Distribution of phenotypic and genotypic antimicrobial resistance and virulence genes in Vibrio parahaemolyticus isolated from cultivated oysters and estuarine water

ABSTRACT

A total of 594 Vibrio parahaemolyticus isolates from cultivated oysters (n = 361) and estuarine water (n = 233) were examined for antimicrobial resistance (AMR) phenotype and genotype and virulence genes. Four hundred forty isolates (74.1%) exhibited resistance to at least one antimicrobial agent and 13.5% of the isolates were multidrug-resistant strains. Most of the V. parahaemolyticus isolates were resistant to erythromycin (54.2%), followed by sulfamethoxazole (34.7%) and trimethoprim (27.9%). The most common resistance genes were qnr (77.8%), strB (27.4%) and tet(A) (22.1%), whereas bla_TEM (0.8%) was rarely found. Four isolates (0.7%) from oysters (n = 2) and estuarine water (n = 2) were positive to tdh, whereas no trh-positive isolates were observed. Significantly positive associations among AMR genes were observed. The SXT elements and class 1, 2 and 3 integrons were absent in all isolates. The results indicated that V. parahaemolyticus isolates from oysters and estuarine water were potential reservoirs of resistance determinants in the environment. This increasing threat of resistant bacteria in the environment potentially affects human health. A ‘One Health’ approach involved in multidisciplinary collaborations must be implemented to effectively manage antimicrobial resistance.

Oyster meat and estuarine water are potential sources of phenotypic and genotypic antimicrobial resistance of Vibrio parahaemolyticus.

INTRODUCTION

Antimicrobial resistance (AMR) has rapidly emerged and become a serious public health challenge worldwide. This problematic issue involves several sectors (i.e. humans, animals and environment) and is referred to as One Health issue. Contamination of multidrug-resistant (MDR) bacteria and AMR determinants in the environment, including the aquaculture and agricultural environments, has been discovered (Singer et al. 2016). Previous studies reported that seafood serves as potential reservoirs for the dissemination of MDR Vibrio parahaemolyticus, which could negatively impact human health (Letchumanan et al. 2014). Evidently, anthropogenic activities (including those performed for municipal sewage, aquaculture discharges, waste from agriculture and run-off water) can introduce MDR bacteria and AMR determinants into the water and environment (Marti, Variatza and Balcazar 2014).

Mobile genetic elements (e.g. plasmids, transposons, integrons or persistent mutations in chromosomal genes) have played a major role in the widespread AMR problem (Christaki, Marcou and Tofarides 2020). Among these, integrons have been extensively studied, and their role in the acquisition and distribution of AMR genes has been well characterized (Ceccarelli et al. 2006; Kitiyodom et al. 2010; Liu, Wong and Chen 2013). These elements can be transferred either within or between species, contributing to the spread of AMR. Among nine integron types, class 1 integrons are most relevant to MDR pathogens (Rowe-Magnus and Mazel 2002) and are known as central players in the dissemination of resistance genes among bacteria. Integrative and conjugative elements (ICEs) or SXT elements containing a number of AMR genes have been reported in V. parahaemolyticus (Kitiyodom et al. 2010). ICEs are self-transmissible mobile genetic elements that can integrate into and excise from the bacterial chromosome using their own integrase gene and can be horizontally transferred to a new host (Johnson and Alan 2015).

Thailand has been ranked among the top five largest exporters of fish and fishery products for decades, with estimated production of bivalves at 197 200 tons in 2016 (FAO 2016). Shellfish aquaculture in Thailand has started for over a century and has rapidly grown to serve both domestic and international demands. Southern Thailand is the most famous location for marine aquaculture, especially bivalves. However, the contamination of pathogenic bacteria in shellfish may negatively affect human health, and the major seafood-borne bacteria are the genus Vibrio. Among halophilic vibrios, V. parahaemolyticus and Vibrio vulnificus are major pathogens of public health concern. As oysters are usually consumed either raw or partially cooked, they can pose a risk for pathogenic Vibrio infection in humans.

