Prevalence of Salmonella Isolates from Chicken and Pig Slaughterhouses and Emergence of Ciprofloxacin and Cefotaxime Co-Resistant S. enterica Serovar Indiana in Henan, China

The prevalence of Salmonella from chicken and pig slaughterhouses in Henan, China and antimicrobial susceptibility of these isolates to antibiotics was determined. From 283 chicken samples and 240 pig samples collected, 128 and 70 Salmonella isolates were recovered with an isolation rate of 45.2 and 29.2% respectively. The predominant serovars in chicken samples were S. enterica serovar Enteritidis, S. enterica serovar Hadar and S. enterica serovar Indiana, while those in pig samples were S. enterica serovar Typhimurium, S. enterica serovar Derby and S. enterica serovar Enteritidis. Resistance to ciprofloxacin was 8.6 and 10.0% for isolates from chickens and pigs respectively, whereas resistance to cefotaxime was 5.5 and 8.6%, respectively. Multidrug resistance (resistance to three or more classes of antimicrobial agent) was markedly higher in pig isolates (57.1%) than in chicken isolates (39.8%). Of particular concern was the detection of ciprofloxacin and cefotaxime co-resistant S. enterica serovar Indiana isolates, which pose risk to public health. All 16 S. enterica serovar Indiana isolates detected were resistant to ciprofloxacin, among which 11 were co-resistant to cefotaxime. The S. enterica serovar Indiana isolates accumulated point mutations in quinolone resistance determination regions of gyrA (S83F/D87G or S83F/D87N) and parC (T57S/S80R). Two plasmid mediated quinolone resistant determinants were found with aac (6')-Ib-cr and oqxAB in 16 and 12 S. enterica serovar Indiana isolates respectively. Cefotaxime-resistance of S. enterica serovar Indiana was associated with the acquisition of a bla CTX-M-65 gene. The potential risk of ciprofloxacin and cefotaxime co-resistant S. enterica serovar Indiana infection is a significant concern due to limited alternative treatment options. Reduction of Salmonella in chicken and pig slaughterhouses, in particular, ciprofloxacin and cefotaxime co-resistant S. enterica serovar Indiana will be an important measure to reduce the public health burden of Salmonella infections.


Introduction
Foodborne salmonellosis is an important public health problem worldwide [1]. There is an estimated 94 million annual cases of non-typhoidal Salmonella infections with 155,000 deaths and 85% of these cases are foodborne [2]. Salmonellosis is also one of the main causes of morbidity in China, accounting for 75% (30 million cases) of the foodborne diseases [3]. Children younger than 5-years were the group most affected by Salmonella-related infection in several cities in China [4,5,6]. Chickens and pigs are the main reservoirs of human foodborne nontyphoidal Salmonella infections. Epidemiological investigations have shown that the contaminated raw or undercooked chicken-and pig-meat was the primary vehicle for transmission to humans [7,8]. Fluoroquinolones and extended-spectrum cephalosporins are the most frequently used antimicrobial agents for treating invasive salmonellosis in humans, especially children and the elderly [1,5]. On the other hand, antimicrobials are also used for animal growth promotion and this application can select for and disseminate antimicrobial-resistant Salmonella. Multidrug-resistant (MDR) Salmonella strains that emerged in animal populations may pass from food animals to humans via the food chain [1]. Recent studies have shown that food-producing animals and retail food products might be important reservoirs of fluoroquinolone and/or extended-spectrum cephalosporin-resistant Salmonella which have already been reported in many countries [5,9]. Salmonella enterica serovar Indiana (S. enterica serovar Indiana) is one of the most frequently isolated Salmonella serovars from many sources [10,11]. And more importantly, the ciprofloxacin and ceftriaxone co-resistant S. enterica serovar Indiana have been isolated from retail chickens in Beijing and sick animals (ducks and pigs) in Gangdong province, China [10,12].
The prevalence of these co-resistant isolates could be amplified through cross-contamination during the slaughtering process [13]. However, limited data are available for the distribution and characteristics of ciprofloxacin and cefotaxime co-resistant Salmonella isolates from chicken and pig slaughterhouses.
In Salmonella, ciprofloxacin resistance is mainly attributed to mutations in gyrA and parC genes [14]. Plasmid-mediated quinolone resistance (PMQR) determinants such as qnrA, qnrB, qnrD, qnrS, and oqxAB may also contribute to ciprofloxacin resistance [12,15]. Resistance to cefotaxime or other extended spectrum beta-lactams is usually due to intracellular production of extended spectrum β-lactamases (ESBLs). The most commonly found ESBL types in Asia are the CTX-M group, which are usually located on transmissible plasmids that tend to disseminate among members of Enterobacteriacea [16]. In this study, Salmonella isolated from chicken and pig slaughterhouses were tested for their susceptibility to antimicrobial compounds commonly used in human medicine. The genetic basis of resistance to ciprofloxacin and cefotaxime resistance and the genetic relationships among these isolates were also determined.

