Distribution of Human Norovirus in the Coastal Waters of South Korea

The presence of human norovirus in the aquatic environment can cause outbreaks related to recreational activities and the consumption of norovirus-contaminated clams. In this study, we investigated the prevalence of norovirus genogroups I (GI) and II (GII) in the coastal aquatic environment in South Korea (March 2014 to February 2015). A total of 504 water samples were collected periodically from four coastal areas (total sites = 63), of which 44 sites were in estuaries (clam fisheries) and 19 were in inflow streams. RT-PCR analysis targeting ORF2 region C revealed that 20.6% of the water samples were contaminated by GI (13.3%) or GII (16.6%). The prevalence of human norovirus was higher in winter/spring than in summer/fall, and higher in inflow streams (50.0%) than in estuaries (7.9%). A total of 229 human norovirus sequences were identified from the water samples, and phylogenetic analysis showed that the sequences clustered into eight GI genotypes (GI.1, 2, 3, 4, 5, 6, 7, and 9) and nine GII genotypes (GII.2, 3, 4, 5, 6, 11, 13, 17, and 21). This study highlighted three issues: 1) a strong correlation between norovirus contamination via inflow streams and coastal areas used in clam fisheries; 2) increased prevalence of certain non-GII.4 genotypes, exceeding that of the GII.4 pandemic variants; 3) seasonal shifts in the dominant genotypes of both GI and GII.


Introduction
Acute gastroenteritis causes the second greatest burden of all infectious diseases, estimated at 89.5 million disability-adjusted life-years and 1.45 million deaths worldwide every year [1]. In particular, human norovirus (HNoV) has been reported as the major cause of non-bacterial acute gastroenteritis in patients of all ages, responsible for approximately 90% of all outbreaks of viral gastroenteritis in the world [2][3][4][5][6]. HNoV can infect via multiple routes, and is transmitted through contact with gastroenteric effluents originating from infected individuals [7]. At least 70% of outbreaks have occurred in semi-closed communities [8][9][10].
Noroviruses (NoVs) are small non-enveloped viruses in the Caliciviridae family with a positive single-stranded RNA genome of 7.5-7.7 kb in length, which is organized into three or four open reading frames (ORFs) [11][12][13][14]. ORF1 encodes six non-structural proteins, including the viral RNA-dependent RNA polymerase, while ORF2 and ORF3 encode the major (VP1) and minor (VP2) capsid proteins, respectively [15]. After a viral incubation period of 12 hours to 2 days, a patient generally experiences acute symptoms, such as vomiting, diarrhea, nausea, abdominal cramps, and low-grade fever [10]; immunocompromised patients are susceptible to chronic gastroenteritis [16]. NoVs are genetically diverse and are classified into six established genogroups (GI-GVI) based on VP1 sequences [10,14]. Of the six genogroups, GI, GII, and GIV infect humans, and GII is the most common threat, causing 75-90% of all HNoV-related outbreaks [11,12,17,18]. To date, nine capsid genotypes have been identified in GI, 22 in GII, and three genotypes of GII (GII.11, GII.18, and GII. 19) have been uniquely detected in swine. Of the two genotypes of GIV identified to date, GIV.1 can infect humans [10].
The objective of this study was to investigate the distribution over time of the GI and GII genotypes of NoV in estuaries and inflow streams in four different geographical areas in South Korea. This study is the first nationwide study conducted in South Korea focusing on detection of HNoV contamination in coastal environments that are utilized as clam fisheries.

Ethics statement
Sample collection was approved by the Korean Food and Drug Administration (KFDA, Project No. 14162-973). This study did not require additional permissions because samples were not collected on private land or in protected areas. We confirm that this study did not involve endangered or protected species.