Vibrio parahaemolyticus is a Gram-negative bacterium globally distributed in marine and estuarine environment, and one of the leading causes of seafood-borne diseases, e.g. vibriosis (Xu et al. 2014). It is considered one of the important target bacteria for integrated monitoring and surveillance of AMR in aquatic animals for human consumption (OIE 2015). Vibriosis usually occurs through the consumption of raw or insufficiently cooked seafood products, drinking of contaminated water and direct contact with contaminated environment through open wounds. In 2016, the US Center for Disease Control and Prevention revealed that V. parahaemolyticus is a major foodborne pathogen in the United States, accounting for ∼34 664 foodborne cases each year (Scallan et al. 2011). According to the Annual Epidemiological Surveillance in Thailand, V. parahaemolyticus was one of the leading bacterial isolates of food poisoning associated with seafood contamination (Bureau of Epidemiology 2018). Foodborne outbreaks have been commonly associated with V. parahaemolyticus carrying virulence genes, particularly thermostable direct hemolysin (tdh) and thermostable direct hemolysin-related hemolysin (trh) (Wang et al. 2017; Park et al. 2018). The tdh and trh genes are major virulence genes in V. parahaemolyticus and considered the most important markers related to hemolytic, enterotoxin and cytotoxic activities in host cells. In particular, V. parahaemolyticus harboring trh can produce urease that increases inflammation of cytokines and has ability to colonize the intestine (Letchumanan et al. 2014).

Despite Thailand being a leading exporter of fish and fishery products, data on distribution and genetic characteristics of AMR are still limited, and AMR is not systematically monitored in aquatic environments. As target bacteria for AMR monitoring in aquaculture, elucidating the phenotypic and genotypic AMR of V. parahaemolyticus will facilitate a better understanding of the dissemination of AMR in the aquatic environment. Therefore, the objectives of this study were to (i) examine the phenotypic and genotypic characteristics of AMR among the Vibrio isolates and (ii) determine the prevalence of virulence genes of V. parahaemolyticus isolates from cultivated oysters and estuarine waters in Southern Thailand.

MATERIALS AND METHODS

Sample collection

Oyster (Crassostrea lugubris and Crassostrea belcheri; n = 144) and estuarine water (n = 96) samples were obtained from Thap Put district, Phang Nga province, Southern Thailand, during March 2016–February 2017. The study location was selected based on a relatively popular oyster cultivation site in the country, and the oysters are usually served for communities and tourists. Cultivated oyster samples were retrieved from mature oysters at market size ∼10–12 months of age for human consumption. The oyster and estuarine water samples were collected from four oyster cultivated farms in the following seasons: summer (from mid-February to mid-May), rainy (from mid-May to mid-October) and winter (from mid-October to mid-February) based on Thailand Meteorological Department (www.tmd.go.th). These cultivated oysters were naturally raised without antimicrobial usage. The estuarine water samples were collected from the same area when harvesting oysters. All samples were transported on ice within 24 h to the microbiological laboratory, the Department of Veterinary Public Health, Faculty of Veterinary Science, Chulalongkorn University, for subsequent analyses.

Vibrio parahaemolyticus isolation and confirmation

The V. parahaemolyticus strains were isolated according to the US Food and Drug Administration Bacteriological Analytical Manual (Kaysner and DePaola 2004). Presumptive colonies were streaked onto tryptic soy agar (TSA). Briefly, alkaline peptone water (Difco, Sparks, MD, USA) was used for sample preparation. Thiosulfate-citrate-bile salts-sucrose (TCBS) (Difco) and CHROMagar Vibrio (HiMedia Laboratories Ltd, L.B.S. Marg, Mumbai, India) agar plates were used for isolating bacteria. On TCBS agar plates, colonies of V. parahaemolyticus were opaque and green with 2–3 mm in diameter, whereas on CHROMagarVibrio plates, V. parahaemolyticus colonies had mauve color. Presumptive colonies were streaked onto TSA (Difco) plates supplemented with 2% NaCl and then biochemically confirmed. Three to four V. parahaemolyticus isolates were collected per sample type per collection time. Total bacterial isolates (n = 594) originated from 361 pooled oyster meat and 233 estuarine water samples. All bacterial strains were purified and stored at −80°C in 20% glycerol.

Antimicrobial susceptibility testing

All V. parahaemolyticus strains were examined for minimum inhibitory concentrations by using the 2-fold agar dilution method according to the Clinical and Laboratory Standards Institute (CLSI 2016). Antimicrobials were selected as representative antibiotics of different antibiotic classes and based on their importance in human and veterinary medicine. Eight antimicrobials were tested and their breakpoints with ranges of tested concentrations were as follows: ampicillin (32, 0.25 to 1024 μg mL⁻¹), chloramphenicol (32, 0.25 to 1024, μg mL⁻¹), ciprofloxacin (4, 0.01563 to 64 μg mL⁻¹), erythromycin (8, 0.0625 to 128 μg mL⁻¹), streptomycin (64, 1 to 1024 μg mL⁻¹), sulfamethoxazole (76, 1.1875 to 1216 μg mL⁻¹), tetracycline (16, 0.0625 to 512 μg mL⁻¹) and trimethoprim (4, 0.125 to 128 μg mL⁻¹). Escherichia coli ATCC 25922 and Staphylococcus aureus ATCC 29213 were used as quality control strains.

