Water Sampling and Screening for qnr Genes

Water samples (40 mL each) were collected in November 2006, ≈0.2 m below the water surface, by immersion of 50-mL sterile, screw-capped tubes. Six distinct urban sites located on the Seine River in Paris were sampled: center (site 1), north (site 2), east (site 3), south (site 4), and west (sites 5 and 6). The Seine crosses the city of Paris (2. 4 million inhabitants) from southeast to northwest.

Samples were stored at 4°C before DNA extraction. To estimate the total number of viable bacteria, we diluted all samples in 0.85% saline solution and plated them (0.2 mL) on chromogenic URISelect4 agar (Bio-Rad, Marnes-la-Coquette, France) without any antimicrobial agent, which allowed visual differentiation of gram-negative species. After enrichment of 1 mL of water sample in 10 mL of trypticase soy (TS) broth for 48 h at 30°C and 37°C, qnrA-, qnrB- and qnrS genes were detected by using a multiplex PCR-based strategy (see below).

Isolation and Identification of qnr-Positive Isolates

To identify the qnr-positive isolates, 1 mL of each qnr-positive water sample was grown in 10 mL of TS broth containing nalidixic acid (16 mg/L) for 24–48 h at 30°C and 37°C. After dilution in 0.85% NaCl, each broth was plated (0.2 mL) on chromogenic URISelect4 agar and 5% horse-blood TS agar, both containing nalidixic acid (16 mg/L). Finally, each colony was tested by multiplex PCR for detection of qnr genes (see below). Further identification of qnr-positive isolates was performed by conventional biochemical techniques (API-20E and API-NE systems [bioMérieux, Marcy-l’Etoile, France]), and by sequencing of 16S rDNA and gyrB genes, as previously described (14,15).

Isolation of Total DNA and PCR Amplification

Whole-cell DNA was extracted from water samples by using QIAamp DNA Mini Kit (QIAGEN, Courtaboeuf, France), according to the manufacturer’s recommendations. Whole-cell DNA from isolated colonies identified after culture was extracted by using the boiling technique, which includes a heating step at 100°C of a single colony in a volume of 100 μL of distilled water followed by centrifugation of the cell suspension.

Screening of the qnrA-, qnrB-, and qnrS genes in water samples and from isolated colonies was conducted by using a multiplex PCR-based technique able to amplify the known Qnr variants, as previously described (16). E. coli Lo, K. pneumoniae B1, and E. coli S7 strains were used as qnrA-, qnrB-, and qnrS-positive controls, respectively (16,17). PCR amplicons were sequenced on both strands (see below). The chromosome-encoded quinolone-resistance determining regions (QRDRs) of gyrA and parC genes were sequenced after PCR amplification by using whole-cell DNA of qnr-positive isolates, as previously described (18).

PCR assays were also performed by using standard conditions of amplification (17) to detect genes coding for the replication protein of IncU-type plasmids (rep gene) and the tetracycline resistance determinant (encoded by the tetA gene). The primers, designed for this study, are as follows: for rep, Rep-IncU-F (5′-CTGGCTGAAATGCTGTTGCC-3′) and Rep-IncU-R (5′-GCTTCATAGGCTTCACGCTC-3′) to give a 1,199-bp product, and for tetA, TetA-1 (5′-GTGAAACCCAACAGACCCC-3′) and TetA-2 (5′-TCAGCGATCGGCTCGTTGC-3′) to give a 589-bp product.

Antimicrobial Drug–Susceptibility Testing

The antimicrobial drug susceptibility of Aeromonas isolates was first determined by the disk diffusion technique on Mueller-Hinton (MH) agar plates according to Clinical and Laboratory Standards Institute guidelines (19). The disks were supplied by Bio-Rad Laboratories, and the following antimicrobial agents were tested: amoxicillin (25 μg), ticarcillin (75 μg), amoxicillin-clavulanate (20/10 μg), ticarcillin-clavulanate (75/10 μg), piperacillin (75 µg), piperacillin-tazobactam (75/10 μg), cephalothin (30 μg), cefuroxime (30 μg), cefoxitin (30 μg), cefotaxime (30 μg), ceftazidime (30 μg), aztrem (30 μg), latamoxef (30 μg), cefepime (30 μg), imipenem (10 μg), kanamycin (30 IU), tobramycin (10 μg), gentamicin (15 μg), netilmicin (30 μg), amikacin (30 μg), nalidixic acid (10 μg), norfloxacin (5 μg), ofloxacin (5 μg), ciprofloxacin (5 μg), chloramphenicol (30 μg), tetracycline 30 (IU), fosfomycin (50 μg), rifampicin (30 μg), sulfonamide (200 μg), trimethoprim (5 μg), and colistin (50 μg).

