Molecular Epidemiology of Antimicrobial Resistance and Virulence Profiles of Escherichia coli, Salmonella spp., and Vibrio spp. Isolated from Coastal Seawater for Aquaculture

The occurrence of waterborne antimicrobial-resistant (AMR) bacteria in areas of high-density oyster cultivation is an ongoing environmental and public health threat given the popularity of shellfish consumption, water-related human recreation throughout coastal Thailand, and the geographical expansion of Thailand’s shellfish industry. This study characterized the association of phenotypic and genotypic AMR, including extended-spectrum β-lactamase (ESBL) production, and virulence genes isolated from waterborne Escherichia coli (E. coli) (n = 84), Salmonella enterica (S. enterica) subsp. enterica (n = 12), Vibrio parahaemolyticus (V. parahaemolyticus) (n = 249), and Vibrio cholerae (V. cholerae) (n = 39) from Thailand’s coastal aquaculture regions. All Salmonella (100.0%) and half of V. cholerae (51.3%) isolates harbored their unique virulence gene, invA and ompW, respectively. The majority of isolates of V. parahaemolyticus and E. coli, ~25% of S. enterica subsp. enterica, and ~12% of V. cholerae, exhibited phenotypic AMR to multiple antimicrobials, with 8.9% of all coastal water isolates exhibiting multidrug resistance (MDR). Taken together, we recommend that coastal water quality surveillance programs include monitoring for bacterial AMR for food safety and recreational water exposure to water for Thailand’s coastal water resources.


Introduction
AMR is an important One Health concept involving interconnectedness between humans, animals, and their shared environment, that can pose serious health threats to humans and animals. Every year, more than 700,000 deaths are attributed to AMR infection, and it is estimated that the number of deaths from MDR bacteria will increase to 10 million people by 2050 [1]. The extensive use of antimicrobials in human medicine, veterinary medicine, agriculture, and aquaculture has been implicated in the emergence and dissemination of AMR in the environment. The global consumption of antimicrobials in humans increased more than 60% between 2000 and 2015, especially in low-and middleincome countries, due to simple accessibility and irrational use [2]. The excessive use of antimicrobials directly affects the bacterial community and contributes to the overall selective pressure for AMR in aquatic environments [3]. Moreover, contamination from biocides and heavy metals in aquatic environments can also contribute to AMR selection due to their cross-resistance [4].
for tlh, which is a species-specific gene for V. parahaemolyticus. The two virulence factors tdh and trh were absent from this group of V. parahaemolyticus isolates. More than half of the V. cholerae isolates (51.3%, n = 20) were positive for ompW, which is a species-specific gene and virulence factor; in contrast, all 20 isolates were negative for the virulence gene ctx, which encodes the cholera toxin. Neither stx1 nor stx2 were detected in the 84 E. coli isolates.

ESBL Production
No ESBL production was detected in any bacterial isolate from the coastal water samples. However, the prevalence of resistance to ceftazidime (2.3%) was higher than for cefotaxime and cefpodoxime (0.5%).

ESBL Production
No ESBL production was detected in any bacterial isolate from the coastal water samples. However, the prevalence of resistance to ceftazidime (2.3%) was higher than for cefotaxime and cefpodoxime (0.5%).
Integrative and conjugative elements