Phenotypic determination of virulent V. parahaemolyticus strains

Phenotypic determination of thermostable direct hemolysin (tdh) and tdh-related hemolysin (trh) virulence genes was carried out in all isolates by using Kanagawa phenomenon (KP) test as previously described (Honda et al. 1980). Briefly, a single colony of each isolate from TSA (Difco) plate supplemented with 3% NaCl was streaked on Wagatsuma agar (HiMedia, Mumbai, India) containing 2% fresh sheep red blood cells. After overnight incubation at 37°C, the isolates that produce β-hemolytic zone were observed and recorded as positive to KP+. A KP+ isolate containing both tdh and trh was used as a positive control.

PCR

DNA template was prepared by the whole cell boiled lysate method as previously described (Scarano et al. 2014). Briefly, the V. parahaemolyticus isolates were grown on TSA (Difco) supplemented with 2% NaCl. A single colony was inoculated into 2 mL of Luria–Bertani broth (Difco) containing 3% NaCl. After incubation at 37°C overnight, 1 mL of broth culture was centrifuged at 13 000 × g for 5 min, and the pellet was resuspended in 500 µL of sterile distilled water. The samples were boiled for 10–15 min and immediately placed on ice for 10 min. Then, the cell lysate was removed by recentrifugation at 5000 × g for 5 min. A total of 200 µL of the supernatant was transferred to a fresh microtube and stored at −20°C until used.

All primers used in this study are listed in Table 1. The presence of eight resistance genes was detected in all (n = 594) isolates with corresponding resistance phenotype: bla_TEM, ampicillin resistance; qnr, fluoroquinolone resistance; erm(B), erythromycin resistance; strB, streptomycin resistance; sul2, sulfamethoxazole resistance; tet(A), tetracycline resistance; and dfrA1 and dfrA18, trimethoprim resistance. The qnr encoding quinolone resistance was examined in all isolates. The presence of virulence genes (i.e. tdh and trh) was investigated in all isolates using multiplex polymerase chain reaction (multiplex PCR) (Bej et al. 1999). ICEs were determined using simple PCR. The presence of class 1, 2 and 3 integrons by multiplex PCR determined the integrase genes intI1, intI2 and intI3, as previously described (Kitiyodom et al. 2010). The resistance determinants were selected based on their common presence in food animals as shown in previous studies (Wu et al. 2011; Fang et al. 2019).

Table 1.

PCR primers used in this study.

Primer	Sequence (5′–3′)	Gene	Amplicon size (bp)	References
Virulence genes
tdhF	GTAAAGGTCTCTGACTTTTGGAC	tdh	269	Bej et al. (1999)
tdhR	TGGAATAGAACCTTCATCTTCACC
trhF	TTGGCTTCGATATTTTCAGTATCT	trh	500	Bej et al. (1999)
trhR	CATAACAAACATATGCCCATTTCCG
AMR genes
blaTEMF	ATAAAATTCTTGAAGAC	bla TEM	1075	Letchumanan et al. (2015)
blaTEMR	TTACCAATGCTTAATCA
qnrF	TCTCGCTAAGGCTCGTAGC	qnr	902	Poirel et al. (2005)
qnrR	TTCCTCGTCGAGGTTATTCG
ermBF	AGACACCTCGTCTAACCTTCGCTC	erm(B)	640	Raissy et al. (2012)
ermBR	TCCATGTACTACCATGCCACAGG
strBF	GGCAGCATCAGCCTTATAATTT	strB	470	Mala et al. (2016)
strBR	GTGGATCCGTCATTCATTGTT
sul2F	TGCGGATGAAGTCAGCTCC	sul2	623	Mala et al. (2016)
sul2R	GGGGGCAGATGTGATCGAC
tetAF	GTAATTCTGAGCACTGTCGC	tet(A)	957	Mala et al. (2016)
tetAR	CTGCCTGGACAACATTGCTT
dfrA1F	CAAGTTTACATCTGACAATGAGAACGTAT	dfrA1	277	Mala et al. (2016)
dfrA1R	ACCCTTTTGCCAGATTTGGTA
dfrA18R	ACTGCCGTTTTCGATAATGTGG	dfrA18	389	Mala et al. (2016)
dfrA18F	TGGGTAAGACACTCGTCATGGG
Integrons
int1F	CCTGCACGGTTCGAATG	intl1	497	Kitiyodom et al. (2010)
int1R	TCGTTTGTTCGCCCAGC
int2F	GGCAGACAGTTGCAAGACAA	intl2	247	Kitiyodom et al. (2010)
int2R	AAGCGATTTTCTGCGTGTTT
int3F	CCGGTTCAGTCTTTCCTCAA	intl3	155	Kitiyodom et al. (2010)
int3R	GAGGCGTGTACTTGCCTCAT
ICEs
intSXTF	GCTGGATAGGTTAAGGGCGG	int SXT	592	Kitiyodom et al. (2010)
intSXTR	CTCTATGGGCACTGTCCACATTG