Water sample collection and processing
From March 2014 to February 2015, water samples were collected from four peri-urban coastal regions located in the eastern (area A), southern (area B), southwestern (area C), and western (area D) areas of South Korea (Fig 1). Each study area included one estuary and between one and six inflow streams (one stream in area A; six streams in each of areas B, C, and D). Domestic sewage from dwellings flows into each of the neighboring streams. The estuary in area A was sampled at two different sites, and each estuary in areas B, C, and D was sampled at 14 different sites. The estuary in area A includes a limited number of inflow streams and sampling sites owing to its geographical features. Each inflow stream was sampled at one site. Thus, samples were collected from a total of 63 sites in the four areas: 44 sites in four estuaries which have been used as clam fisheries and 19 sites in 19 inflow streams. The 63 sites were sampled eight times during the study period (Mar-Apr, May, Jul, Aug, Sep, Oct, Jan, and Feb). Thus, a total of 504 samples were collected.
According to the standard procedure [33,34], less than 100 L of water was filtrated through NanoCeram cartridge filters (Argonide, Sanford, FL, USA). The filters were stored at 4°C and concentrated according to method 1615 within 72 h of sample collection [35]. Final concentrates were stored at -70°C until analysis. Additional water samples were collected using sterile polypropylene bottles for analysis of physicochemical characteristics and fecal indicator bacteria.

RNA extraction and nucleic acid amplification
Viral RNA was extracted from the water concentrate using a QIAamp viral RNA mini kit (Qiagen, Hilden, Germany) according to the manufacturer's protocol, in a final volume of 60 μL. The RNA was stored at -70°C until RT-PCR analysis.
The primer sets used to amplify the NoV GI and GII target regions are listed in Table 1 [36]. Region C [37] of NoV GI and GII was first amplified using a Verso 1-step RT-PCR ReddyMix kit (Thermo Fisher Scientific, Waltham, MA, USA) according to the manufacturer's protocol, with minor modifications. Briefly, the RNA (5 μL) extracted from each water sample was mixed with 1-Step PCR ReddyMix (9.5 μL), the forward and reverse primers (40 μM each), and Verso Enzyme Mix (0.5 μL). Deionized sterile water was added up to a final reaction volume of 25 μL. The one-step RT-PCR conditions were as follows: reverse transcription at 45°C for 30 min, denaturation at 95°C for 5 min, and 35 cycles of amplification (94°C for 30 s, 55°C for 30 s, and 72°C for 90 s), plus a final extension at 72°C for 7 min. For the second semi-nested PCR amplification, Top DNA polymerase (Bioneer, Daejeon, South Korea) was used according to the manufacturer's protocol, with minor modifications. Briefly, each RT-PCR product (2 μL) was mixed with 10× buffer (5 μL), 10 mM dNTP mix (4 μL), forward and reverse primers (50 μM each), and DNA polymerase (5 U). Deionized sterile water was added up to a final reaction volume of 50 μL. The semi-nested PCR conditions were as follows: initial denaturation at 94°C for 5 min and 25 cycles of amplification (94°C for 30 s, 55°C for 30 s, and 72°C for 90 s), plus a final extension at 72°C for 7 min.

Cloning and sequence analysis
Target amplicons were separated by electrophoresis and purified using a MiniBEST Agarose Gel DNA Extraction Kit Ver. 4.0 (Takara, Kusatsu, Japan) following the manufacturer's protocol. The purified amplicons were cloned using both a Mighty TA-cloning Kit (Takara) and chemically competent DH5 α (Enzynomics, Daejeon, South Korea). Six colonies for each amplicon were chosen for seeding in liquid medium for culture. Cloned genes were purified using a HiYield™ Plasmid Mini Kit (RBC, Banqiao, Taiwan) and sequenced (Macrogen Seoul, South Korea) using a 3730xl DNA analyzer (Thermo Fisher Scientific).

Measurement of physicochemical parameters and fecal indicator bacteria
The physicochemical parameters of temperature (°C), conductivity (uS/cm), pH, and turbidity (NTU) were measured for each sample using an Orion 4-star pH/conductivity meter (Thermo Fisher Scientific). Fecal coliform and E. coli were measured using the 5-tube most probable number (MPN) method [38]. Briefly, each diluted water sample was first inoculated into lauryl tryptose broth and incubated at 35°C for 48 h. Cultures that generated gas were transferred into both brilliant green bile broth (Oxoid, Hampshire, UK) and E. coli broth (BD, Franklin Lakes, NJ, USA) to test for fecal coliform and E. coli. After incubation, positive tubes were counted and assessed using MPN tables.

