Prevalence, Antimicrobial Susceptibility, and Molecular Characterization of Escherichia coli Isolated From Raw Milk in Dairy Herds in Northern China

Escherichia coli is a common bacterium in the intestines of animals, and it is also the major important cause of toxic mastitis, which is an acute or peracute disease that causes a higher incidence of death and culling of cattle. The purpose of this study was to investigate E. coli strains isolated from the raw milk of dairy cattle in Northern China, and the antibacterial susceptibility of these strains and essential virulence genes. From May to September 2015, 195 raw milk samples were collected from 195 dairy farms located in Northern China. Among the samples, 67 (34.4%) samples were positive for E. coli. About 67 E. coli strains were isolated from these 67 samples. The prevalence of Shiga toxin-producing E. coli (STEC), enterotoxigenic E. coli (ETEC), enteropathogenic E. coli (EPEC), and enteroinvasive E. coli (EIEC) were 9, 6, 4.5, and 1.5%, respectively. Among the virulence genes detected, stx1 was the most prevalent (6/67, 9%) gene, followed by eae (3/67, 4.5%), and estB (2/67, 3%). Moreover, the strains exhibited different resistance levels to ampicillin (46.3%), amoxicillin-clavulanic acid (16.4%), trimethoprim-sulfamethoxazole (13.4%), tetracycline (13.4%), cefoxitin (11.9%), chloramphenicol (7.5%), kanamycin (7.5%), streptomycin (6.0%), tobramycin (4.5%), azithromycin (4.5%), and ciprofloxacin (1.5%). All of the E. coli isolates were susceptible to gentamicin. The prevalence of β-lactamase-encoding genes was 34.3% in 67 E. coli isolates and 45% in 40 β-lactam-resistance E. coli isolates. The overall prevalence of blaSHV, blaTEM, blaCMY, and blaCTX-M genes were 1.5, 20.9, 10.4, and 1.5%, respectively. Nine non-pathogenic E. coli isolates also carried β-lactamase resistance genes, which may transfer to other pathogenic E. coli and pose a threat to the farm’s mastitis management projects. Our results showed that most of E. coli were multidrug resistant and possessed multiple virulence genes, which may have a huge potential hazard with public health, and antibiotic resistance of E. coli was prevalent in dairy herds in Northern China, and ampicillin should be used cautiously for mastitis caused by E. coli in Northern China.


INTRODUCTION
Escherichia coli was a common inhabitant of the intestine of animals (Tark et al., 2016). During parturition and early lactation period, E. coli was found to usually infect mammary gland of cows, which may cause acute and local mastitis (Hinthong et al., 2017). Escherichia coli is the main cause of bacterial mastitis in cows. It is usually short-lived, causing the infection that lasts 2-3 days. However, E. coli has been displayed to cause persistent infections in a few cases (Lippolis et al., 2017). Pathogenic E. coli can cause disease in animals and humans due to virious virulence (Ntuli et al., 2016). Based on the epidemiological, clinical, and pathogenic characteristics, E. coli is classified into different pathotypes: Shiga toxin-producing E. coli (STEC), enteroaggregative E. coli (EAEC), enterotoxigenic E. coli (ETEC), enteropathogenic E. coli (EPEC), and enteroinvasive E. coli (EIEC; Rugeles et al., 2010). Numerous outbreaks associated with E. coli in milk and other foods have been reported recently (EFSA-ECDC, 2012;EFSA, 2015;Ombarak et al., 2016). For example, STEC can generate two types of Shiga toxins (stx1 and stx2), and EPEC can produce bfp gene, which were involved in pathogenicity of gastrointestinal tract (Hernandes et al., 2009;Douellou et al., 2016). ETEC can express heat-stable est genes that can cause severe diarrhea. EAEC can produce aggR gene, which were associated with the generation of biofilm (Medeiros et al., 2013). The ipaH gene from EIEC can lead to the occurrence of fever, vomiting, and dehydration in infected children. The higher prevalence of E. coli is closely associated with hygiene in raw milk (Radostits et al., 2007). Therefore, the study on E. coli in raw milk is significant.
Escherichia coli is not only with the potential occurrence, but also with the rapid development of antibiotic resistance bacteria (Ntuli et al., 2016). Inappropriate selection and abuse of antibiotics could lead to antibiotic resistance in bacteria (Da Silva and Mendonça, 2012). Moreover, E. coli may develop acquired resistance to other antibiotics by carrying various resistance characteristics on mutation, plasmids, or transposons . For example, extended-spectrum β-lactamases E. coli, resistant to β-lactam antibiotics including third-and fourth-generation cephalosporins, acquires ESBL by mutation or plasmid-mediated horizontal gene transfer (Freitag et al., 2016). Acquired antibiotic resistance also has a transmission potential to humans and other animals (Ruegg et al., 2014). Raw milk can also facilitate the transmission of antibiotic resistance genes to the human gastrointestinal tract, In addition to the presence of pathogenic bacteria. A better understanding on the resistance profile of E. coli isolates will improve our understanding of appropriate treatments for pathogen-related management (Tark et al., 2016). Therefore, monitoring the antibiotic resistance of E. coli in raw milk may show the trend or specific characteristics of antibiotic resistance and help to better prevent or more effectively treat mastitis on dairy farm.
Antimicrobial resistance and virulence types in E. coli have been studied on raw milk of healthy dairy cattle and of bovine mastitis in a variety of countries, including Northern Italy, Romania, Brazil, Egypt, South Korea, and Thailand (Trevisani et al., 2014;Ombarak et al., 2016;Ribeiro et al., 2016;Tark et al., 2016;Hinthong et al., 2017;Tabaran et al., 2017). However, incidence on antibiotic resistance of E. coli from raw milk in Northern China were very limited. Continuous monitoring of the antibiotic resistance and virulence type of E. coli could be necessary to evaluate E. coli risk in raw milk. Therefore, the objective of the work was to investigate the rate of E. coli strains isolated from raw milk in Northern China, and to characterize the antimicrobial susceptibility and key virulence genes of these strains.