Statistical analyses

The presence of resistance phenotype and resistance genotype including AMR genes, integrons, SXT element and virulence genes was described using descriptive statistics. Pearson's Chi-square test was used to determine the association between phenotypic and genotypic AMR among oyster and estuarine water samples. The association among resistance genes and the associations between seasonal trends and phenotypic and genotypic resistance were analyzed using logistic regression models. Odds ratio (OR) > 1 is considered as positive relationship, while OR < 1 is negative relationship. A P-value < 0.05 was considered as statistically significant difference under the two-sided hypothesis test. All statistical analyses were conducted using SPSS version 22.0 (IBM Corp., NY, USA).

RESULTS

Antimicrobial susceptibility of V. parahaemolyticus isolates

Different resistance rates were observed among the V. parahaemolyticus isolates (n = 594) in this study (Fig. 1A). Four hundred forty isolates (74.1%) were resistant to at least one antimicrobial agent tested, whereas 154 isolates (25.9%) were susceptible to all agents. The majority of the isolates were resistant to erythromycin (54.2%), followed by sulfamethoxazole (34.7%) and trimethoprim (27.9%). In contrast, resistance to streptomycin (0.8%) and tetracycline (0.5%) was limited. No chloramphenicol and ciprofloxacin resistance was detected. MDR bacteria being resistant to at least three different classes of antimicrobial agents were observed (13.5%).

Figure 1.

Distribution of AMR among the V. parahaemolyticus isolates from pooled oyster meat (n = 361) and estuarine water (n = 233) samples. (A) All isolates and (B) isolates classified by sample type (ERY, erythromycin; SUL, sulfamethoxazole; TRI, trimethoprim; AMP, ampicillin; STR, streptomycin; TET, tetracycline; MDR, multidrug resistance).

The AMR prevalence classified by the sample type is shown in Fig. 1B. Among the oyster isolates (n = 361), resistance rates to erythromycin, sulfamethoxazole, trimethoprim, ampicillin and streptomycin were 53.7, 33.5, 27.4, 10.2 and 1.4%, respectively. Resistance rates to erythromycin, sulfamethoxazole, trimethoprim, ampicillin and tetracycline among the isolates from estuarine waters (n = 233) were 54.9, 36.5, 28.8, 12.4 and 1.3%, respectively. The MDR V. parahaemolyticus isolates were detected in 13.0% of pooled oyster meat and 14.2% of estuarine water samples. No marked statistical difference in the prevalence of resistant V. parahaemolyticus was observed among oysters and estuarine water samples (P > 0.05). Different AMR patterns are shown in Table 2. The seasonal trend was associated with the presence of the isolates resistant to erythromycin, sulfamethoxazole and trimethoprim (P < 0.05). The isolates resistant to erythromycin and sulfamethoxazole were mainly found during winter. In contrast, the V. parahaemolyticus isolates resistant to trimethoprim were commonly observed in the rainy season.

Table 2.

AMR patterns of V. parahaemolyticus isolates (n = 594).

Antimicrobial resistance pattern		No. of V. parahaemolyticus isolates (%)
		Oyster (n = 361)	Estuarine water (n = 233)	Total (n = 594)
1	ERY	76 (21.0)	54 (23.3)	130 (21.9)
2	SUl	13 (3.6)	8 (3.4)	21 (3.5)
3	TRI	29 (8.0)	12 (5.2)	41 (6.9)
4	AMP	2 (0.6)	7 (3.0)	9 (1.5)
5	STR	1 (0.3)	0 (0)	1 (0.2)
6	ERY-SUL	43 (11.9)	27 (11.6)	70 (12.0)
7	ERY-TRI	19 (5.3)	6 (2.6)	25 (4.2)
8	ERY-AMP	8 (2.2)	11 (4.7)	19 (3.2)
9	ERY-STR	1 (0.3)	0 (0)	1 (0.2)
10	SUL-TRI	12 (3.3)	16 (6.9)	28 (4.7)
11	SUL-AMP	7 (1.9)	2 (0.9)	9 (1.5)
12	TRI-AMP	3 (0.8)	1 (0.4)	4 (0.7)
13	TRI-STR	1 (0.3)	0 (0)	1 (0.2)
14	ERY-SUL-TRI	28 (7.8)	22 (9.4)	50 (8.4)
15	ERY-SUL-STR	2 (0.6)	0 (0)	2 (0.3)
16	ERY-SUL-AMP	10 (2.8)	1 (0.4)	11 (1.9)
17	ERY-TRI-AMP	2 (0.6)	1 (0.4)	3 (0.5)
18	SUL-TRI-AMP	1 (0.3)	3 (1.3)	4 (0.7)
19	ERY-SUL-TRI-AMP	4 (1.1)	3 (1.3)	7 (1.2)
20	ERY-SUL-TRI-TET	0 (0)	3 (1.3)	3 (0.5)
	Total	262 (72.6)	177 (76.0)	439 (73.9)

