An outbreak caused by GII.17 norovirus with a wide spectrum of HBGA-associated susceptibility

During the past norovirus (NoV) epidemic season, a new GII.17 variant emerged as a predominant NoV strain, surpassed the GII.4 NoVs, causing outbreaks of acute gastroenteritis (AGE) in China. Here we report a study of an AGE outbreak in an elementary school in December 2014 caused by the new GII.17 NoV to explore the potential mechanism behind the sudden epidemics of the GII.17 NoV. A total of 276 individuals were sick with typical NoV infection symptoms of vomiting (93.4%), abdominal pain (90.4%), nausea (60.0%), and diarrhea (10.4%) at an attack rate of 5.7–16.9%. Genotyping of the symptomatic patients showed that individuals with a secretor positive status, including those with A, B, and O secretors and Lewis positive blood types, were sensitive to the virus, while the non-secretors and the Lewis negative individual were not. Accordingly, the recombinant capsid P protein of the GII.17 isolate showed a wide binding spectrum to saliva samples of all A, B, and O secretors. Thus, the broad binding spectrum of the new GII.17 variant could explain its widely spread nature in China and surrounding areas in the past two years.

Our data showed that this new GII.17 variant can infect all A, B and O secretors and accordingly, the recombinant P domain protein of the variant bound saliva of these individuals. Thus our data established the association between the HBGA phenotypes and infection of this new GII.17 variant.

Results
An AGE outbreak caused by a single GII.17 NoV. The outbreak occurred in a primary school with 2355 students and 80 teachers. One canteen with seven food workers provided breakfast for about 600 students and lunch for about 800 students. A total of 273 students, two food workers and one teacher were sick with AGE that met the NoV outbreak disease criteria. The first symptom onset (the index case) started on December 17 th , while final case occurred on December 24 th of 2014 (Fig. 1). The median age of cases was 7 years (range 6-50 years), the majority of the cases (98.9%) were students. Attack rates among student respondents were 16.9% (first grade), 11.6% (second grade), 12.2% (third grade), 10.2% (fourth grade), 6.5% (fifth grade), and 5.7% (sixth grade) with significant decrease by grade (Χ 2 = 34.8, df = 5, P < 0.001), in consistent with decreasing of the fraction of students per school-year eating lunch and breakfast in the canteen.
Stool swab and vomitus samples were collected from 18 students, two food workers, one teacher with AGE on December 22 nd for laboratory tests. GII NoVs were detected in 14 (66.7%) of the 21 specimens by real-time RT-PCR, among which 13 samples were sequenced and identified as GII.17 NoVs (Table 1). All the specimens were also tested for Staphylococcus aureus and Bacillus cereus but neither bacteria were found. The 13 GII.17 NoV sequences showed 97-100% nucleotide identity, belonging to the same genetic cluster of the recently emerged GII.17 variant found in the Guangdong province 7 (data not shown). Thus the studied AGE outbreak was caused by the single GII.17 variant.
HBGA phenotypes and host susceptibility to symptomatic infection. Saliva samples from 69 symptomatic individuals and 123 asymptomatic controls were tested for the ABO, Lewis, and secretor phenotypes. Typical distribution of ABO bloodtypes was seen among the asymptomatic controls and comparable with the general Chinese population (http://www.bloodbook.com/world-abo.html) ( Table 2). In the present study, individuals with Le a−b+ and/or Le x−y+ and those with Le a+b+ or Le x+y+ phenotypes were grouped into secretors. In contrast, individuals with Le a+b− and/or Le x+y− , Le a−b−x−y− but lacking A, B, H phenotypes were sorted into non-secretors 14 . As a result, 67 secretors and two nonsecretors were found among symptomatic individuals; while 102 secretors and 21 nonsecretors were found among asymptomatic controls. Except the Lewis negative non-secretor group (n = 5) that were found only in asymptomatic controls, all other investigated HBGA phenotypes were found in the symptomatic individuals ( Fig. 2 and Table 2).
Compared with controls, a significantly higher proportion of secretors and lower proportion of nonsecretors were found in the symptomatic group than the asymptomatic group (P = 0.003). No significant differences were seen in the proportions of blood types between symptomatic vs. asymptomatic individuals (P = 0.340). These data indicated that the secretor status correlated with infection of the GII.17 variant, the cause of this outbreak.
Similar scenario were seen, when the odds ratios were analyzed. Secretor individuals were at significantly higher infection risk than nonsecretors (OR = 6.90, 95% CI 1.57-30.38 vs. OR = 0.15, 95% CI 0.03-0.64, P = 0.003), including those with Le a+b+ or Le x+y+ (OR = 3.60, 95% CI 1.32-9.84) and without Le a+b+ or Le x+y+ (OR = 0.20, 95% CI 0.04-0.90) ( Table 3). For symptomatic infection risk, no significant difference was found among individuals with different secretor bloodtypes (Table 3). In addition, no correlations were found between blood types and secretor status with clinical symptoms (     (Fig. 4). Two major mutations occurred in the GII.17 HBGA binding site 17 of the newly emerged variant compared with the earlier strains. In addition, four residue mutations in the surrounding regions of the binding site were also seen (Figs 4 and 5).

