Introduction

Quinolones, a class of synthetic broad-spectrum antimicrobial agents, have been used for treatment of bacterial infections for several decades. They bind specifically to the bacterial type II topoisomerases and form a drug-enzyme-DNA complex to interfere with DNA replication and kill the bacteria [31]. Extensive and excessive use of quinolones in medication stimulates the propagation of antibiotic-resistant clinical pathogens. More than 50% of Escherichia coli strains isolated from Chinese hospitals in Shanghai were resistant to ciprofloxacin [80]. Quinolones are also used in veterinary, agriculture, and aquaculture in some developing countries. In Chile, more than 100 tons of quinolones were used annually as veterinary medicines, nearly ten times of those used as human medicines [6]. Most of the applied quinolones may end up into the environment eventually [21, 44, 62, 76], creating a huge selective pressure on environmental bacteria, which may, in turn, form a huge reservoir and dissemination source of quinolone resistance genetic determinants.

The molecular mechanisms of quinolone resistance involve both chromosome- and plasmid-mediated resistance. Chromosomal mutations of gyrases, type IV topoisomerases, outer membrane proteins, membrane efflux pumps, and other regulatory proteins may reduce the affinity of quinolones to their targets or reduce the accumulation of quinolones inside bacterial cells [30]. Plasmid-mediated quinolone resistance (PMQR) involves target protection by pentapeptide-repeat Qnr (plasmid-mediated quinolone resistance) proteins, enzymatic inactivation by aminoglycoside acetyltransferase AAC(6′)-Ib-cr, and active efflux by QepA or OqxAB pump, with the qnr genes recognized as the most common PMQR determinants [63]. PMQR dramatically accelerates the global propagation of quinolone-resistant pathogens. Similarly, PMQR may also constitute an important mechanism for the transmission of quinolone resistance in environment.

The first qnr gene, designated qnrA, was discovered on a transferable plasmid pMG252 in a multidrug-resistant clinical Klebsiella pneumoniae isolate [47]. This gene could be transferred into different Enterobacteriaceae and Pseudomonas aeruginosa recipient strains by conjugation, reducing quinolone susceptibility in the transconjugants. The qnrA gene encodes a pentapeptide-repeat protein that binds to the type II topoisomerases to prevent these enzymes from quinolone inhibition [67]. Plasmid-borne genes qnrS, qnrB, qnrD, and qnrC were subsequently identified in Shigella flexneri, K. pneumoniae, Salmonella enterica, and Proteus mirabilis, respectively [11, 28, 36, 72]. The differences between these qnr genes were more than 30% in their amino acid sequences [33]. Meanwhile, diverse qnr gene alleles and their transferable genetic elements were described in different clinical bacterial isolates. Among the five groups of qnr genes, qnrB with 42 gene alleles appeared more prevalent in the qnr-positive bacterial isolates reported [63]. Some qnr genes were detected in E. coli or K. pneumoniae isolates from pediatric patients without receiving quinolone treatment [70]. Broad distribution of quinolone-resistant environmental strains and cross-selection by other antimicrobial agents, such as β-lactam antibiotics, may play an important role in the global dissemination of the qnr genes [29].

Investigations on epidemiology of qnr genes were carried out in more than 30 countries, and most surveys focused on clinical Enterobacteriaceae isolates. Beyond clinical settings, quinolone-resistant bacteria with elevated abundance were identified from several mariculture ponds in China [17–20]. Furthermore, the qnrA, qnrB, and qnrS genes were detected in bacterial isolates from fish farms in Egypt [32]. The qnrS genes were also detected in Aeromonas spp. isolates from the Seine River in France and the Lugano Lake in Swiss, and in E. coli and K. pneumoniae isolates from northern rivers in Turkey [9, 49, 53]. Studies in the last five years indicate that environmental bacteria may form a natural source of antibiotic resistance gene pool (the “resistome”) [7, 75], and the qnr genes may originate in the chromosomes of some aquatic bacteria, such as Shewanella and Vibrio [40, 54, 55, 57]. In aquatic environment, bacteria from different origins could mix without geographic limits, promoting frequent exchange of genetic materials, such as antimicrobial resistance genes and transferable genetic elements [15, 16, 71]. Aquatic environment is thus recognized as a reservoir, natural source, and dissemination vehicle of antimicrobial resistance determinants [3]. The exchange of antibiotic resistance determinants between natural environment and clinical setting may promote the creation of new resistance determinants and the distribution of resistance in clinically important pathogens, aggravating the environmental quality and the difficulty and cost of bacterial infection disease control and treatment. Nowadays, antibiotic resistance determinants have been recognized as important environmental contaminants [45]. However, environmental PMQR bacteria were seldom studied.

