Dissemination and Characterization of Plasmids Carrying oqxAB-bla CTX-M Genes in Escherichia coli Isolates from Food-Producing Animals

Background The association of PMQR and ESBLs in negative-bacteria isolates has been of great concern. The present study was performed to investigate the prevalence of co-transferability of oqxAB and bla CTX-M genes among the 696 Escherichia coli (E. coli) isolates from food-producing animals in South China, and to characterize these plasmids. Methods The ESBL-encoding genes (bla CTX-M, bla TEM and bla SHV), and PMQR (qnrA, qnrB, qnrS, qnrC, qnrD, aac(6’)-Ib-cr, qepA, and oqxAB) of these 696 isolates were determined by PCR and sequenced directionally. Conjugation, S1 nuclease pulsed-field gel electrophoresis (PFGE) and Southern blotting experiments were performed to investigate the co-transferability and location of oqxAB and bla CTX-M. The EcoRI digestion profiles of the plasmids with oqxAB-bla CTX-M were also analyzed. The clonal relatedness was investigated by PFGE. Results Of the 696 isolates, 429 harbored at least one PMQR gene, with oqxAB (328) being the most common type; 191 carried bla CTX-M, with bla CTX-M-14 the most common. We observed a significant higher prevalence of bla CTX-M among the oqxAB-positive isolates (38.7%) than that (17.4%) in the oqxAB-negative isolates. Co-transferability of oqxAB and bla CTX-M was found in 18 of the 127 isolates carrying oqxAB-bla CTX-M. These two genes were located on the same plasmid in all the 18 isolates, with floR being on these plasmids in 13 isolates. The co-dissemination of these genes was mainly mediated by F33:A-: B- and HI2 plasmids with highly similar EcoRI digestion profiles. Diverse PFGE patterns indicated the high prevalence of oqxAB was not caused by clonal dissemination. Conclusion bla CTX-M was highly prevalent among the oqxAB-positive isolates. The co-dissemination of oqxAB-bla CTX-M genes in E. coli isolates from food-producing animals is mediated mainly by similar F33:A-: B- and HI2 plasmids. This is the first report of the co-existence of oqxAB, bla CTX-M, and floR on the same plasmids in E. coli.


Introduction
Quinolone resistance was thought to be mediated only by chromosomal mutations, until plasmid-mediated quinolone resistance (PMQR) was described in 1998 [1]. Since then, a number of plasmid-mediated quinolone resistance (PMQR) mechanisms have been described: the pentapeptide repeat family Qnr proteins (QnrA, QnrB, QnrS, QnrC, and QnrD) [1,2,3,4,5], AAC(6')-Ib-cr, an aminoglycoside acetyl-transferase that is responsible for reduced susceptibility to ciprofloxacin by modifying ciprofloxacin [6], QepA, an efflux pump belonging to the major facilitator subfamily [7], and OqxAB, a multidrug efflux pump that confers resistance to multiple agents, which has been recently reported to reduce susceptibility to ciprofloxacin and nalidixic acid [8]. The PMQR genes confer only low-level resistance to quinolones; however, they can be spread horizontally among enterobacteria and facilitate the selection of resistant mutants following exposure to ciprofloxacin [9]. Fluoroquinolones, and cephalosporins are commonly used to treat gram-negative bacterial infections, especially some intestinal or extraintestinal infections caused by E. coli. Increasing resistant isolates, especially multidrugresistant E. coli isolates, have been observed [10,11], due to the use of these antimicrobials, both in human and animal diseases over the past decades. The presence of multidrug-resistant isolates harboring multiple resistance genes on the same plasmid has been of great concern, as it expands the subset of drugs that may select for the dissemination of multidrug resistance plasmids and poses a serious risk to both animal and human health. Except for oqxAB, other PMQR genes were often found to be strongly associated with extended-spectrum β-lactamase (ESBL) genes, and some were often found to be located on the same plasmid [9]. Since oqxAB was reported to be related to reduced susceptibility to ciprofloxacin and nalidixic acid [8], it has been found among E. coli isolates from animals and humans [12,13,14]. Reports on the prevalence of coexistence of PMQR (including oqxAB) and ESBL genes in the same isolate have increased in the past years [12,15]. However, there is a paucity of data with regard to the prevalence and characterization of plasmids co-carrying oqxAB-bla CTX-M genes in bacteria, except only one E. coli isolate in our previous report [16]. Because antibiotic resistant bacteria from food-producing animals can be transferred to humans through the food chain or other routes [17,18], monitoring antimicrobial resistance in bacteria from the food-producing animals is important for ensuring food safety and public health.
The present study was conducted to investigate the clinical E. coli isolates from food-producing animals in China for the prevalence and dissemination of plasmids harboring oqxABbla CTX-M genes, but also the characterization of these plasmids.

