Prevalence and Characteristics of Extended-Spectrum-β-Lactamase-Producing and Carbapenemase-Producing Enterobacteriaceae from Freshwater Fish and Pork in Wet Markets of Hong Kong

Extended-spectrum-β-lactamase-producing Enterobacteriaceae (ESBL-E) and carbapenemase-producing Enterobacteriaceae (CPE) are of global health importance, yet there is a paucity of surveillance studies on food animals in Hong Kong. Here, we report a high prevalence of ESBL-E (ranging from 0.5% to 52.4%) and CPE (0% to 9.9%) from various food animal samples procured from wet markets across Hong Kong. All CPE strains were characterized by whole-genome sequencing and possessed NDM-1 and -5 genes and other resistance determinants. Given the increased resistance profile of these strains, this study highlights the emerging threat of ESBL-E and CPE disseminated in farmed animals. Furthermore, our data enriched our understanding of antibiotic resistance reservoirs from a One Health perspective that can widely spread across various niches, beyond health care settings.

KEYWORDS ESBL, CPE, Enterobacteriaceae, freshwater fish, pigs, livestock, One Health, aquaculture, antimicrobial resistance E xtended-spectrum-␤-lactamase-producing Enterobacteriaceae (ESBL-E) and carbapenemase-producing Enterobacteriaceae (CPE) are making a significant impact on antimicrobial resistance (AMR) due to their capability of horizontal gene transfer to other bacteria (1)(2)(3)(4). The isolation rate of ESBL-E from humans soared in the past few decades, and the emergence of CPE around the globe has been described as foreshowing the end of the antibiotic era, with an urgent threat to public health worldwide (1)(2)(3)(4)(5)(6). ESBL-E and CPE in food animals and their environment have been considered potential sources of resistant bacterial infections in the community (1). Increased consumption of antimicrobials, such as ␤-lactams, macrolides, aminoglycosides, polymyxins, and carbapenems, is observed on animal farms (1-3, 7, 8). Misuse of antibiotics in food animals, especially those used in the treatment of human infection, may lead to dissemination of highly mobile genetic elements in pathogenic bacteria that confer resistance to antimicrobials beyond the realm of health care settings to the community (1,7,8).
Animal-based food production is growing rapidly in Asia, with China as the world's biggest producer of farmed fish in 2016, claiming over 60% (49.2 million tonnes) of the world's production (80.0 million tonnes). The country has also been providing 54.5 million tonnes of pork around the world, supporting 50% of the global demand (9,10). A consequence of the extensive use of antibiotics in aquaculture and pig farms could be the isolation of multidrug-resistant bacteria in these farms and their produce, which can eventually interfere with human gut commensals when animal food is consumed raw or undercooked.
Apart from therapeutic purposes, antibiotics are also administered to healthy animals for growth promotion or prevention of disease to improve production yields in regions round the globe (11). As a result, drug-resistant bacteria are frequently found in surveillance programs worldwide in farm animals, which led to hypotheses of potential spread of AMR bacteria to humans via the food chain and contaminated water (11,12). There is limited understanding of the impact of the transmission of drugresistant bacteria from animals to humans, yet there is a rising threat of AMR to global public health that requires action from societies and government sectors (10,11). In addition to increased AMR awareness among professionals and consumers, surveillance systems for animal antibiotic use and antimicrobial resistance improve animal husbandry and are cornerstones to promote rational antibiotic use in animals (11).
In Hong Kong, fresh food products are often purchased from traditional wet markets. Currently, there are 97 public markets which are distributed across all districts in Hong Kong. They are run by the Food and Environmental Hygiene Department of Hong Kong SAR (13). Generally, the wet markets sell fresh meat, including poultry, beef, pork, etc., supplied from three licensed slaughterhouses (14) and include animal parts such as offal, head, tail, and feet, which are ingredients of the local cuisine. Live poultry are sold only at certain markets, with interventions in place to minimize zoonotic influenza transmission (15), while live marine fish and seafood may also be kept in "aquariums" before being sold.
Our present study sought to isolate and characterize ESBL-E and CPE from freshwater fish and pig gastrointestinal tract (GIT) organs procured from wet markets across Hong Kong during the period of April 2018 to January 2019 in order to understand the prevalence and significance of antimicrobial resistance in Hong Kong.