Collection of Samples
In total, 195 raw milk were collected from 195 dairy farms from four cities, which was the major dairy-production cities of Northern China (herd size ≥300, no clinical mastitis cow, milking frequency two or three times per day), from May to September in 2015 (average daily temperature >20°C). There were 30 raw milk samples from Jinan, 40 samples from Harbin, 50 samples from Beijing, and 75 samples from Hohhot (Figure 1). The raw milk samples were collected from the top, middle, and bottom of bulk tank, mixed well, and then transferred into sterile bottles and transported to laboratory at 4°C immediately.

Isolation and Identification of E. coli
Aliquots (25 ml) of each sample were added to 225 ml tryptic soy broth, and then incubated at 37°C for 16 h with shaking for E. coli detection. The samples were placed onto Eosin Methylene Blue agar plates (Beijing Land Bridge Technology Ltd., Beijing, China). The agar plates were incubated at 37°C for 18-24 h. The presumptive colonies (typical blue-black appearance with a metallic green sheen) were picked. All the colonies were sub-cultured onto nutrient agar slants at 37°C for 16 h, and then used for biochemical identification. The colonies initially identified as E. coli were examined by Voges-Proskauer negative, methyl-red positive and citrate negative. All isolates were stored at −80°C until use.
All the presumptive colonies were confirmed by PCR on 16S rRNA gene detection (Supplementary

Detection of Virulence Determinants
Seven virulence genes for each diarrheagenic E. coli were detected by PCR method: stx1 and stx2 for STEC, estA, estB, and eltB for ETEC, aggR for EAEC, bfp and eae for EPEC, and ipaH for EIEC. Amplified products were analyzed by agarose gel electrophoresis, and then visualized by SYBR Safe DNA Stain gel staining. All the primers were shown in Supplementary Table S1.

Antimicrobial Resistance Genes
Four β-lactamase resistance related genes (bla CMY , bla SHV , bla CTX-M , and bla TEM ) and two tetracycline genes (tetA and tetB) were detected by multiplex PCR in E. coli strains (Supplementary Table S1). The amplification conditions were as follows: 95°C for 5 min, 30 cycles of 94°C for 30 s, 63°C for 90 s, and 72°C for 90 s, and 72°C for 7 min for a final extension step (Ribeiro et al., 2016). Escherichia coli strains ATCC 25922 was used as a positive control in each run.

Screening of Antibiotic Resistance Genes
The β-lactamase-encoding genes results were presented in Table 3. The prevalence of β-lactamase-encoding genes were 34.3% in 67 E. coli isolates and 45% in 40 β-lactam resistance E. coli isolates. The overall prevalences of bla SHV , bla TEM , bla CMY , and bla CTX-M genes among E. coli isolates, which was narrow spectrum extended-spectrum β-lactamase-encoding genes, β-lactamase-encoding genes, AmpC, and β-lactamase-encoding genes, were 1.5, 1.5, 10.4, and 20.9%, respectively. In total, Frontiers in Microbiology | www.frontiersin.org 5 September 2021 | Volume 12 | Article 730656 71.4% of the isolates, which possessed the bla TEM gene, were resistant to ampicillin. Around 57.1% of bla CMY positive isolates were resistant to amoxicillin-clavulanic acid. Five (7.5%) isolates possessing bla TEM or bla CMY did not suggest β-lactamase antibiotic resistance. Moreover, the presence of the tet genes, which were conferring resistance to tetracycline, were confirmed in seven tetracyclineresistance strains. None of the studied strains possessed tetA ( Table 4).