Discussion
Investigation of AMR in the environment is challenging because the bacteria have acquired multiple mechanisms to confer AMR and MDR. This study focused on the distribution of AMR phenotypes and genotypes, and the virulence of E. coli, S. enterica subsp. enterica, V. parahaemolyticus, and V. cholerae isolated from coastal seawater in dense cultivation areas. Tracking resistant bacteria from coastal seawater is important under a One Health perspective, especially for aquatic products that are usually consumed raw or partially cooked, such as oysters. Therefore, contaminated water with resistant bacteria that is used for coastal aquaculture can pose a serious health risk to human health. Among the 384 bacterial isolates from coastal waters from Thailand's oyster-producing regions, the most common resistance phenotype was to ampicillin (52.1%), which is consistent with previous studies that also found a high prevalence of ampicillin resistance in coastal water [25][26][27].
The overall prevalence of MDR in this study was less than 10%, with the majority of the MDR bacteria being E. coli (25.0%) and S. enterica subsp. enterica (16.7%). In this study, E. coli also had the highest diversity of resistance patterns compared to other bacterial species, indicating that different bacterial species from the same coastal environment can acquire and/or maintain different resistance traits. This finding indicates that E. coli is potentially a reservoir and/or mode of introduction of drug resistance for coastal environments used for shellfish production in Thailand. A recent study indicated that the sources of bacterial contamination that causes marine pollution were anthropogenic activities, aquaculture, agriculture, industry, etc. [28]. In low-and middle-income countries, untreated water from inadequate and ineffective facilities has been shown to be a significant source of bacterial contamination [29]. Thermotolerant coliforms are the main bacteria in fecal coliforms that are usually present in the intestinal tracts of human and warm-blooded animals, which can indicate fecal contamination in humans, animals, water, food, and the environment [30,31]. In Thailand, the Department of Pollution Control sets the microbiological standards of coastal seawater (total coliforms, thermotolerant coliforms, and Enterococci), but does not regulate for the presence of resistant bacteria. Therefore, guidelines may be needed to reduce AMR bacterial contamination in coastal environments, especially for regions used for shellfish production and human recreation.
Among the 384 isolates, all S. enterica subsp. enterica isolates (n = 12) harbored invA. A diversity of S. enterica subsp. enterica serovars was observed in this study, indicating that multiple sources of contamination may exist for this bacterial pathogen. For example, S. enterica subsp. enterica serovars Lamberhurst and Othmarschen were reported in poultry and humans [32,33]. S. enterica subsp. enterica serovars Braenderup, Bruebach, Chester, Paratyphi B, Wentworth, Litchfield, and Orion have been isolated from shell eggs, papaya, wastewater, companion animals, livestock animals, and humans [34][35][36][37][38]. The stx1 and stx2 genes are bacterial toxins found in various serogroups of E. coli, but neither stx1 nor stx2 was detected in this study. Given our sample size of 84 E. coli, perhaps it is not surprising that stx genes are not found in this bank of E. coli, given that it has been estimated that one in a 1000 fecal coliform isolates harbors the stx gene [39,40].
The tdh (thermostable direct hemolysin) and trh (TDH-related hemolysin) are major virulence factors of V. parahaemolyticus. In this study, both genes were not present in our coastal water isolates. However, other virulence indicators, including type 3 secretion systems T3SS1 and T3SS2β found in pathogenic V. parahaemolyticus, have been reported in seafood samples [41]. For V. cholerae, more than half (51.3%) of the isolates were positive for ompW, which codes for an outer membrane protein, but none harbored the ctx gene which codes for the cholera toxin. This result agrees with previous studies showing that isolates of environmental V. cholerae are often not positive for the ompW gene [42,43]. The V. cholerae serogroups O1, O139, and O141 were also not found in this study. The distribution of serogroup O1 has been reported in environmental samples in northern Thailand, which contrasts with the results of this study [44]. However, non O1/O139 isolated from coastal area of southern Thailand was reported in humans with gastroenteritis [45].
ESBL-producing bacteria have been implicated in impacts on public health due to limited therapeutic options following human infections. The rapid spread of ESBL-producing bacteria in coastal environments increases concerns because of the widespread occurrence of gram-negative bacteria. In this study, ESBL producers were not detected in any of our coastal bacterial isolates. However, the cefotaxime-hydrolysing β-lactamase isolated in Munich (CTX-M) families, including CTX-M-15, CTX-M-14, and CTX-M-27, have been commonly found worldwide [46]. Previous studies reported ESBL producers in lagoon, recreational water, and wastewater [47,48]. Moreover, the spread of carbapenem-resistant Enterobacterales has been an emerging threat in coastal and estuarine water [49]. The bla TEM (5.5%) was the main resistance gene detected in this study, which agrees with a previous study [50]. Inversely, a previous study of E. coli isolated from coastal water contained a variety of bla CTX-M genes [51]. Other genes, including tetA (3.7%), qnrS (1.8%), strA (1.6%), and floR (1.3%), were detected at low prevalence this study. On the other hand, sul1 and sul2 were the most abundant AMR genes in the coastal mariculture system in China [52]. The absence of mcr-1, mcr-2, and mcr-3 was observed in this study, which is in contrast to a recent study in Brazil which found mcr-1 in E. coli isolated from coastal waters [53]. These differences in the geographical distribution of AMR contamination for coastal environments suggest that the processes of AMR contamination and persistence can vary widely from region to region.
Novel resistance mechanisms are associated with mobile genetic elements, which can facilitate the widespread dissemination of resistance determinants in the environment [54]. In this study, integrase (int1) was observed at a low level (1.8%), limited mostly to isolates of MDR E. coli. This observation raises concern of transferable MDR genes between intra-and inter-bacterial species, because integrons are located in transferable plasmids and conserved DNA sequences carrying gene cassettes with resistance genes, which can facilitate the spread of multiple resistant genes simultaneously [55]. In this study, one isolate of V. cholerae had int SXT . As a consequence, it may be prudent for coastal water quality monitoring programs to also include AMR surveillance for these mobile genetic elements.
Cohen's kappa analyses are generally used to test the agreement between two test methods. This study found strong associations (kappa agreement: 0.61-0.80) between MDR vs. TET, STR vs. strA, CHL vs. floR, and MDR vs. floR, each with statistical significance. For example, the MDR isolates were associated with the presence of resistance to tetracycline and floR, while streptomycin-and chloramphenicol-resistant bacteria were associated with their corresponding genes (strA and floR). Numerous other pairs of association were also observed, indicating relatively common association between resistant phenotypes, genotypes, and their determinants. Regarding the inferences from the logistic regression analyses, ampicillin-resistant isolates exhibited 20-times higher odds of carrying bla TEM (OR = 20.3) compared to ampicillin-susceptible isolates (p < 0.0001). This finding indicates that ampicillin-resistant isolates collected from Thailand's coastal aquaculture regions may preferentially harbor bla TEM . A previous study recommended that bla TEM might be a good indicator for AMR resistance genes in wastewater [56]. Additional regression analyses found that the trimethoprim-resistant isolates were positively associated with MDR (OR = 5.7; p < 0.0001) and int1 (OR = 4.7; p = 0.015). This observation suggests that mobile genetic elements may play a significant role in MDR bacterial development. Further studies using whole genome sequencing are recommended in order to more fully characterize the genomic basis of bacterial AMR resistance in Thailand's coastal regions used for shellfish cultivation.