Bacterial isolates
A total of 696 non-duplicate E. coli isolates (318 avian including 177 from ducks, 110 from chickens and 31 from geese, and 378 from pigs) were isolated from diseased foodproducing animals between March 2002 and August 2012. The animals were from more than 80 farms all over Guangdong province. Animals we chose liver or heart tissues to obtain isolates were infected with E. coli, and the other animals showed diarrhea. Further information about these animals, the underlying disease and possible antimicrobial pretreatment were unfortunately not available. Cotton swabs of the liver and heart tissues or faeces from these animals were streaked onto MacConkey agar. After 16h incubation at 37°C, one colony with typical E. coli morphology was selected and purified on MacConkey agar. One colony was selected from each sample and all E. coli isolates were identified by classical biochemical methods and confirmed using the API 20E system (bioMé rieux). All identified isolates were stored at -80°C in Luria-Bertani broth containing 30% glycerol. E. coli C600, resistant to streptomycin, was used as the recipient strain in the conjugation experiments.

Antimicrobial susceptibility testing
Susceptibilities to enrofloxacin, ciprofloxacin, levofloxacin, nalidixic acid, amikacin, gentamicin, florfenicol, chloramphenicol, ampicillin, ceftiofur, cefotaxime, doxycycline, and tetracycline of the 696 isolates were assayed by the agar dilution method, according to the guidelines of the Clinical and Laboratory Standards Institute (CLSI) [19]. Escherichia coli ATCC 25922 was used as the control strain. Isolates were classified as either susceptible or resistant according to the interpretative standards recommended by the CLSI [19]. As there are no CLSI breakpoints for florfenicol and ceftiofur that are applicable to E. coli of animal origin, the breakpoints of florfenicol (≥16mgL -1 ) and ceftiofur (≥ 8mgL -1 ) were sourced from the Danish Integrated Antimicrobial Resistance Monitoring and Research Program [20] and a previous report [21]. ESBLproducing isolates were screened by double-disk synergy test using both cefotaxime and ceftazidime in the presence or absence of clavulanic acid as recommended by the CLSI.
The DNA sequences obtained after direct sequencing of the amplification products were confirmed using the BLAST algorithm available through the National Center for Biotechnology Information (NCBI).

Detection of ESBL-encoding genes
ESBL-encoding genes (bla TEM , bla SHV , bla CTX-M-1G , bla CTX-M-9G , bla CTX-M-2G , and bla CTX-M-25G ) among the ESBL-producing isolates were analyzed by PCR amplification using previously published primers and protocols [24,25,26]. Among the 136 isolates harboring bla CTX-M-9G , 89 were selected randomly to be directly sequenced using the PCR products, and 33 out of the 63 bla CTX-M-1G -positive isolates were randomly selected to be sequenced. The DNA sequences obtained were compared with genes in GeneBank (http://www.ncbi.nlm.nih.gov/) to confirm the subtypes of ESBL-encoding genes.

Conjugation experiment
Isolates harboring oqxAB and genes encoding ESBLs, were selected for conjugation experiments by the broth-mating method using E. coli C600 as the recipient [27]. Transconjugants were selected on MacConkey agar plates containing streptomycin (1,000mg/L) and cefotaxime (2mg/L). The transconjugants harboring oqxAB and ESBL-encoding genes mentioned above were confirmed by PCR as previously and antimicrobial susceptibility testing for the transconjugants, recipient, and donors.

