Surveillance of antimicrobial-resistant Escherichia coli in Sheltered dogs in the Kanto Region of Japan

There is a lack of an established antimicrobial resistance (AMR) surveillance system in animal welfare centers. Therefore, the AMR prevalence in shelter dogs is rarely known. Herein, we conducted a survey in animal shelters in Chiba and Kanagawa prefectures, in the Kanto Region, Japan, to ascertain the AMR status of Escherichia coli (E. coli) prevalent in shelter dogs. E. coli was detected in the fecal samples of all 61 and 77 shelter dogs tested in Chiba and Kanagawa, respectively. The AMR was tested against 20 antibiotics. E. coli isolates derived from 16.4% and 26.0% of samples from Chiba and Kanagawa exhibited resistance to at least one antibiotic, respectively. E. coli in samples from Chiba and Kanagawa prefectures were commonly resistant to ampicillin, piperacillin, streptomycin, kanamycin, tetracycline, and nalidixic acid; that from the Kanagawa Prefecture to cefazolin, cefotaxime, aztreonam, ciprofloxacin, and levofloxacin and that from Chiba Prefecture to chloramphenicol and imipenem. Multidrug-resistant bacteria were detected in 18 dogs from both regions; β-lactamase genes (blaTEM, blaDHA-1, blaCTX-M-9 group CTX-M-14), quinolone-resistance protein genes (qnrB and qnrS), and mutations in quinolone-resistance-determining regions (gyrA and parC) were detected. These results could partially represent the AMR data in shelter dogs in the Kanto Region of Japan.

Drug-susceptibility testing in the 77 E. coli isolates from Kanagawa revealed that the isolates derived from 20 dogs (26.0%) were resistant to at least one antibacterial drug among ABPC, PIPC, CEZ, CTX, AZT, SM, KM, TC, CPFX, and NA ( Table 1) The chi-square test of sex-related differences in the ratio of susceptible (S), intermediate (I), and resistant (R) results of the antimicrobial susceptibility test revealed no significant differences between males and females for any of the antibacterial agents.
Multidrug-resistant E. coli was detected in 18 dogs, with resistance to as many as six drugs in 1 dog and five drugs in 5 dogs. The patterns of multidrug resistance are shown in Table 2a.   www.nature.com/scientificreports/ KM, and GM)-resistant or intermediate-resistant isolates (originally 15 samples, but one sample could not be tested due to poor growth). In seven quinolone-resistant or intermediate-resistant isolates, qnrB (1 sample) and qnrS (2 samples) were detected. Mutations in quinolone-resistance-determining regions (QRDR), 83 serine (S) and 87 aspartic acid (D) of the gyrA sequence and 80S of the parC sequence (2 samples), were detected. The first sample showed mutations of 83S to leucine (L) and 87D to tyrosine (Y) in gyrA and 80S to isoleucine (I) in parC. In the second sample, 83S was mutated to L and 87D was mutated to asparagine (N) in gyrA, and 80S to isoleucine (I) in parC. blaTEM was commonly detected in Chiba and Kanagawa prefectures. qnrB and qnrS were detected only in Chiba Prefecture, and the blaCTX-M-9 group CTX-M-14, blaDHA-1, and quinolone-resistant mutations were detected only in Kanagawa Prefecture.