Sample collection and Salmonella serotyping
From February to December in 2011, 283 whole chicken carcasses post cooling bath samples from chicken slaughterhouses with processing capacity of 10,000~50,000 chickens per day were collected from 2 cities (Hebi and Zhoukou) in Henan province, China. The chicken slaughterhouses were Qixian TIYA and Tangying YUDA in Hebi and Zhoukou TIYA and Shaizai TIYA in Zhoukou. For pig samples, 120 surface swabs of four sites of pig carcasses after evisceration and before chilling and 120 ileocaecal lymph nodes without fat or connective tissue samples from pig slaughterhouses with processing capacity of 2,000~3,000 pigs per day were collected from 2 cities (Hebi and Luohe) in Henan province, China. The pig slaughterhouses were Chengsheng and Xihui in Hebi and Luohe respectively. Both the chicken and pig slaughterhouses were the largest-scale slaughtering and process lines in the regions sampled. No more than five samples were collected from the same slaughterhouse per visit during the study. Each sample was put into an aseptic plastic bag, marked and shipped to the laboratory within 24 h on ice. These chickens and pigs were killed as part of the slaughterhouse routine. The whole chicken carcasses and ileocaecal lymph nodes from pigs were purchased during the project.
For whole chicken carcasses samples, each sample was immediately removed from the bag in which it was transported and then placed into a 3500 stomacher bag (Seward, UK) followed by the addition of 500 mL buffered peptone water (BPW; Becton-Dickinson, USA) per kilogram and thoroughly manually massaged for 3-5 minutes to ensure the surface, internal and external of the chicken carcass were fully in contact with the rinse. Then samples were incubated at 37°C for 22-26 h.
Surface swabs for sampling of the pig carcasses were performed after evisceration and before chilling. A surface of about 100 cm 2 per site was swabbed using one single abrasive sponge for 4 sites (hind limb, abdomen, mid-dorsal region, jowl). The swabs were transferred to 100 mL BPW and incubated at 37°C for 18-20 h. For Lymph nodes samples from pigs, lymph nodes from each pig were removed and pooled, surface de-contaminated before analysis by dipping into 75% (v/v) alcohol and drying in air for 3~5 minutes. Twenty-five grams of lymph nodes were placed into a plastic bag with 225 mL BPW added and mechanistically homogenized by banging with a hammer and then incubated at 37°C for 18-20 h.
For all the samples, 0.5-and 0.1-mL of the pre-enrichment culture were transferred to 10 mL tetrathionate broth (TT; Becton-Dickinson, USA) and Rappaport-Vassiliadis (RV; Becton-Dickinson, USA) broth and incubated at 42±1°C with shaking at 100 rpm for 22-24 h. After selective enrichment, a loopful of TT or RV broth culture was streaked onto xylose lysine tergitol 4 (XLT4; Becton-Dickinson, USA) agar, and incubated at 37±1°C overnight. Three presumptive Salmonella colonies on each XLT4 plate were picked and inoculated onto a triple sugar iron slant (Becton-Dickinson, USA) and incubated at 35±1°C for 24 h. Isolates with typical Salmonella morphologies were confirmed by amplification of the invA gene by PCR [17] and the API 20E biochemical identification system (BioMérieux, Beijing, China). For all confirmed Salmonella isolates, serovars were determined by slide agglutination with commercial Salmonella antisera (Statens Serum Institute, Denmark) following the Kauffmann-White scheme.

