Quinolone Resistance Mechanisms among Extended-Spectrum Beta-Lactamase (ESBL) Producing Escherichia coli Isolated from Rivers and Lakes in Switzerland

Sixty extended-spectrum β-lactamase (ESBL)-producing Escherichia coli isolated from rivers and lakes in Switzerland were screened for individual strains additionally exhibiting a reduced quinolone susceptibility phenotype. Totally, 42 such isolates were found and further characterized for their molecular (fluoro)quinolone resistance mechanisms. PCR and sequence analysis were performed to identify chromosomal mutations in the quinolone resistance-determining regions (QRDR) of gyrA, gyrB, parC and parE and to describe the occurrence of the following plasmid-mediated quinolone resistance genes: qepA, aac-6′-Ib-cr, qnrA, qnrB, qnrC, qnrD and qnrS. The contribution of efflux pumps to the resistance phenotype of selected strains was further determined by the broth microdilution method in the presence and absence of the efflux pump inhibitor phe-arg-β-naphthylamide (PAβN). Almost all strains, except two isolates, showed at least one mutation in the QRDR of gyrA. Ten strains showed only one mutation in gyrA, whereas thirty isolates exhibited up to four mutations in the QRDR of gyrA, parC and/or parE. No mutations were detected in gyrB. Most frequently the amino-acid substitution Ser83→Leu was detected in GyrA followed by Asp87→Asn in GyrA, Ser80→Ile in ParC, Glu84→Val in ParC and Ser458→Ala in ParE. Plasmid-mediated quinolone resistance mechanisms were found in twenty isolates bearing QnrS1 (4/20), AAC-6′-Ib-cr (15/20) and QepA (1/20) determinants, respectively. No qnrA, qnrB, qnrC and qnrD were found. In the presence of PAβN, the MICs of nalidixic acid were decreased 4- to 32-fold. (Fluoro) quinolone resistance is due to various mechanisms frequently associated with ESBL-production in E. coli from surface waters in Switzerland.


Introduction
Quinolones and fluoroquinolones (FQ) are broad-spectrum antimicrobial compounds and of critical importance in clinical use. They are the antimicrobials of choice in the treatment of acute gastrointestinal infections, urinary tract infections and respiratory tract infection. Resistance to FQ drugs increased in past years following the extensive and widespread use.
The mechanisms of action of (F)Qs is due to inhibition of DNA replication and resistance to (F)Qs is either based on (i) mutational alterations in the FQ target enzymes: DNA topoisomerase II (DNA gyrase) and topoisomerase IV, or (ii) plasmid-mediated quinolone resistance (PMQR) mechanisms or simply (iii) the decreased uptake of the drug due to the loss of a membrane-bound porin and/or drug extrusion via efflux pumps [1,2]. PMQR mechanisms include target protection proteins of the qnr gene family, enzymatic modification gene aac-(69)-Ib-cr and the efflux pump gene qepA. So far, five different qnr genes with a number of variations have been described: qnrA, qnrB, qnrC, qnrD and qnrS [3][4][5][6][7]. The Qnr proteins belong to the pentapeptide family of proteins and share common structural properties. AAC(69)-Ib-cr differs from the aminoglycoside acetyltransferase, AAC(69)-Ib, by two specific codon exchanges Trp102RArg and Asp179RTyr, which have been found to be necessary and sufficient for the ciprofloxacin (CIP) resistance phenotype [8]. The PMQR mechanisms confer only low level resistance to quinolones and thus, they were thought to enable the occurrence of chromosomal mutation which then lead to clinically relevant resistance levels [9].
In a previous study [10], 42 of 60 extended-spectrum blactamase-producing Escherichia coli strains isolated from rivers and lakes additionally displayed increased levels of nalidixic acid (NA) and/or CIP resistance.
The spread of E. coli co-expressing (fluoro)quinolone resistance along with extended-spectrum b-lactamases into the environment, especially into water bodies such as rivers, streams, wastewater effluents and lakes is worrisome. Rivers are considered to be of special importance as a reservoir of resistance genes since they are recipients of bacteria from different sources, e.g., wastewater plants, water of urban or industrial effluents, or agricultural activities [11].
The aim of this study was to identify and characterize quinolone and fluoroquinolone resistance mechanisms in higher generation cephalosporin resistant E. coli isolated from rivers and lakes in Switzerland.