Jiaozhou Bay is a typical semi-enclosed coastal bay in China, surrounded by several sewage processing plants, small rivers, maricultural zones, and bathing beaches. Due to rapid urbanization and development of marine economics in the surrounding areas, environmental quality of this coastal bay was dramatically deteriorated by various chemical and biological contaminations [13–15]. This study was carried out to investigate the current status of PMQR in this coastal environment. To explore the possible mechanisms of persistence and dissemination of quinolone resistance in Jiaozhou Bay, molecular techniques were employed to determine the diversity and prevalence of PMQR bacteria and their gene determinants.

Materials and Methods

Strains and Culture Conditions

Seven hundred four tetracycline-resistant bacteria and 348 chloramphenicol-resistant bacteria were isolated from the surface seawater of Jiaozhou Bay in September and October of 2004 [15, 16]. These isolates were collected from ten sampling stations associated with different anthropogenic disturbances as described in previous publications [13–16]. From these antibiotic-resistant bacteria collections, quinolone-resistant strains were screened on tryptic soy agar (TSA, Difco formula) plates supplemented with 3% NaCl and 32-μg ml−1 nalidixic acid and further cultivated at 25°C in tryptic soy broth (TSB) containing 30-μg ml−1 tetracycline or chloramphenicol to avoid possible mutations induced by quinolone. The antimicrobial agents used in this study were purchased from Sigma, USA.

Detection of PMQR Gene Determinants

A simple boiling method was used for rapid bacterial genomic DNA extraction [17]. The qnrA, qnrB, and qnrS genes were screened using a multiplex PCR method as described previously [10]. The qnrC, aac(6′)-Ib-cr, and qepA genes were also screened [39, 43, 51]. A pair of specific primers was designed for detection of the qnrD gene. For the bacterial isolates carrying a qnrA, qnrB, or qnrS gene, further qnr gene allele determination was performed by PCR amplification and gene sequencing. The primers used in this study were listed in Table 1.

Table 1 Primers used for screening the PMQR and gyrA genes in this study
Full size table

Primers 27 F and 1492R [74] were used for bacterial 16S rRNA gene amplification and sequencing for phylogenetic analysis of the bacterial isolates bearing PMQR genes.

Multiple Antibiotic Resistance Assay

Nalidixic-acid-resistant isolates that harbored the qnr genes were selected for further determination of their susceptibility to other typical antimicrobial agents. These isolates were screened on TSA plates supplemented with 30-μg ml−1 tetracycline (TET30), 30-μg ml−1 chloramphenicol (CHL30), 30-μg ml−1 streptomycin (STR30), 30-μg ml−1 kanamycin (KAN30), 30-μg ml−1 gentamicin (GEN30), 15-μg ml−1 erythromycin (ERY15), 5-μg ml−1 ciprofloxacin (CIP5), and 100-μg ml−1 ampicillin (AMP100), respectively, for 48-h incubation at 25°C, based on the method described previously [17].

Molecular Typing

Enterobacterial repetitive intergenic consensus (ERIC)-PCR assay was carried out to analyze the clonal relatedness of the Proteus vulgaris isolates that harbored the qnr genes. Bacterial genomic DNA was extracted by TIANamp bacteria DNA kit (Tiangen, China) according to the manufacture’s protocol. Primers ERIC1 and ERIC2 [69] were used in combination for PCR with Ex Taq polymerase (TaKaRa, Japan). The PCR program consisted of an initial denaturation step at 94°C for 5 min,followed by 30 cycles of DNA denaturation at 94°C for 1 min, primer annealing at 46°C for 1 min, and primer extension at 72°C for 3 min. After the last cycle, a final extension step at 72°C for 10 min was performed. PCR products of 8 μl each were then electrophoresed directly on 1.5% agarose gel containing 0.5-μg ml−1 ethidium bromide at 70 V for 3 h. The gel image was photographed by an AlphaImager HP system (Alpha Innotech, USA), and band detection and normalization were analyzed by software Quantity One (version 4.6.2, Bio-Rad, USA). The presence or absence of each band was presented by a binary code (1 or 0), and a binary data sheet was generated according to the band distribution. Similarity between the ERIC-PCR profiles was determined by using the Jaccard coefficient, and a dendrogram was produced by UPGMA method using the software NTSYSPC (version 2.1, Exeter Software, USA) [56].

