High Prevalence of Plasmid-Mediated Quinolone Resistance and IncQ Plasmids Carrying qnrS2 Gene in Bacteria from Rivers near Hospitals and Aquaculture in China

Effluents from hospital and aquaculture are considered important sources of quinolone resistance. However, little information is available on the impact of this effluent on nearby rivers. In this study, 188 ciprofloxacin-resistant bacterial isolates obtained from rivers near hospitals and aquaculture were screened for plasmid-mediated quinolone resistance (PMQR) genes. Species identification, antibiotic susceptibility testing, and PMQR gene transferability assessment were conducted for PMQR-positive bacteria. Representative qnrS2-encoding plasmids were subsequently sequenced using a primer-walking approach. In total, 44 isolates (23.4%) were positive for qnr genes (16 qnrB2, 3 qnrS1, and 25 qnrS2) and 32 isolates (17.0%) were positive for aac(6′)-Ib-cr. Other PMQR genes were not detected. The qnrB2 and aac(6′)-Ib-cr genes had a higher prevalence in aquaculture samples than in hospital samples, and were significantly associated with Enterobacteriaceae (p < 0.05). In contrast, the prevalence of qnrS2 was not site-related, but was significantly associated with Aeromonas spp. (p < 0.05). All PMQR isolates were resistant to three or more classes of antibiotics. Eleven qnrS2-harboring plasmids from Aeromonas spp., including a novel conjugative plasmid pHP18, were selected for sequencing. These plasmids were small in size (6,388–16,197 bp) and belonged to the IncQ or IncU plasmid family, with qnrS2 being part of a mobile insertion cassette. Taken together, our findings suggest that aquaculture is a possible source for aac(6′)-Ib-cr and qnrB2 dissemination, and demonstrate the ubiquity of qnrS2 in aquatic environments. Finally, Aeromonas spp. served as vectors for qnrS2 with the help of IncQ-type plasmids.


Introduction
The dissemination of quinolone-resistant bacterial isolates has become a major problem in infection control worldwide. In Portugal, 23.8% of clinical Escherichia coli isolates were ciprofloxacin. Individual colonies were picked up based on morphology. All experiments in this study were performed in parallel.

Characterization of PMQR Bacteria and Their Gene Determinants
Total DNA was extracted using a simple boiling method. The qnrA, qnrB, and qnrS genes were screened using a multiplex PCR method as described previously [18]. The genes qnrC, qnrD, aac(6 0 )-Ib-cr, and qepA were also detected [19]. The 16S rRNA, gyrB, and ropD genes were amplified and sequenced for phylogenetic analysis [20].

Transferability of PMQR Genes
Transferability of PMQR genes from the environmental isolates was tested, with E. coli J53 Azi R and E. coli DH5α used as recipients for conjugation and transformation experiments [21], respectively. The transconjugants were screened on Luria-Bertani (LB) agar plates supplemented with sodium azide (100 μg/mL) plus ciprofloxacin (0.05 μg/mL). The transformants were selected on LB agar plates with ciprofloxacin (0.05 μg/mL). PCR was performed to identify the PMQR genes acquired by the E. coli transconjugants and transformants, using primers and conditions described elsewhere [18].

Antibiotic Susceptibility Testing
Antibiotic susceptibility to amikacin, ampicillin, cephalothin, chloramphenicol, ciprofloxacin, erythromycin, nalidixic acid, streptomycin, trimethoprim/sulphamethoxazole, and tetracycline was assayed by the disc diffusion method, according to the guidelines of the Clinical and Laboratory Standards Institute (CLSI, 2012). E. coli ATCC 25922 was used as a control strain. The minimal inhibitory concentrations (MICs) of ciprofloxacin were determined using a broth microdilution method in accordance with CLSI guidelines.

Plasmid Analysis
The E. coli transconjugants and transformants were grown overnight at 37°C in LB broth containing ciprofloxacin (0.05 μg/mL). Plasmid DNA was extracted using the Qiagen Plasmid Midi Kit (Qiagen Science Inc., MD, USA). Complete nucleotide sequences were determined using a primer walking strategy. The primers P1-F (5 0 -AACTCAATACCGTAGCAAT-3 0 ) and P1-R (5 0 -TTTATGTCACGCCGAACT-3 0 ), which target the qnrS2 gene and read outward, were applied to long-range PCR. Thermal cycling conditions were as follows: 1 min at 94°C; 35 cycles of 15 s at 93°C, 30 s at 55°C, and 7 min at 72°C,followed by a 7 min extension at 72°C. Direct sequencing was carried out using ABI PRISM Big Dye Terminator Cycle Sequencing technology (Applied BioSystems, Foster City, USA). Results were analyzed using the software DNA Sequencing Analysis V5.1 (ABI). A sequence analysis was carried out using BLASTn, BLASTx, DNASTAR Lasergene 8.0, and the open reading frame (ORF) finder program (http:// www.ncbi.nlm.nih.gov/gorf/gorf.html). The plasmids were named according to their hosts.

