Antimicrobial resistance of Escherichia coli isolated from retail foods in northern Xinjiang, China.

Abstract To determine antimicrobial resistance, 431 samples of retail foods purchased at different supermarkets in Northern Xinjiang were examined in this study. There were 112 Escherichia coli strains that were isolated, with approximately 26% of the samples contaminated by E. coli. The detection rate of E. coli isolated from pork was the highest (59.6%), followed by mutton (52.6%), retail fresh milk (52.4%), duck (36.4%), beef (35.3%), chicken (33.3%), and ready‐to‐eat food (12.9%); the E. coli detection rate for fish and vegetables was <11%. The result showed that the 112 isolates were mostly resistant to tetracycline (52%), followed by ampicillin (42%), compound trimethoprim/sulfamethoxazole (37%), amoxicillin (33%), and nalidixic acid (32%), imipenem resistance was not detected. One hundred isolates carried at least one antimicrobial resistance gene. The detection rate of resistance genes of our study was as follows: tetA (38%), tetB (27%), bla OXA (40%), bla TEM (20%), floR (20%), sul1 (16%), sul2 (27%), aad Ala (19%), aadB (11%), strA (28%), and strB (24%); tetC and bla PSE were not detected. Virulence genes fimC, agg, stx2, fimA, fyuA, papA, stx1, and eaeA were found in 52, 34, 21, 19, 6, 3, 2, and 2 isolates, respectively; papC was not detected. There was a statistically significant association between fimC and resistance to ciprofloxacin (p = .001), gentamicin (p = .001), amikacin (p = .001), levofloxacin (p = .001), and streptomycin (p = .001); between fimA and resistance to tetracycline (p = .001), ampicillin (p = .001), compound trimethoprim/sulfamethoxazole (p = .001), and amoxicillin (p = .003); between agg and resistance to gentamicin (p = .001), tetracycline (p = .001), ciprofloxacin (p = .017), and levofloxacin (p = .001); and between stx2 and resistance to ampicillin (p = .001), tetracycline (p = .001), compound trimethoprim/sulfamethoxazole (p = .002), and amoxicillin (p = .015).


| INTRODUC TI ON
It is well known that Escherichia coli mainly exists in the human and animal gastrointestinal tract. It also occurs in the natural environment, especially in soil, water, and plants (Katarzyna & Anna, 2016). Therefore, it is not surprising that some of the E. coli in the environment reinfects humans through vegetable-or animal-derived foods.
Escherichia coli is a highly diverse virulent species that is widely distributed in open systems, is easy to spread in the environment, and can be harmful to human health (Tenaillon, Skurnik, Picard, & Denamur, 2010). Drug resistance genes carried by E. coli can be transferred to other pathogenic bacteria, and, due to the excessive use of antibiotics, selection pressure is very high, resulting in bacterial strains resistant to a variety of drugs. Multi-drug-resistant strains are characterized by the presence of multiple genes conferring drug resistance, which results in insensitivity to many different drug groups (Hu, Yang, & Li, 2016;Rasheed, Thajuddin, Ahamed, Teklemariam, & Jamil, 2014).
Genetic mutations or genetic acquisition of antibiotic resistance genes (ARG) through horizontal gene transfer might also result in the occurrence of antibiotic-resistant bacteria (ARB) throughout the environment (Céline & David, 2015). This has resulted in the emergence of many different ARG, including the dfr and sul genes related to trimethoprim and sulfamethoxazole resistance, respectively (Chang, Lin, Chang, & Lu, 2007;Ho, Wang, Chow, & Que, 2009), and other genes, such as ampC, oxa2, and tetA.
The ever-increasing threat of ARB may be associated with enhanced virulence (Guillard, Pons, Roux, Pier, & Skurnik, 2016;Roux et al., 2015), and with the increase in antibiotic resistance, an increase in virulence may naturally evolve. Therefore, when controlling the spread of antibiotic resistance, we must also control the spread of virulence (Meredith, Brooks, & Brooks, 2017).
Although the profile of virulence and antimicrobial resistance genes of E. coli from foods has been reported (Luo, Ji, & Wang, 2016), the data elucidating the association between these two gene sets are lacking.
In Xinjiang, China, a previous study conducted antibiotic resistance research on foodborne E. coli based on samples from slaughterhouses, butcher shops, and farms (Xia, Xiang, & Guo, 2014;Yao, Long, Kuerbannaimu, Wang, & Xia, 2017). However, little is known about the resistance of those bacteria in retail foods.
There have been some reports describing the antimicrobial resistance and virulence of E. coli, such as Arisoy, Rad, Akin, and Akar (2008), who showed that the virulence genes afaI, pap, hly, aer, and sfa were increased in sensitive strains. However, detailed information on the relationship between antimicrobial resistance genes and virulence genes of E. coli isolated from retail foods in Xinjiang is scarce.
The purpose of this study was to evaluate the drug resistance of E. coli strains isolated from retail foods in northern Xinjiang, identify their virulence genes, and determine the possible relationship between the virulence genes and drug resistance.

