Unexpected Occurrence of Plasmid-Mediated Quinolone Resistance Determinants in Environmental Aeromonas spp.

Identification of plasmid-mediated qnr genes outside Enterobacteriaceae underlines a possible diffusion of those resistance determinants within gram-negative rods.

Q uinolones are broad-spectrum antibacterial agents used in human and veterinary medicine. Their extensive use has been associated with a rising level of quinolone resistance (1). The 2 main mechanisms of quinolone resistance are chromosomally encoded, either a modifi cation of the quinolone targets with changes of DNA gyrase (gyrA) and/or of topoisomerase IV (parC) genes; or a decreased intracellular concentration due to impermeability of the membrane or to overexpression of effl ux pump systems (2). Plasmid-mediated quinolone resistance was fi rst identifi ed in a Klebsiella pneumoniae clinical isolate from the United States (3). It is mediated by a 218-aa protein, Qnr (lately termed QnrA), which belongs to the pentapeptide repeat family of proteins that protects DNA from quinolone binding to topoisomerases (4,5). QnrA confers re-sistance to quinolones such as nalidixic acid and increases MICs of fl uoroquinolones up to 32-fold in Escherichia coli (6). In addition, it enhances selection of associated chromosome-encoded quinolone resistance determinants that confer additional resistance to fl uoroquinolones (7). The QnrA determinants have been reported worldwide in many enterobacterial species, and 6 of them are known so far (QnrA1 to QnrA6) (8). Other plasmid-mediated quinolone resistance determinants, QnrB (QnrB1 to QnrB10) and QnrS (QnrS1 and QnrS2), have been identifi ed in enterobacterial species, sharing 41% and 60% amino acid identity with QnrA, respectively (8)(9)(10). The plasmid-mediated qnr genes have been identifi ed so far only in Enterobacteriaceae (6,8). Recent fi ndings indicated that those genes originate from environmental gram-negative bacterial species, such as Shewanella algae, the progenitor of the qnrA genes (11), and Vibrio splendidus, the progenitor of qnrS genes (12). We have shown that many Vibrionaceae species may harbor chromosome-encoded qnr-type genes (13).
To further evaluate the spread of plasmid-mediated resistance determinants in the environment, we have searched for those genes in water samples drawn from the Seine River in Paris, France. We identifi ed QnrS determinants in Aeromonas species in uncommon genetic environments.

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, Marnesla-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 Identifi cation of qnr-Positive Isolates
To identify the qnr-positive isolates, 1 mL of each qnrpositive 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 identifi cation 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 Amplifi cation
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 identifi ed 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 amplifi cation by using whole-cell DNA of qnr-positive isolates, as previously described (18).

Antimicrobial Drug-Susceptibility Testing
The antimicrobial drug susceptibility of Aeromonas isolates was fi rst determined by the disk diffusion technique on Mueller-Hinton (MH) agar plates according to Clinical and Laboratory Standards Institute guidelines (19). Thus, MICs of quinolones and fl uoroquinolones 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 ciprofl oxacin 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 pre-viously described (20). All transformants were confi rmed 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, purifi ed 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 identifi ed 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 10 3 to 10 5 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 identifi ed 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 identifi cation was fi rst attempted by sequencing of a ≈1,200-bp fragment of the 16S rDNA gene, but sequence analysis did not allow a precise identifi cation. Isolate A37 was identifi ed 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 identifi cation of members of the genus Aeromonas [15]) identifi ed 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 narrowspectrum 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 fl uoroquinolones (Table). However, A. media 42 exhibited higher resistance levels to fl uoroquinolones, 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
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 identifi ed 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 fl uoroquinolones (Table). No other antimicrobial resistance marker was carried by those natural plasmids. 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 identifi ed 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 identifi ed 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 amplifi cation that used specifi c 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 identifi ed in plasmid pFBAOT6 (23).

qnrS2 Gene as Part of a Mobile Insertion Cassette
The qnrS2 gene was identifi ed 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 identifi ed 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 defi nes 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 identifi ed 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.