Antimicrobial susceptibility testing
Antimicrobial susceptibility of all Salmonella isolates was determined by the agar dilution method and interpreted according to the Clinical and Laboratory Standards Institute guidelines (CLSI) with the current MIC breakpoints being used [18] . Multidrug resistance was defined as resistance to three or more classes of antimicrobial agent. Isolates with an MIC 1 mg/L for ceftriaxone or ceftazidime were further screened for extended-spectrum β-lactamase (ESBL) production by determination of synergy between 0.25 and128 mg/L ceftazidime or cefotaxime and 4 mg/L clavulanate. The tigecycline MIC interpretive breakpoint was recommended by the European Committee on Antimicrobial Susceptibility Testing (EUCAST-2012; www.eucast.org). Escherichia coli ATCC 25922 and Klebsiella pneumoniae ATCC 700603 were used as quality control organisms in antimicrobial susceptibility tests.
PCR amplification and DNA sequencing of S. enterica serovar Indiana isolates The quinolone resistance determination regions (QRDRs) of gyrA, gyrB, parC and parE in Salmonella isolates were amplified by PCR as described previously [14], sequenced, and then compared to the genes from Salmonella enterica serovar Typhimurium LT2 to identify any mutated positions in the target gene. The presence of one or more PMQR genes qnrA, qnrB, qnrC, qnrD, qnrS, qepA, oqxAB and aac(6')Ib-cr was determined by PCR and sequencing using primers described previously [12,15]. The genotypes of ESBL-positive isolates were determined by PCR targeting known β-lactamase genes as previously described (bla TEM , bla SHV , bla CMY , bla CTX-M , and bla OXA ) [12,19]. All the PCR products were either directly sequenced or cloned into pMD18-T plasmid (Takara Biotechnology Cooperation, Dalian, China) for sequence analysis at Takara Biotechnology Cooperation. The sequences obtained were analyzed by Sequencher 4.6 software (Gene Codes Corporation, Ann Arbor, MI, USA). A search for homologous sequences was performed using the BLASTn program at the U.S. National Center for Biotechnology Information website (http://www.ncbi.nlm.nih.gov/BLAST/).

Pulsed-Field Gel Electrophoresis (PFGE) and Multilocus sequence typing (MLST) of S. enterica serovar Indiana isolates
Genomic DNA for PFGE was prepared in accordance with the method described previously [20]. Briefly, the genomic DNA of the test strains were digested with 50 U XbaI for 2 hours and separated on 1% agarose SeaKem Gold gel with the CHEF DR III system (Bio-Rad, Hercules, California). The PFGE patterns were interpreted with Bionumerics software (Applied Math, Belgium) using Unweighted Pair Group Method with Arithmetic Mean (UPGMA). Bands which were smaller than 20.5 kb were not included in the analysis. Patterns indistinguishable by computer and visual inspection were assigned the same pattern designation, position tolerance of 1%. Clusters were defined as DNA patterns sharing 80% similarity (A, B, C).

Plasmid replicon typing and conjugation experiment of S. enterica serovar Indiana isolates
Plasmids were isolated using Plasmid Plus Midi kit (Qiagen, Germany). Replicon typing of plasmids was performed using a PCR-based method as previously described [22]. Conjugation experiments were performed for all Salmonella isolates using a sodium azide-resistant E. coli J53 as recipient strain as previously described [23]. Transconjugants were selected on LB agar plates containing cefotaxime (2 μg/mL) and sodium azide (100 μg/mL).