Discussion
As a rarely reported genotype, GII.17 remained silent for the past decades 18,19 . However, in the past NoV epidemic season, several most recent report showed that a newly emerged GII.17 variant caused major outbreaks in China and Japan, making it a dominant strain surpassing GII.4 in these regions [5][6][7][8] . This sudden increased prevalence of GII.17 may be caused by changes of certain virus-related factor(s) that remain elusive. For example, changes at the antigenic epitopes of the viral capsid that may lead to new adaptive advantage for rapid spread 5,6 . In addition, changes at the HBGA binding site and/or its surrounding region may result in host range expansion. To explore such change as possible mechanism to explain the increased prevalence of the new GII.17 variant, we investigated an AGE outbreak caused by the same GII.17 variant, focusing on the host susceptibility in association with the HBGA phenotypes. Our data showed that the infection source of the outbreak could be the chief cook (food worker 1) who was infected and shedded NoVs at 3 days before the outbreak. Another food worker might be infected from him. Students were the major implicated population. Previous studies have shown a strong association between HBGA phenotypes and NoV infection, which is believed to be an important factor affecting the prevalence of GII NoVs 20,21 . For example, the GII.4 NoVs recognize HBGAs of all A, B, O secretors 22 that represent over 80% of the general population, and some strains can even recognize nonsecretor 13,14 , which explains their predominance in causing NoV epidemics worldwide. Further studies showed that GII.4 NoV P domains can accommodate numerous HBGA types 23,24 . In the present study, we found that the newly emerged GII.17 variant can also infect individuals with all A, B and O secretors, including a low rate of non-secretors, suggesting that this GII.17 variant also has a wide spectrum of target populations. This finding not only explains the emerging epidemic caused by this new GII.17 variant in China, but also alert the global surveillance system for a potential pandemic of this newly emerged GII.17 variant in the near future 16 .
This new GII.17 variant was also detected in neighboring regions and countries of Asia, including Japan 7 , Hong Kong, Taiwan 25 , and United States 26 . GII.17 NoV was also found in groundwater in Kenya 27 , suggesting that this variant of GII.17 has been active in these areas in recent years. Compared with other GII.17 variants that circulated previously, the new variant found in this study had significant sequence variations and was classified into cluster C 5,7,26 . Two amino acid mutations were noted in the HBGA binding site, while four other mutations were seen in the surrounding regions of the binding site (Figs 4 and 5). As indicated by previous studies 17, [28][29][30] , these mutations may result most likely in changes of HBGA binding property of the new GII.17 variant, which in turn, may expand its host range and become more prevalent 31 . One limitation of our study is that some controls may be asymptomatically infected or have not been exposed to the virus in this case-control study. However, it is unlikely to influence the main finding that GII.17 can infect individuals with a wide spectrum of different HBGA phenotypes. Continued monitoring of emerging GII.17 strains is highly demanded for a better understanding of their evolution and epidemiology.