RESULTS
Prevalence of ESBL-E and CPE in freshwater fish and pig organs. A total of 171 ESBL-E and 28 CPE strains were isolated in our study from 411 fish and 339 pig organs that were purchased in wet markets across all 18 districts of Hong Kong. Thirty-nine of the samples contained 2 ESBL-E species and/or CPE species. The percentages of food samples with ESBL-E isolated were 52.4% (44/84) of pig's small intestine, 50% (32/64) of large intestine, 25.1% (43/171) of pig tongue, 0.9% (2/213) of snakehead fish, and 0.5% (1/198) of black carp samples. In addition to the low isolation rates of ESBL-E in fish, no CPE were identified from freshwater fish samples. However, 28 CPE strains were isolated from pig gastrointestinal tract (GIT) organs, with a prevalence rate of 8.15% (26/319). No ESBL-E or CPE were isolated from pigs' livers, kidneys, tails, minced meat, or snout during our preliminary examination; thus, samples from these sites were not further investigated.
The genetic characteristics of CP-K. pneumoniae are described in Table 3. One ST127 strain had the bla NDM-5 gene on an IncX3 plasmid together with bla SHV and bla TEM . The other 2 strains (serotypes K154:O2 and K152:OL102) showed bla NDM-1 genes on IncX3 plasmids as well as the presence of bla SHV . Similarly to ESBL-E. coli, genes inferring trimethoprim (dfrA gene), sulfonamide (sul gene), and aminoglycoside (aac genes) resistance were found. In addition, fluoroquinolone resistance gene oqxAB was found in all 3 K. pneumoniae strains. The presence of oqxAB was associated with low to intermediate resistance to quinoxalines, quinolones, tigecycline, nitrofurantoin, and several detergents and disinfectants (16). One of the CP-K. pneumoniae strains also contained the aac(6=)-Ib-cr gene with associated ciprofloxacin resistance in accordance  with previous reports (17,18). The KP23 strain was found to carry the fosA gene in the whole-genome analysis, which may explain the phenotypic resistance to fosfomycin.

DISCUSSION
There are limited data on the surveillance of ESBL-E and CPE in aquaculture and food animals, albeit similar data have been extensively reported in health care settings (19). We thus conducted this study to investigate the distribution and characteristics of ESBL-E and CPE in our food chain. This territory-wide surveillance study not only provides an update to the burden of ESBL-E and CPE in food products from wet markets but also highlights the possible exposure of these resistant bacteria in the community.   (20,21), and ESBL genes were less heterogeneous than in another Chinese study (bla TEM , bla SHV , bla CTX-M , and bla LEN in ESBL-E. coli) (18). The predominant ESBL gene identified from fish sampled in Saudi Arabia was bla CTX-M , which was detected in all tilapias imported from Thailand (6). Bla CTX-M was also observed in all carfu fish, 60% of milkfish, 52.3% of catfish, and 34.8% of tilapia, which were all imported from India (6). The original source of our fish is believed to be local as well as from Guangdong Province (southern part of China), while enquiries with the hawkers revealed also importation from Southeast Asia, such as Vietnam. Hence, CTX-M is still the most prevalent ESBL type among freshwater fish in Southeast Asia.  ESBL-E and CPE isolation rates were much higher in pigs than in freshwater fish in Hong Kong. Our isolation rates were higher than those reported in Thailand (36.7% of 588 pig farms) (22), Portugal (24.6% of 65 pig fecal samples; bla CTX-M-1 predominance) (23), and Cameroon (11.26% [8/71] of pigs) (24). A study in Denmark identified ESBL-producing E. coli in 79% (15/19) of pig farms with high consumption of cephalosporin compared to 20% (4/20) of the pig farms with no consumption (25). The former showed a predominance of bla CTX-M followed by bla SHV genes (25). On the other hand, our results were noticeably lower than a previous surveillance study during 2008 to 2010 from pig feces (63.6% [136/214]) in Hong Kong (26). This discrepancy might be due to the differences in sample types and processing methods. It may also be due to the recently introduced food-related initiatives under the Hong Kong Strategy and Action Plan on antimicrobial resistance (AMR) and licensing control of livestock keeping, regulating the feeding of drugs and chemicals to food animals in 2017 (2,27). Under this regulation, seven chemicals, including two antibiotics (avoparcin, clenbuterol, chloramphenicol, dienestrol, diethylstilbestrol, hexestrol, and salbutamol) are prohibited for use in food animals. Moreover, chemicals, including 36 antibiotics, have restricted usage on animals to address the concern of proper antibiotic usage and non-exceedance of drug residue levels for food safety purposes and AMR issues in Hong Kong (2,3,28).
Similarly to a Turkish study on fish of the Eastern Mediterranean (29), no CPE were detected in our freshwater fish. However, scanty reports of CPE in seafood were published during 2011 to 2016, with a bla VIM-1 -expressing E. coli (ST10) isolated from a Venus clam (Ruditapes philippinarum) in the Mediterranean Sea (Italy) (30) and carbapenem-resistant Enterobacteriaceae in 0.6% (8/1,328)  a Canadian study, where all the samples were imported from Southeast Asian countries, specifically, Vietnam and Bangladesh (31). This is the first study in Hong Kong to report NDM-1 and NDM-5 in pig offal from local farms. However, bla NDM-1 and bla NDM-5 were also reported in China, where the former subtype was found in E. coli from diseased pigs lung samples (0.89% [3/334]) in 2013 in Guangdong Province (32) and the latter subtype from imported pigs in Hong Kong originating from Guangdong, Henan, and Hunan provinces during 2015 and 2017, where the carbapenem-resistant Enterobacteriaceae (CRE) isolation rate was 0.7% from 856 samples (33). The CRE isolation rate was as high as 60% (18/30) in environmental samples collected from pig farms in the United States (34). The increased isolation of CRE urges an immediate and sustainable plan of action to overcome the dissemination of AMR in all sectors, including agricultural, veterinary, and public health sectors, worldwide.
There is a substantial number of overseas studies on the efficacy and cost efficiency of interventions to reduce AMR (35,36). However, local surveillance studies are limited, and it is important to investigate the possible effect and feasibility of new measures. From surveillance of AMR, the knowledge of the associated bacteria and molecular   elements are important in our aim to control the multidrug-resistant (MDR) Gramnegative pathogenic infection burden in the veterinary and public health sectors (2,3,10,11). Thus, our results will be valuable as a baseline to guide interventions in reducing AMR in agriculture, farms, and the community.
In conclusion, our study is the first to demonstrate the presence of ESBL genes in fish purchased from Hong Kong wet markets. Efforts should be made worldwide to closely monitor and introduce control of antibiotic resistance in aquaculture as well as pig farms. Our results depict a major reservoir of resistance genes that extend beyond our health care environments and threaten our dwindling options of effective antibiotics in future human medicine. Further epidemiological studies and detailed analyses of the mobile genetic elements encoding these genes should also be conducted to assess the full extent of zoonotic transmission and dissemination of these AMR genes between animal and humans.

