Vibrios from the Norwegian marine environment: Characterization of associated antibiotic resistance and virulence genes

Abstract A total of 116 Vibrio isolates comprising V. alginolyticus (n = 53), V. metschnikovii (n = 38), V. anguillarum (n = 21), V. antiquarius (n = 2), and V. fujianensis (n = 2) were obtained from seawater, fish, or bivalve molluscs from temperate Oceanic and Polar Oceanic area around Norway. Antibiotic sensitivity testing revealed resistance or reduced susceptibility to ampicillin (74%), oxolinic acid (33%), imipenem (21%), aztreonam (19%), and tobramycin (17%). Whole‐genome sequence analysis of eighteen drug‐resistant isolates revealed the presence of genes like β‐lactamases, chloramphenicol‐acetyltransferases, and genes conferring tetracycline and quinolone resistance. The strains also carried virulence genes like hlyA, tlh, rtxA to D and aceA, E and F. The genes for cholerae toxin (ctx), thermostable direct hemolysin (tdh), or zonula occludens toxin (zot) were not detected in any of the isolates. The present study shows low prevalence of multidrug resistance and absence of virulence genes of high global concern among environmental vibrios in Norway. However, in the light of climate change, and projected rising sea surface temperatures, even in the cold temperate areas, there is a need for frequent monitoring of resistance and virulence in vibrios to be prepared for future public health challenges.

non-O139 V. cholerae can also cause infections. The virulence factors of non-O1 and non-O139 include a heat-stable enterotoxin, repeat in toxin (rtx) and El Tor hemolysin (hlyA) (Kumar, Peter, & Thomas, 2010). In contrast, the pathogenicity of V. parahaemolyticus is linked to their ability to produce a thermostable direct hemolysin (TDH), or a TDH-related hemolysin (TRH), encoded by tdh and trh genes (Raghunath, 2015). For V. vulnificus, virulence is related to the production of a polysaccharide capsule and lipopolysaccharide (LPS), flagellum, hemolysin, and proteases (Roig et al., 2018). The genetic basis for human virulence is only partially known, although several studies suggest that all strains of V. vulnificus, regardless of their origin, may be able to cause infections in humans (Roig et al., 2018). Several other Vibrio spp., such as V. alginolyticus, V. fluvialis, V. mimicus, V. metschnikovii, V. furnissii, V. hollisae, and V. damsela, can occasionally cause infections in humans (Austin, 2010;Baker-Austin et al., 2018).
Vibrio infections in humans typically occur as a result of ingestion of contaminated seafood, through the handling of raw seafood or by exposure of wounds to seawater during recreation (Iwamoto, Ayers, Mahon, & Swerdlow, 2010). The human pathogenic vibrios show strong seasonality and are more abundant when the water temperature exceeds 18°C and the salinity drops below 25 ‰ (Vezzulli et al., 2013). In the last decades, an increase in infections caused by Vibrio spp. has been reported, also in colder regions of South America and Northern Europe, including Norway, where this was previously rare (Baker- Austin et al., 2016). One of the primary effects of climate change is increased sea surface temperatures (SSTs), and this may facilitate the spread of seawater associated diseases (EEA, 2017). The temperature is predicted to increase further in northern temperate waters (EEA, 2017), and new areas may become more favorable for the pathogenic vibrios.
The role of the marine environment in the development and dissemination of antimicrobial resistance is largely unknown. Vibrios are indigenous to the sea (Banerjee & Farber, 2018), and in recent years, the occurrence of resistance genes in Vibrio spp. has been examined. Genes encoding resistance to β-lactams like penA, bla   (Letchumanan, Chan, & Lee, 2015), and bla VCC-1 (Hammerl et al., 2017;Mangat et al., 2016), chloramphenicol resistance genes, such as floR, catI, and catII, and several tet genes encoding resistance to tetracycline (Letchumanan et al., 2015), have been detected in Vibrio spp. Clinically important mobile resistance genes like qnrVC and qnrS have originated in Vibrio spp. (Fonseca, Dos Santos Freitas, Vieira, & Vicente, 2008). This makes Vibrio spp. a good model organism for the studying antibiotic resistance in the marine environment.
Although V. parahaemolyticus, V. cholerae, and V. vulnificus have previously been isolated from Norway (Bauer, Ostensvik, Florvag, Ormen, & Rorvik, 2006), there is limited knowledge on the prevalence of different Vibrio spp. and associated resistance and virulence markers in the Norwegian marine environment. This study aimed to examine the prevalence of different Vibrio spp. in the Norwegian marine environment and to characterize associated virulence and antibiotic resistance genes among these. We here present a detailed account of taxonomy, resistance, and virulence genes detected based on phenotypic culture-based methods and whole-genome sequence (WGS) analysis.

