Genotypic and Phenotypic Characterization of Antimicrobial-Resistant Escherichia coli from Farm-Raised Diarrheic Sika Deer in Northeastern China

In China, overuse and/or abuse of antimicrobials are common in stockbreeding, which possess high risks of antimicrobial-resistant contaminations. The serogroups, major virulence genes, and antimicrobial resistant patterns of the antimicrobial-resistant Escherichia coli (E. coli) were investigated in the feces of diarrheic farm-raised sika deer from 50 farms in three Northeastern provinces of China. A total of 220 E. coli isolates were obtained and characterized. Twenty-eight O serogroups were identified from the obtained E. coli isolates with O2, O26, O128, O142 and O154 being dominant. Nearly all the isolates were resistant to at least four of the tested antimicrobials. More than 90% of the E. coli isolates carried at least one of the tested virulence genes. About 85% of the E. coli isolates carried one or more antimicrobial-resistant genes responsible for resistant phenotypes of sulfonamides, streptomycin/spectionomycin or tetracycline. The antimicrobial resistant level and pathogenic group occurrences of the obtained E. coli isolates were higher than that of livestock and wild animals reported in some developed countries. Thus, the fecal-carrying antimicrobial-resistant E. coli from the farm-raised sika deer is potentially a significant contamination source for freshwater systems and food chain, and may pose great health risks for human and animals in Northeastern China.


Introduction
Antimicrobial resistance (AMR) in enteropathogens has become a major public health problem due to its potential infections on human and animals [1]. The domestic animals are usually considered as major reservoirs for antimicrobial-resistant bacteria. Recently, increasing interest has been given to antimicrobialresistant pathogenic bacteria from various domestic animals and their habitats. Antimicrobial-resistant bacteria have been detected in a variety of domestic animals and the environments that are affected by stockbreeding [2,3]. E. coli, a type of bacteria common in the intestine of warm-blooded animals, was widely used as an indicator of fecal contamination in drinking water system assessment and food safety evaluation [4]. Pathogenic E. coli is an important pathogen that can infect humans and animals. Various pathotypes of E. coli can be distinguished by the virulence genes [5]. Infection by pathogenic E. coli mainly cause diarrhea in domestic livestock, especially in young animals with clinical syndromes including acute severe watery diarrhea, haemorrhage, and sudden death [6]. Among the identified pathotypes, EPEC (enteropathogenic E. coli), ETEC (enterotoxigenic E. coli), and STEC (Shiga-like toxin-producing E. coli) strains represented three major classes of enteric pathogens leading to diarrhoea in humans and animals [7]. Previous studies showed that feces from wild deer could contaminate surface water that may be used as drinking water for humans and/or domestic animals [8]. Thus, the pathogenic E. coli strains from domesticated wild animals (e.g., domesticated wild deer) can also be transmitted to humans [9,10].
As domesticated wild animals, China has a large population of farm-raised sika deer (Cervus nippon), mainly distributed in three Northeastern provinces (Liaoning, Jilin and Heilongjiang) with about 80% sika deer production in China [11]. Although the farming pattern of farm-raised sika deer is similar to the domestic ruminants (e.g., cattle, sheep); there are still some uniqueness in its farming production. For example, the mainly purpose of farmraised sika deer in China is to obtain pilose autler rather than for meat production, and the feeding cycle of farm-raised sika deer is usually longer than two years. Therefore, antimicrobials are not added (or only a small amount of antimicrobials are added) into daily feed, but they are used heavily when disease outbreaks. In comparison, the mainly purpose of domestic ruminants in China are meat production with a feeding cycle of less than one year; excessive antimicrobials are applied in the daily feed for improving growth performance. Compare to wild animals (including wild deer), the diarrheic farm-raised deer accepting heavy amounts of antimicrobials for disease treatment may be a larger reservoir of potential AMR pathogens. However, to our knowledge, few studies have been focused on AMR pathogens of farm-raised diarrheic sika deer and their potential risk to public health. The objective of this study is to characterize the diversity of serological types, the distribution of antimicrobial-resistant patterns and virulence genes of AMR E. coli in the feces from farm-raised diarrheic sika deer in Northeastern China.