Phylogenetic and statistical analysis
HNoV sequences from the water samples were aligned using Clustal W, version 1.81. Phylogenetic relationships among the sequences were determined using Molecular Evolutionary Genetics Analysis software (version 6.0). The Kimura 2-parameter model was used as the substitution method, and neighbor-joining was used to construct phylogenetic trees with 1000 bootstrap replicates. Statistical analysis was performed using the Statistical Package for the Social Sciences software (version 22.0.0). Independent samples t-tests were used to compare Table 1. Primers used in this study.

Genogroup
Primer ID Sequences (5 0 !3 0 ) c Position d Polarity the mean value of each environmental factor in NoV-positive samples and the mean value of the same environmental factor in NoV-negative samples [39]. P-values < 0.05 were considered to indicate significance.

Accession numbers for isolated sequences
The nucleotide sequences were deposited in the GenBank database with the following accession numbers: KT383850-KT384078.

HNoV detection in estuaries and inflow streams
From March 2014 to February 2015, we collected a total of 504 water samples (352 estuary samples and 152 inflow stream samples) from a total of 63 sites in four coastal areas in South Korea (Fig 1). Each area for water sampling contained one estuary and its inflow streams. From estuaries, we collected 16 samples from area A and 112 samples from each of areas B-D.
From inflow streams, we collected eight samples from area A and 48 samples from each of areas B-D. Using RT-PCR targeting NoV GI and GII, HNoVs were detected in 104 of the total 504 samples (7.9% of the estuarine water samples and 50.0% of the inflow stream water samples; Table 2). The rates of HNoV detection revealed regional differences within each of the two water types. Among the estuary samples, those from area B showed the highest rate of HNoV detection (11.6%), followed by area D (8.0%), area A (6.2%), and area C (4.5%). Among the inflow stream samples, those from area A showed the highest rate of HNoV detection (87.5%), followed by area B (64.6%), area D (47.9%), and area C (31.2%). Both estuary and stream water samples collected in spring (Mar to May) and winter (Jan to Feb) showed higher rates of HNoV detection (Fig 2a and 2b). Synthetically, detection rates of GII (28.2%) and the combination of two genogroups (GI and GII, 32.1%) in winter/spring (Jan to Feb and Mar to May) samples were significantly (P < 0.05) higher than in summer/fall (July to Aug and Sep to Oct) samples (5.2% for GII, and 9.1% for the combination of the two genogroups) (Fig 2c). Although it was not significant (P > 0.05), the detection rate of GI (19.0%) in winter/spring samples was higher than in summer/fall (7.5%).

Correlation between environmental factors and HNoV occurrence in water samples
The measured values of both physicochemical factors and fecal indicator bacteria are listed according to season and water type (Table 3). Water temperature was lowest in winter, followed by spring, fall, and summer in both water types. The MPN values of total coliform and E. coli were highest in summer and lowest in winter in both water types. A t-test to detect associations between environmental factors and HNoV occurrences revealed that water temperature was significantly correlated with HNoV occurrence in three of the four areas, as well as overall (P < 0.05; Table 4). In estuaries, conductivity was correlated with HNoV occurrence overall; but significant correlations (P < 0.05) were observed in only two of the four individual areas.
Genotypic diversity and prevalence of NoV GI and GII in environmental water samples We conducted a phylogenetic analysis of region C (5 0 -end of ORF2, 0.3 kb) to analyze the relationships between the 229 HNoV sequences (107 GI sequences and 122 GII sequences) and genotypic reference sequences, using the neighbor-joining method. Except for one unclassified