Plasmids analysis of transconjugants
Incompatibility (Inc) groups were assigned by PCR-based replicon typing of transconjugants [28]. To better clarify IncF plasmids, replicon sequence typing of IncF plasmids was carried out according to the method reported previously [29]. Alleles were assigned by submitting the amplicon sequence to the plasmid multilocus sequence typing (pMLST) database (http://www.pubmlst.org/plasmid).
To analyze the location of the oqxAB gene and ESBLencoding genes of transconjugants, S1 nuclease-PFGE and Southern blot analysis were performed. Briefly, whole-cell DNA of the transconjugants co-harboring oqxAB and ESBLencoding genes embedded in agarose gel plugs was treated with S1 nuclease (TaKaRa, Dalian, China) and separated by PFGE alongside a standard lambda ladder PFG Marker (NEB, UK). Subsequently, Southern blot hybridization was performed with DNA probes specific for oqxB, bla CTX-M-1G or bla CTX-M-9G , which were non-radioactively labeled with a DIG High Prime DNA labeling and detection kit (Roche Diagnostics, Mannheim, Germany). Plasmid DNA extraction was performed using a QIAGEN Plasmid Midi kit (QIAGEN, Germany). Plasmids of transconjugants were digested with the endonuclease EcoRI (TaKaRa Biotechnology, Dalian, China) to analyze the restriction fragment length polymorphism (RFLP) profiles.

Molecular typing
To determine their genetic relatedness, chromosomal DNAs of 109 E. coli isolates randomly selected from the oqxABharboring isolates were digested with XbaI and subjected to pulsed-field gel electrophoresis (PFGE) according to a protocol described previously [30]. The DNA banding patterns were analysed using BioNumerics software version 2.5 (Applied Maths), and a cut-off value of 95% of the similarity values was chosen to indicate identical PFGE types. Salmonella enterica serotype Braenderup H9812 standards served as size markers.

Prevalence of PMQR genes
Four hundred and twenty-nine (61.6%) of the 696 isolates were found to have at least one PMQR gene by PCR and sequencing of the PCR products. oqxAB, found in 328 isolates (47.1% of the total), was the most prevalent PMQR gene, followed by qnrS (14.5%) and aac(6')-Ib-cr (14.4%). The number of isolates harboring qnrB and qepA was 48 (7.0%) and 21 (3.0%), respectively, but no isolate was positive for qnrA, qnrC or qnrD. In addition, 180 of the 328 oqxAB-positive isolates were detected to be linked with IS26. The combination types of PMQR genes in E. coli of different origin were listed in Table 1.

ESBL-encoding genes detection
ESBL-encoding genes were detected in most E. coli isolates with a cefotaxime MIC ≥ 2 μg/mL. CTX-M-type genes were found to be dominant in the isolates with ESBL production, and 191 isolates carried one or two CTX-M genes, representing 27.4% of the total 696 clinical E. coli isolates. bla SHV and bla TEM type ESBL-encoding genes were not found in any of these isolates. Among the 191 bla CTX-M -positive isolates, the number of isolates carrying bla CTX-M-1G and bla CTX-M-9G were 63 and 136, respectively, with 8 isolates carrying both bla CTX-M-1G and bla CTX-M-9G included. As the data of randomly sequenced bla CTX-M that shown in Table 2, the most predominant CTX-M-encoding gene was bla CTX-M-14 (n=36), followed by bla CTX-M-55 (n=29). The most common CTX-M type in isolates from pigs was bla CTX-M-14 , whereas bla CTX-M-27 was the most common type among isolates from avian (Table 2).

Molecular typing
Of the 109 isolates, 97 were successfully typed by PFGE, and a total of 88 different PFGE profiles were obtained, suggesting that most of the isolates in the study were from epidemiologically unrelated E. coli clones.