Statistical analysis
The Pearson Chi-Square test were used to determine the significant difference (p < 0.05) of prevalence of Salmonella in chicken and pigs with different characteristics and the rates of resistance between ciprofloxacin and cefotaxime co-resistant Salmonella isolates and ciprofloxacin and cefotaxime none-co-resistant Salmonella isolates. Statistical software SPSS for Windows (version 17.0, SPSS, Inc., Chicago, IL) was used for descriptive analysis.
In total, 11 S. enterica serovar Indiana isolates were identified as being co-resistant to ciprofloxacin and cefotaxime, with 6 and 5 of these being isolated from chickens and pigs, respectively. These isolates were also resistant to 7 other antimicrobial agents (AMP, CAZ, CHL, GEN, NAL, SXT, TET) ( Table 5). In contrast, the ciprofloxacin and cefotaxime -co-resistant Salmonella isolates alone showed high rate of resistance to most of the antibiotics except nalidixic acid (Table 5). Nine isolates were ciprofloxacin and cefotaxime non-co-resistant with 7 and 2 Salmonella isolates being resistant to ciprofloxacin only and cefotaxime only respectively. The ciprofloxacin resistant only isolates were from two different serovars, and 5 of these were S. enterica serovar Indiana isolates from chickens along with 2 S. enterica serovar Typhimurium isolates from pigs. The cefotaxime resistant only isolates were from two different serovars, a S. enterica serovar Infant isolate from a chicken sample and a S. enterica serovar Enteritidis isolate from a pig sample.

Identification of quinolone resistance determinants of S. enterica serovar Indiana isolates
Sixteen isolates were resistant to ciprofloxacin with minimum inhibitory concentration (MIC) values between 64 and 128 μg /mL. PCR and sequencing was used to identify point mutations   We also determined the carriage of the PMQR determinants and found that all 16 isolates carried aac(6')-Ib-cr, among which 12 isolates also carried oqxAB. None of the isolates carried qnrA, qnrB, qnrC, qnrD, qnrS or qepA.
Occurrence of β-lactamase-encoding genes of S. enterica serovar Indiana isolates Among the 16 S. enterica serovar Indiana isolates, 11 carried genes encoding CTX-M-type ESBLs and the subtype of bla CTX-M was identified as bla CTX-M-65 for all 11 isolates. The ESBL genotypes, bla OXA-1 and bla TEM-1 were also identified in 16 and 5 isolates respectively.

PFGE and MLST analysis of S. enterica serovar Indiana isolates
To further elucidate the genetic relationships of the 16 ciprofloxacin and/or cefotaxime resistant S. enterica serovar Indiana isolates by MLST and PFGE. All 16 isolates belonged to the same sequence type ST17. By PFGE, the 16 isolates were clustered into 12 pulsotypes. A dendrogram was constructed using these PFGE patterns. Isolates grouped into the same PFGE patterns mostly were recovered from the same regions. The predominant pulsotype was denoted as CN001, which included five isolates from the same pig slaughterhouse in Hebi on the same day (23 rd , February 2011). The remaining 11 pulsotypes contained a single isolate (Fig 1). Cluster A consisted of mainly ciprofloxacin and cefotaxime co-resistant Salmonella isolates of which the isolates from chicken and pig slaughterhouses were different. Cluster C contained only ciprofloxacin resistant Salmonella isolates with the exception of isolate 11C28. Those isolates containing point mutations in gyrA (S83F/D87N) and parC (T57S/S80R) were all grouped together in cluster A whilst those with point mutations in gyrA (S83F/D87G) and parC (T57S/ S80R) were mostly grouped into clusters B and C. Table 5. Resistance phenotypes of ciprofloxacin and cefotaxime co-resistance Salmonella isolates and none ciprofloxacin and cefotaxime coresistance Salmonella isolates from chicken and pig slaughterhouses.

Antimicrobials
No. of resistant isolates Plasmid replicon typing and conjugation experiment of ESBL-S. enterica serovar Indiana isolates Conjugation experiments were conducted on these isolates using E. coli J53 as a recipient strain. However, no transconjugants were obtained. Seven isolates contained plasmids belonging to the IncN group and none of the 18 plasmids was detected in the rest of 4 ESBL-Salmonella isolates.