Discussion
Human activity influences the presence and prevalence of human norovirus in the aquatic environment [40], and outbreaks caused by waterborne viruses have caused serious health and socio-economic impacts [41]. Although the presence of HNoVs in clams has been widely reported [9,[42][43][44][45][46], HNoV monitoring in coastal environments is relatively limited [2,17,24]. In this study, we investigated the prevalence, seasonal dependence, and genotypic diversity of   both NoV GI and GII in four different estuaries (clam fisheries) and their inflow streams, which were in the vicinity of human dwellings in South Korea. In this study, HNoVs were detected in 7.9% of estuary water samples and 50.0% of inflow stream water samples. The difference in detection rates between estuaries and streams might be due to the greater dispersion/dilution effects and lower stability of viruses in the marine environment [24]. The detection rate did not differ significantly between the two genogroups in streams (P > 0.05; Table 2), concordant with the results of previous studies of NoV in rivers [20,25]. GII was slightly more prevalent in estuaries than GI, although the difference was not significant (P > 0.05), suggesting the possible presence of a genotype-specific PCR inhibitor or water type-dependent GI capsid instability. Nevertheless, these environmental observations did not coincide with the overwhelming dominance of GII in hospitalized cases [47][48][49][50], which might suggest that large-scale asymptomatic infection with NoV GI occurs in South Korea.
GII.4 has been the dominant genotype in HNoV-associated outbreaks worldwide [45,50,[52][53][54][55]. Recent reports, however, showed that novel GII.17 NoVs, not GII.4 pandemic variants, have been the major causative agent of recent outbreaks in China and Japan [56][57][58][59]. Interestingly, GII.4 in this study was not the dominant GII genotype; GII.6 was dominant in 2014 and GII.17 was dominant in 2015 (Figs 6 and 7). The GII.17 sequences identified in this study were clustered with the previously reported novel GII.17 variants in the phylogenetic analysis. The novel GII.17 genotype detected in this study was first detected in July 2014 (20140714FL01_1, reference sequence: LC037415 , Fig 4c), and the GII.17 sequences were detected at highest frequency in the following winter. Thus, the GII.17 NoV that is dominant in East Asia could pose an additional major threat on other continents, and we recommend that current studies,   including vaccine development, concentrating on GII.4 pandemic variants should be extended to other genotypes. Because current in vivo/in vitro infection systems for study of HNoV are limited, the underlying cause of the disproportionate detection of one or two genotypes in most clinical samples cannot be fully investigated at present.
It is commonly accepted that HNoV-related diseases tend to occur in winter [24], and seasonal HNoV occurrence was higher in spring/winter than in summer/fall (Fig 2c). A previous study showed that environmental NoVs were more frequently detected in winter/spring [21]. The correlation between water temperature and HNoV occurrence support the seasonality of HNoV (Table 4). Although the correlation between water temperature and HNoV detection in area A was not significant (P > 0.05), this may have been due to the relatively small sample size. Because winter is the main season for harvest of clams in South Korea, the highest viral prevalence observed in winter estuary samples may increase the norovirus load in clams. In fact, the major genotypes detected in inflow streams were also frequently detected in all four estuaries (S1 and S2 Figs), suggesting that streams could affect norovirus inflow into the marine environment; thus, continuous surveillance efforts and construction of sanitation facilities at the lower streams that affect clam fisheries will help to reduce seafood-related HNoV outbreaks.
Thus far, few reports have investigated the prevalence of HNoVs in coastal environments, and to the best of our knowledge, this is the first comprehensive report on HNoV prevalence in estuaries and their inflow streams in South Korea. Environmental monitoring for HNoV will provide data that will be helpful for resolution of seasonal epidemics in future. Although this study preferentially focused on correlations between HNoV contamination of estuaries and their inflow streams, a recent report demonstrated that the novel GII.17 variant was second in predominance among all NoV-positive children diagnosed with acute gastroenteritis from 2013-2015 in South Korea, indicating the importance of monitoring the aquatic environment [60].
This study provides important basic data on 1) the inflow of HNoV into estuaries harboring clam fisheries and 2) the temporal distribution/transition of HNoV in the aquatic environment, which can provide important data for identifying novel genotypes that may cause epidemics in the near future. The numbers of isolated NoV GII sequences are described as horizontal stacked bar charts. Charts are illustrated by combination between study areas (a and b = area A; c and d = area B; e and f = area C; g and h = area D) and water types (a, c, e, and g = stream; b, d, f, and h = estuary). In each panel, NoV GII sequences are sorted by sampling months (vertical axis), and each horizontal bar (total NoV GII sequences in a sampling month) is consist of internal patterns (the number of sequences for identified genotypes). (TIF)