Bacterial Isolates
Forty-two of 60 extended-spectrum b-lactamase-producing Escherichia coli strains isolated from rivers and lakes during a previous study and additionally displaying increased levels of nalidixic acid (NA) and/or CIP resistance were selected for this study. The protocol for collecting the water samples and isolating the strains as well as the species identification, antibiotic sensitivity testing and the characterization of the ESBL-types were described in the previous study [10].

MIC Determination of Nalidixic Acid and Ciprofloxacin
MICs for NA and CIP of all 42 strains were determined using Etest strips (bioMérieux, Marcy l'Etoile, France). Of selected strains, MICs for NA (Sigma-Aldrich, Switzerland) were determined by the broth microdilution method according the Clinical and Laboratory Standards Institute (CLSI) [12], in the absence and presence of 40 mg/ml phe-arg-b-naphthylamide (PAbN) (Sigma-Aldrich, Switzerland). As recommended by the CLSI, the E. coli ATCC25922 was included as quality control strain.

Identification of Quinolone Resistance Determinants and DNA Sequencing
DNA was extracted by a standard heat lysis protocol. Amplification of the quinolone resistance-determining regions (QRDRs) of the gyrA, gyrB, parC and parE genes and PCR based detection of the plasmid-mediated quinolone resistance (PMQR) markers (qnrA, qnrB, qnrC, qnrD, qnrS, aac(69)-Ib-cr and qepA) was performed using primers listed in Table 1. Custom-sequencing was performed by Microsynth (Balgach, Switzerland) and the nucleotide and translated protein sequences were analyzed with CLC Main Workbench 6.6.1. For database searches the BLASTN and BLASTP programs of NCBI (http://www.ncbi.nlm.nih.gov/ blast/) were used.

Replicon Typing
Plasmids were assigned to incompatibility groups on the basis of the presence of specific replicon sequences identified by PCR [14].

Results
Of the total of 60 ESBL-producing E. coli isolates, 42 coexpressing (F)Q resistance were further characterized. Based on the Etest assay, two strains showed intermediate resistance to NA, nine were classified as resistant to NA, two were resistant to NA and showed intermediate resistance to CIP and 29 were resistant to NA and CIP.
In 40/42 (95.2%) of the investigated isolates, at least one point mutation leading to amino acid substitution in any of the quinolone resistance-determining regions (QRDR) of gyrA, parC and parE was detected ( Table 2). No mutations were identified in gyrB. High-level resistance to NA was always attributed to a single mutation in gyrA, whereas strains resistant to both tested (F)Qs exhibited at least one point mutation in two different topoisomerase genes; most frequently in gyrA and parC. A total of ten strains showed only one point mutation, nine had the predominate substitution in GyrA Ser83RLeu and one a Ser83RVal exchange. One strain (OW55E1) showed one substitution in GyrA (Ser83RLeu) and one in ParC (Ser80RArg). Six strains had the following set of substitutions: GyrA (Ser83RLeu and Asp87RAsn), and ParC (Ser80RIle). OW86E2 showed the same substitutions in GyrA, but in ParC (Glu84RVal). Twenty-one strains showed four substitutions in at least two different proteins: fourteen had substitutions in GyrA (Ser83RLeu, Asp87RAsn) in ParC (Ser80RIle) and in ParE (Ser458R Ala), five strains in GyrA (Ser83RLeu, Asp87RAsn) in ParC (Ser80RIle, Glu84RVal), OW54E1 in GyrA (Ser83RLeu, Asp87RAsn) in ParC (Ser80-RIle, Glu84RGly) and OW60C2 in GyrA (Ser83RLeu, Asp87RAsn) in ParC (Ser80RIIe) and ParE (Glu460RAsp), respectively. Strain OW1C2 exhibited unique substitutions in GyrA: besides the frequently substituted position Ser83, here (Ser83RTre), also positions Ile112RVal, Leu127RMet, Ala128R Ser and Lys154RArg were affected. Additionally, substitutions in ParC (Ser80RIle) and in ParE (Ser458R Ala) were detected.
Intermediate resistance to NA was associated either with the presence of the plasmid-mediated resistance protein QnrS1 since in two strains (OW54E2 and OW60E1) chromosomal mutations and other PMQR determinants were absent, or with the presence of general efflux pumps.
Conjugation experiments revealed that two of the four qnrS1, three of five selected aac-69-Ib-cr but not the qepA bearing plasmid were transferable to the recipient strain E. coli HK225.
Plasmid incompatibility typing of the transconjugants revealed that both the qnrS1 and aac-69-Ib-cr genes were present on IncF plasmids. Furthermore, all transferred PMQR determinants were detected on the same plasmids as the bla ESBL genes. The contribution to FQ resistance of the plasmids bearing the PMQR determinates was measured by the difference in the MICs (NA and CIP) of the recipient strain (E. coli HK225) compared to the Table 2. Cont. transconjugants. The effect on the resistance to NA was minimal since the MICs increased only 1-3-fold. However, the resistance to CIP increased by factors of 4 to 16 in the presence of PMQR determinant harboring plasmids (Table 3).
In the presence of the efflux pump inhibitor PAbN, the MICs of NA decreased 4 to 32-fold. This finding confirmed that also efflux pump activity is involved in quinolone resistance (Table 4).