ERY, erythromycin; SUL, sulfamethoxazole; TRI, trimethoprim; AMP, ampicillin; STR, streptomycin; TET, tetracycline.

AMR genes of V. parahaemolyticus isolates

Of all eight resistance genes tested, qnr was the most common (77.8%), followed by those encoding resistance to streptomycin (strB, 27.4%), tetracycline [tet(A) 22.1%] and trimethoprim (dfrA18, 19.5%) (Fig. 2A).

Figure 2.

Distribution of AMR genes in the V. parahaemolyticus isolates among oyster meat (n = 361) and estuarine water (n = 233). (A) All isolates and(B) isolates classified by samples. *P < 0.05.

The prevalence of AMR genes in different sample types is shown in Fig. 2B. Among the Vibrio isolates from oyster (n = 361), the rates of AMR genes for qnr, strB, tet(A) and dfrA18 were 81.2, 27.1, 23.3 and 16.6%, respectively. Among the isolates from estuarine water (n = 233), the most frequently found resistance gene was qnr (72.5%), followed by strB (27.9%), dfrA18 (24.0%) and tet(A) (20.2%), respectively. The significant difference was observed only between the presence of qnr and dfrA18 in oyster meat and estuarine water samples (P < 0.05). The presence of qnr was significantly higher in oyster meat than in estuarine water samples, while a significant higher rate of dfrA18 was observed in estuarine water samples than oyster meat (P < 0.05). The resistance genes [tet(A) and dfrA18] were more frequently found in the rainy season (P < 0.05).

Associations between AMR phenotype and genotype

To observe the association between AMR phenotype and genotypes (Table 3), the statistical associations between erythromycin [erm(B)], sulfamethoxazole (sul2) and trimethoprim (dfrA18) in V. parahaemolyticus isolates were analyzed using logistic regression models. Five different logistic regression models [i.e. bla_TEM vs strB; qnr vs strB; strB vs tet(A), bla_TEM and qnr; sul2 vs tet(A); tet(A) vs strB and sul2] were used and are presented in Table 4. The statistical significance under different logistic regression models revealed that there are associations among various AMR genes. For example, a positive association was observed between the ampicillin resistance gene bla_TEM and the streptomycin resistance gene strB.

Table 3.

Statistical association between phenotypic and genotypic AMR in V. parahaemolyticus based on the logistic regression models.

Predictor	Parameter	ORa	S.E.b	95% C.I.c	P-value	AICd
1. ERY	erm(B)	1.65	0.39	1.03–2.63	0.036	818.70
	constant	1.10	0.10	0.92–1.31	0.285
2. SUL	sul2	1.62	0.38	1.02–2.56	0.041	775.45
	constant	0.49	0.05	0.41–0.59	<0.0001
3. TRI	dfrA18	1.78	0.39	1.16–2.74	0.008	701.00
	constant	0.34	0.04	0.28–0.42	<0.0001

ERY, erythromycin; SUL, sulfamethoxazole; TRI, trimethoprim.

^aOR, odds ratio; ^bS.E., standard error; ^cC.I., confidence interval; ^dAIC, Akaike information criterion.

Table 4.

Multivariate logistic regression models for the association among AMR genes in V. parahaemolyticus.

Model	Predictor	Parameters	ORa	S.E.b	95% C.I.c	P-value	AICd
1	bla TEM	strB constant	10.82 0.002	12.14 0.002	1.20–95.52 0.0–0.02	0.034 <0.0001	55.69
2	qnr	strB constant	1.62 3.10	0.39 0.35	1.02–2.59 2.49–3.87	0.043 <0.0001	628.94
3	strB	tet(A) blaTEMqnr constant	1.72 12.75 1.63 0.22	0.37 14.5 0.40 0.05	1.14–2.62 1.37–118.33 1.01–2.62 0.14–0.34	0.010 0.025 0.044 <0.0001	688.74
4	sul2	tet(A) constant	2.65 0.13	0.65 0.02	1.64–4.30 0.10–0.18	<0.0001 <0.0001	487.52
5	tet(A)	strB sul2 constant	1.68 2.57 0.20	0.36 0.64 0.03	1.11–2.56 1.58–4.12 0.16–0.26	0.015 <0.0001 <0.0001	612.18

^aOR, odds ratio; ^bS.E., standard error; ^cC.I., confidence interval; ^dAIC, Akaike information criterion.