Assay of qnr Gene Transfer

Transfer of the qnr genes from quinolone-resistant environmental isolates to E. coli was attempted, with E. coli J53 AziR (resistant to sodium azide) and E. coli TOP10 used as recipient strains for conjugation and transformation experiments, respectively.

Conjugation experiments between the qnr-positive Enterobacteriaceae isolates and E. coli J53 AziR were performed by the liquid mating assay as previously described, with minor modifications [70]. The transconjugants were selected on TSA plates supplemented with 100-μg ml−1 sodium azide and 6-μg ml−1 nalidixic acid. For the other qnr-positive isolates, the donor and recipient strains were grown in TSB medium to logarithmic phase (OD600 = 1), respectively. Donor cells (0.1 ml) and recipient cells (1 ml) were mixed together and added to fresh TSB medium (3.9 ml) and then incubated overnight at 25°C without shaking. The transconjugants were selected on TSA plates containing 300-μg ml−1 sodium azide and 6-μg ml−1 nalidixic acid, at which the growth of the donor bacterial cells was suppressed, and then replica-plated onto Chromocult® coliform agar (Merck, USA) plates with 6-μg ml−1 nalidixic acid.

For transformation experiments, plasmids were extracted individually from the qnr-positive environmental isolates by alkaline lysis method [58], checked by gel electrophoresis, and then electroporated into E. coli TOP10 cells in a 0.2-cm cuvette at a capacity of 25 μF, resistance of 200 Ω, and current at 2.5 kV. Transformants were selected on Luria-Bertani agar (LB, Difco formula) plates containing 3-μg ml−1 nalidixic acid or 0.06-μg ml−1 ciprofloxacin.

The E. coli transconjugants and transformants were confirmed to carry the same qnr gene as that from their donors by PCR experiments.

Screening of Mutations in gyrA Gene

For the qnr-positive environmental isolates, PCR assay was carried out to amplify the quinolone resistance-determining regions (QRDRs) of the gyrA gene. Based on known gyrA gene sequences of the closest-match bacterial species in GenBank, primers (Table 1) were designed and used for PCR gene amplification and DNA sequencing to analyze possible mutations in QRDRs.

Screening of bla Genes

Various extended-spectrum β-lactamase (ESBL) and AmpC β-lactamase gene variants, such as bla SHV , bla TEM , bla OXA , bla MOX , bla CMY , bla LAT , bla BIL , bla DHA , bla ACC , bla FOX , bla ACT , bla MIR , bla KPC , and bla CTX-M , were screened for the ampicillin-resistant qnr-positive isolates using multiplex PCR-based techniques described previously [4, 12, 22, 52]. The amplified gene products were sequenced to determine the bla genes.

Bioinformatic Analysis

DNA sequence alignments were processed using Blast [1] and ClustalX [66] programs. Phylogeny of qnr gene or partial 16S rRNA gene sequences from the qnr-positive isolates was constructed using MEGA5 software [65].

GenBank Accession Numbers

The bacterial 16S rRNA gene sequences were submitted to GenBank with the accession nos. JN384129–JN384161 and HM371197. The representative qnr gene sequences of quinolone-resistant isolates were submitted to GenBank with the accession nos. JN384125–JN384126 and JN384196–JN384207. The gyrA gene sequences were submitted to GenBank with the accession nos. JN384162–JN384195. The bla gene sequences were submitted to GenBank with the accession nos. JN384127–JN384128, JN384208–JN384210, and JN587513.

Results

qnr Genes in Multidrug-Resistant Environmental Isolates

In the current study, 71 tetracycline-resistant and 53 chloramphenicol-resistant bacterial isolates obtained from Jiaozhou Bay were screened for quinolone resistance, and 82 isolates (66%) were found to be resistant to 32-μg ml−1 nalidixic acid. These nalidixic-acid-resistant isolates were screened for the qnr, aac(6′)-Ib-cr, and qepA genes. The qnr genes were detected in 34 isolates, including the qnrA gene in 1 isolate, the qnrB gene in 1 isolate, the qnrS gene in 10 isolates, and the qnrD gene in 22 isolates. Other PMQR genes, such as qnrC, aac(6′)-Ib-cr, and qepA, could not be detected in this study. Therefore, 41% of the nalidixic-acid-resistant environmental isolates carried the qnr genes, 27% carried the predominant qnrD gene, 12% carried the qnrS gene, and 1% carried the qnrA or qnrB gene.