| Sampling
Water samples were collected from four different locations (A-D) at the West coast of Norway (Oceanic temperate zones) at five different depths (0, 2, 5, 7, and

| Isolation of Vibrio spp.
From each water sample, three aliquots of 100-250 ml were filtered through 0.45 µm filters (Merck Millipore, Germany) using the EZ-fit Manifold 3-place system (Merck Millipore, Germany) connected to a vacuum pump. Each filter was transferred to thiosulfate-citratebile-sucrose (TCBS) agar (Oxoid, UK) plates and incubated at 37°C for 24-48 hrs. Also, an enrichment step was performed in duplicates on 500 ml water adding 50 ml concentrated (360 mg/ml) alkaline peptone water (APW) with 2% sodium chloride (NaCl). The enrichment cultures were incubated at 42 ˚C for 18 hr. After incubation, 100 µl of the enrichment cultures was streaked on TCBS agar and incubated at 37 ˚C for 24-48 hr. Typical colonies were picked from the plates and restreaked for obtaining pure cultures.
Isolation of Vibrio spp. from fish and bivalve molluscs followed a method based on NMKL method no. 156 (NMKL, 1997). The method takes advantage of the vibrios alkaline and halophilic properties (Vezzulli et al., 2013) and applies APW supplemented with 2% NaCl and 42°C as incubation temperature for selective enrichment of human pathogenic species (NMKL, 1997). For isolation of Vibrio spp., TCBS is a widely used medium. The alkaline pH (8.6), bile salts, and NaCl concentration in the agar inhibit the growth of Enterobacteriaceae and Gram-positive organisms (Donovan & van Netten, 1995). From herring collected in June 2018, samples were taken from the skin with muscle, gills, and intestine. From each tissue type, 20 g was homogenized in 180 ml APW with 2% NaCl and APW with 2% NaCl supplemented with polymyxin B (250 IU/ml) for 30 s. using a stomacher. The homogenate was incubated at 42 ± 1°C for 18 ± 2 hrs. After incubation, 10 µl of the enrichment cultures was streaked on TCBS agar and incubated at 37 ± 1°C for 24 ± 3 hrs.
From mackerel collected in September, samples were taken from the skin with mussel following the same protocol as described previously. Samples were also collected from gut content and homogenized in phosphate-buffered saline (PBS) (Sigma-Aldrich), and tenfold dilution series were made. From each sample, 100 µl was spread on TCBS and incubated at 37 ± 1°C for 24 ± 3 hrs. From herring collected in November, samples were collected from the skin with muscle and prepared following the same method as described previously.
From bivalve molluscs, 100 g soft tissue and intravalvular fluid from at least 10 individual bivalves were homogenized in sterile plastic bags and 20 g was transferred to new sterile bags. Enrichment followed the same protocol as for fish samples. Additionally, from the homogenate tenfold dilution series were made using peptone water (bioMerièux, France). From dilutions and undiluted samples, 100 µl was spread on TCBS and Vibrio ChromoSelect agar (VCS; Sigma-Aldrich) and incubated at 37°C for 24-48 hrs followed by a selection of typical colonies.