MATERIALS AND METHODS
Sample collection. Food animal samples were purchased from 18 wet markets to include one each from a district and to represent three geographical regions (New Territories, Kowloon, and Hong Kong Island) of the city. Food animals, including gut samples of 213 snakehead fish (Channa spp.) and 198 black carp (Mylopharyngodon spp.) and 339 pig organs (171 tongues, 84 small intestines, 64 large intestines, 9 minced meat samples, 4 tails, 3 livers, 3 kidneys, and 1 snout). They were obtained between April 2018 and January 2019. Samples were transported and stored at 4°C after purchase and were processed within 24 h.
Isolation of ESBL-E and CPE. Deep tissues, where applicable in the food sample, were dissected using sterile equipment to avoid handling and environmental contamination (37). A small piece of tissue was transferred to 3 ml of normal saline and homogenized prior to seeding 10 l of the homogenized sample on Chromid ESBL agar (bioMérieux, France), which was then incubated at 37°C for 18 to 24 h (38). Presumptive Enterobacteriaceae colonies were selected from each plate and identified by matrix-assisted laser desorption ionization-time of flight mass spectrometry (MALDI-TOF-MS) (Bruker Daltonics, Inc., Germany) followed by phenotypic confirmation via double disc synergy test according to CLSI guidelines (39). All confirmed ESBL-E strains were saved at Ϫ80°C in 10% (vol/vol) glycerol-brain heart infusion (BHI) broth (Oxoid, UK) for further analysis. CPE were isolated by transferring 2 ml of homogenized normal saline suspension (mentioned above) to sterile tubes containing 8 ml of Trypticase soy broth (TSB) (Oxoid, UK) enriched with 1 mg/liter meropenem (Oxoid, UK) prior to incubation at 37°C for 18 to 24 h. Ten microliters of incubated broth was seeded on CARBA SMART agar (bioMérieux, France) and incubated at 37°C for 18 to 24 h (38,40). Presumptive Enterobacteriaceae colonies were selected from each plate and identified by MALDI-TOF-MS followed by phenotypic confirmation using a carbapenem inactivation method (CIM) as described previously (41). All confirmed CPE strains were saved at Ϫ80°C in 10% (vol/vol) glycerol-BHI broth (Oxoid, UK) for further analysis.
Screening for ESBL-producing isolates. The combination disk method was used to confirm ESBL-E strains. In brief, pairs of disks containing cefotaxime (30 g) and ceftazidime (30 g) were used with and without clavulanic acid (10 g) on the same inoculated plate containing Muller-Hinton agar (Oxoid, UK). A positive test result was defined as a Ն5-mm increase in the zone diameter compared to that of a disk without clavulanic acid (42).
Molecular characterization of ESBL-E and CPE. (i) DNA extraction, PCR amplification, and amplicon sequencing. Two to four bacterial colonies were emulsified in 200 l of distilled water, heated at 95°C for 15 min, and centrifuged at 16,000 ϫ g for 5 min (43). The supernatants were directly used as the template DNA and stored at Ϫ20°C until use. All confirmed ESBL-E strains were screened using multiplex PCRs as previously described for the detection of bla CTX-M genotype groups 1, 2, and 9, bla TEM ,