So far, 7 qnrA, 42 qnrB, and 5 qnrS gene alleles with a few amino acid substitutions have been identified (http://www.lahey.org/qnrStudies/). The qnrA-, qnrB-, and qnrS-positive isolates in this study were selected for further determining their qnr gene subtypes. The complete qnr gene sequences of these isolates were obtained by PCR amplification and DNA sequencing. As a result of the amino acid sequence alignments with known Qnr proteins (Fig. 1 and Table 2), qnrA3 gene was identified in isolate QC39, qnrB9 gene was identified in isolate Q15, qnrS1 genes were identified in isolates Q6, Q11, Q13, Q14, and Q19, and qnrS2 genes were identified in isolates Q24, Q29, QC43, QC44, and QC45.

Figure 1
figure 1

Phylogenetic tree of the representative Qnr protein sequences from the qnr-positive isolates with their closest-match GenBank Qnr protein sequences. The reconstruction was computed by the distance method (Neighbor-Joining Poisson correction distance model) with interior branch length supports from 1,000 replicates using MEGA5 software. The reference Qnr protein sequences are according to Jacoby et al. [33], with their GenBank accession numbers labeled in parentheses

Full size image
Table 2 Representative qnr and bla gene sequences determined in the current study and their closest-match GenBank sequences
Full size table

These 34 qnr-positive isolates were also analyzed for their susceptibility to other antibiotics (Table 3). All isolates were resistant to at least five different classes of antimicrobial agents, and 23 isolates (68%) were resistant to 9 different antimicrobial agents tested, including the qnrS1-positive isolate Q11 and all of the qnrD-positive isolates. Thirty-one isolates (91%) were resistant to the fluoroquinolone class antibiotic ciprofloxacin. Twenty-nine isolates (85%) were resistant to ampicillin, excluding the five qnrS2-positive isolates.

Table 3 Bacterial taxonomic affiliation, antibiotic resistance spectra, and resistance genes detected of the qnr-positive isolates
Full size table

As a result of the 16S rRNA gene sequence alignments, all of the 34 qnr-positive isolates shared more than 99% sequence identities of the 16S rRNA genes with their closest-match sequences retrieved from the GenBank database. The phylogenetic tree constructed verified their phylogenetic affiliations (Fig. 2). All isolates belonged to the γ-Proteobacteria subdivision. Bacteria affiliated to Enterobacteriaceae contributed to 82% of the 34 isolates, including species closely related to Citrobacter freundii, Enterobacter sp., K. pneumoniae, and Proteus vulgaris. The remaining six isolates are indigenous estuarine or marine bacteria, affiliated with Pseudoalteromonas sp., Pseudomonas sp., and Shewanella algae. The Shewanella algae isolate carried the qnrA3 gene, the C. freundii isolate carried the qnrB9 gene, and the Proteus vulgaris isolates carried the qnrD genes. The qnrS-positive isolates were quite diverse. The Enterobacter sp. and Klebsiella spp. isolates harbored the qnrS1 genes, and the Pseudoalteromonas sp. and Pseudomonas sp. isolates harbored the qnrS2 genes.

Figure 2
figure 2

Phylogeny of partial 16S rRNA gene sequences from the qnr-positive isolates. The reconstruction was computed by the distance method (Neighbor-Joining Poisson correction distance model) with interior branch length supports from 1,000 replicates using MEGA5 software. The GenBank accession numbers of the reference sequences were labeled in parentheses. The qnr genes detected were labeled in brackets for the corresponding isolates. Asterisk, the 16S rRNA gene sequences of isolates Q5, Q28, QC14, QC15, QC17, QC29, QC32, QC33, QC34, QC35, QC36, QC46, QC50, QC52, and QC53 were identical to that of isolate Q3

Full size image

Molecular Types of Proteus vulgaris Isolates

Molecular typing was carried out to investigate the genetic diversity of the 22 qnrD-positive Proteus vulgaris isolates. A series of PCR amplification conditions were tested to optimize the ERIC-PCR result. Eventually, 46°C was chosen as the optimal annealing temperature. Several different PCR product profiles were observed by electrophoresis (Fig. 3). Cluster analysis was made based on the banding types of ERIC-PCR fingerprinting. These 22 Proteus vulgaris isolates were classified into four distinct groups. The isolates QC46, QC48, and QC51 were unique individually, and the remaining 19 isolates were highly related (Fig. 4).