Results and Discussion
Serogroup differences between farm-raised sika deer and other wild animals/domestic ruminants In this study, 10 isolates with morphology of E. coli were randomly picked up from each deer farm (also from each fecal sample) and subjected to biochemical identification. Among a total of 500 suspect bacterial isolates, 220 of them were identified as typical E. coli strains. One hundred ninety of the obtained E. coli isolates were classified into 28 different types of O serogroups, and the remaining did not belong to any known serogroups (untypable or O-rough) ( Figure 1). Around 60% of the identified O serogroups belonged to twelve major groups: O2, O128, O26, O142, O154, O55, O9, O27, O126, O45, O111, and O125, with the former five O serogroups being dominant. Four E. coli isolates were affiliated with serogroup O157, which is known to be associated with life threatening diseases [10].
More than a half of the identified O serogroups in this study have never been reported from wild deer to date, and a few peculiar O serogroups identified here (e.g., O9, O26, O138 and O157) might be associated with human and livestock infections [5,12,13]. Prevalence rate of E. coli O157 from diarrheic farmraised sika deer was 1.8%, which was significantly higher than 0.3% and 0.25% of wild animals (including wild deer) in Louisiana and Nebraska USA, respectively [14,15]. This indicated that farm-raised sika deer has become natural reservoir for pathogenic E. coli O157 in Northeastern China. The dominant serogroups of the E. coli isolates in this study were markedly different from those of diarrheic sheep flocks (O5, O6, O76, O87, O91, O123, O128, O146, O166 and O176) and cattle calves (O56, O78, O8 and O164) [3,16]. The observed difference in serogroups among farm-raised sika deer, livestock and wild animals might be ascribed to their different feeding managements, habitats and antimicrobials ingests. The underlying reasons await further investigation.
The astA gene and typical pathogenic E. coli (EPEC, ETEC, and STEC) from the farm-raised sika deer occurred more frequently than that of wild animals (including wide deer), and even than that of some domestic livestock. The astA gene encodes the toxin EAST1, which is associated with diarrhoea of postweaning pigs [5,17]. Among the E. coli isolates from farm-raised sika deer, a high frequency of 60.9% was observed for the astA gene. The frequency of the gene eaeA (expressing the virulence of EPEC) was 19.5% in the obtained E. coli isolates, in comparison with 1.4% and 0.9% for Danish and Slovakia postweaning diarrhea pigs, respectively [5,12]. Fifty-one isolates (23.2% of total obtained) contained stx1 and/or stx2 genes, in contrast with lower frequencies (16.3% and 10.5% ) of stx-gene containing E. coli isolates obtained from wild deer [18,19]. The prevalence rates of the EPEC (20.9%) and STEC (23.2%) strains in this study were much higher than those (1.5% for EPEC and 5.5% for STEC) of wildlife in the south Belgium [7]. ETEC causes travelers diarrhoea by producing different combinations of heat labile (LT) and heat stable (ST) enterotoxins. The prevalence rate of ETEC strains was 31.4% in this study, compared with that of 55.3% and 33.2% from diarrheal pigs and calves, respectively [20,21].
Antimicrobial resistance of the E. coli isolates from the farm-raised sika deer All the obtained E. coli isolates showed resistance to at least one of the tested antimicrobials (Table 2). Two hundred and eight (94.5%) E. coli isolates were resistant to 4 or more antimicrobials and 170 (77.3%) E. coli isolates were resistant to 7 or more antimicrobials. Forty-eight (21.8%) E. coli isolates were capable of resisting 9 antimicrobials. The most frequently resisted antimicrobials were sulfadiazine, sulfanethazine, tetracycline, ampicillin, amoxicillin, chloramphenicol, gentamicin, and ceftriaxone, which were the main antimicrobials used for sika deer diarrhea treatment. These results suggested that the antimicrobials resistance of the E. coli might be derived from the antimicrobials overuse of their hosts. Similar to the livestock in China, farm-raised sika deers are usually supplied with heavy antimicrobials (e.g., cephalosporin, fluoroquinolone, aminoglycoside and sulfonamides) for disease treatment. Nearly all of the E. coli isolates in this study resisted four or more antimicrobials, which was similar to that of