Occurrence of virulence genes, ICEs, and class 1, 2 and 3 integrons

None of the samples displayed the Kanagawa phenomenon. Only four isolates carried tdh (0.7%), including two isolates from pooled oyster meat and two isolates from estuarine water. None of the Vibrio isolates harbored the virulence genes, ICEs (trh, int_SXT) and integrons (intI1, intI2 and intI3) tested.

DISCUSSION

This study highlights the wide distribution of resistant V. parahaemolyticus isolates among oyster meat and estuarine water (74.1%). More than half of the isolates (54.2%) were resistant to erythromycin in agreement with previous studies conducted in Spain, Poland and India (Lozano-Leon et al. 2003; Lopatek, Wieczorek and Osek 2015; Silvester, Alexander and Ammanamveetil 2015). Erythromycin has been widely used in human and veterinary medicine for the treatment of diarrheal diseases because of its broad-spectrum activity against Gram-positive and Gram-negative bacteria (Kanfer, Skinner and Walker 1998). To date, erythromycin analogs (e.g. azithromycin and clarithromycin) have replaced erythromycin in clinical practice owing to their longer duration of effect, broader spectrum activity and fewer adverse gastrointestinal effects (Hof 1994). A study of global erythromycin resistance distribution in the environment revealed that erythromycin resistance is common, with a highest prevalence in Asia (Schafhauser et al. 2018), which appeared to mainly derive from anthropogenic activities (Liu and Wong 2013). The rates of resistance to sulfamethoxazole (34.7%) and trimethoprim (27.9%) were high, which were inconsistent to previous studies using isolates from seafood, including marine and freshwater fish (Ottaviani et al. 2001; Lee et al. 2018). However, these results agreed with a previous study showing high sulfamethoxazole and trimethoprim resistance in marine food webs (Liu et al. 2017). Sulfamethoxazole and trimethoprim have been widely used to treat respiratory and urinary tract infections in human. Their frequent use may lead to their high concentrations in the environment, which is supported by a study revealing high contamination rate of sulfamethoxazole and trimethoprim in surface, ground and drinking water, possibly leading to an ecological risk (Liu et al. 2017). Taken together, these results support our analysis that the most common resistance pattern was ERY (21.9%), followed by ERY-SUL (12%), ERY-SUL-TRI (8.4%) and TRI (6.9%).

Currently, knowledge of AMR in V. parahaemolyticus from oysters and estuarine water remains limited. Therefore, some findings were compared to results from different aquatic animals. None of the isolates in this collection were resistant to chloramphenicol and ciprofloxacin in contrast to previous studies where vibrios from environmental isolates were resistant to chloramphenicol (Daramola, Williams and Dixon 2009). Another study using V. parahaemolyticus isolates from shrimp also showed that chloramphenicol resistance was high (35%), while ciprofloxacin resistance was low (6%) (Wong et al. 2012). However, the findings of the present study did agree with previous studies on oyster, mussel and seafood (Ottaviani et al. 2001; Silva et al. 2018). A similar phenomenon was observed for the limited ampicillin resistance results obtained (11.1%), which were inconsistent to a report with V. parahaemolyticus isolates from marine freshwater fish (Lee et al. 2018); as V. parahaemolyticus is intrinsically resistant to ampicillin, the low resistance to this antibiotic remains unclear (Chiou, Li and Chen 2015). In addition, tetracycline resistance rate was low in agreement with a previous study performed with isolates from cultured sea cucumbers in China (Jiang et al. 2014). Previous studies indicated that prevalence of V. parahaemolyticus in seafood from China and shrimps from India was more during summer (Zarie et al. 2012; Yang et al. 2017), which is inconsistent with our study where high distribution was observed during the rainy season and winter. The high resistance rates to tetracycline and trimethoprim were commonly found in the rainy season. This may be because of resistant bacterial contaminants in wastewater from either sewage or anthropogenic activities that are washed out with the rain to the harvest site. These observations indicate that AMR prevalence in V. parahaemolyticus varies with the bacterial originating source and geographical location, and may reflect different antimicrobial agents used in different regions. The samples in this study were originated from oysters raised without antimicrobial agents, while the AMR contamination in either oyster or estuarine water was derived from diverse sources, particularly from agriculture, aquaculture, wastewater, run-off water and the community. Distribution and drivers of AMR in the environment have been previously demonstrated (Singer et al. 2016). In Thailand, the Department of Fisheries (DOF) is authorized to control quality and safety of fish and fisheries products. Good aquaculture practices and Code of Conduct (CoC) standards have been implemented in Thai aquaculture to ensure that farm standards are qualified for exportation (DOF 2010, 2014). According to the Thai Food and Drug Administration, lists of antimicrobials for shrimp and fish aquaculture are currently available. However, none of the antimicrobials are used in oyster cultivation in Thailand, including the harvest area in this study. Therefore, the contamination of AMR bacteria in the cultivation site is mainly derived from the environment. Hence, AMR mitigation strategies should be addressed to reduce AMR contamination in the environment.