Discussion
So far, there are only few studies that have investigated the occurrence of (fluoro)quinolone resistant Enterobacteriaceae in surface waters [15][16][17][18]. Rivers and lakes are known to be a relevant putative reservoir of multiresistant bacteria since they collect surface waters containing material from different origins, e.g., wastewater plants, water from urban or industrial effluents, agricultural activities, and they harbor an immense antibiotic resistome, distributed to pathogenic and nonpathogenic bacteria [19].
FQ resistance in E. coli is associated predominantly with DNA topoisomerase II (DNA gyrase) and topoisomerase IV involving specific mutations in the QRDRs of gyrA and parC genes [2]. A single mutation in the QRDR of gyrA seems to be sufficient to generate a high-level quinolone, and a double mutation even a high-level fluoroquinolone resistance phenotype [20]. It is supposed that these mutations occur in a stepwise manner [2]. The effect of these mutations may alter the formation of hydrogen bonds, and the negative charge of the amino acids seem to be important for quinolone interactions with the DNA-gyrase-DNA complex [21].
As in previous studies, we mainly detected the substitutions in terms of amino acid positions GyrA Ser83RLeu and GyrA Asp87RAsn, which are located within the QRDR. Interestingly, OW1C2, besides harboring a substitution at position GyrA Ser83RVal, contained additional substitutions at position 112, 127, 128 and 154. So far, these four substitutions have been found in a Raoultella planticola and a Klebsiella oxytoca isolate (except for position 154), as shown by sequence alignments using the NCBI BLASTP program (accession-numbers: BAN58234.1 and CAQ76691.1). Hitherto, very few studies have focused upon the role of amino acid substitutions outside the QRDR [22], hence little is known about their effect on resistance levels. Our finding suggests that in addition to the standard QRDR, additional sites in GyrA seem to play a role in FQ resistance. Further studies thus need to be conducted in order to further characterize the role of rare substituions in GyrA.
In clinical E. coli isolates, quinolone resistance is mainly associated with mutations in gyrA whereas mutations in the topoisomerase IV genes, parC and parE, play a secondary role [2]. To date, mutations within parC and parE are always linked to mutations in gyrA, and it is supposed that mutations within these genes only occur if the strain already posses a reduced sensitivity to fluoroquinolones by mutations in DNA gyrase. Therefore, additional mutations in parC and parE are supposed to play an important role in the formation of highly resistant strains [2,23]. As many as 71% of the isolates described in this study, in addition to showing one to two substitutions in GyrA, harbored substitutions in ParC. Interestingly, the substitutions in ParC were found predominantly at position Ser80, which, apart from substitutions in GyrA, is one of the most common alterations in FQ-resistant clinical isolates [24]. This finding is suggestive of the anthropogenic origin of these strains that were isolated from the aquatic environment.
PMQR is a phenomenon of increasing importance. The most relevant PMQR genes known to date are the acetyltransferase gene aac(69)-Ib-cr, the qnr-like genes and genes encoding efflux pumps, such as qepA [25,26].
In the isolates described in this study, PMQR determinant aac(69)-Ib-cr was the most prevalent (36%), followed by qnrS1 (10%) and qepA (2%). The relatively high prevalence of aac(69)-Ib-cr is not surprising since several reports showed that this gene is often present on the same plasmid as bla ESBL [27,28]. QnrS-producing strains were frequently described in aquatic environments, and -as shown by a study by Cattoir and Co-workers [29] -Vibrio splendidus, an aquatic Gram-negative bacteria, was identified as a possible source for qnrS-like genes. Interestingly, in a previous study by Picão et al. [30], a qnrS2 bearing Aeromonas allosaccharophila strain was isolated from Lake Lugano in the southern part of Switzerland. QnrS is found very rarely and only sporadically in clinical isolates in Europe [31]. Further, Liassine et al. [32] investigated human ESBL-producing Enterobacteriacae isolates for qnrA, qnrB and qnrS collected by a Swiss clinical microbiology laboratory. However, from 155 analyzed isolates, they found only qnrB4 (3 Citrobacter freundii, 1 Klebsiella pneumonia and 1 E. coli) and qnrB2 (1 E. coli). In contrast to our data, no qnrS variants were found in that study, indicating that qnrS is not characteristic for FQ-resistant human isolates in Switzerland.