foodproducing animals reported in China and other countries [1,22,23,24]. Due to less contacted with antimicrobials, much lower abundances (7.3% and 8.8%, respectively) of the E. coli isolates from deer in Norway and USA were resistant to one or more of antimicrobials [8,25]. A few previous studies showed that more than 60% of the E. coli isolates from humans and foodproducing animals in China were resistant to fluoroquinolone drugs [23,26]. In this study, 34.1%, 36.8%, and 35.0% of the E. coli isolates resisted ciprofloxacin, enrofloxacin, and norfloxacin, respectively. However, fluoroquinolone resistance was relatively low in other countries. For example, only 2.9% of the E. coli isolates from Danish dogs were resistant to ciprofloxacin and 8.0% of the E. coli isolates from Korean pigs were resistant to enrofloxacin [27,28]. The E. coli isolates from wild deer were very sensitive to cephems [7]. Whereas, the cephems resistance of E. coli isolates this study was even much higher (26.4-48.2%) than that of food-producing animals (8.4% and 6.3%) [29,30]. These significant differences may be ascribed to the fact that the antimicrobials are frequently overused on the deer farms in China.
The pathotypes and antimicrobials resistant genotypes of the obtained E. coli isolates resisting tetracycline, sulfonamides, and streptomycin-spectinomycin are shown in Table 3. The genes of aadA, tetA, strA, strB and sul2 were the dominant antimicrobialresistant genotypes in this study. For the analysis of tetracyclineresistant genes, tetA, tetB and tetC were found in 72, 29, and 10 strains, respectively. In contrast, 86 tetracycline-resistant isolates did not carry any of the tested tetracycline-resistant genes, indicating that other tetracycline-resistant genes (e.g., tetD, tetE or tetM) might be present, or other novel genetic resistant determinants exist [31]. Interestingly, at least one of the tested tetracycline-resistant genes (e.g., tetA, teB or teC) was detected in 6 tetracycline susceptible isolates. For the sulfonamides resistance analysis, genes of sul1, sul2, and sul3 were found in 12, 98, and 5 isolates, respectively. Similar to the results of the tetracyclineresistant genes, at least one of the sulfonamides resistant genes (e.g., sul1, sul2 or sul3) was detected in 20 sulfonamides susceptible isolates. For the aminoglycoside resistance, genes of aadA, strA, and strB were found in 55, 114, and 123 isolates, respectively. Twentyfive aminoglycoside resistant isolates did not carry any of the tested aminoglycoside resistant genes. However, at least one of the aminoglycoside resistant genes (e.g., strA, strB or aadA) was detected in 73 streptomycin and spectinomycin susceptible isolates.
It is notable that majority (187 of 220) of the obtained E. coli isolates carried at least one of antimicrobial resistant genes that encodes resistant phenotypes to tetracycline, sulfonamides and streptomycin/spectionomycin, corresponding to high abundances of the genes of tetA, sul2 and strA/strB. The tetracycline resistance in E. coli isolates from farm-raised sika deer was mostly due to tetA and teB, and the frequency of tetA (32.7%) was obviously higher than tetB (13.2%). However, previous studies showed that the frequency of tetB (49.8%) was higher than tetA (24.0%) for the E. coli isolates from pigs raised under overuse of antimicrobials in China [32]. The sulfonamides resistant E. coli is generally attributed to the presence of sul1, sul2 and/or sul3 genes [33]. The sul2 gene displayed much higher frequency (44.5%) than that of the sul1 and sul3 (5.5% and 2.3%, respectively) in the E. coli isolates from this study. Other studies showed that the genes of sul1, sul2 and sul3 showed equal importance for sulfonamidesresistance in E. coli strains from food-producing animals in China [34,35]. Among streptomycin/spectionomycin resistant genes, the strA and strB genes were detected at the highest frequency (51.8% and 55.9%, respectively). One previous study indicated that the strA and strB genes might be present together to make E. coli strains streptomycin resistance [36]. In addition, twelve streptomycin/ spectionomycin susceptible E. coli isolates carried aadA, and the findings are consistent with former studies in which a large reservoir of nonintegrated gene cassettes could exist, but might not be expressed in some streptomy/spectionomycin sensitive E. coli strains [20,36].

