-Lourmat), based on the Dice similarity coefficients. A dendrogram was deduced from the matrix of similarities using the unweighted pair group method with the arithmetic mean (UPGMA) clustering algorithm.

	RESULTS

TOP INTRODUCTION METHODS RESULTS DISCUSSION ACKNOWLEDGEMENTS REFERENCES

 
Characterization of 12 selected A. salmonicida isolates

(i) Quinolone MICs. In order to study the mechanisms of resistance to quinolones of A. salmonicida, we selected 12 clinical isolates from our strain collection whose level of resistance to quinolones ranged from full susceptibility to clinically significant resistance, with all the intermediate levels represented. Oxolinic acid MICs ranged from 0.03 to 8 µg ml^–1, flumequine MICs from 0.06 to 32 µg ml^–1 and ciprofloxacin MICs from 0.015 to 1 µg ml^–1 (Table 2). The fluoroquinolone ciprofloxacin was the most active against all isolates, followed by oxolinic acid and flumequine.

(ii) Nucleotide sequence of the gyrA and parC gene QRDRs. The nucleotide sequences of the gyrA QRDR of 12 clinical isolates and A. salmonicida ATCC 14174 were determined by PCR amplification and direct sequencing. The isolates could be separated into two groups: (i) those that had a wild-type gyrA gene QRDR (identical to that of A. salmonicida ATCC 14174) and showed oxolinic acid MICs 0.5 and flumequine MICs 1 µg ml^–1 and (ii) those that had an identical single nucleotide mutation in gyrA leading to an Asp-87 Asn (GAC AAC) substitution and showed oxolinic acid MICs 2 and flumequine MICs 4 µg ml^–1 (Table 2). There was no clear distinction between the ciprofloxacin MIC of isolates harbouring the gyrA mutation and those of isolates with a wild-type gyrA QRDR. Indeed, isolate 56, which had the gyrA mutation, appeared equally or more susceptible to ciprofloxacin than some other isolates (2, 3 and 5) with no gyrA mutation.

Using degenerate primers (Table 1), we amplified, cloned and sequenced the parC QRDR of A. salmonicida ATCC 14174. This nucleotide sequence served to define primers specific to A. salmonicida, which made possible the amplification and direct sequencing of the parC QRDR of the 12 selected clinical isolates. The parC QRDR sequences of all the isolates were 100 % identical to that of A. salmonicida ATCC 14174. The deduced amino acid sequences showed high identity with the corresponding ParC sequences of E. coli (88.1 %) and P. aeruginosa (90.3 %). Alignments over a 68 amino acid overlap with the ParC QRDR sequences of Aeromonas sobria and Aeromonas hydrophila (Goni-Urriza et al., 2002) revealed 100 and 98.5 % identity (one substitution), respectively.

(iii) Active efflux analysis. To evaluate the contribution of active efflux to the quinolone-resistant phenotype of the 12 selected isolates, we compared MICs obtained in the presence and absence of the EPI Phe-Arg ß-naphthylamide (Lomovskaya et al., 2001) (Table 2). We assumed that active efflux and target gene mutations were contributing independently to phenotypic quinolone resistance. Consequently, we considered that E_max (i.e. the largest quinolone MIC reduction induced by the EPI) provided a measure of the contribution of active efflux to quinolone resistance, independently of the gyrA gene status.

Depending on the isolate, E_max was obtained with concentrations of EPI between 64 and 256 µg ml^–1 and we verified that EPI alone had no inhibitory activity within this range (data not shown). E_max calculated for flumequine and oxolinic acid reached values of 16 and 8, respectively. This was true for the most resistant isolates with the Asp-87 Asn substitution (isolates 74, 44 and 41) as well as for the most resistant isolates harbouring a wild-type gyrA QRDR (isolates 2, 3 and 15). The EPI showed a weaker effect with fully susceptible strains (A. salmonicida ATCC 14174 and isolate 12) although it still induced hypersusceptibility to flumequine and oxolinic acid. In contrast, E_max for ciprofloxacin was as low as 2 for 10 out of the 12 isolates, including the most resistant. Surprisingly, this value was 4 for the two most susceptible strains, which were thus also made hypersusceptible to ciprofloxacin by the EPI.

Flumequine uptake experiments performed on four strains showed a correlation between E_max and levels of accumulation (Fig. 3). Indeed, strains 15 and 74, which had an E_max of 16, accumulated 10- to 20-fold less flumequine than A. salmonicida ATCC 14174, which had an E_max of 4. Intermediately, strain 56, with an E_max of 8, accumulated 3- to 4-fold less flumequine than did strain ATCC 14174.

View larger version (15K): [in this window] [in a new window]   Fig. 3. Accumulation of flumequine by strains ATCC 14174 (•), 15 (), 56 () and 74 ().

Detection of mutations at codon 87 by PCR/RFLP

Considering the correlation between the presence of a mutation at codon 87 of gyrA and the level of resistance to quinolones that occurred among the 12 initially selected isolates, we developed a PCR/RFLP assay to detect mutations rapidly at this codon (Fig. 1). When performed on the clinical and reference strains whose gyrA QRDR had been sequenced, the PCR/RFLP assay provided the expected restriction products (Fig. 2). The assay was further applied to a set of 34 clinically resistant or susceptible strains. The assay indicated the presence of a mutation at codon 87 for 13 of the 14 resistant strains. In contrast, the assay was negative for the 20 susceptible strains.

