Characteristics of Extended-Spectrum β-Lactamase–Producing Escherichia coli From Dogs and Cats Admitted to a Veterinary Teaching Hospital in Taipei, Taiwan From 2014 to 2017

Extended-spectrum β-lactamases (ESBLs) are enzymes that mediate resistance to newer β-lactam antibiotics, including extended-spectrum cephalosporins and monobactams. The production of ESBL is primarily plasmid mediated, and such plasmids often comprise the genes that encode resistance to other classes of antimicrobials, such as aminoglycosides and fluoroquinolones. Therefore, ESBL-producing microorganisms leave clinicians with limited therapeutic options in both human and veterinary medicine. Compared with human medicine, information regarding ESBL-producing microorganisms is limited in veterinary medicine. We screened for ESBL-producing Escherichia coli in dogs and cats admitted to National Taiwan University Veterinary Hospital, Taipei, from 2014 to 2017 and further analyzed the genotypes and phylogenetic traits of these ESBL producers. Double disk tests specified by the Clinical and Laboratory Standards Institute were performed on 283 E. coli isolates and revealed a total of 65 E. coli (54 from dogs and 11 from cats) with the ESBL phenotype (22.8%). blaCTX−M−1 group and blaCTX−M−2group were the most commonly identified ESBL gene groups. blaCTX−M−55 was the main ESBL gene within the blaCTX−M−1group, whereas the blaCTX−M−2group contained only blaCTX−M−124. The ESBL-producing E. coli were all resistant to ampicillin. The resistance rate to ceftiofur, doxycycline, enrofloxacin, and ciprofloxacin was 93.8, 73.8, 80, and 78.5%, respectively. Of the antibiotics tested, greater sensitivity to imipenem and gentamicin was noted. Multilocus sequence typing indicated that ST457, ST131, and ST648 were the most common sequence types. Our study identified eight ST131/O25b isolates, which is a global zoonotic clone of public health concern. The major ESBL genes of these clones were blaCTX−M−174 and blaCTX−M−194. Because companion animals such as dogs and cats are in close contact with humans, the characterization of ESBL producers originating from them is crucial from the perspective of both public health and veterinary medicine.


INTRODUCTION
Escherichia coli, a type of Gram-negative bacteria is a ubiquitous inhabitant of the gastrointestinal tract of both humans and animals. This microorganism frequently causes urinary tract, skin, or soft tissue infections in cats and dogs (1). Commonly prescribed medications to treat E. coli infection in companion animals include ampicillin, amoxicillin-clavulanic acid, fluoroquinolones, or cephalosporins. However, the emergence of drug-resistant bacteria encountered in clinical practice decreases the therapeutic efficacy of these antimicrobial agents. One major mechanism of this drug resistance is the production of enzymes by microbes to inactivate antimicrobial agents. For example, β-lactam agents are widely used to treat bacterial infections in veterinary medicine, whereas extended-spectrum β-lactamases (ESBLs) are a group of enzymes that mediate resistance to most β-lactam antibiotics, including extendedspectrum cephalosporins and monobactams but excluding carbapenems and cephamycins (2). ESBLs are inhibited by clavulanic acid, sulbactam, and tazobactam; this fact is used as a criterion to classify β-lactamases and for ESBL diagnosis purposes (3). TEM, SHV, and CTX-M-group enzymes are examples of commonly encountered ESBLs (2). ESBL producers usually exhibit a multi-drug-resistant phenotype. In addition, the ESBL genes are mainly plasmid mediated, thus facilitating the transmission of drug-resistant genes to other bacteria. Such a situation poses a challenge for infection management in clinical practice. ESBLs have been previously documented primarily in human clinical cases (4). Because companion animals such as dogs and cats are in close contact with humans, they could contract ESBL-producing microorganisms from humans and then possibly transmit them back to humans, which represents a public health concern (5).
Information regarding the prevalence of ESBL producers or the genotypes of these clinical isolates from cats and dogs is limited in Taiwan. It is imperative to investigate related matters from both a veterinary medicine and public health perspective (6). The present study analyzed a collection of E. coli isolates obtained from National Taiwan University Veterinary Hospital (NTUVH), a universitybased veterinary teaching hospital in Taipei, from 2014 to 2017 to determine the prevalence of ESBL-producing E. coli, assess their antimicrobial profile, and characterize the strains phylogenetically through multilocus sequence typing (MLST). The results obtained should provide insights into the role of ESBL-producing E. coli in companion animals. Some of the data herein have previously been reported at a conference (7).