Co-transferability of oqxAB and bla CTX-M genes and plasmid analysis
In this study, 84 transconjugants carrying bla CTX-M were obtained from the 127 isolates with oqxAB-bla CTX-M . Eighteen of the 84 transconjugants were found to be also positive for oqxAB. Among the 18 transconjugants, 8 carried oqxAB, bla CTX-  very small plasmid not detected using S1-PFGE). As listed in Table 3, the IncFII (5 different alleles) replicon types were detected in 13 transconjugants, 7 of them carrying two Swine ( replicons (FII in combination with FIB, HI2, or N). HI2 replicon type was found in 6 transconjugants, with 3 carrying other replicons. All transconjugants showed extremely high-level resistance to ampicillin, ceftiofur and cefotaxime, at the same level as the donor strains. For quinolones, the transconjugants showed 2~16-, and 4~16-fold increases in the MICs of nalidixic acid, and ciprofloxacin, respectively, when compared with the recipient C600. As shown in Table 3, all the transconjugants were multidrug-resistant and showed resistance to more than two non-β-lactam antimicrobial agents. Notably, 13 transconjugants were found to show high-level resistance to florfenicol, a veterinary antibiotic commonly used in veterinary medicine and aquaculture. The floR gene, which confers resistance to florfenicol, was found in all the 13 transconjugants. The results of Southern blot hybridization revealed that oqxAB and bla CTX-M were located on the same plasmid in the all 18 transconjugants (Figure 1). And floR was also located on these plasmids in the 13 transconjugants resistant to florfenicol. Interestingly, isolates FS3Z3C and FS3Z3G sharing the same PFGE pattern were both resistant to flofenicol, however, the two F18: A-: B1 plasmids of their transconjugants were different (floR in FS3Z3GT, none in FS3Z3CT) (Table 3 and Figure 1E) As shown in Figure 2, the two F33:A-: B-plasmids p42-2 and pFS5E1DT shared the same EcoRI digestion profiles. This result could be confirmed by the positions of the bands in lane 7 and 8 in Figure 1. The EcoRI digestion profile of plasmid from FS2Y1XT was only one band different from that of p42-2 and pFS5E1DT ( Figure 2). As shown in Figure 2, the 6 HI2 plasmids carrying both oqxAB and bla CTX-M-9G also showed the same or highly similar EcoRI digestion profiles. The plasmids