Our conjugation experiments revealed that qnrS1, and aac-69-Ibcr harboring plasmids were transferable to FQ-susceptible strains. The resistance phenotypes of the resulting transconjugants confirmed previous observations that the contribution of PMQR  Table 4. MIC values of nalidixic acid in the presence and absence of Phe-Arg-b-naphthylamide in selected E. coli isolates. determinants to fluoroquinolone resistance levels of the plasmidbearing strains is minor compared with the chromosomal mutations in DNA gyrase. The proposed role of these genes is that they enhance resistance to fluoroquinolones, supplementing chromosomal resistance genes [33]. Poor transfer of qnrS-encoding plasmids is confirmative of a previous study describing a qnrSharboring clinical E. coli isolate [31]. Nevertheless, our results demonstrate that qnrS determinants are indeed transferable and capable of conferring very low level resistance, also in the absence of chromosomally-encoded (F)Q-resistance determinants. Replicon typing of the transconjugants revealed that both qnrS1 and aac-69-Ib-cr are present on IncF plasmids. Additionally, the genes coding for ESBL were detected on the same plasmids. CTX-M-15, the most prevalent CTX-M-type isolated from water bodies in Switzerland, is often associated with aac(69)-Ib-cr resistance genes and has been located frequently on IncF group plasmids [34]. IncF type plasmids carrying CTX-M-15 are an inhomogeneous group of plasmids with sizes that vary between 50-200 kb and often carry more than one replicon (e.g., repFII, repFIA and/ or repFIB) [34]. This multi-replicon plasmid-genotype was also observed in our study.
Besides chromosomal mutations and PMQR determinants, efflux pumps can also contribute in a more unspecific way to the (F)Q resistance phenotype. The effect of members of the resistance-nodulation-division (RND) family of membrane transporters can be investigated by inhibiting them with PAbN [35,36]. In this study, the MICs of NA decreased 4 to 32-fold in the presence of the inhibitor, which clearly confirmed that these efflux pumps play an important role in quinolone resistance. Sáenz et al. [36] could already show that a single amino acid exchange in GyrA was not enough to confer a high level NA resistance phenotype in E. coli.
Recently, Sato et al. [37] investigated an E. coli isolate whose MICs of CIP, levofloxacin and norfloxacin exceeded the respective CLSI breakpoints, although it had no mutations in the QRDRs of either DNA gyrase or topoisomerase. The conclusion of their study was that FQ resistance can also arise due to concomitant acquisition of several PMQR determinates (in their study: QnrS1 and OqxAB, in combination with overexpression of acrAB-TolC and other chromosome-encoded efflux pumps). Our study yielded no isolate of this kind.
In summary, the acquisition of a high-level (F)Q resistance phenotype often includes combination of multiple mechanisms such as acquisition of mutations in DNA gyrase and topoisomerase genes as well as enhanced efflux and these processes can be accelerated by the presence of PMQR-determinants [2]. Our findings suggest that this holds true also for our FQ-resistant strains isolated from the aquatic environment.
It is possible that as a consequence of excretion of unmetabolized quinolones into wastewater by humans and animals, the environmental quinolone accumulation has contributed to the high prevalence of quinolone resistant E. coli in the water environment [9,38].
This study provides further evidence for the importance of the water-related environment as a reservoir for this emerging threat of FQ resistance. Moreover, the resistance characteristics of the isolates described in this study indicate that multilateral exchange of genetic material between bacteria of both anthropogenic and environmental origins is currently occurring and presents a phenomenon of growing importance. It is, finally, worrisome that more than two thirds of ESBL producers carry resistance determinants against (F)Qs, thus defying the two currently most popular antibiotic options for human therapy, b-lactams and FQs.