DISCUSSION
In this research, 34.4% (67/195) of samples were positive for E. coli in raw milk. These results are significantly lower than that in previous studies. The incidence of E. coli in raw milk in India was 81.1% (Bhoomika et al., 2016), 75% in Bangladesh (Islam et al., 2016), 64.5% in Malaysia (Jayarao and Henning, 2001), and 45% in Northern China . In contrast, a much lower incidence (22.4%) of E. coli was discovered in raw milk in Sharkia Governorate (Awadallah et al., 2016). Moreover, our results are comparable with the findings of Ntuli et al. (2016), who reported 36% prevalence rate in bulk milk in South Africa, and Sharma et al. (2015), who reported 35.63% occurrence rate in raw milk in the Jaipur city of Rajasthan. Overall, the results indicated that E. coli is a common strain in raw milk collected from dairy herds of Northern China. The high prevalence of E. coli in raw milk and dairy products is a cause of concern because it is related to contamination from fecal sources and the consequent risk of enteric pathogenic microorganisms in food (Ombarak et al., 2016).
An important factor of E. coli infections is virulence factors. When E. coli carried some virulence genes, they could be potentially harmful to public consumers (Hinthong et al., 2017). In the study, 20.9% (14/67) of the tested raw milk possessing more than one virulence gene tested, may carried potentially pathogenic E. coli, as shown in Table 3. STEC, cause a life-threatening sequel, such as neurological disorder and hemolytic syndrome or HUS (Kaper et al., 2004), was found to be the most common pathogenic E. coli strain in raw milk. It has been reported that the virulence genes of STEC isolates were commonly implicated in many foodborne STEC outbreaks in the world (Beutin and Fach, 2015). In this study, the most common virulence genes in raw milk samples in Northern China were stx genes. The result was in agreement with Suojala et al. (2011), who reported the STEC (stx-positive isolates) was the most common E. coli type of raw milk with subclinical mastitis in Southern Finland, and by Lambertini et al. (2015), who found that the most frequently detected gene in raw milk of the United States northeastern was stx1. However, STEC or stx factors has been detected in the farms  .3) 1 (5.9) 8 (11.9) Chloramphenicol 2 (8.7) 0 (0) 2 (12.5) 1 (5.9) 5 (7.5) Kanamycin 2 (8.7) 1 (9.1) 1 (6.3) 1 (5.9) 5 (7.5) Streptomycin 0 (0) 0 (0) 2 (12.5) 2 (11.8) 4 (6.0) Tobramycin 0 (0) 0 (0) 2 (12.5) 1 (5.9) 3 (4.5) Azithromycin 1 (4.3) 1 (9.1) 1 (6.3) 0 (0) 3 (4.5) Ciprofloxacin 0 (0) 1 (9.1) of United States and European at a low prevalence (Jayarao et al., 2006;Pradel et al., 2008;Van Kessel et al., 2011;Claeys et al., 2013;Ombarak et al., 2016). Enteropathogenic E. coli is responsible for diarrhea in both developing and developed countries. As an important foodborne pathogen, EPEC has high isolation rate in retail foods in China . EPEC were isolated from many animals, such as cattle, goat, sheep, chicken, gull, and pigeon (Gomez-Aldapa et al., 2016). In the study, three strains were eae genespositive and bfp gene-negative, which could be classified as EPEC. Cortés et al. (2005) and Gomez-Aldapa et al. (2016) found that atypical EPEC strains were found in raw milk in Egypt, Saudi Arabia, and Slovakia. However, there is no report on the eae-positive E. coli strains found in mastitis cows in Iran and Thailand (Ghanbarpour and Oswald, 2010;Hinthong et al., 2017). Moreover, an increasing frequency of eae-negative isolates were postulated to have other putative adherence and virulence associated factors (Gomez-Aldapa et al., 2016). ETEC strains are usually transmitted by contaminated food. In the study, EPEC and ETEC strains were isolated from Hohhot and Jinan. EPEC/ETEC hybrid isolates were related to EPEC strain, and appeared to have acquired virulence genes by horizontal gene transfer (Hazen et al., 2017).
In the study, antimicrobial resistance was most frequently observed to ampicillin (46.3%). The susceptibility to amoxicillin can be predicted by antimicrobial resistance to ampicillin (CLSI, 2012). So the tested E. coli isolates may showed a high resistance to amoxicillin. Nam et al. (2009) reported that 32.2% E. coli strains from mastitis cow were resistant to ampicillin. However, the resistant rates in the study were much higher than those in South Korea from 2012 to 2015 (Tark et al., 2016) andin Northern Colorado (McConnel et al., 2016). Antibiotic susceptibility of E. coli was more important on choosing a suitable antibiotic for mastitis . The information of antibiotic use for dairy in Northern China has been investigated in our previous survey. Ampicillin was commonly used in dairy mastitis therapy . So, ampicillin is not a suitable treatment for mastitis caused by E. coli in Northern China.