Thus, MICs of quinolones and fluoroquinolones were determined by using the E test technique, according to the manufacturer’s recommendations (AB Biodisk, Solna, Sweden). MIC breakpoints were retained for determining susceptibility and resistance ranges to nalidixic acid and ciprofloxacin were <8 and >32 mg/L and <1 and >4 mg/L, respectively, as recommended for Enterobacteriaceae (19).

Transfer of Resistance and Plasmid Analysis

Transfer of the plasmid-mediated quinolone resistance markers from qnr-positive Aeromonas isolates to E. coli TOP10 was attempted by using electroporation and conjugation techniques (20). E. coli TOP10 and azide-resistant E. coli J53 strains were recipient strains for transformation and conjugation experiments, respectively (20). Conjugation experiments were performed at 22°C and 37°C, as previously described (21). Transformants and transconjugants were selected on MH agar plates containing nalidixic acid (3 mg/L) only or containing azide (100 mg/L) plus nalidixic acid (6 mg/L), respectively. Plasmid extraction was performed from each qnr-positive isolate and its corresponding transformants by using the Kieser technique (22). Plasmid sizes were determined by electrophoresis on an agarose gel and comparison with sizes of reference plasmids (164, 66, 38, and 7 kb) of E. coli NCTC 50192, as previously described (20). All transformants were confirmed to be qnr positive by mulitplex PCR (see above).

Cloning experiments were performed with EcoRI-restricted whole-cell DNA of Aeromonas isolates 37 and 42, followed by ligation of DNA fragments in the EcoRI-site of cloning vector pBK-CMV. Recombinant plasmids were then transformed by electroporation into E. coli TOP10 electrocompetent cells. E. coli TOP10 harboring recombinant plasmids were selected on MH agar plates containing kanamycin (30 mg/L) and nalidixic acid (3 mg/L). All clones were tested as carrying qnr gene by multiplex PCR (see above), and cloned fragments of recombinant plasmids were sequenced by primer walking (see below).

Sequencing and Bioinformatic Analysis

PCR products, purified with a Qiaquick PCR Purification Kit (QIAGEN), and recombinant plasmids were sequenced by using an Applied Biosystems sequencer (ABI377, Foster City, CA, USA). The nucleotide sequences and the deduced protein sequences were analyzed with BLAST software (www.ncbi.nlm.nih.gov/BLAST).

QnrS2 Determinant from Aeromonas spp.

Samples extracted from 2 of the 6 sites (sites 3 and 4) from the Seine River in autumn 2006 were PCR positive for qnrS genes. Sequencing of the amplicons identified a qnrS2 gene in both cases. Further samples were therefore collected 1 week later from the same collecting sites and assessed the permanent occurrence of QnrS2- positive isolates in that river at that time. No qnrA and qnrB genes were detected.

The total number of gram-negative bacteria from water collection varied from 10³ to 10⁵ CFU per mL, depending on the sampling site, and included ≈10%–50% of isolates belonging to Aeromonas species (data not shown). The QnrS2-positive isolates were identified as Aeromonas species according to their phenotypic characterization. Strain 37 (site 4, pink colonies on URI4Select agar) and strain 42 (site 3, blue colonies on URI4Select agar) were selected as qnrS2-positive Aeromonas isolates to be further characterized. Genotypic identification was first attempted by sequencing of a ≈1,200-bp fragment of the 16S rDNA gene, but sequence analysis did not allow a precise identification. Isolate A37 was identified as A. trota or A. punctata (formerly A. caviae) (only 1 nt difference) and isolate A42 as A. hydrophila or A. media (only 1 nt change). Because gyrB sequence divergence is greater than that of 16S rDNA, phylogenetic analysis based on gyrB sequence (allowing more reliable identification of members of the genus Aeromonas [15]) identified isolate A37 as A. punctata subsp. punctata and isolate A39 as A. media.