Seawater Sample Collection and Bacterial Isolation
Coastal water samples of 500 mL were collected in sterile bottles from different oyster cultivation areas in Thailand, including Surat Thani (9 •  Bacterial confirmation followed standard methods. E. coli was determined using Levine's Eosin Methylene Blue (L-EMB) agar (Difco, Becton, Dickinson and Company, Sparks, MD, USA). The colonies showing flat, dark center, with or without metallic sheen, were collected and confirmed by biochemical tests, including triple sugar iron (TSI) agar and indole test [57]. Xylose Lysine Deoxycholate (XLD) (Difco) and MacConkey Agar (Difco) were used for Salmonella determination. Presumptive colonies showing pink color with or without black center were confirmed biochemically as Salmonella using TSI and citrate test [58]. CHROMagar™ Vibrio Agar (HiMedia Laboratories Ltd., Mumbai, India) was used for Vibrio spp. determination with colonies showing mauve and green-blue color identified as V. parahaemolyticus and V. cholerae, respectively. Oxidase test and arginine glucose slants were used to confirm isolates [59]. The positive control strains were E. coli ATCC™ 25,922, S. enterica serovar Typhimurium ATCC™ 14,028, V. parahaemolyticus ATCC™ 17,802, and V. cholerae non-O1/non-O139 ATCC™ 14,733, respectively.
One confirmed bacterial isolate of E. coli, V. parahaemolyticus, and V. cholerae per one positive sample, and up to five Salmonella isolates per one positive sample were sub-cultured and stored in 20% glycerol at −80 • C in the Department of Veterinary Public health, Faculty of Veterinary Science from Chulalongkorn University.