Figure 3
figure 3

Enterobacterial repetitive intergenic consensus (ERIC)-PCR assay of the 22 Proteus vulgaris isolates with ERIC primers ERIC1 and ERIC2. No template DNA was added to the negative control. The DNA molecular weight marker was 1-kb DNA Ladder (Fermentas, USA)

Full size image
Figure 4
figure 4

Dendrogram of the cluster analysis based on ERIC-PCR fingerprinting of the 22 Proteus vulgaris isolates. The numbers on the horizontal axis indicate the percentage similarities as determined by the Jaccard coefficient and UPGMA method using the software NTSYSPC2.1

Full size image

Transfer of qnr Genes

Due to multidrug resistance of the 34 qnr-positive isolates, nalidixic acid and ciprofloxacin were used as selective pressure to study the transfer capacity of qnr genes between nalidixic-acid-resistant isolates and E. coli recipients. The qnrS1 gene could be transferred from isolates Q6, Q11, Q13, Q14, and Q19 to E. coli J53 AziR by conjugation and to E. coli TOP10 by electroporation. The qnrD gene in the 22 Proteus vulgaris isolates could be transferred to E. coli TOP10 by electroporation, but conjugation experiments failed. The same qnr gene as that from their donors could be verified in the E. coli transconjugants or transformants by PCR amplification.

None plasmid could be extracted from the qnrA-positive isolate QC39 or the qnrS2-positive isolates Q24, Q29, QC43, QC44, and QC45 using alkaline lysis method, and these qnr genes failed to be transferred into E. coli recipient strains. Although five plasmids were isolated from the qnrB9-positive isolate Q15 (data not shown), no E. coli transconjugant or transformant could be obtained in this study.

Mutations in GyrA Protein

Mutations in the QRDRs of type II topoisomerases are frequently associated with the qnr genes in quinolone-resistant or less susceptible bacteria [5]. For quinolone resistance, the most common point mutations of Gram-negative bacteria occur in the gyrA gene. In the current study, the gyrA genes of the 34 qnr-positive isolates were PCR amplified and sequenced. Compared with the amino acid sequences in the QRDRs of GyrA in the closest-match bacterial species, the isolates Q11, Q24, Q29, QC43, QC44, and QC45 and the 22 Proteus vulgaris isolates had a Ser83Ile substitution, the isolate Q15 had a Ser83Thr substitution, and the isolate QC39 had an Ala67Ser substitution. No point mutation in the QRDRs of GyrA was found in isolates Q6, Q13, Q14, and Q19 (Table 3). Therefore, 88% of the qnr-positive isolates had a single point mutation in the QRDRs of GyrA protein.

bla Gene Screening

Considering the possibility of cross-selection of environmental antimicrobial-resistant strains by quinolones and β-lactam antibiotics, 29 ampicillin-resistant isolates that harbored the qnr genes were screened for the ESBL and AmpC β-lactamase genes by PCR amplification. Four different types of bla genes were detected in five environmental isolates, including the bla CMY gene in isolate Q15, the bla OXY and bla TEM genes in isolate Q11, and the bla SHV gene in isolates Q13, Q14, and Q19. The bla OXY , bla SHV , and bla TEM genes combined with the qnrS1 gene were present in four Klebsiella isolates, and the bla CMY gene combined with the qnrB9 gene was present in the C. freundii isolate. The other bla gene variants tested in this study could not be detected.

Discussion

The PMQR determinants have been identified in a number of clinically important pathogens and environmental bacteria in different geographic regions of the world [9, 32, 41, 49, 53]. Studies of PMQR determinants as a new type of biological contaminants in aquatic environment, especially in coastal areas that receive high anthropogenic activities, are very limited and need more efforts. Previous studies on the incidence and persistence of tetracycline-, chloramphenicol-, ampicillin-, and streptomycin-resistant bacteria and their resistance determinants in Jiaozhou Bay found that the antimicrobial resistance status in the coastal seawater was very severe and complicated [15, 16, 71, 79]. Our current study adds more evidences to the exacerbated antimicrobial resistance status by providing new data about the diversity and prevalence of PMQR determinants in multidrug-resistant bacterial isolates from this coastal bay.