Sample Collection
NTUVH is a teaching hospital affiliated with the College of Bioresources and Agriculture at National Taiwan University located in Taipei, Taiwan. Between 2014 and 2017, 283 E. coli isolates obtained from dogs (n = 224) and cats (n = 59) that were admitted to NTUVH were screened for ESBL producers. These E. coli isolates were cultured from different sources of the animals and identified using a Vitek 2 Compact (Biomérieux, Marcy-I'Etoile, France) to the species level and stored at −80 • C. Urine and pus samples from the uterus or wounds comprised almost 70% (47 and 22%, respectively) of the E. coli sources. These samples were collected from the animals to facilitate diagnosis and treatment. An ethical review was not required for this study.

ESBL Phenotype Testing
The ESBL producers of E. coli were tested using combination disk tests with cefotaxime and ceftazidime (30 µg), with and without clavulanic acid (10 µg), as specified by the Clinical and Laboratory Standards Institute (8). Briefly, the tested E. coli were plated on Muller-Hinton agar at a concentration of 0.5 McFarland standards and incubated at 35 • C for 16-18 h. A difference of 5 mm or more in the inhibition zones for either cefotaxime or the ceftazidime-clavulanic acid combination vs. the corresponding cefotaxime or ceftazidime alone was defined as an ESBL-producing E. coli. Klebsiella pneumoniae ATCC 700603 and E. coli ATCC 25922 were used as the positive and negative controls, respectively.

Detection of bla Genes
The E. coli isolates that were phenotypically ESBL producers were analyzed using polymerase chain reaction (PCR) to detect their bla genes. Bacterial DNA was extracted using the boiling method (9). Briefly, bacterial strains were cultured overnight at 37 • C on tryptic soy agar plates (Difco/Becton Dickinson, Franklin Lakes, NJ), and a loopful of cells was boiled in 200 µL of ddH 2 O for 10 min. The supernatant was saved after centrifugation at 12,000 × g for 10 min and used as the source of template DNA for PCR. The primers used to amplify bla CTX−M−1−group , bla CTX−M−2−group , bla CTX−M−8−group , bla CTX−M−9−group , bla CTX−M−25−group , bla SHV , bla TEM , and the expected PCR product sizes are listed in Table 1. The PCR cycling conditions were as follows: initial denaturation at 95 • C for 5 min, followed by 35 cycles at 95 • C for 30 s, annealing at 52-55 • C (as specified in Table 1) for 30 s, and a 72 • C extension for 1 min. Ten microliters of each PCR sample were loaded onto a 1.5% agarose gel and electrophoresed at 100 V for 30 min. The gels were then

Antibiotic Susceptibility Test
The ESBL-producing E. coli isolates were tested for susceptibility to antimicrobial agents used in clinical settings using the standard Kirby-Bauer disk diffusion method (8). The antimicrobial agents tested included β-lactams (amoxycillin/clavulanic acid, ampicillin, imipenem, and ceftiofur), tetracyclines (doxycycline), quinolones (enrofloxacin and ciprofloxacin), aminoglycosides (gentamicin), and sulfonamides (sulfamethoxazole/trimethoprim). The isolates were classified as susceptible, intermediate resistant, or resistant to the antimicrobial agents.