Serotyping of S. enterica subsp. enterica and V. cholerae
All Salmonella isolates were serotyped by detection of somatic (O) and flagella (H) antigens using slide agglutination, according to the Kauffmann-White scheme [60] with available commercial antiserum (S&A Reagents Lab, Bangkok, Thailand).

Determination of ESBL Production
The disc diffusion method was performed on all bacterial isolates according to the Clinical and Laboratory Standards Institute (CLSI) guidelines [61]. The susceptibility to ceftazidime (30 µg), cefotaxime (30 µg), and cefpodoxime (10 µg) (Oxoid, Basingstoke, UK) was used for the screening test. Resistance to at least one of these cephalosporins was then confirmed using a combination disk diffusion method of ceftazidime (30 µg), cefotaxime (30 µg), and these two disks combined with clavulanic acid. A difference in the inhibition zone between single cephalosporin and cephalosporins containing clavulanic acid greater than 5 mm was classified as a positive ESBL-producing isolate.

Genotypic Characterization of AMR and Virulence Genes by Polymerase Chain Reaction (PCR)
The DNA template was prepared using the whole cell boiling method [63]. Briefly, the bacterial isolate was streaked onto nutrient agar (Difco) and incubated overnight at 37 • C. An individual colony was transferred to an Eppendorf tube containing 150 µL of rNase-free water. The suspension was mixed, heated, and immediately placed on ice, after which the suspension was centrifuged at 11,000 rpm for 5 min. DNA-containing supernatant was collected for PCR template.

Statistical Analyses
Descriptive statistics were used to characterize the prevalence of resistance, AMR distribution, ESBL production, virulence genes, integrons, and SXT element in E. coli, Salmonella, V. parahaemolyticus, and V. cholerae. Cohen's kappa coefficient was used to determine the agreement between pairs of phenotypes and genotypes of AMR and resistance determinants for all isolates. The interpretation of the kappa coefficient, expressed as a strength of agreement, was: <0.00: poor; 0.00-0.20: slight; 0.21-0.40: fair; 0.41-0.60: moderate; 0.61-0.80: substantial; 0.81-1.00: almost perfect [85].
Multivariate logistic regression analysis was performed to characterize the association between the most common resistant phenotypes, AMR determinants, virulence genes, and ESBL production. Odds ratios (OR) were used to identify the magnitude of the observed association. The interpretation of ORs was indicated as OR > 1: positive association; OR < 1: negative association; OR = 1: no association. Two-sided hypothesis testing together with likelihood ratio test were used with a p ≤ 0.05 to decide statistical significance. All statistical analyses were performed with Stata version 14.0 (StataCorp, College Station, TX, USA).

Conclusions
The occurrence of waterborne AMR bacteria in areas of high-density oyster cultivation is an ongoing environmental and public health threat consistent with the One Health concept, especially given the popularity of shellfish consumption, water-related human recreation throughout coastal Thailand, and geographical expansion of the shellfish industry. Waterborne isolates of E. coli, S. enterica subsp. enterica, V. parahaemolyticus, and, to a lesser extent, V. cholerae from coastal Thailand exhibited phenotypic AMR for a wide variety of antimicrobials and for E. coli possessing concurrent genomic AMR. Although coastal seawater is regulated for excessive total coliforms, fecal coliforms, and Enterococci, current water quality monitoring does not include bacteria surveillance for phenotypic and/or genotypic AMR. Therefore, in addition to tracking and preventing key sources of bacterial contamination and promoting proper treatment of wastewater before release to Thailand's coastal water resources, we recommend that water quality surveillance programs also include monitoring excessive levels of bacterial AMR to better protect shellfish food safety, water-related body contact recreation, and to reduce AMR contamination for Thailand's coastal water resources.