Phylogenetic Analysis
Multiple sequence alignments were performed using ClustalW in MEGA 5.10 [22]. Phylogenetic trees were constructed using the neighbor-joining method, and the robustness was evaluated by bootstrap analyses based on 1000 resamplings.

Statistics
Statistical analyses were performed using SPSS 22.0 for Windows (SPSS Inc., Chicago, IL). The prevalence of PMQR genes or antibiotic resistance phenotypes was compared among isolates from different origins or taxonomic groups using the chi-square test at a significance level of 0.05.

Nucleotide Sequence Accession Numbers
Representative 16S rRNA, gyrB, and rpoD gene sequences were deposited in GenBank under accession numbers KU644678 to KU644707. The aac(6 0 )-Ib, aac(6 0 )-Ib-cr, and qnr gene sequences were submitted to GenBank with accession numbers KU644708 to KU644712. The complete nucleotide sequences of plasmids were deposited in GenBank under accession numbers KU644672 to KU644677.
The prevalence of PMQR in isolates from hospital-associated samples, aquaculture-associated samples, or clean water was 28.3% (24/85), 37.3% (31/83), and 5% (1/20), respectively. The prevalence of qnrS1 and qnrS2 genes was not associated with a specific sampling site (p > 0.05). The aac(6 0 )-Ib-cr and qnrB2 genes showed a significantly higher prevalence in aquaculture samples than in hospital samples (p < 0.05), but their prevalence was not significantly different between hospital samples and clean water (p > 0.05).

Antibiotic Susceptibility of Environmental Isolates
All PMQR isolates (n = 56) presented multidrug resistance (MDR), i.e., resistance to antibiotics belonging to three or more classes (Fig 1). The most commonly encountered resistances were  (Fig 2). For most antibiotic classes, the prevalence of resistance did not significantly differ (p > 0.05) by sampling site. The exception was the significantly higher (p < 0.05) prevalence of resistance to chloramphenicol, erythromycin, and sulfamethoxazole/trimethoprim in isolates
Characterization of qnrS2-Harboring Plasmids in Aeromonas spp.
Eleven E. coli transformants derived from clonally unrelated Aeromonas strains were selected for plasmid analysis. These plasmids were small in size (6,388-8,663 bp) and belonged to the IncQ family except for an IncU plasmid, pHP2 (16,197 bp). All plasmids contained qnrS2 inserted at either side of the oriV region (IncQ) or the gene mpR coding for a putative zincmetalloprotease (IncU) in the form of a mobile insertion cassette (mic). The mic was bracketed by 22-bp imperfect inverted repeats and a 5-bp duplication of the target site (Fig 3), suggesting the acquisition of this structure by transposition. Based on genetic structure, three IncQ plasmid groups were identified.
The first IncQ plasmid group was isolated from Aeromonas caviae and included two identical plasmids, pHP18 and pAB4, and one variant pHP1. Plasmid pHP18, the sole plasmid found in this study, which was capable of mobilization through conjugation, was 6,388 bp in size with a G+C content of 57%. It contained 6 ORFs: repA and repC genes for plasmid replication, mobA and mobC genes for plasmid mobilization, the qnrS2 gene for quinolone resistance, and a repressor gene ( Table 2). The qnrS2 gene was 100% identical with its Aeromonas spp. counterpart. The plasmid backbone showed high similarity with two IncQ-1 plasmids, pCHE-A (66% coverage, 85% identity) and pMS260 (65% coverage, 80% identity). Plasmid pCHE-A originated from Enterobacter cloacae from Canada, and harbored the bla GES-5 gene [11], whereas pMS260 was a streptomycin and sulfonamide resistance-coding plasmid isolated   from Actinobacillus pleuropneumoniae in Japan [23]. Despite sharing three common 22-bp iterons with pMS260, the backbone of pHP18 showed characteristics that were distinct from those of other IncQ-family plasmids, including entirely different A+T-rich and G+C-rich regions, and the absence of a highly conserved 15-bp region. A codon bias analysis indicated that the backbone genes of pHP18 had a strong preference for a G or C in the third position (72.8%), while only 39.3% of the codons of qnrS2 had a G or C residue in that position. Compared with pHP18, pHP1 had an additional 22-bp iteron insertion within oriV (Fig 4). Phylogenetic analysis of the replication protein sequences showed that pHP18 and pHP1 grouped with the IncQ-1 family (Fig 5).
The plasmid pHP2, harboring the qnrS2 and aac(6 0 )-Ib-cr genes in A. caviae, had a typical IncU backbone for plasmid replication and maintenance functions (Fig 3). It was 16,197 bp in size, with an average G+C content of 55%. qnrS2 was inserted into mpR as a mic. aac(6 0 )-Ib-cr, located on an incomplete class 1 integron, was inserted into the gene nic, encoding VirD2 relaxase. pHP2 was highly related to the PMQR-carrying IncU plasmid pAH6 (98% coverage, 100% identity) and pAC3 (100% coverage, 99% identity) found in Aeromonas spp. from Czech Republic and Korea [27]. The divergence was mainly due to different repetitive sequences surrounding the rep genes. In pHP2, the repetitive sequences located upstream of repA (319 bp) and downstream of repB (739 bp) were organized as (r3r1) 3 r3 and (r1) 8 r2, respectively. The corresponding sequences in pAH6 were organized as (r3r1r3) 2 and (r1) 8 r2, whereas in pAC3 the counterparts were organized as (r3r1) 3 r3 and r1(partial)r1(partial)r1r2. Furthermore, the left inverted repeat of mic in pHP2, as well as a partial truncated region of mpR, was missing in pAH6. This difference may have resulted from plasmid rearrangements subsequent to the acquisition of the qnrS2-carrying mic.