Genotyping and Phylogenetic Analysis
The ESBL-producing E. coli strains were genotyped using MLST (15). Internal fragments of adk, fumC, gyrB, icd, mdh, purA, and recA were amplified through a PCR by using the primers listed in Table 1 and sequenced. They were then uploaded to the EnteroBase MLST website (http://enterobase.warwick.ac.uk/) for comparison. Phylogenetic analysis of the strains was performed using BioNumerics version 7.0 (Applied Maths, Sint-Martens-Latem, Belgium).

E. coli ST131 O25b Detection
The PCR-based detection of E. coli ST131/O25b was based on the method described by Clermont et al. (16). The trpA and pabB primers and annealing temperature used are listed in Table 1. The PCR cycling conditions were as follows: initial denaturation at 94 • C for 4 min, followed by 30 cycles at 94 • C for 5 s, annealing at 65 • C for 10 s, and 72 • C extension for 5 min. Ten microliters of each PCR sample was loaded onto 2.0% agarose gel and electrophoresed at 100 V for 30 min. The gels were then stained with a fluorescent nucleic acid dye (Biotium) and examined under ultraviolet illumination.

RESULTS
A total of 283 E. coli isolates (59 from cats and 224 from dogs) were obtained during our study period (2014-2017). Table 2 lists the prevalence of ESBL-producing E. coli from dogs and cats.
The ESBL-producing E. coli isolates from cats were all resistant to ampicillin, ceftiofur, enrofloxacin, and ciprofloxacin, whereas those from dogs were all resistant to ampicillin. All the ESBLproducing E. coli were susceptible to imipenem, and more than 50% of the isolates were susceptible to gentamicin. Overall, most strains exhibited a multidrug resistant phenotype ( Table 6).
PCR detection to target trpA and pabB was performed on 10 E. coli ST131 isolates, and 8 isolates were identified as E. coli ST131/O25b clones (Figure 2). The ESBL-producing E. coli        possessed only the trpA specific DNA fragment, whereas the ESBL-producing E. coli ST131/O25b clones contained both the trpA and pabB DNA fragments. Among the 10 ESBL-producing E. coli, only one ST131/O25b clone was from a cat (E. coli 1942), whereas the others were from dogs. The two non-ST131/O25b clones were both from dogs.