Strain typing by PFGE

PFGE patterns were obtained for all the clinical strains described in Table 2, except strain 41, which repetitively gave smeared patterns. All the strains displayed similar PFGE patterns, thus confirming the genetic homogeneity of typical A. salmonicida strains (Garcia et al., 2000) (Fig. 4). However, as previously shown by Chomarat et al. (1998), restriction by SpeI was discriminative, providing unique patterns for all the strains, except strains 47 and 77, which were indistinguishable. Cluster analysis clearly distinguished the patterns of clinical strains from that of strain ATCC 14174 at a similarity level of about 50 % (Fig. 4). Among the clinical strains, those harbouring the mutation at codon 87 of gyrA and those with the wild-type gyrA gene clustered in two separate groups at a similarity level of 70 %.

View larger version (59K): [in this window] [in a new window]   Fig. 4. SpeI PFGE patterns of the A. salmonicida isolates described in Table 1 and dendrogram based on the UPGMA clustering of Dice similarity coefficient (expressed as percentages).

	DISCUSSION

TOP INTRODUCTION METHODS RESULTS DISCUSSION ACKNOWLEDGEMENTS REFERENCES

 
Our study shows that the acquisition of resistance to quinolones by A. salmonicida is, as for other Gram-negative bacteria, mainly linked to the acquisition of mutations in the gyrA gene (Akasaka et al., 2001; Bachoual et al., 1998; Bagel et al., 1999; Cloeckaert & Chaslus-Dancla, 2001; Deguchi et al., 1997; Giraud et al., 1999; Nakano et al., 1997). However, it appears that the level of resistance can be significantly modulated by increases in efflux activity. Isolates sharing an identical status for the gyrA and parC gene QRDRs exhibited very different levels of resistance to quinolones. Although we cannot exclude the presence of quinolone resistance mutations outside the sequenced QRDR, our results show that these differences are probably due mainly to active efflux. Indeed, in strains with no mutation in the gyrA and parC QRDR, MICs measured in the presence of an EPI were restored to a full level of susceptibility, and even to a lesser level for some strains. This indicates that active efflux is not only involved in acquired quinolone resistance but also contributes significantly to the intrinsic level of activity of quinolones against susceptible strains. It has also been demonstrated that inhibition of efflux pumps significantly decreases the level of intrinsic resistance in P. aeruginosa (Lomovskaya et al., 2001). In resistant strains possessing the gyrA mutation, EPI could not decrease MIC below values that may be interpreted as the MIC caused by the presence of this mutation alone. It is interesting to note that MICs observed in strains with no mutations in the gyrA and parC QRDR but with high E_max values are close to or even higher than MICs theoretically conferred by the presence of this gyrA mutation. This suggests that efflux upregulation, which is associated with a substitution in the GyrA QRDR in most resistant strains, may also be responsible by itself for significant levels of resistance to quinolones. More directly, flumequine accumulation experiments performed on selected strains confirmed the much lower level of accumulation of strains showing high E_max values (i.e. high efflux activity).

Mutations at codon 83 and 87 of gyrA (or homologous codons) are the most commonly observed in quinolone-resistant strains of Gram-negative bacteria. In the most resistant strains, they are often combined with parC mutations (Akasaka et al., 2001; Bachoual et al., 1998; Bagel et al., 1999; Deguchi et al., 1997; Gibreel et al., 1998; Nakano et al., 1997). Furthermore, different mutations affecting codons 83 and 87 and thus resulting in different amino acid substitutions may be responsible for the resistance to quinolones. However, the mutation that we found in the gyrA QRDR of most of the resistant isolates was of a single type, leading to an Asp-87 Asn substitution in GyrA. This substitution has previously been identified in resistant strains of P. aeruginosa (Akasaka et al., 2001). However, only mutations at codon 83 have been found in the gyrA QRDR of Aeromonas strains so far (Goni-Urriza et al., 2002; Oppegaard & Sorum, 1994). In a recent study, mutations associated with quinolone resistance were investigated in riverine Aeromonas strains of mesophilic species other than A. salmonicida (Aeromonas caviae, A. hydrophila and A. sobria) (Goni-Urriza et al., 2002). All the resistant strains carried a point mutation leading to Ser-83 Arg or Ser-83 Ile substitutions. In addition, some strains harboured a parC mutation at codon 80 or 84. Oppegaard & Sorum (1994) identified a Ser-83 Ile substitution in quinolone-resistant A. salmonicida strains isolated from fish in Norway. The diversity of quinolone-resistance mutations (or combinations of mutations) in all the genera studied so far, and particularly in environmental Aeromonas strains, contrasts with the homogeneity of quinolone-resistant alleles that we observed in our study and that Oppegaard and Sorum observed in their study.

This homogeneity of quinolone-resistance mutations, together with the high similarity of the PFGE patterns, reveals a probable close epidemiological relationship between our isolates. Fry or fingerlings imported in fish farms from the same region are often purchased from a single producer. As a possible scenario, a clone possessing the Asp-87 Asn substitution may have spread from a single source and then evolved differently in the different fish farms, through the occurrence of novel selection events. This could explain the various quinolone-resistance phenotypes observed among these isolates. Also, repeated isolation of resistant A. salmonicida strains showing the same gyrA mutation in the same fish farms and over periods spanning several years should raise questions about the persistence of resistant strains in the farm structures. Confirmation of epidemiological relatedness may have considerable implications in terms of improvement of sanitary practices on fish farms.

	ACKNOWLEDGEMENTS

TOP INTRODUCTION METHODS RESULTS DISCUSSION ACKNOWLEDGEMENTS REFERENCES

 
This work was performed as part of the transversal project ‘Utilisation raisonnée des antibiotiques en élevage', supported by the General Management of INRA and led by Dr E. Chaslus-Dancla. We thank M. L. Morvan and Dr H. Pouliquen for their expertise in HPLC.

	REFERENCES

TOP INTRODUCTION METHODS RESULTS DISCUSSION ACKNOWLEDGEMENTS REFERENCES