Discussion
The PMQR determinants have been identified in a number of environmental bacteria worldwide. Their prevalence appears to vary considerably depending on the selection criteria of studied strains, ranging from 16.7 to up to 58.0% [9,28]. In this study, the prevalence of PMQR genes in isolates from rivers was 29.8%, with aac(6 0 )-Ib-cr and qnrS2 as the predominant genes. This conclusion was supported by previous studies demonstrating the frequent occurrences of these genes in environmental isolates [9,15,28]. The qnrS-type genes seem to be the most commonly identified acquired qnr genes in the environment [4]. They have been mainly identified from waterborne species, and in particular Aeromonas spp. [29][30][31]. aac(6 0 )-Ib-cr is widespread geographically and stable over time [32]. It has often been more common than qnr alleles. The occurrence of qepA is quite limited among environmental isolates, but in animals might be significant, with a strong association with rmtB genes [33]. Noticeably, some studies identified a high prevalence of other PMQR genes such as qnrD or qnrB in aquatic environments [2,28]. Factors, such as bacterial species and microbial habitat, may play important roles in the development and spread of antibiotic resistance.
In this study, the prevalence of the genes aac(6 0 )-Ib-cr and qnrB2 was significantly higher in aquaculture samples than in hospital samples. However, there was no significant difference between hospital samples and clean water. This suggests a site-related dissemination route and a non-clinical origin for the two genes. Rivers near aquaculture may serve as a source of aac (6 0 )-Ib-cr and qnrB2 genes in the urban water cycle, as indicated by other studies focused on fish farms [34,35]. A site-related prevalence may be related to a given habitat, where the success of resistance acquisition depends on the fitness of the target bacterium (receivers of horizontal gene transfer) [14]. In agreement with this observation, aac(6 0 )-Ib-cr and qnrB2 genes were mainly detected in a specific taxonomic group, Enterobacteriaceae. There may be environmental conditions or selective pressures imposed in rivers near aquaculture that contribute to this site-related dissemination pattern.
Unlike aac(6 0 )-Ib-cr and qnrB2, the prevalence of qnrS2 was not site-related; rather, it was species-related. The fact that qnrS2 was mainly identified from Aeromonas spp., typical inhabitants of aquatic environments, suggested a non-clinical source for this gene in bacteria from urban rivers. This conclusion is supported by previous studies demonstrating the frequent occurrence of qnrS2 in municipal wastewater and rare detection in hospital effluent [9,14,16]. The ubiquity and genomic plasticity of Aeromonas spp. may contribute to the homogeneous distribution of qnrS2 in water.
All PMQR bacteria examined showed resistance to more than three classes of antibiotics. In Enterobacteriaceae, multidrug resistance may be related to the high prevalence of aac(6 0 )-Ib-cr, which confers resistance to kanamycin, tobramycin, netilmicin, amikacin, and ciprofloxacin, and is associated with an MDR phenotype [6,36]. Moreover, qnr genes have been found to exist in many extended-spectrum beta-lactamase-and AmpC-producing Enterobacteriaceae [37][38][39], suggesting their co-selection with other resistance elements. In Aeromonadaceae, because of their intrinsic resistance to some beta-lactams and susceptibility to antibiotic resistance acquisition [9], Aeromonas spp. are frequently multidrug resistant [16,17]. The significantly higher prevalence of resistance to chloramphenicol, erythromycin, and sulfamethoxazole/trimethoprim in isolates from aquaculture samples may be associated with the wide range of antibiotics used in aquaculture [40]. High rates of resistance to the older, heavily used antibiotics (chloramphenicol, sulfonamides, and tetracyclines) have frequently been detected in fish food and fish pathogens such as Aeromonas salmonicida [34,35,41].