| Biochemical identification
Isolates were grown overnight on plate count agar (PCA) (Oxoid, UK) supplemented with 2% NaCl and characterized biochemical using the

| Identification by MALDI-TOF-MS
All isolates were grown overnight on PCA supplemented with 2% NaCl and sent to the Norwegian Veterinary Institute (NVI) in Bergen for identification by matrix-assisted laser desorption ionization timeof-flight mass spectrometry (MALDI-TOF-MS) (Bruker, Germany). The obtained peptide mass fingerprints (PMFs) were compared to spectra in the commercial MALDI-TOF-MS database (MALDI Biotyper, Bruker, Germany) and to spectra in an in-house generated database containing spectra from Vibrio spp. known to be associated with marine fish.

| Whole-genome sequencing and sequence analysis
Eighteen isolates were subjected to whole-genome sequencing (WGS). DNA was extracted from isolates using the DNeasy Blood & Tissue kit (Qiagen, Germany). An additional lysis step was performed by resuspending the samples in 180 µl lysis buffer and incubating them at 37°C overnight. After incubation, DNA extraction was done as described by the manufacturer (Quiagen, 2006). The purity (260/280 and 260/230 ratios) and concentration in the DNA was measured using Nanodrop ND-1000 (NanoDrop Technologies, USA) and Qubit 2.0 broad range dsDNA kit (Invitrogen, USA).

| Species identification of WGS
Raw forward and reverse reads in the FastQ format were uploaded to The Microbial Genomes Atlas (MiGA) (Rodriguez et al., 2018) web server in the TypeMat mode. In this mode, the sequences are trimmed, assembled, and aligned to give the closest relatives found in the MiGA Reference database.

| Antimicrobial susceptibility testing
Antimicrobial susceptibility testing of isolated Vibrio spp. was con-

| CarbaNP test
Isolates showing reduced susceptibility to imipenem by the disk diffusion method were grown overnight on tryptic soy agar (TSA; Merck, Germany) at 37°C and examined for carbapenemase production by the CarbaNP test as described by Dortet, Poirel, Errera, and Nordmann (2014).

| Prevalence and identification of Vibrio spp
Among the species considered to be opportunistic human pathogens (Austin, 2010;Baker-Austin et al. 2018), V. alginolyticus was isolated from water, herring, and bivalves, while V. metschnikovii was isolated from herring and water samples. On the other hand, species harboring virulence genes but not known to cause human disease, like V. antiquarius (Dahanayake, De Silva, Hossain, Shin, & Heo, 2018;Nur et al., 2015) and V. fujianensis (Fang et al., 2018), were isolated from water only. V. anguillarum, a well-known fish pathogen F I G U R E 2 Physical parameters in seawater samples collected during herring fisheries, locations A, B, C, and D ( Figure 1). (a) Measured temperature (°C). (b) Measured salinity (‰), note: missing measurement at 10 m from location B. (c) Number of colony-forming units (cfu)/100 ml water on TCBS plates incubated at 37°C for 24-48 hrs.
Global mapping of the sequenced isolates of V. alginolyticus and

V. anguillarum (Figures A1 and A2) showed that Vibrio isolates from
Norway had high similarity to strains from other countries and continents, including the United States and China, indicating a global presence of these strains.

| Hemolytic activity on blood agar
None of the 53 V. alginolyticus isolates displayed hemolysis on blood agar. All 38 V. metschnikovii isolates were hemolytic on both sheep and human blood. On sheep blood, five V. metschnikovii isolates were β-hemolytic, while the remaining isolates were α-hemolytic on both media.
The most prominent virulence genes detected in this study were related to hemolysins. All Vibrio species examined had genes coding for the Aeromonas-related hemolysin type III (Hemolysin  (Table 2).

| Examination of carbapenemase production
Among the 116 Vibrio isolates examined, resistance to imipenem was observed in all V. anguillarum isolates, while two V. alginolyticus isolates and one V. fujianensis isolate were intermediately susceptible to the agent. These imipenem-resistant isolates were also resistant to ampicillin but susceptible to meropenem. All but one V. anguillarum isolate (B4-12) was susceptible to cefotaxime. CarbaNP test was negative for all isolates, suggesting the absence of carbapenemase with high hydrolytic activity.