Methods
Outbreak investigation and sample collection. In December 2014, a NoV gastroenteritis outbreak occurred at a primary school (December 17-24) in DongGuan city, GuangDong Province, China. Illnesses were reported among students, food workers, and teacher. Epidemiologic investigations indicated that one chief cook (food worker 1, Table 1) was ill with 8-10 episodes diarrhea per day since December 14 th 10 pm, suggesting the cook as the probable source of NoV contamination in the food.
Description of symptoms and exposure history were obtained through a questionnaire sent to all students with symptoms in the study. Cases with NoV infection and disease were defined by at least one of the following signs or symptoms 11 : vomiting, diarrhea, or nausea combined with stomach cramp after December 17 th . 21 samples from individuals with NoV diseases were collected, including 14 stool swab and 4 vomitus samples from students, 1 stool swab sample from teacher and 2 stool swab samples from food workers.
To test the role of HBGAs in the host susceptibility to the outbreak strain, saliva samples were collected using multiple-stage sampling, as follow: 1) symptomatic students distributed in 28 classes out of totally 42 classes, among which 12 classes with more than 10 cases per class. To improve the efficiency Figure 3. Saliva-based HBGA binding assay of P particle of GII.17 strain DG42. (A) 8 boiled wellcharacterized saliva samples were coated onto 96-well plates prior to the addition P particle. The P particle was tested in a series of 3-fold dilutions (15 μ g/ml, 5 μ g/ml and 1.7 μ g/ml) by enzyme-linked immunosorbent assay (ELISA). (B) The binding activities of 90 saliva samples (all 69 symptomatic and 21 asymptomatic nonsecretor individuals) from the outbreak were sorted by blood types. The P particle was diluted at 5 μ g/ ml. "A, " "B, " "O, " and "N" represent the type A, B, O and nonsecretor saliva, respectively.
Scientific RepoRts | 5:17687 | DOI: 10.1038/srep17687 of this study, these 12 classes were selected in the first stage. 2) 6 out of 12 classes with more than 10 cases per class were randomly selected and 339 saliva samples were collected from all students in these 6 classes at second stage. 3) All 69 saliva samples from symptomatic students and almost half systematic sampling saliva samples (123 out of 270) from asymptomatic students were used to test for HBGA phenotyping at third stage. Together, 192 students in the 6 classes were included in this study. Stool  NoV detection, cloning and sequencing of the P domain. Stool specimens were tested for Staphylococcus aureus and Bacillus cereus by culture. NoV was firstly tested by one-step real-time reverse transcription-polymerase chain reaction assay (RT-PCR, using GI and GII primers) in DongGuan Center for Disease Control and Prevention 32 . To further genotyping, One-Step RT-PCR (QIAGEN, CA, USA) was performed with region C-specific primers 33 . The positive PCR products were sequenced and genetic identity of the viruses was determined using the NoV Automated Genotyping Tool 34 .
The P domain sequences of positive samples were amplified using primers based on sequences of a previously reported GII.17 strain Kawasaki 323 (AB983218), and then cloned into T-vector (Thermo Fisher, US) before being sequenced (Invitrogen, CN). Alignment of P domains was performed using MegAlign of DNAstar 7.0. HBGAs binding residues of GII.17 viruses were determined by the high conservation of the residues constituting the CBPs (central binding pockets) in sequence alignments of NoV P domains 12,35,36 . Preparation of P particles. The P proteins of GII.17 DG42 strains were made as described previously 37,38 . A cysteine-containing peptide was linked to the C (P-CDCRGDCFC) terminus of the P domains to enhance P-particle formation. The cDNAs encoding the capsid P domain were cloned into the expression vector pGEX-4T-1 (Amersham Biosciences, Piscataway, NJ) between Sal I and Not I sites. After sequence confirmation, the P proteins were expressed in E. coli. Briefly, the BL21 cultures were induced by IPTG (isopropyl-β -D-thiogalactopyranoside) (0.4 mM) at room temperature (22 °C) overnight. The recombinant P domain-GST fusion proteins were purified using Glutathione Sepharose 4 Fast Flow resin (7 Sea Pharmatech Co., Ltd, CN) according to the manufacturer's instructions. GST was removed from the P proteins by thrombin (GE Healthcare life Sciences, NJ, USA) cleavage on beads at room temperature overnight. The P-particle formation was confirmed by gel filtration, using a Superdex 200 (GE Healthcare Life-Sciences, Piscataway, NJ) size exclusion column, during which the P particles formed a peak at ~830 kDa. Detection of Histo-Blood Group Antigens in Saliva. The HBGA phenotypes of A, B, H, Le a , Le b , Le x , and Le y antigens of the saliva samples were determined by EIA assays using the corresponding monoclonal antibodies against individual HBGAs, as described previously 39 . Briefly, boiled saliva (1:1000) was coated on high binding ELISA plates (Costar, Corning, NY, US). After blocking with 5% nonfat milk-PBS, 100μ l diluted (1:300) monoclonal antibodies specific for A (Z2A), B (Z5H-2), H (87-N) (Santa Cruz, CA), Le a (BG-5), Le b (BG-6), Le x (BG-7) and Le y (BG-8) antigens (Signet Laboratories Inc., Dedham, MA) were added. Then HRP conjugated goat anti-mice IgG or IgM (1:3000) (Boster biological Techology, Pleasanton, CA) were added. The signal intensities were displayed by adding HRP substrate reagents for 10 min (Tiangen biotech co. ltd, Beijing, CN), and stopped by addition of 2M H 3 PO 4 . The cut-off of a positive signal was OD 450 = 0.1. Well-characterized positive and negative samples were added in each plate as quality control.
HBGA binding assay. Saliva-based P particle-HBGA binding assay was performed. Boiled saliva samples were diluted 1:1000 with 1XPBS and coated onto 96-well microtiter plates at 4 °C overnight. After blocking with 5% nonfat dry milk, P particles of GII.17 variant DG42 were added and incubated at 37 °C for 1 hr. The bound P particles were detected using guinea pig anti-NoV (1:2000) sera 39 , followed by HRP-conjugated goat anti-guinea pig IgG (AB clonal Biotechnology Co., Ltd, CN). The signals were developed using a TMB substrate kit (Beyotime Biotechnology Co., Ltd, Shanghai, CN).