The qnrS2 genes from Aeromonas spp. were often identified in IncU, IncQ, and ColE plasmids [24,25,27,42]. In this study, IncQ plasmids were highly prevalent (90.9%). The wide distribution of IncQ plasmids carrying PMQR genes can be attributed to several mechanisms, including their ability to readily mobilize and replicate in a broad range of hosts, their low molecular weight that minimizes metabolic load, and their high copy numbers that ensure stability [43]. The ubiquity of Aeromonas species may have facilitated evolution by recombination of IncQ plasmids with a large number of genetic elements or resistance gene insertions.
Among qnrS2-carrying plasmids in Aeromonas spp., pHP18 is the first identified plasmid that can be mobilized by conjugation. However, it did not carry the mating pair formation (Mpf) components essential for efficient transfer by conjugation. A larger conjugative plasmid may be co-resident with pHP18, providing it with the Mpf components. A resistance gene reservoir not only allows resistant genes to stably exist, but also facilitates transfer of these genes to other species in a natural state. Our findings highlighted the important role of Aeromonas species as a PMQR gene reservoir. Plasmid pHP18 shows >80% identity with plasmids from A. pleuropneumoniae and E. cloacae in the backbone region, suggesting a possible origin from bacterial pathogens of humans or animals. The codon bias analysis indicated that qnrS2 was likely inserted after a functional replicon had evolved in IncQ family plasmids.
In addition to the accessory genes, three 22-bp iterons in IncQ plasmids were subject to insertion or mutation, as shown in the pHP18 and pAB5 variants. Iterons were the major incompatibility determinants in IncQ plasmids, so point mutations within the three 20-bp conserved repeats of iterons could result in an inability or significantly reduced ability to exert incompatibility, initiate replication, or transfer genes by conjugation [43]. Meanwhile, the iteron number had a negative effect on plasmid copy number, as well as the antibiotic resistance level, and imposed metabolic burden [44]. Hence, variations in the iteron sequence may reflect the need for balance between high antibiotic resistance and a stable existence [45]. This may explain why pHP18 conferred a higher level of quinolone resistance to E. coli recipients than pHP1 did.
In IncU plasmids, a fragment encompassing repB with r1/r2 repetitive regions was essential for minireplicon functionality [46]. Other repetitive regions seemed to be a hotspot for sequence elimination, as shown in the pHP2 variants. Plasmid pHP2 is one of the smallest IncU plasmids identified so far. Recently, a smaller IncU plasmid pPA-2 (7,995 bp) was identified in Pseudomonas aeruginosa clinical isolates [47]. Both these plasmids have an elimination of conjugative transfer region. Despite loss of conjugative abilities, small IncU plasmids can be efficiently transferred in aquatic environments by transformation, due to the general ability of Aeromonas environmental isolates to acquire free DNA [48].
Overall, this study suggested aquaculture as a possible source of aac(6 0 )-Ib-cr and qnrB2 in aquatic environments. Enterobacteriaceae, monitored in water quality controls, were important hosts of these two genes. The ubiquitous bacteria, Aeromonas spp., served as vectors for qnrS2 with the help of IncQ-type plasmids. Evidence for water-human transmission of Aeromonas species was available [49,50], and a qnrS-containing plasmid was identified in an Aeromonas sp. clinical isolate [51]. Given to the patterns of acquired antimicrobial resistance, more attentions should be paid to the antibiotic resistance surveillance of both clinical and environmental Aeromonas isolates.