| Sampling and E. coli isolation
A total of 431 food samples were purchased at supermarkets in Shihezi, Kuitun, and Urumqi, in northern Xinjiang, China, from 2014 to 2016, and each type of sample and its number are listed in Table 1.
Each sample weighed 25 g and was placed in a sterile plastic bag containing 225 ml of sterilized sodium chloride solution (0.85%) and then homogenized for 90 s using a BagMixer 400 CC beating homogenizer. Lauryl Sulfate Tryptose (LST) broth was inoculated with 1 ml of homogenate and incubated for 48 hr at 37 ± 1°C. Gas-positive tubes were inoculated into 100 ml of E. coli (EC) broth and incubated at 44 ± 0.5°C for 48 hr (Wang, Sun, & Ji, 2014). After that, one loopful from each gas-positive tube was streaked onto eosin methylene blue agar. Presumptive E. coli colonies were streaked onto Luria-Bertani nutrient agar and incubated for 12-48 hr at 36 ± 1°C. Each culture was confirmed as E. coli through an IMViC test. E. coli ATCC 25922 was used as a positive control for polymerase chain reaction (PCR) of UidA. Template was prepared via the boiling method, for the amplification of selected UidA genes in E. coli using PCR (Heijnen & Medema, 2006). The oligonucleotide sequences used and the predicted sizes of PCR amplification products of genes are listed in Table 2.

| Antimicrobial susceptibility testing
Antimicrobial susceptibility testing was performed utilizing the disk-

| PCR amplification of antimicrobial resistance and virulence genes
Genomic DNA for PCR was extracted by the boiling method.
Tables 2 and 3 list the oligonucleotide sequences of different antimicrobial genes and virulence genes in E. coli and the predicted sizes after PCR amplification.

| Statistical analysis
SPSS v.17.0 software was used to analyze the data. Logistical regression analysis was used to analyze the correlation between variables.
p < .05 was considered statistically significant.

| E. coli isolated from retail foods
A total of 112 strains of E. coli were isolated from 431 random samples, with 26% of the samples testing positive for contamination.
The overall incidence was higher than 14.7% reported elsewhere (Rasheed et al., 2014). As shown in  (Alharbi & Khaled, 2018), and vegetables (Rasheed et al., 2014). Whether there is a link between high contamination rates and high antibiotic resistance rates for E. coli in food remains to be determined.
In both developed and developing countries, antibiotic resistance has been recognized as a problem in the field of human and veterinary medicine (Bottacini et al., 2018;Zhang et al., 2017). There is ample evidence that the widespread use of antibiotics in agriculture and medicine is the main reason for the high resistance rate of Gram-negative bacteria (Bothyna & Randa, 2018). Various food and environmental sources contain bacteria resistant to one or more antimicrobial agents used in human or veterinary medicine and animal food production (Hinthong, Pumipuntu, & Santajit, 2017).

| Antimicrobial resistance profiles of E. coli isolates
Antibiotic resistance in E. coli is of particular concern because it is the most common Gram-negative pathogen in humans, the most common cause of urinary tract infections, and a frequent cause of community and hospital-acquired bacteremia (Bothyna & Randa, 2018) and diarrhea (Jessica, Lashaunda, & Levens, 2016).
Worldwide data have shown that resistance to traditional drugs is increasing, and resistance is also being encountered against newer and more effective antibiotics (Sara, Mohammad, & Sadegh, 2014). As in this study, the most frequent resistance was seen for third-generation cephalosporin-ceftazidime (22%) and tetracyclines (52%; Table 5). A comparative study by Dominguez et al. (2018) showed that high resistance rates (76.5%-79.4%) were observed in oxyimino-cephalosporins (cefotaxime, ceftriaxone, and ceftiofur) and cefepime (70.6%). This phenomenon requires additional study and sustained data support.
As shown in  Rosengren et al. (2009) multiple drug resistance rate is shown in Table 6, and the pattern of antibiotic resistance in those isolates is shown in Table 7.
The incidence of multidrug resistance is a compelling issue, as there is a repository of antimicrobial resistance genes in the community, and drug resistance genes and plasmids can easily be transferred to other strains. The high resistance to tetracycline and ampicillin may be due to the easy availability and low cost of those medications. Although these antibiotics have been banned, the bans have not been effectively implemented by the relevant regulatory bodies. Another explanation for a strain's high resistance rate is its contact with environmental microorganisms that produce natural antibiotics, or with soil contaminated by wildlife feces carrying antibiotic-resistant microorganisms.