Correlations between resistant and virulence genes in the farm-raised sika deer
Significant correlations (P,0.05) were found between a few virulence and resistant genes (Table 4). For example, the correlation coefficients between the resistant gene sul2 and virulence genes of aidA, elt and stx1 were 0.350, 0.318, and 20.400, respectively. The correlation coefficient was 0.333  between the resistant gene strA and the virulence gene aidA, and 0.316 between the resistant gene aadA and the virulence gene fedA. Besides of the above genes, no significant correlation was observed between the remaining resistant genes (tetA, tetB, tetC, sul1, sul3 and strB) and the virulence genes (eaeA, faeG, fasA, astA, estA, estB and stx2). Such weak correlations between the selected resistant and virulence genes suggested that the presence of some virulence genes does not necessarily possess resistant characteristics for the E. coli of farm-raised sika deer. The results here was inconsistent with traditional view that frequent exposure to heavy antimicro-bials might drive the distribution, reassortment and co-location of both resistant and virulence genes onto conjugative plasmids or pathogenicity islands in the pathogens, and the antimicrobialresisting bacteria are more frequent as the pathogens than that of commensal bacteria [20]. Therefore, other factors may responsible for the observed prevalence and the associations of antimicrobial resistant and virulence genes.

Conclusions
Antimicrobials are used heavily for the farm-raised sika deer when disease outbreaks. The results of this study indicated that the pathogenic groups (EPEC, ETEC and STEC) of E. coli strains occurred at much higher frequency than that of wild life (including wild deer), and even higher than that of livestock in some developed countries. Furthermore, the antimicrobial resistance of the E. coli strains from farm-raised sika deer was also significantly higher than those reported in wild animals and certain livestock in some countries. Other than domestic livestock, antimicrobialresisting E. coli strains from domesticated wildlife have become a new heavy contamination source, and already posed high potential risks to public health in Northeastern China. Feasible measurements should be taken for prudently antimicrobials use in domesticated wildlife and livestock to prevent the increasingly antimicrobial resistance problem from worsening. In addition, comprehensive surveys on domesticated wildlife for the antimicrobial-resisting bacteria are also strongly recommended to ensure the safety of food products and environments.

Ethics Statement
The fecal samples of the present study were collected from sika deer farms, no specific permissions were required for these  Table 3. Antimicrobial resistant genotypes and pathotypes of E. coli strains from farm-raised sika deer (n = 220). locations/activities, the sika deer farms are the public open place in China, and the sample activities did not involve any endangered or protected species. This study focuses on the microbial antimicrobial-resistant characteristics of fecal samples from farm-raised sika deer, and the vertebrate materials are not included.
Sample collection and E. coli strain isolation E. coli isolates of this study were obtained from farm-raised sika deer with clinic signs of yellow/white diarrhea. All fecal samples were collected from 50 sika deer farms located in three Northeastern provinces of China, including 18 farms from Jilin province, 15 farms from Heilongjiang province and 17 farms from Liaoning province during the period of March to October in 2009 ( Figure 2). In China, the deer farms are relatively small and usually raise less than 20 sika deers in each farm. One diarrheic fecal sample was collected from one location, and the antimicrobials usage information was obtained from the owners and from the medical records in each farm. The information of the antimicrobials usage in the studied farms within the last 12 months before sampling is shown in Table S1 and Fig. S1. The antimicrobials of sulfonamides, tetracycline and amino-glycosides were the most frequently used drugs for sika deer diarrhea treatment, followed by chloramphenicol, penicillins, cephems and fluoroquinolones (Fig. S1). Five types of antimicrobials (sulfonamides, tetracycline, amino-glycosides, chloramphenicol and penicillins) javascript:;were used in more than half of the investigated farms (Fig. S1).
All fecal samples were kept in dark, transported to the laboratory on ice, and analyzed for E. coli strains within 24 hrs. The fecal material was suspended in 0.9% (w/v) sterile NaCl (physiological salinity), spread on Petri dishes containing Mac-Conkey agar (Oxoid, UK), followed by incubation at 37uC for 18-24 hrs [37]. Bacterial colonies with morphology of E. coli were randomly picked and identified by biochemical methods (Gram reaction; activities of catalase, oxidase and urease; indole production, methyl red reaction, Voges-Proskauer test and citrate utilization), and by the API 20E strips (BioMérieux, France) according to the manufacture instructions. The isolates identified as E. coli were maintained on Luria-Bertani (LB, Oxoid) slants at 4uC, and as glycerol suspension (20%, v/v) at -80uC for long-term preservation.