Discussion
Our study identifi ed plasmid-mediated quinolone resistance QnrS determinants from water samples collected in different sites in a Paris river. To our knowledge, this is the fi rst identifi cation of plasmid-mediated QnrS determinants in nonenterobacterial species. Previous studies did not identify such qnr genes from tested gram-negative isolates that represented Campylobacter jejuni (27), Aeromonadaceae, Pseudomonadaceae, Xanthomonadaceae, Moraxellaceae, and Shewanellaceae (11). Identifi cation of QnrS-positive isolates at 2 collection sites in different water samples may highlight their relative persistence in the environment, at least in this area at that time.
The high-level resistance to quinolones and fl uoroquinolones might be due to mutations in type II topoisomerase genes because the mutations described in type II topoisomerases in A. media and A. punctata subsp. punctata have already been associated with resistance in Aeromonas spp (18). QnrS2 may confer low-level resistance to quinolones, as known in E. coli.
The qnrS1 gene has been identifi ed now from several enterobacterial isolates from Japan (9), Germany (28), the United Kingdom (29), the United States (25), France (17), Vietnam (30), Taiwan (31,32), and Denmark (33). The qnrS2 gene (92% amino acid identity with QnrS1) was identifi ed from a transferable IncQ-related plasmid (pGNB2) isolated from an activated sludge bacterial community of a wastewater treatment plant in Germany (24) and in a single non-Typhi Salmonella clinical isolate from the United States (25). Identifi cation of QnrS determinants in Aeromonas spp. indicates that those bacterial species may play a role as a reservoir of the qnrS genes in an aquatic environment, as already evidenced for tet genes (34,35). However, whether Aeromonas species are a main or an accessory reservoir of plasmid-mediated quinolone resistance determinants in regard to Enterobacteriaceae remains to be determined. For Aeromonas spp. to act as a reservoir of qnr genes it must be capable of acquiring these resistance genes from their progenitors (36) and transferring this genetic information to Enterobacteriaceae. QnrS2-positive plasmids p37 and p42 were not able to be transferred by conjugation to an E. coli host in vitro, but they were able to replicate in E. coli, indicating their broad host spectrum. In addition, our study demonstrated that this IncU-type plasmid-mediated qnrS2 gene was expressed and able to confer reduced susceptibility to quinolones, at least in E. coli. Aeromonas species and IncU plasmids, which are ubiquitous in a wide range of environments, might therefore act as important vectors for transfer of plasmid-mediated quinolone resistance determinants (23). As opposed to most qnrA and qnrB genes, qnrS genes have never been reported to be associated with sul1-type class 1 integrons (6,8). The qnrS1 gene has been identifi ed either upstream of Tn3-like transposon (9,28) or upstream of the insertion sequence ISEcl2 (17,37). In IncQ-related plasmid pGNB2 and in pMG308 from a Salmonella isolate, surrounding genetic structures of the qnrS2 genes were similar, with 2 open reading frames located immediately downstream of qnrS2, similar to repC and repA genes involved in plasmid replication (24,25) (Figure 2). In plasmid p37 from A. punctata 37, the genetic structure was different since the qnrS2 was part of a transposon-like structure and inserted in an open reading frame coding for a zinc MpR.
We have shown that plasmid integration of the qnrS2 gene may result from a peculiar transposition process that likely corresponds to trans-transposition. The qnrS2 gene was inserted in a peculiar mic. Such mic elements have been identifi ed rarely, e.g., mic231-like elements in Bacillus cereus, carrying in only 1 instance an antibiotic resistance gene, the fos gene encoding resistance to fosfomycin (26). This fi nding may indicate that mic elements might be clinically relevant and the origin of an additional gene plasticity in a bacterial species. These elements containing genetic features involved in gene dissemination (with a transposase likely acting in-trans) and expression may be also vehicles  for antibiotic resistance genes. This structure type may be added to the list of genetic tools at the origin of dissemination and expression of antibiotic resistance determinants. As previously described for the spread of a carbapenemase gene (bla IMI-2 ) in US rivers (20), this report underlines that the aquatic environment is an important reservoir of novel antibiotic resistance determinants. Quinolones are antimicrobial agents extensively used in aquaculture and are stable molecules in water (as opposed to β-lactams) (38). Thus, they may be the source of an important driving force for selection of quinolone resistance, which explains why QnrS2-positive plasmids did not possess any additional resistance determinants. Further studies might focus on the particular effect of quinolone use for inducing the qnrS2 gene mobility because it is known those molecules may induce bacterial repair systems and antibiotic resistance gene transfer (39).
We have previously shown that the qnrA and qnrS genes originate from water-borne bacterial species, S. algae and Vibrio splendidus, respectively. This identifi cation of a qnrS gene in another water-borne species, Aeromonas, further strenghens the role of water as a vehicle for spread of those resistance determinants.