Discussion
Drug-resistant E. coli was detected in some of the shelter dogs surveyed in this study. In addition, resistance genes related to the resistance mechanism were identified. First, we compared drug-susceptibility testing results with data available in Japan. Most of the canine AMR data currently reported in Japan are from animal patients who visited veterinary clinics for the treatment of some diseases. Other than those released by the MAFF in 2020 15 , almost no AMR survey data are available for non-patient companion animals. Table 1 compares the results of our study with the drug-resistance rates of dog rectal swab-isolated E. coli reported by the MAFF. The MAFF survey also included dogs taken to a veterinary hospital in 2017 (ill dogs) and 2018 (healthy dogs), which overlaps with our survey period (2016-2017). Regarding common antibacterial agents tested in our study and the MAFF survey (ABPC, CEZ, CTX, MEPM, SM, KM, GM, TC, CPFX, NA, and CP), the antibiotic resistance rate observed in sheltered dogs was mostly lower than that in healthy dogs in the MAFF survey. In the samples obtained from Chiba, the 95% confidence interval (95% CI) range of the antibiotic resistance rates against ABPC, CEZ, CTX, MEPM, SM, GM, CPFX, and NA in sheltered dogs was lower than that in healthy dogs in the MAFF survey ( Table 1). The 95% CI range of the resistance rates against KM, TC, and CP in sheltered dogs overlapped with that in healthy dogs in the MAFF survey. In the samples obtained from Kanagawa, the 95% CI range of the antibiotic resistance rates against ABPC, CEZ, CTX, MEPM, GM, TC, CPFX, NA, and CP in sheltered dogs was lower than that in healthy dogs in the MAFF survey ( Table 1). The 95% CI range of the resistance rates against KM and SM in sheltered dogs overlapped with that in healthy dogs in the MAFF survey. In the MAFF survey, the resistance rates in healthy dogs were lower than those in sick dogs 15 . The 95% CI range of the resistance rates against KM and CP in the samples from Chiba and against KM in the samples from Kanagawa overlapped with that in the sick dogs in the MAFF survey ( Table 1). The use of β-lactam antibiotics and fluoroquinolone antibiotics in veterinary medicine has been reported to promote an increase in the number of drug-resistant E. coli isolates 1,16 . Sheltered dogs include abandoned and stray dogs; presumably, these dogs are less exposed to veterinary medical facilities and the administration of antibacterial drugs than dogs in households. This may explain the lower drug-resistance rate observed in our study than in the MAFF survey. Next, the results of the identification of drug-resistance genes were compared with data from Japan and other countries. Several types of β-lactamase genes, QRDR mutations, and quinolone-resistant protein genes were detected in E. coli from shelter dogs. β-Lactamase genes, blaTEM, blaCTX-MTX-M-14, and blaDHA, were detected. These are genes that are reportedly detected in the intestinal bacteria of humans, farm animals, and companion animals [17][18][19][20][21] . A 2016 study of sheltered dogs and cats in Osaka, Japan, reported that many of these resistance genes are detected in cephalosporin-resistant E. coli 22 . As quinolone-resistance mechanisms, QRDR mutations and quinolone-resistance proteins (qnrB and qnrS) were detected. Furthermore, β-lactamase genes, which are also involved in resistance mechanisms, have been detected in humans, farm animals, and companion animals 17,23-25 . The quinolone-resistant mechanisms have been predominantly detected in a survey of E. coli in shelter dogs and cats in Osaka from 2016 to 2017 26 . Therefore, the drug-resistance mechanism in E. coli detected in this study was of the type that has been reportedly detected in the intestinal bacteria of dogs in Japan and abroad.
In conclusion, the rates of resistance to various antibiotics among the E. coli isolated from shelter dogs in the animal welfare centers in Chiba and Kanagawa prefectures were mostly lower than those in the healthy and sick domestic dogs in Japan, surveyed at almost the same time 15 . The detected resistance genes presented the same trend as those reported in shelter dogs in the same years in Japan 22,26 . As several studies have already mentioned, drug-resistant bacteria in companion animals can be a health risk to humans [6][7][8][9] . AMR surveillance in companion animals, including shelter dogs, for which there is a lack of data, needs to be widely conducted to accurately assess the AMR prevalence in Japan. The present results will make up for the lack of AMR data in shelter dogs.

Methods
Sampling of dog feces. This study was conducted in accordance with the principles of the ARRIVE guidelines. Feces from sheltered dogs were used, and no invasive treatment was performed on the dogs; therefore, the study did not require ethics approval.
The required sample size (n) was calculated at a 95% confidence level using the formula and parameters below. The proportion of AMR (P) in the population was estimated as 10%, based on the results of the preliminary survey. The margin of error (δ) was 0.08. The required sample size was estimated to be 54.
Between 2016 and 2017, we collected feces from 61 and 77 dogs housed in two public animal welfare centers in Chiba and Kanagawa prefectures, in the Kanto Region of Japan. None of the dogs exhibited any specific veterinary health abnormalities in their medical data. The age was not known for most animals, but samples were The fecal samples were collected using a sterilized swab from naturally excreted feces. The portion in contact with the ground was not collected. Duplicate samples from the same animal were not collected. The fecal samples were preserved in Carry-Blair transport medium (Nissui Pharmaceutical Co., Ltd., Tokyo, Japan), stored at 4 °C, and transported to the laboratory for E. coli culture immediately.
Detection of E. coli. The fecal samples were resuspended in sterilized saline solution and smeared onto an XM-G agar plate (Nissui Pharmaceutical Co., Ltd.) using a platinum loop. The plates were cultured under aerobic conditions at 35 °C for 20 h. After incubation, β-glucuronidase-positive colonies (a biochemical characteristic of E. coli) were selected and purified in nutrient agar (Eiken Chemical Co., Ltd., Tokyo, Japan). The selected colonies were identified as E. coli by polymerase chain reaction according to an established method 27 .