Determination of O serogroups of E. coli isolates
Sika deer E. coli isolates proliferated on a nutrient agar (NA, BD) were suspended in 0.9% (w/v) NaCl, and then autoclaved at 1.05 kg f/cm 2 for 1 h to extract somatic antigens. The serogroups of the E. coli isolates were examined by slide agglutination using O antisera commercially available from China Institute of Veterinary Drugs Control (IVDC, Beijing), and a NaCl control was performed to eliminate false positive results. Positive reactors were confirmed by tube agglutination test [38].

Antimicrobial susceptibility test
Minimal inhibitory concentration (MIC) determination was performed by using the broth micro-dilution method according to the Clinical and Laboratory Standards Institute (CLSI, formerly NCCLS) Guideline (CLSI, 2009). The following 16 antimicrobials were selected for susceptibility test: amikacin, gentamicin, kanamycin, streptomycin, spectinomycin, ceftiofur, ceftriaxone, ciprofloxacin, enrofloxacin, norfloxacin, sulfadiazine, sulfamethazine, amoxicillin, ampicillin, chloramphenicol and tetracycline. The breakpoints for each antimicrobial resistance were outlined in table 2 according to the CLSI guidelines (CLSI, 2009). The isolates showing resistance to one or more antimicrobials were characterized for antibiotic resistant genes. Strain E. coli ATCC 25922 was applied as quality control for the susceptibility testing procedure.
Polymerase chain reaction (PCR) detection of virulence genes in E. coli isolates The PCR was applied to detect whether the obtained E. coli isolates harboring toxins (LT, STa, STb, Stx1, Stx2, and EAST1) and adhesions (F4, F5, F6, F18, AIDA, and EaeA), known as virulence genes for E. coli pathotypes causing intestinal diseases in humans and animals. The information of specific oligonucleotide primers, amplicons predicted sizes and pathotypes definition (EPEC, STEC, ETEC and EHEC) for the tested virulence genes are shown in Table 5. The PCR conditions for each virulence gene were performed as described previously [39,40,41,42]. Three strains from Zhejiang Province Key Laboratory for Food Safety were selected as positive controls for determining pathotypes of EPEC, ETEC and STEC.

Antimicrobial-resisting gene detection
Antimicrobial-resisting genes were detected and identified following the protocols as described previously (Table 6). Briefly, the E. coli strains were grown in 500 ml LB broth overnight, and 20 ml of the culture was transferred to 200 ml lysis buffer [0.1 M Tris-HCl (pH 8.5), 0.05% Tween 20, and 0.24 mg/ml proteinase K]. The sample was incubated at 60uC for 1 hour and subsequently heated at 97uC for 15 min. The PCR primers and annealing temperatures for major resisting genes of tetracycline (tetA, tetB and tetC), sulfonamides (sul1, sul2 and sul3), and streptomycin-spectinomycin (strA/strB and aadA) were detailed in Table 6, and the major resisting genes were amplified by a set of multiplex PCR protocols [36,43,44]. The multiplex PCRs were all performed with a total 25-ml reaction mixture and a Qiagen multiplex PCR kit (Qiagen, Shanghai) with 1 ml Qiagen multiplex PCR master mixture, 16 Qsolution, and 16 primer mixture according to the manufacturer's instructions. The PCRs were performed as follows: 1 cycle of 4 min at 95uC; 35 cycles, each consisting of 1 min at 95uC, 1 min at annealing temperature, and 1 min at 72uC; and 1 cycle of 7 min at 72uC.

Statistical analysis
For the purpose of statistical analysis, isolates with reduced susceptibility were classified into resistant groups. The pairwise statistical associations between major antimicrobial-resisting and virulence genes were determined by using the Statistical Package  Figure S1 Frequency of antimicrobials usage in 50 sika deer farms. Note: Antimicrobials abbreviations are the same as Table S1.

Supporting Information
(TIF) Note: ł, the antimicrobial was used; #, the antimicrobial was not used. Note: Sample locations of LS, JS and HS are the same as Figure 2. (DOC)