The Aeromonas isolates both displayed a wild-type resistance phenotype to β-lactams with resistance to narrow-spectrum penicillins and remained susceptible to several β-lactam molecules, including broad-spectrum cephalosporins and carbapenems. A. media 42 was also resistant to several aminoglycosides (kanamycin, tobramycin), chloramphenicol, and tetracycline, whereas A. punctata 37 was fully susceptible.

Aeromonas isolates 37 and 42 were resistant to nalidixic acid (MIC >256 mg/L) and to fluoroquinolones (Table). However, A. media 42 exhibited higher resistance levels to fluoroquinolones, with MICs 2- to 8-fold higher than those for A. punctata 37 (Table). Sequence analysis of the QRDR regions of gyrA and parC genes showed that A. punctata 37 had 1 aa substitution, Ser83Ile in GyrA, whereas A. media 42 had 2 aa substitutions, Ser83Ile in GyrA and Ser80Ile in ParC, as compared with the wild-type proteins of Aeromonas species (18).

qnrS2 Gene on IncU-Plasmid Backbone

Figure 1

Figure 1. Plasmid DNAs from Aeromonas punctata 37 and A. media 42 and their Escherichia coli TOP10 transformants (TF) carrying plasmids p37 or p42. Lanes: 1, A. punctata 37; 2, E. coli TOP10/p37...

The plasmid-mediated quinolone resistance QnrS2 determinant was transferred from Aeromonas isolates 37 and 42 to E. coli TOP10 recipient strain by electrotransformation, but repeated conjugation experiments failed. Plasmid analysis identified a single 55-kb plasmid (p37) and a single 20-kb plasmid (p42) from E. coli transformants from A. punctata 37 and A. media 42 isolates, respectively (Figure 1). After they were transferred into E. coli TOP10, plasmids p37 and p42 conferred increased MICs of quinolones and fluoroquinolones (Table). No other antimicrobial resistance marker was carried by those natural plasmids.

Figure 2

Figure 2. Genetic environments of the qnrS2 gene in plasmid p37 from Aeromonas punctata 37 and comparison with related plasmid structures. Plasmid pFBAOT6 is from A. punctata from the United Kingdom (

Cloning of EcoRI-restricted DNA from whole-cell DNA of A. punctata 37 produced a recombinant plasmid, pAS37, containing a 19,050-bp insert that contained the qnrS2 gene. Sequencing showed that the qnrS2 gene was located in a plasmid-encoded genetic structure previously identified in an IncU-related plasmid, pFBAOT6 (84,748 bp), isolated from an A. punctata strain recovered from hospital sewage in Kendal (United Kingdom) in 1997 (23). This region consisted of 15 open reading frames (Figure 2). Detailed analysis of pAS37 showed that a fragment of 1,375 bp containing the qnrS2 gene (657 bp) was inserted within the mpR gene coding for a putative zinc metalloprotease (MpR) (Figure 2). Sequencing of the full insert in plasmid pAS37 identified the same rep region as reported from the IncU-related pFBAOT6 plasmid, which indicated that p37 was a member of the IncU incompatibility group. PCR amplification that used specific primers of this rep gene also gave positive results from whole-cell DNA of E. coli (p42) transformant, indicating that p42 also belonged to IncU-type plasmid family. In addition, cloning of EcoRI-restricted DNA from whole-cell DNA of A. media 42 showed that the qnrS2 gene was inserted into the mpR gene in plasmid p42. No tetA gene encoding resistance to tetracycline was detected by PCR in plasmids p37 and p42, whereas it was identified in plasmid pFBAOT6 (23).

qnrS2 Gene as Part of a Mobile Insertion Cassette

The qnrS2 gene was identified inside a 1,375-bp structure bracketed by two 22-bp imperfect inverted repeats (4 mismatches) (Figure 2). This structure ressembled a transposon; however, no transposase encoding gene was associated with the qnrS2 gene, the single gene identified in this genetic structure. Acquisition of that structure was likely the result of a transposition process because it was bracketed by a 5-bp duplication of the target site (CCTCC) that might be likely considered as the signature of transposition. Thus, this genetic structure defines a mobile insertion cassette (mic) containing the qnrS2 gene instead of a transposase, as observed in insertion sequences. Those mic elements are not related to the gene cassette associated with class 1 integrons (26). The mic-qnrS2 element identified here carried putative promoter sequences able to enhance the qnrS2 expression made of a –35 box (TTCTCT) and a –10 box (TAACTT) separated by a 17-bp sequence.