| Genetic characterization of resistance determinants
The sequenced genomes revealed the presence of β-lactamases

| D ISCUSS I ON
To the best of our knowledge, this study is the most comprehensive assessment of vibrios from the Norwegian marine environment describing the prevalence of Vibrio spp. in Norwegian pelagic fish, bivalves, and seawater, and their characteristics concerning antimicrobial resistance and virulence.

| Prevalence of Vibrio spp. in the Norwegian marine environment
The highest plate count of aquatic bacteria was observed in the water samples collected closest to the shore, where the measured temperature was highest and the salinity lowest (Location A). A total of 67% of isolated V. alginolyticus were isolated from these samples, where the temperature was measured to above 15°C and the salinity to ≤25 ‰, close to the preferred conditions for vibrios (Vezzulli et al., 2013;Vezzulli, Pezzati, Brettar, Höfle, & Pruzzo, 2015 where the seas are influenced by the North and Atlantic Ocean. As a result, the sea temperature in these areas is normally low and the salinity is high. It is well known that the human pathogenic vibrios are most abundant at elevated sea temperatures, >18°C, and at lower salinity levels, <25‰ (Vezzulli et al., 2013). This may explain the absence of the major human pathogenic Vibrio spp. in this study. The risk of increased numbers of vibrios due to elevated temperatures is greater in the east coast of Norway and closer toward the Baltic sea (Escobar et al., 2015) where the seas are less affected by the open oceans. isolated during this study were phenotypically susceptible to tetracycline, doxycycline, meropenem, sulfamethoxazole/trimethoprim, ciprofloxacin, florfenicol, mecillinam, and azithromycin.

| Antimicrobial susceptibility
Consistent with previous reports, a high prevalence of resistance to ampicillin was observed in all Vibrio spp. isolates in our study (Banerjee & Farber, 2018;Chiou, Li, & Chen, 2015;Hernández-Robles et al., 2016;Li et al., 1999;Pan et al., 2013), and this resistance is usually due to the presence of a bla CARB gene (Chiou et al., 2015;Li et al., 2016). The bla CARB -like genes have been found in V. cholerae predating the introduction of penicillins (Dorman et al., 2019).
In this study, the bla CARB genes were detected in V. alginolyticus, V. metschnikovii, and V. antiquarius. Genes encoding ampC β-lactamase were found in V. alginolyticus, V. anguillarum, and V. fujianensis, which is conflicting to the results from phenotypic susceptibility testing as all these isolates were susceptible to cephalosporins. This may indicate that the breakpoints used in this study are insufficient for detection of these enzymes by a phenotypic method. This also highlights the need for establishing breakpoints for environmental Vibrio species. However, differences between phenotype and genotype may also be caused by a variable expression of genes in tested isolates (Sundsfjord et al., 2004). Although all isolates in our study were susceptible to both tetracycline and doxycycline, the tetracycline enzymatic inactivation gene tet34 (Akinbowale, Peng, & Barton, 2007) and efflux encoding gene tet35 were frequently detected within the examined genomes in the current study.
Resistance to oxolinic acid has been reported in V. alginolyticus (Scarano et al., 2014), and the prevalence of reduced susceptibility was quite high in this study. All examined isolates of V. alginolyticus carried the qnr gene. It has been suggested that the marine bacteria may constitute the origin of plasmid-mediated quinolone resistance (PMQR) genes (Poirel, Cattoir, & Nordmann, 2012) and vibrios might act as a reservoir for these genes (Poirel, Liard, Rodriguez-Martinez, & Nordmann, 2005).
Genes encoding chloramphenicol resistance are frequently found in examined Vibrio spp. (Letchumanan et al., 2015), and in the current study, V. metschnikovii and V. anguillarum harbored the catB-like acetyltransferase able to inactivate chloramphenicol. This gene, however, does not give resistance to florfenicol (Schwarz, Kehrenberg, Doublet, & Cloeckaert, 2004), which was the only amphenicol tested in our study. However, none of these isolates produced positive results in the car-baNP test indicating another resistance mechanism than the production of a carbapenemase, or an imipenem hydrolyzing enzyme with a slow turnover rate (Verma et al., 2011). The observed resistance is likely caused by an alteration in porins, the presence of low-affinity penicillin-binding proteins or overexpression of ampC (El Amin et al., 2001;Nordmann, Dortet, & Poirel, 2012;Zapun, Contreras-Martel, & Vernet, 2008). One V. anguillarum isolate carried gene encoding a VarG subclass B1-like lactamase, an enzyme with the ability to hydrolyze most β-lactam antibiotic, including cephalosporins and carbapenems (Lin et al., 2017). This isolate was, however, susceptible to both meropenem and cephalosporins.