Diverse qnr genes, including qnrA, qnrB, qnrS, and qnrD, were identified in seawater bacterial isolates of Jiaozhou Bay, and surprisingly, the predominant PMQR gene was qnrD that was carried by 27% quinolone-resistant isolates. The prevalence of different PMQR genes in Jiaozhou Bay was different from the statistical data of the global prevalence of PMQR genes, as the qnrA, qnrB, qnrS, and aac(6′)-Ib-cr genes were found to be more common in clinical Enterobacteriaceae isolates [63]. The qnrD gene is relatively a new member of the PMQR determinants, and it has been identified in a few isolates from human and animals [11, 48, 68, 78]. Our study is the first report about the occurrence of qnrD in marine environment, indicating the importance of this newly discovered quinolone resistance mechanism, especially in coastal settings. Given that some bacteria with qnr genes could not be selected via antimicrobial susceptibility tests, especially with nalidixic acid [26], higher prevalence and diversity of PMQR genes should be expected from Jiaozhou Bay.

In our current study, the qnrA3 gene was found to be carried by a Shewanella algae isolate with no plasmid detected, and this gene could not be transferred into E. coli recipients, consistent to the speculation of other studies that this gene may be located on the chromosome of Shewanella algae stains [55]. The ubiquitous distribution of qnrA-bearing Shewanella algae in aquatic environments suggests their important role in qnr gene origination and evolution in a global scale. The qnrB9 gene in Jiaozhou Bay was found to be carried by C. freundii isolate Q15 that carried five plasmids, but this gene could not be transferred into E. coli cells using conjugation or transformation technique. It is likely that the qnrB9 gene may be situated on the bacterial chromosome or a plasmid that could not replicate in E. coli recipients. Currently, a great number of qnrB gene allelic variants, located on both plasmids and chromosomes, have been identified in C. freundii strains [2, 8, 35, 37, 61, 63, 64]. In China, most C. freundii isolates collected from other aquatic environments or clinical settings were found to carry the qnrB gene, and qnrB9 was a dominant quinolone resistance determinant for this bacterial species [77]. Therefore, C. freundii bacteria may provide a large gene pool with diverse qnrB gene alleles for the evolution and transmission of the qnr genes.

In Jiaozhou Bay, two qnrS gene alleles, qnrS1 and qnrS2, were identified in ten isolates affiliated to Enterobacter, Klebsiella, Pseudoalteromonas, and Pseudomonas. The qnrS1 gene in Enterobacter and Klebsiella isolates could be transferred into E. coli recipients, suggesting a potential role of plasmids in the spread of qnrS1 gene in Jiaozhou Bay. The qnrS2 genes were identified in four Pseudoalteromonas and one Pseudomonas isolates, which were found to be without plasmids using alkaline lysis method. As these genes could not be transferred into E. coli recipients, the qnrS2 genes may be situated on the bacterial chromosomes or large plasmids that were difficult to transfer or detect. Our study is the first report of qnrS2 in Pseudoalteromonas and Pseudomonas strains, which may serve as an origin of the qnrS gene. Our study also indicates the importance of the aquatic Pseudoalteromonas and Pseudomonas strains in the evolution and dissemination of quinolone resistance in coastal environments and related clinical settings.

Our study is the first identification of qnrD gene in Proteus vulgaris isolates that were collected from different months and sampling stations in Jiaozhou Bay (Table 2). The ERIC-PCR analysis revealed that these Proteus vulgaris isolates could be divided into four distinct groups. The major group consisted of 19 isolates, the close relatedness of which suggested that the prevalence of qnrD gene in Jiaozhou Bay might originate from a common Proteus vulgaris ancestor strain that was well adapted to the survival, persistence, and dissemination in this coastal environment.

Acquirement of PMQR genes may only result in reduced susceptibility or low-level resistance to quinolones. However, this process may facilitate the recovery of mutants with high level of quinolone resistance [63]. In our current study, 88% of the qnr-positive isolates had a single point mutation at codon 67 or codon 83 in the QRDRs of GyrA protein. As these isolates were previously selected on TSA plates supplemented with tetracycline or chloramphenicol, their gyrA genes retained the original statuses and avoided additional mutations induced by the quinolone selection pressure. The combination of chromosome- and plasmid-mediated quinolone resistance may play an important role in the persistence and dissemination of quinolone resistance in Jiaozhou Bay.