Drug-susceptibility profile testing. The disk diffusion method, based on the performance standards issued by the CLSI 14 , was used to test the susceptibility of E. coli isolates toward all drugs except LVFX. Mueller-Hinton agar and antimicrobial susceptibility test discs (Sencsi-Disc) were purchased from BD Biosciences (Franklin Lakes, NJ, USA). The dry Eiken plate (Eiken Chemical Co., Ltd.), which uses the broth microdilution method based on the performance standards issued by the CLSI, was used for susceptibility testing of only LVFX ( Table 1). Results of the antimicrobial susceptibility test were indicated as S, I, or R. E. coli ATCC25922 and Pseudomonas aeruginosa ATCC27853 (both from American Type Culture Collection, Manassas, VA, USA) were used as control strains.
Chromosomal DNA and plasmid DNA extraction. PrepManUltra sample preparation reagent (Thermo Fisher Scientific, Waltham, MA, USA) was used for chromosomal DNA extraction. The Mini Plus Plasmid DNA Extraction System (Viogen-Bio Tek Corporation, Taipei, Taiwan) was used for plasmid DNA extraction.
Detection of drug-resistance genes by PCR and DNA sequencing. Eighteen samples of multidrugresistant E. coli were subjected to genetic testing to predict the mechanism of drug resistance. One of the strains (sample No. 16C1) presented poor growth; therefore, 17 samples were tested. E. coli that showed resistance or intermediate resistance to β-lactam antibiotics were analyzed for blaTEM, blaSHV, and AmpC (bla CMY/MOX, bla CMY/LAT, bla DHA, bla ACC , bla ACT-1/MIR-1, and bla FOX) genes 28,29 . In addition to this, we analyzed the CTX-M genes (bla CTX-M-1-group, bla CTX-M-2-group, blaCTX-M-8-group, and bla CTX-M-9-group) in E. coli that showed third-generation cephalosporin resistance or intermediate resistance 30 and carbapenemase genes (bla IMP-1, bla IMP-2, bla VIM-2, bla KPC-2, bla GES, and bla NDM-1) in carbapenem-resistant E. coli [31][32][33][34][35] . Aminoglycoside antibiotic resistance and intermediate E. coli were analyzed for aminoglycoside resistance 16S rRNA methylases genes (armA and rmtB) and aminoglycoside-modifying enzyme genes (Aac(6′)-Ib, Ant(3″)-Ia, Aph(3′)-Ia, and Aac(3)-II) 36,37 . Quinolone-resistant and intermediate-resistant E. coli were analyzed for quinolone-resistance genes (qnrA, qnrB, qnrC, qnrD, qnrS, qepA, oqxAB, and aac(6')-lb-cr) 38 . The antibiotic resistance genes mentioned above were analyzed using the extracted plasmid DNA as a template to amplify the target region by PCR, followed by sequencing to decipher the nucleotide sequence and homology search by BLAST (https:// blast. ncbi. nlm. nih. gov/ Blast. cgi).). PCR and DNA sequencing analysis using chromosomal DNA as the template were performed to examine mutations in QRDR in quinolone-resistant and intermediateresistant strains. In the DNA gyrase subunit A gene (gyrA), the mutations at 83S and 87D were analyzed 39 . In topoisomerase IV gene (parC), the mutations at 80S and 84 glutamic acid (E) were analyzed 40 . The primers used for the amplification of each gene and the references are shown in Table 3. The PCR conditions were based on the conditions described in the references, and the Multiplex PCR Kit (Takara Bio, Kyoto, Japan) was used for PCR. The ProFlex PCR System (Thermo Fisher Scientific) was used as the thermal cycler for PCR. The PCR amplification product was treated with Illustra ExoProStar (Cytiva, Marlborough, MA, USA) to remove unwanted nucleotides. The primers used for sequencing were the primers used for PCR amplification. DNA sequencing was outsourced to a specialized external organization (Fasmac Co., Ltd., Kanagawa, Japan). The nucleotide sequences were determined by the direct sequencing of PCR products, performed by Sanger sequencing on a 3730xl DNA Analyzer (Thermo Fisher Scientific) using the BigDye Terminator and BigDye XTerminator Purification Kit (Thermo Fisher Scientific) 41 .
Statistical analysis. The sex differences in the rate of S, I, and R were evaluated using the chi-square test.
SPSS (version 19, IBM Japan, Tokyo, Japan) was used for the analysis. The statistical significance level was set to 5%.
The 95% CI of resistance rates were calculated using the Agresti-Coull method.