The qnr gene was the most prevalent in this study. The sources of qnr in the environment might be associated with selective pressure caused by quinolone contamination in the environment or horizontal gene transfer among bacterial species (Kaplan et al. 2013). It has been observed in other bacteria (e.g. Salmonella spp. and E. coli) that the presence of qnr does not always contribute to resistance to quinolone but promotes co-selection of other resistance genes, potentially resulting in MDR bacteria. Knowledge on the qnr effects on Vibrio isolates is still limited, particularly in V. parahaemolyticus.

The trimethoprim-resistant genes, dfrA18 and dfrA1, and the sulfamethoxazole-resistant gene, sul2, were observed in our results, and these agreed with a previous study analyzing drug-resistant genes in V. cholerae found in Thailand and India (Dalsgaard et al. 2000). The bla_TEM gene was additionally detected at a very low rate (0.8%) in contrast to a previous report on oysters and mussels in Brazil (Rojas et al. 2011). These differences showcase the different antimicrobial contamination results obtained from different geographical regions. Regardless, our findings indicate that oyster and estuarine water can carry AMR genes and possibly disseminate them into the environment.

Effluents and wastewater from various sources play a crucial role in disseminating AMR to the environment, negatively impacting the habitat of aquatic animals and their surrounding water. Several resistant bacteria have been detected in multiple environment of sewage, effluents, surface water and wastewater treatment plants (Lundborg and Tamhankar 2017). Pathogenic V. parahaemolyticus with high AMR rates were previously isolated from fish, coastal water, sediment and estuarine environments (Baker-Austin et al. 2008; Alaboudi et al. 2016; Menezes et al. 2017). Interestingly, the resistance rate and the percentage of resistance genes in oyster and estuarine water were approximately equal. This implies that estuarine water may be used as a proxy to study AMR in oysters because it is cost effective and does not require invasive samples, and is an indicator for AMR monitoring in other aquatic animals. The occurrence of tdh was low (0.7%, n = 4/594), which was lower than a previous study in New Zealand (3.4%) (Kirs et al. 2011). However, the findings were similar to a study conducted in oysters from Brazil (0.04%, n = 1/2243) (Sobrinho Pde et al. 2010). A study on isolates from patients (n = 34) with vibriosis in Maryland showed that isolates harbored tdh (8.0%) and trh (11.0%), and a majority carried both tdh and trh simultaneously (Haendiges et al. 2015). Taken together, these results indicate that V. parahaemolyticus carrying tdh and/or trh is more common among clinical cases than in the environment.

The high prevalence of V. parahaemolyticus in oysters has been previously reported in many countries (89.3–100%) (Chen and Ge 2010; Sobrinho Pde et al. 2010; Jeamsripong, Chuanchuen and Atwill 2018). In general, a Vibrio concentration of ∼10⁵–10⁷ colony forming unit (CFU) g⁻¹ is considered significant and can cause infection in humans. Our previous study showed that the concentrations of all V. parahaemolyticus isolates were ∼10⁵–10⁸ CFU g⁻¹ in oysters (Jeamsripong, Chuanchuen and Atwill 2018). Although a low prevalence of virulence genes was identified among the isolates in the collection, all virulence strains were resistant to at least one antimicrobial agent. Of the four tdh+ isolates, three were resistant to erythromycin, one to tetracycline and one to sulfamethoxazole, and all four harbored qnr, two harbored strB and one harbored dfr18 (n = 1). In this regard, the V. parahaemolyticus isolates carrying both resistance and virulence genes pose a serious threat to humans who consume raw or insufficiently cooked oysters. Therefore, reduction or elimination of V. parahaemolyticus in seafood products could improve safety of seafood and reduce public health risk. The decontamination methods to eliminate this bacterial pathogen include physical methods (e.g. oyster suspension, UV radiation, refrigeration), chemical methods (e.g. antimicrobials, disinfectants) and biological methods (e.g. use of phage, probiotics) (Su, Yang and Häse 2010; Wang et al. 2010; Cole et al. 2015; Zhang et al. 2018).