Discussion
In this study, we screened processed chickens and pigs in slaughterhouses collected from 3 districts of Henan province which was one of the largest food-animal processing provinces in China for the presence of Salmonella, and further characterized the isolates using serotyping and of antimicrobial susceptibility testing. Specifically the genotypic mechanisms of ciprofloxacin-resistant and cefotaxime-resistant isolates were identified. There had been a few studies on prevalence of Salmonella in poultry in retail markets in China wherein values varied from 38.9 to 65.3% depending on the geographical [24]. Our isolation rate of 45.2% from chickens in slaughterhouses was similar to that of 47.9% reported earlier from retail markets in the same (Henan) province [24], suggesting similar levels of contamination in the entire supply chain. However, our isolation rate was much higher than that (22.1%) from a slaughterhouse in Anhui province reported by Wang et al. [11]. The top serovars found in this study were S. enterica serovar Enteritidis, S. enterica serovar Indiana and S. enterica serovar Typhimurium. Two of the three (S. enterica serovar Enteritidis, S. enterica serovar Indiana) were also counted in the top three Salmonella serovars (S. enterica serovar Enteritidis, S. enterica serovar Hadar and S. enterica serovar Indiana) in chickens in China as reported by other studies [10,25,26]. These serovars were within the top four serovars (S. enterica serovar Typhimurium, S. enterica serovar Enteritidis, S. enterica serovar Derby, S. enterica serovar Indiana) isolated from patients in Henan from 2006 to 2007 [27], consistent with poultry as a major source of Salmonella infection in humans.
The isolation of Salmonella from pigs was much lower and the prevalent serovars were also different. The isolation rate of 29.2% was higher than the isolation rate from slaughterhouses reported by the EFSA (10.3%) [28] and Li et al. (12.0%) [29], but similar to that from pork in retail market (31%) reported by Yang et al. [30]. The differences were likely due to sources or regional difference. The common serovars found in this study (S. enterica serovar Typhimurium, S. enterica serovar Derby and S. enterica serovar Enteritidis) were also reported to be common in pigs by the EFSA [28].
Salmonella is still the leading cause of hospitalization in USA [31] and MDR Salmonella has become a significant public health concern worldwide [32]. Our data showed that nearly half of the isolates (46.0%) from Henan province were multidrug resistant which was the same as that (46.0%) from patients in Guangdong province [4] and a somewhat higher than that (34.9%) in Beijing [10] but was sharply lower than that (93%) in Anhui province [11]. The high occurrence of multidrug resistant Salmonella in slaughterhouses indicates that crosscontamination in the subsequent retail process poses serious risk to public health. With the large-scale use, and often abuse, of antimicrobials over time, fluoroquinolone and/or Prevalence of Salmonella from Slaughterhouses in Henan, China extended-spectrum cephalosporin resistant Salmonella isolates have been detected from patients in numerous locations with variable prevalence in China, such as 6.7% in Gangdong province [4], 12.2% in Wuhan [5], 17.2% in Shanghai [6] and 26.9% in Henan province [27]. Importantly, 10.1% Salmonella isolates exhibited resistance to ciprofloxacin or third generation cephalosporin in this study. Fluoroquinolone resistance in Salmonella is a unique problem in China and other Asian countries where there is no strict control on its availability [5,33], in contrast to countries with strict antimicrobial controls, such as the United States, where less than 3% of Salmonella isolates were considered resistant to ciprofloxacin [34] and in Belgium, wherein resistance to ciprofloxacxin was undetectable in isolates from both healthy pigs and chickens [35].
In this study, we found 11 Salmonella isolates that were co-resistant to ciprofloxacin and cefotaxime and all of which were S. enterica serovar Indiana. The appearance of ciprofloxacin and cefotaxime co-resistant S. enterica serovar Indiana is alarming. It seems that the co-resistance only developed recently since the study by Xia et al. in 2009 [27] found that S. enterica serovar Indiana from Henan province were susceptible to cefotaxime [27]. The emergence of cefotaxime resistant S. enterica serovar Indiana could either be due to acquisition of cefotaxime resistance genes by local strains or through spread from other regions where ciprofloxacin and cefotaxime co-resistant S. enterica serovar Indiana has been reported such as Beijing [10] or by cross-genus spread from ciprofloxacin and cefotaxime co-resistant E. coli which has been reported in Gangdong province, China [36].