In our previous survey, we found that five antibiotics (penicillin, ciprofloxacin, sulfamethoxazole-trimethoprim, streptomycin, and gentamicin) were commonly used in mastitis cow. In the study, most of tested strains showed an obvious antimicrobial resistance to ciprofloxacin, sulfamethoxazoletrimethoprim, and streptomycin. These results also indicated that there was a correlation between antibiotic use and antimicrobial resistance.
In the study, there were four β-lactamase resistance genes detected. The β-lactamase-encoding genes prevalence was 34.3% in 67 E. coli isolates. β-lactamase resistance genes, such as bla CMY , bla SHV , bla CTX-M , and bla TEM were detected in nine non-pathogenic E. coli isolates. So non-pathogenic E. coli can serve as an antibiotic resistance reservoir and could possibly transfer genes to other pathogenic E. coli strains, which can pose a threat to mastitis management programs of farm (Hu et al., 2016). The rate of bla CTX-M, bla CMY , bla TEM , and bla SHV genes among E. coli was 1.5, 1.5, 10.4, and 20.9% in the study, respectively. The bla TEM and bla CMY genes were the most common, which is similar to several previous studies (Navajas-Benito et al., 2016;Gomi et al., 2017;Hinthong et al., 2017). The cephalosporins treatment in mastitis cattle also raised the proportion of bla TEM in milk samples at the period of withdrawal (p < 0.05; Dong et al., 2021). The bla CTX-M , which was the most important ESBL-related gene, it was associated with the geographic area (Su et al., 2016). However, bla CTX-M was the most popular gene in Japan, United Kingdom, France, Netherlands, and Germany (Dahmen et al., 2013;Ohnishi et al., 2013;Timofte et al., 2014;Freitag et al., 2016;Santman-Berends et al., 2016).
Around 11.8% of E. coli stains showed resistance to tetracycline in the study. However, Su et al. (2016) reported that the tetracycline-resistance prevalence was 51%. Navajas-Benito et al. (2016) reported that antimicrobial resistance for tetracycline was detected in 19.2% of E. coli strains, which recovered from air and its surroundings in Spain. Antimicrobial resistance genes to tetracycline were tested in all the tetracycline-resistant isolates, and three tetracycline-resistant isolates harbored one tetracycline resistance gene tetB, which was the most frequent gene, and the studied E. coli did not possess tetA. However, Gomi et al. (2017) found that the prevalent of tetA was more than tetB in E. coli isolates. It was reported that one representative E. coli strain (No. JXLQYF114666) contained nine ARGs including aph(3'')-Ib, bla TEM-1B , bla CMY-2 , aph(6)-Id, mdfA, sul2, tetB, catA2, and dfrA14, which result in resistance to seven important antibiotics classes (Liu et al., 2020). Moreover, the phenotype-genotype discrepancies on the tetracycline-resistant E. coli were observed in the study. However, resistance genotypes on tetracycline, gentamicin, kanamycin, and oxacillin correlated well with resistance phenotypes in E. coli and S. aureus (Gomi et al., 2017). Therefore, it was still necessary to fully account of testing phenotypic susceptibility for resistance (Zhao et al., 2015). Further research should be carried out to analyze the genetic characteristics on antibiotic resistance by whole-genome approach, which may explain the phenotype-genotype discrepancies observed for many strains.

CONCLUSION
In conclusion, the antibiotic resistance on E. coli isolated from raw milk in Northern China was assessed for the first time. Our data indicated that E. coli isolates were widely present in raw milk samples in Northern China. A total 20.9% of the tested E. coli possessed one or more virulence genes, which showed a potential pathogenicity. Escherichia coli strains exhibited different levels of antimicrobial resistance, except gentamicin. Ampicillin should not be a suitable treatment of dairy herds for mastitis by E. coli in Northern China. Majority of E. coli were multiple-antibiotic resistant and co-carried many virulence genes, and it may pose great potential risk to public health. The possibility of transferring and transmitting resistance genes, between non-pathogenic and pathogenic E. coli isolates, should be evaluated in further studies.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author.

AUTHOR CONTRIBUTIONS
HL, LM, and LD designed and performed the research. YZ helped with the data analysis. JW gave advices to the researchers. NZ gave the opinions on the research design. All authors contributed to the article and approved the submitted version.