Significant correlations were observed between phenotypes and genotypes that were resistant to erythromycin [erm(B)], sulfamethoxazole (sul2) and trimethoprim (dfrA18). The presence of AMR genes and the corresponding phenotypes indicated the expression of the identified resistance genes. However, the inconsistency between phenotypic and genotypic resistance was observed for ampicillin, ciprofloxacin and tetracycline, similar to a previous study of pathogenic V. parahaemolyticus isolated from seafood (Lou et al2016). The results indicate the existence of other resistance genes that were not examined in this study. It is also possible that the genes were unexpressed (silent) or confer low resistance level not above the interpretive breakpoint used. Despite the inconsistency, the resistance genes still contribute to resistance phenotype. These genes could be expressed under appropriate conditions and/or horizontally transferred, and therefore, pose the risk of AMR.

Logistic regression models were built in 594 isolates and five different logistic regression models among AMR genes, including bla_TEM, qnr, strB, sul2 and tet(A) presented in Table 4. For example, the Vibrio isolates harboring bla_TEM were more 10.82 times more likely to carry strB than those without bla_TEM in model 1 (P = 0.034). Similarly, the V. parahaemolyticus isolates harboring qnr were 1.62 times more likely to contain strB than the qnr negative isolates. These findings agreed with previous studies that showed that the presence of qnr was associated with other resistance genes, including aminoglycosides and sulfonamides, and might be carried in class 1 integron (Robicsek et al. 2006; Strahilevitz et al. 2009; Kaplan et al. 2013). The qnr gene is responsible for plasmid-mediated quinolone resistance and can be located on transferable plasmid. The plasmid can carry additional genes encoding resistance to other antimicrobial classes (Aldred, Kerns and Osheroff 2014). This could facilitate co-selection of multiple genes, resulting in multidrug resistance bacteria.

In the present study, none of the isolates were positive to intI1, intI2 and intI3, in contrast to previous studies conducted in other seafood types in Thailand and China (Kitiyodom et al. 2010; Jiang et al. 2014). However, our results were similar to a previous study in Chile (Dauros et al. 2010). Class 1 and 2 integrons were predominantly found in freshwater and soil, whereas class 3 integrons were present in the marine environment (Gillings 2014). However, class 2 integrons have never been reported in V. parahaemolyticus. The absence of class 1, 2 and 3 integrons among V. parahaemolyticus strains in the present study highlighted that there was no selective pressure for this type of genetic element. AMR genes in the area of the present study are non-integron-borne and likely located on other mobile genetic elements (e.g. transposons). The existence of other resistance mechanisms, such as active efflux pumps and mutations, should be noted. However, the role of these additional mechanisms was not pursued in the present study.

ICEs carrying AMR genes, known as the SXT element, have been reported in Vibrio, and are associated with the multidrug resistance phenotype, including chloramphenicol (floR), streptomycin (strA and strB), sulfamethoxazole (sul2), trimethoprim (dfrA1 and dfrA18) and tetracycline [tet(A)] (Ceccarelli et al. 2006; Kitiyodom et al. 2010). In the present study, none of the isolates were found to carry the SXT integrase gene. However, a previous study reported that V. parahaemolyticus isolates from wastewater in South Africa carried the SXT element harboring AMR genes (Okoh and Igbinosa 2010). These discrepancies are possibly associated with different antibiotic exposure and spatial distribution. Therefore, further studies of genetic relatedness and genetics underlying AMR in V. parahaemolyticus in the environment based on geographical distribution are required.

CONCLUSIONS

The present study highlighted the role of oysters and estuarine water as potential sources of resistant V. parahaemolyticus that possibly disseminates to humans, seafood and the environment. High rates of AMR were observed among the isolates from oysters and estuarine waters, indicating that resistance strains of Vibrio have been already disseminated in the environment. Further studies of the sources and mechanisms of resistant bacteria contaminating the environment are required to understand the root causes of the distribution of AMR. Rational antimicrobial use may decrease the emergence and spread of AMR bacteria in the environment. Continuous AMR monitoring in aquaculture and the environment is essential for the assessment of risks to human health.

FUNDING

This work was financially supported by the Ratchadaphiseksomphot Endowment Fund (Grant Ref no: RGN_2559_038_02_31), Chulalongkorn University, Bangkok, Thailand. It was partically supported by the Faculty of Veterinary Science Research Fund (Grant Ref no: RG 6/2562), Chulalongkorn University. WK was a recipient of a scholarship program for ASEAN/neighboring countries from Chulalongkorn University.

Conflicts of interest

None declared.