DISCUSSION
The overall prevalence of ESBL-producing E. coli in dogs and cats was 23.0% in our study. A comparable prevalence was also reported in Japan, China, and Switzerland (18)(19)(20). However, this prevalence is considerably higher than that reported in France (3.7%) and the Netherlands (2%) (21,22). The medication strategy employed by first-line veterinarians from different countries or regions is a potential explanation for this difference. High prevalence of ESBL-producing E. coli threatens the efficacy of third-generation cephalosporins, such as cefovecin, approved for use in veterinary medicine (23). The E. coli isolates were obtained from several sample types in cats and dogs. The most common source of ESBL-producing E. coli in cats and dogs was from aspirated urine samples, with prevalence's of 54.5% (6/11) and 68.5% (37/54), respectively. This is unsurprising because urinary tract infection (UTI) is a common diagnosis in companion animals (24). Moreover, UTIs in cats and dogs usually involve a single agent: E. coli (25).
The bla CTX−M−1 group was observed in 58.5% of the bla genes. This bla gene group is also commonly detected in Europe, the Middle East, and Asia (26).    widely reported in food and pets in China, and its geographic distribution is primarily in Asian countries (29)(30)(31). Notably, CTX-M-55 has rarely been encountered outside Asia. However, the recent emergence of CTX-M-55 in companion animals in Switzerland may indicate the spreading of this enzyme due to international food or animal trade, which warrants further attention (18). A study in the United Kingdom also revealed a decreased prevalence of CTX-M-15 producers over some years in favor of new variants, particularly CTX-M-55 (32). CTX-M-124 was another frequently observed β-lactamase in our study. CTX-M-124 was first detected in wild birds (33); the transmission of CTX-M-124 to other animals from the migratory behavior of wild birds may explain, in part, the presence of CTX-M-124 in ESBL-producing E. coli from pets (34). ST457, ST131, and ST648 are the three major STs of ESBLproducing E. coli detected in our study, with ST457 being the most prevalent. This ST has been associated with diseases in companion animals in other studies (21,35). E. coli ST131 and ST648 with CTX-M have been reported worldwide in both human and animal samples. These two clones combine multidrug resistance and virulence; ST131, in particular, is a globally distributed uropathogenic E. coli lineage (36). E. coli ST131 O25b carrying CTX-M-15 is a globally spreading clone with a high virulence potential, making it a public health concern (37), whereas ST131 O25b with CTX-M-14 has predominated in Japan (38). By contrast, CTX-M-174 and CTX-M-194 were the two main β-lactamases in our E. coli ST131 O25b clones. An E. coli ST131 carrying CTX-M-174 was identified in humans in Korea (39). CTX-M-174 is a variant of CTX-M-14 with two amino acid substitutions (Glu-7-Leu and Asp-242-Gly). Regardless of the type of CTX-M present in our ST131 isolates, the presence of these clones in cats and dogs raises concerns about potential zoonotic risks. This finding also justifies the continued investigation of ESBL-producing E. coli to evaluate the persistence of these fast-spreading clones in companion animals in Taiwan. A study in Europe indicated that 1.6% of the diseased dogs and cats carried ESBL-producing Enterobacteriaceae but only 2 E. coli ST131 isolates were identified; therefore, companion animals may be a source of bla genes but may not be the major source of epidemic clones (40).
Previously, LeCuyer et al. (41) revealed a thought-provoking finding regarding uropathogenic E. coli in canines. They found that ST372 was the predominant ST in dogs, whereas ST372 was an infrequent human pathogen. The prevalence of ST372 observed in dogs was similar to that of ST131 in human uropathogenic E. coli and ST73 in feline E. coli that caused urinary tract infections. They therefore concluded that each host species may have a particular ST that comprises most of the E. coli uropathogens. A French study also reached a similar conclusion, identifying ST372 as the major pathogenic E. coli ST in dogs (42). Similar findings in two distinct geographic areas may indicate a dog-specific distribution of pathogenic E. coli clones instead of the effect of regional factors (42). In contrast to LeCuyer's and Valat's reports, ST372 was observed only once in our study. Different criteria for the screening of E. coli in the study design may have contributed to this discrepancy.
Some STs such as ST3429, ST5229, ST5640, ST5685, ST5686, ST5703, and ST5865, to the best of our knowledge, have not been reported before; therefore, the pathogenic potentials of these strains were unknown.
Imipenem reportedly remains relatively active against ESBLproducing bacteria (43), which is consistent with our results ( Table 6). Nonetheless, the use of carbapenems in companion animals should be avoided, since the emergence of carbapenem resistance in companion animals has been reported (44).
The current study had some limitations. AmpC-β-lactamases, which also hydrolyze the third generation of cephalosporins, were not assayed for the E. coli isolates. In addition, resistant plasmids were not characterized using PCR-based replicon typing. Although the results obtained in this study originate from only one veterinary hospital, this university-based teaching hospital is the major referral hospital for local veterinary clinics in Taipei. We believe that the information regarding ESBL in cats and dogs reported herein could be helpful for infection management and prevention.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

ETHICS STATEMENT
The purpose of collecting these samples from animals was for diagnosis and treatment. An ethical review process was not required for this study according to national/local guidelines.

AUTHOR CONTRIBUTIONS
Y-HH conducted the characterization of the phenotype and genotype of the ESBL-producing E. coli and drafted the manuscript. N-LK analyzed the ESBL-producing E. coli through MLST. K-SY conceived and coordinated this research plan. All authors have read and approved the final manuscript.

FUNDING
This work was supported by National Taiwan University grant G049919. Some of the results has been reported in the Chinese Society of Veterinary Science Academic Conference in 2018.