It has been found that the qnr genes exist in a great number of ESBL- and AmpC-producing Enterobacteriaceae [23, 60]. Identification of qnr genes in clinical isolates of E. coli and K. pneumoniae from Chinese pediatric patients without quinolone treatment indicated that the widely used β-lactam antibiotics might contribute to the cross-selection of qnr genes [27, 70]. Coexistence of quinolone- and β-lactam-resistant genes on the same genetic element provides one of the mechanisms for this phenomenon [42]. In our current study, 85% of the qnr-positive isolates were found to be resistant to ampicillin, and the bla CMY , bla OXY , bla TEM , and bla SHV genes were identified in the qnrB- and qnrS1-positive isolates. The common ESBL and AmpC β-lactamase genes were not identified in the qnrA- and qnrD-positive isolates, so some other determinants may be responsible for their resistance to ampicillin. The dominant bla SHV genes were detected in three K. pneumoniae isolates, and the bla OXY and bla TEM genes were detected in Klebsiella sp. isolate Q11, simultaneously. Coexistence of these ESBL genes and qnrS genes was previously reported in Asian and European countries [8]. The plasmid-mediated cephalosporinase gene bla CMY was identified in association with qnrB9 gene in C. freundii isolate Q15 in our current study, in contrary to bla CMY-1 in association with qnrB2 in an E. coli isolate from Korea [50]. Novel combinations may be expected to find in future studies due to the diverse arrays of both genes.

The quinolone-resistant isolates and their qnr genes in Jiaozhou Bay might originate from two distinct sources, the terrestrial bacteria related to anthropogenic activities and the indigenous estuarine or marine bacteria. The qnrB9, qnrS1, and qnrD genes were identified in Enterobacteriaceae isolates affiliated to Citrobacter, Enterobacter, Klebsiella, and Proteus, most of which might potentially relate to human or animal pathogens. Civic wastewater discharges via rivers and sewage processing plants on the seashore may be the major source of quinolone-resistant Enterobacteriaceae in Jiaozhou Bay. The qnrA3 and qnrS2 genes were detected in Shewanella, Pseudoalteromonas, and Pseudomonas isolates, respectively, most of which are common estuarine or marine bacterial species. Several qnr-like genes have been found in the chromosomes of aquatic bacteria and in metagenomes from marine organisms [54, 55, 59]. Expression of some qnr genes in environmental bacteria may be induced by cold shock and regulated by the SOS system [38, 73], indicating that the qnr gene may contribute to low temperature adaptation of bacteria in aquatic environment. Coastal seawater bacteria may thus provide a natural origin and reservoir of diverse qnr genes that may be acquired by or exchanged with pathogenic bacteria, facilitating the creation and transmission of new quinolone resistance determinants. The increasing occurrence of PMQR gene with time indicates that the gene transmission and exchange rate is getting faster recently [34]. Study of antibiotic resistance plasmids isolated from seawater bacteria of Jiaozhou Bay indicates that there is no border for the transmission of antibiotic resistance on a global scale [79]. Our study indicates that coastal environments serve as a dynamic mixture of gene pools from both natural and anthropogenic domains. Antibiotic resistance in these environments should draw a great attention and research effort for winning the battle against the intimidating antibiotic resistance threat [24]. Regional and global antibiotic resistance surveillance programs are present or proposed [25]. However, the focus of these programs is on clinical pathogens. Antibiotic resistance in the environment is seldom concerned. As the clinical setting and natural environment are tightly interlinked in coastal environments [77], the neglect of antibiotic resistance in the environment may be a big mistake in the strategy and implementation of antibiotic resistance surveillance and control [46].

In summary, our pioneering identification of qnrD gene in Proteus vulgaris and qnrS2 gene in Pseudoalteromonas and Pseudomonas enriched the list of microbial diversity of the qnr gene hosts. Surveillance survey on prevalence of qnr genes in Jiaozhou Bay could not only acquire a basic understanding of the abundance and diversity of these resistance determinants in marine environment, but also help to design better strategy and implementation for coastal environment and human health management.