Two PMQR determinants were identified in our S. enterica serovar Indiana isolates withaac (6')-Ib-cr present in all 16 isolates and oqxAB present in 12 isolates. The distribution of the oqxAB-positive S. enterica serovar Indiana isolates on the PFGE dendrogram (Fig 1) suggests that it was acquired independently 3 times. OqxAB, a novel efflux pump, that mediates resistance to olaquindox and also extrudes antibiotics such as chloramphenicol and fluoroquinolones, broadening the resistance phenotype to these latter antibiotics [37]. Olaquindox, a quinoxaline derivative antibiotic, is as a growth promoter that has been in use in animals for decades in China. The latter feature is suggestive of a constant selective pressure for the acquisition and dissemination of oqxAB [12,38]. The co-existence of PMQR genes, aac(6')-Ib-cr and oqxAB, had been detected previously in both human and animal Salmonella isolates in China [12]. Although PMQR confers only low-level quinolone resistance, it can facilitate the emergence of high-level resistance via mutations in a topoisomerase gene gyrA, parC or both [39]. This may be also the case for the 16 S. enterica serovar Indiana isolates in this study, PMQR combined with multiple mutations in QRDRs [gyrA (S83F/D87N) or gyrA (S83F/D87G), parC (T57S/S80R)] resulted in high-level resistance to ciprofloxacin (MIC, 64μg/mL).
Uniquely in China, our study found that bla CTX-M-65 was carried by the 11 cefotaxime resistant S. enterica serovar Indiana isolates. In contrast bla CTX-M-14 and bla CTX-M-27 have been found in S. enterica serovar Indiana isolates from chickens and pigs in Gangdong [12] and bla CTX-M-24 was found in S. enterica serovar Indiana isolates from chickens in Shandong [40]. These bla CTX-M genes are usually located on conjugative plasmids [16]. The bla CTX-M-24 found in S. enterica serovar Indiana by Lai et al. was shown to be encoded on a unique class 1 integron with multiple resistance genes co-located on a IncHI2 plasmid [40] whilst bla CTX-M-14 and bla CTX-M-27 were found in S. enterica serovar Indiana by Jiang et al. and was shown to be encoded on IncHI2 and/ or IncN plasmid [12]. Similarly bla CTX-M-24 ,bla CTX-M-27 and bla CTX-M-65 have been reported in E. coli from both animals and humans in China [36,41]. It is possible that these bla CTX-M genes were acquired by Salmonella from E. coli.
The genetic relationship of the S. enterica serovar Indiana isolates based on PFGE revealed that all except one ciprofloxacin and cefotaxime co-resistant isolate were grouped together in one cluster, denoted as cluster A, suggesting a single origin. The pig isolates were identical and belonged to this cluster, indicating likely the possible spread from chickens to pigs. The ciprofloxacin and cefotaxime co-resistant isolate located outside the cluster was likely to have acquired the bla CTX-M-65 gene independently. The isolate also carried a different gyrA mutation (S83F/D87G), different from those of the isolates in cluster A, lending further support to its independent origin. Therefore both clonal spread and independent acquisition of cefotaxime resistance may have acted to contribute to the emergence and spread of ciprofloxacin and cefotaxime co-resistant S. enterica serovar Indiana in Henan province.
Multidrug resistant S. enterica serovar Indiana has already been reported in human infections in China [6,27]. All S. enterica serovar Indiana isolates from human infections in Henan were resistant to ciprofloxacin but susceptible to cefotaxime [27]. It is inevitable that the ciprofloxacin and cefotaxime co-resistant S. enterica serovar Indiana emerged in Henan will transmit to humans, posing a significant threat to public health. The co-resistance limits the optimal treatment alternatives of salmonellosis in humans.

Conclusion
The prevalence of Salmonella in the chicken and pig slaughterhouses in Henan was found to be high, and represented by a small number of different serovars such as S. enterica serovar Enteritidis, S. enterica serovar Hadar and S. enterica serovar Indiana in chickens and S. enterica serovar Typhimurium, S. enterica serovar Derby and S. enterica serovar Enteritidis in pigs. This finding highlights the fact that food-producing animals at processing plants are a significant reservoir of antimicrobial-resistant Salmonella. Our data may contribute to the development of interventions at the processing level to interrupt the transmission of Salmonella to humans. Ciprofloxacin and cefotaxime co-resistant S. enterica serovar Indiana possibly emerged only recently through acquisition of bla CTX-M-65 . The potential risk of ciprofloxacin and cefotaxime co-resistant S. enterica serovar Indiana infections in humans requires urgent attention. The emergence of ciprofloxacin and cefotaxime co-resistance also underscores the importance of strict control on the use of antibiotics in animal production to prevent the emergence of antibiotic resistance in Salmonella.