| Virulence
Members The hemolysins produced by V. metschnikovii is known to lyse cells from several animals, including humans, sheep, and horse (Miyake, Honda, & Miwatani, 1988). All the V. metschnikovii isolates were α-hemolytic on tryptic soy agar (TSA) with 5% human blood and on TSA with sheep blood, except five isolates that were β-hemolytic on TSA with sheep blood. The results indicate that sheep erythrocytes are more susceptible to these hemolysins, even though a previous study showed the opposite, where human cells were more susceptible to the hemolysins produced by V. metschnikovii (Matté et al., 2007).
RTX is a pore-forming toxin found in several pathogenic Gramnegative bacteria (Lee, Choi, & Kim, 2008), while HlyA, also known as V. cholerae cytolysin (VCC), is a hemolysin and cytolysin with activity against a range of eukaryotic cells (Ruenchit, Reamtong, Siripanichgon, Chaicumpa, & Diraphat, 2017) and is found in both V. cholerae O1 and non-O1/non-O139. The cytotoxic activity has previously been described in V. metschnikovii isolated from a leg wound (Linde et al., 2004). Even though V. metschnikovii have caused infections in humans, it is poorly described with regard to virulence factors, and the presence of these genes may indicate a pathogenic potential.
Horizontal gene transfer can mediate transfer not only antibiotic resistance genes but also virulence factors. V. cholerae virulence encoding genes, for example, zonula occludens toxin (zot), are encoded by prophages, and it has been suggested that the transfer of zot encoding phages occurs frequently in the Vibrio community (Castillo et al., 2018). Similarly, fragments of V. cholerae pathogenicity islands have been detected in V. alginolyticus, V. anguillarum, and V. metschnikovii, indicating that important virulence genes can be present in environmental Vibrio spp. (Gennari, Ghidini, Caburlotto, & Lleo, 2012).
The API20E  between two genomes (Kim, Oh, Park, & Chun, 2014). MiGA can discriminate between closely related species (Rodriguez et al., 2018) and the reference database includes a large number of genomes, including the Vibrio spp. proposed by MALDI-TOF-MS (http://micro bial-genom es.org/proje cts/20). Hence, the results from identification by MiGA should be considered most reliable.

| CON CLUS ION
To the best of our knowledge, this study presents the most comprehensive assessment of vibrios from the Norwegian marine environment, where potentially human pathogenic species like V. alginolyticus and V. metschnikovii were detected. Although the low frequency of multidrug-resistant isolates was observed, several clinically important resistance genes were detected in the Vibrio spp.
isolates. These environmental vibrios could act as a reservoir of resistance genes in the marine environment.

E TH I C S S TATEM ENT
None required.

ACK N OWLED G EM ENTS
We are grateful for samples provided for this study by the Norwegian Food Safety Agency and the research cruises monitoring pelagic fisheries organized by Dr Arne Levsen. We also thank Tone Galluzzi and Hui Shan Tung for help during the processing of samples and analysis. We also want to acknowledge Hanne Nilsen at the Norwegian Veterinary Institute for help with identification of Vibrio spp. by MALDI-TOF-MS.

CO N FLI C T O F I NTE R E S T S
None declared.    F I G U R E A 1 ML phylogenetic inference of Vibrio anguillarum strains included in this study. Genome used as reference is red shaded, while the genomes from this study are in green. Blue dots show nodes with bootstrap values above 85%