| Antimicrobial resistance genotypes of E. coli isolates
We detected 11 of the 13 resistance genes (tetA, tetB, bla tem , bla oxa , floR, aad Ala ,aadB,sul1,sul2,strA,and strB), and one hundred isolates carried one or more antimicrobial genes. Resistance genes were not detected in twelve strains of E. coli. The resistance genotypes of E.
coli isolates are shown in Table 7.   We found that the detection rate of pork was more than that of chicken, duck, and beef, but there are fewer resistance genes in pork as compared to chicken. Ayoyi, Bii, and Okemo (2008) showed that multidrug resistance is closely related to different farm management treatments, and statistical significance (p ≤ .001) was found between them.  We detected fyuA virulence genes in six isolates (5.4%), compared to 83.3% found by Laupland, Gregson, Church, Ross, and Pitout (2008).

| Virulence genes of E. coli isolates
Bacterial pili and fimbriae are important structures for bacterial pathogenicity, and it has been suggested that type I fimbriae function primarily in the initial pathogenic phase of avian pathogenic E. coli (APEC) infection. P-type fimbriae are also thought to contribute to bacterial pathogenicity (Paniagua et al., 2017). The fimC virulence gene encodes a protein necessary for the biosynthesis of type I fimbriae. The papA virulence gene encodes the main protein component of P-type fimbriae, and P-type fimbriae are encoded by the nine-gene pap operon, which includes papA, papB, papC, papD, papE, papF, papG, papH, and papI. Sequence analysis showed that there is sufficienthomology between P fimbriae in humans and chickens to indicate that they share some common antigen (Laupland, Kibsey, & Gregson, 2013). We detected the fimC gene in 46.4% of isolates, and the papA gene was detected in 2.7%; papC was not detected. This suggests that APEC in the Xinjiang region is mainly caused by a type I fimbriae. Arisoy et al. (2008) showed that there was a correlation between antibiotic sensitivity and virulence factors (VFs) of E. coli isolates causing pyelonephritis. They reported an increased presence of virulence genes pap, sfa, afai, hly, and aer in sensitive strains. Horcajada et al. (2005) showed that a significant correlation was found between nalidixic acid resistance and the decreased prevalence of three VFs:

| The relationship between virulence genes and antibiotic resistance
sfa, hly, and cnf-1.
In the current study, strong associations were found between the presence of fimC and resistance to ciprofloxacin, gentamicin, amikacin, levofloxacin, and streptomycin; between the presence of fimA and resistance to tetracycline, ampicillin, compound trimethoprim/sulfamethoxazole, and amoxicillin; between the presence of agg and resistance to gentamicin, tetracycline, ciprofloxacin, and levofloxacin; and between the presence of stx2 and resistance to ampicillin, tetracycline, compound trimethoprim/sulfamethoxazole, and amoxicillin.
Based on statistical analysis, the following correlations were

| CON CLUS IONS
Differences in the pathogenicity of E. coli and its susceptibility to antimicrobial agents were detected in different retail foods. This must be taken into account in developing guidelines for retail food management. Periodic review and formulation of antibiotic consumption policies are required to control the spread and acquisition of antibiotic resistance. Because most isolates express several types of VFs at the same time, it is necessary to further study the interaction between different VFs at the molecular level. In conclusion, E. coli has become a potential source of foodborne illness due to the possibility of horizontal transfer of drug-resistant genes, high drug resistance rate, and the correlation between the resistance to some antibiotics and several virulence factors. As those problems become more and more serious, we need to strengthen the supervision of veterinary drugs used in the raising of livestock.
At the same time, the detection and monitoring of antimicrobial agents in animal foods can help to reveal the ongoing use of prohibited animal husbandry practices.