Discussion
In the present study, the prevalence of ESBLs in PMQRpositive (especially oqxAB) clinical E. coli isolates from foodproducing animals in South China was investigated and the characteristics of plasmids carrying oqxAB-bla CTX-M were also achieved. Surveys on the coexistence of oqxAB and ESBLs among Enterobacteriaceae have been reported [12,15,31], however, the characteristics of plasmids with oqxAB-bla CTX-M were not analyzed in these previous studies. In the present study, a high prevalence (61.6%) of PMQR determinants was found in the 696 E. coli isolates from diseased food-producing animals, similar to our previous work [32]. The positive rate of oqxAB in E. coli from pigs in this study was similar to that found in E. coli from pigs in China [33]. Though olaquindox, the main substrate of efflux pump OqxAB, has been forbidden in poultry since 2000 due to its toxic side effects, the positive rate of oqxAB (46.2%) in E. coli from avian was significantly higher than that in E. coli isolates from chicken in 2002 in China [33]. This indicated the rapid dissemination of oqxAB in E. coli from animals in China in recent years. In this study, oqxA was found to be flanked by IS26 in 54.9% of the oqxAB-carrying isolates, which suggests that the mobile element may play an important role in the dissemination of oqxAB among different E. coli strains. In this study, 322 isolates were resistant to ceftiofur and 70.8% of the ceftiofur-resistant isolates were found to be positive for ESBLs, which was similar to the result (68.3%) of a previous study [21]. There might be two reasons for the higher ceftiofur resistance rate: (i) there might be other β-lactamases not included in this study among these isolates, especially bla CMY-2 , often conferring resistance to ceftiofur rather than cefotaxime; (ii) ceftiofur, one of the good substrates of AcrAB efflux pump [34], is often used to treat animal diseases and this long-term pressure will contribute to the presence of ceftiofurresistant isolates. Among the 696 E. coli isolates studied, 29.6% of them were ESBL-producers and 27.4% carried CTX-M-β-lactamases, which were both much higher than the detection rates (13.1% and 12.4% respectively) in E. coli from healthy animals in a recent report [35] (P<0.001). The different incidence of ESBLs amongst E. coli isolates from food animals may be due to the use of third-generation cephalosporins in the diseased or dead food-producing animals in this study. At least ten types of bla CTX-M genes were found in this study, indicating the bla CTX-M genes in E. coli from food-producing animals in China were diverse. The predominant bla CTX-M type in this study was bla CTX-M-14 , similar result was also reported in E. coli isolates from humans and animals in China [35,36,37], however, bla CTX-M-1 was the most common type in some countries in Europe like England, and France [38,39].
We observed a significantly higher prevalence of CTX-M genes in the 429 PMQR-positive isolates (37.3%) than that (11.6%) in PMQR-negative isolates. This result supports previous findings that PMQR genes are often linked with ESBL production [9]. In addition, the detection rate of CTX-M type genes in oqxAB-positive isolates (38.7%) was significantly higher than that in oqxAB-negative isolates (17.4%) (P<0.001), indicating bla CTX-M might have significant relationship with the new PMQR determinant, oqxAB. Among the 127 isolates with oqxAB-bla CTX-M , plasmids carrying bla CTX-M from 84 isolates (66.1%) were conjugatively transferable, similar to the rate in a previous report in China [35]. Co-transferability of oqxAB and bla CTX-M occurred in 18 (14.2%) of the 127 isolates, providing support for our previous hypothesis that oqxAB has correlation with ESBL. oqxAB and bla CTX-M were confirmed to be located on the same plasmids in all the 18 isolates. To our knowledge, this is the first report of the prevalence of plasmids carrying oqxAB and bla CTX-M . Association of multiple antibiotic resistance genes on the same transferable plasmids has been an important mechanism of dissemination of multidrug resistance, and the transferable oqxAB-bla CTX-M plasmids might explain in part the rapid increasing prevalence of oqxAB in E. coli of foodproducing animals. Because OqxAB has a wide substrate specificity, the existence of its substrates in the environment will increase the resistances of E. coli isolates to fluoroquinolones and cephalosporins. In addition, floR was located on the plasmids carrying oqxAB-bla CTX-M in 13 transconjugants, indicating that the application of florfenicol, commonly used in veterinary medicine and aquaculture, will also increase the resistances of E coli to fluoroquinolones and cephalosporins. In 8 transconjugants, aac(6')-Ib-cr was also located on the plasmids carrying oqxAB-bla CTX-M , consistent with the findings that aac(6')-Ib-cr is linked to bla CTX-M in Enterobacteriaceae [9]. IncFII replicon types were detected in 13 of the 18 transconjugants, consistent with previous findings that most bla CTX-M genes or oqxAB were found to be linked with IncFII plasmids [33,35]. In addition, HI2 plasmids were also often found to be linked with the co-dissemination of oqxABbla CTX-M in this study. Three of the 6 HI2 plasmids were also positive for other replicon types (FII or N), and this might be explained by the presence of a multireplicon fusion of the HI2 plasmid with other replicon type plasmids, similar to a previous report [28]. Though the donors had different PFGE patterns, the F33:A-: B-and HI2 plasmids with oqxAB-bla CTX-M -floR genes shared highly similar RFLP profiles. This indicates that these two type plasmids might mediate the dissemination of oqxAB-bla CTX-M -floR genes in E coli from food-producing animals, and whether this represents global dissemination of these plasmids in the future is unclear and further work is required to clarify this.
In conclusion, we report a high prevalence (37.3%) of bla CTX-M among PMQR-positive E. coli strains from diseased foodproducing animals in China between 2002 and 2012. The codissemination of oqxAB, floR and bla CTX-M genes in E. coli isolates from food-producing animals is mediated mainly by the F33:A-: B-and HI2 plasmids. These plasmids may promote the development of high-level multidrug-resistant isolates. Although a low prevalence (14.2%, 18 of the 127 oqxAB-bla CTX-M -positive isolates) of transferable plasmids carrying oqxAB-bla CTX-M and dissemination of them mediated by similar plasmids were observed in the clinical E. coli isolates from food-producing animals, continued surveillance of the dissemination of these plasmids in Gram-negative bacteria is urgently needed because of the possibility that plasmids can be exchanged between bacteria from animals and those from humans. To our knowledge, this is the first report on the prevalence of the cotransferability of oqxAB and bla CTX-M genes. This is also the first description of the co-existence of the oqxAB, floR, and bla CTX-M on the same plasmid in E. coli.