Large-scale gastroenteritis outbreaks occurring in a university town of Guangzhou city during the global wave of the norovirus in China Authors: Yi Tian, Jun Yuan, Wenfeng Cai, Huaping Xie, Yanhui Yang, Lijiao Cao, Shi Cong, Xiangyu Kong, Qing Zhang, Miao Jin Published: 2026 ABSTRACT Noroviruses are the primary viral pathogens responsible for epidemic acute gastroenteritis. The novel GII.4 Sydney 2012 variant, which emerged during the 2012/13 winter season, led to significant global outbreaks. This study reports large-scale outbreaks occurring in the university town of Guangzhou, China, from January 8 to April 3, 2013. These outbreaks affected 652 individuals across five universities and one middle school. Of the 386 samples tested, 114 were norovirus-positive, and 19 stool or swab samples were genotyped as the GII.4 Sydney 2012 strain. The genome sequence of the GII.4 Sydney 2012 variant, GZ3135, was determined and compared with those of earlier GII.4 variants. This comparison revealed residue substitutions at specific sites, including the HBGA binding pocket and putative antigenic epitopes. The investigation outcomes showed that GII.4 Sydney variant infected individuals with both secretor and nonsecretor phenotypes. This finding was consistent with the HBGA binding results that the recombinant P protein of the GII.4 Sydney variant bound to saliva samples of nonsecretors, as well as the Lewis x and Lewis a oligosaccharides. Taken together, our data suggested that the GII.4 Sydney 2012 variant gained new HBGA binding ability and antigenic features which may contribute to its widespread epidemics.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12985-026-03129-x. Introduction Noroviruses (NoVs) are the leading viral pathogens causing both epidemic and sporadic acute gastroenteritis across all age groups. Transmission primarily occurs through the fecal-oral route, often via a contaminated environment, food, or water [1, 2]. NoV outbreaks are common in large communities and institutions including schools, hospitals, nursing homes, military camps, and cruise ships [3, 4]. The high infectivity of NoVs, their low infectious dose, short-term host immunity, and remarkable environmental stability make controlling NoV gastroenteritis particularly challenging [5]. The NoV RNA genome is organized into three open reading frames (ORF1-3), which encode the nonstructural proteins, including RNA-dependent RNA polymerase (RdRp), the major capsid protein (VP1), and the minor capsid protein (VP2), respectively. Structurally, VP1 comprises two major domains: the shell (S) domain and the protruding (P) domain. The S domain forms the inner shell of the viral capsid, while the P domain, consisting of P1 and P2 subdomains, forms the protrusions, which are the most exposed regions of the viral capsid [6, 7]. Previous studies have shown that the P domain plays a crucial role in the viral immune response and in binding to host carbohydrates, affecting host susceptibility to NoV infection [2, 8, 9]. NoVs are genetically diverse, comprising at least 10 genogroups (GI-GX) and more than 40 genotypes [10]. Despite this significant variation, the GII.4 genotype accounts for approximately 50% of all NoV infections worldwide [6]. Periodic. emergences of GII.4 variants have been observed, with changes in both antigenic sites and histo-blood group antigen (HBGA) binding sites, allowing new variants to escape herd immunity [11, 12]. Since 1995, several GII.4 variants have emerged and caused global epidemics, including US 1995/96 in 1996 [13], Farmington Hills in 2002 [14], Hunter in 2004 [15], DenHaag in 2006 [16], New Orleans in 2009 [17], and Sydney in 2012 [18]. The GII.4 Sydney 2012 variant has remained consistently predominant globally since its emergence in 2012, despite changing its ORF1 from GII.P31 (previously called GII.Pe) to a novel GII.P16 [19]. It has been shown that there are differences in NoV susceptibility which are associated with histo-blood group antigens (HBGAs) (i.e. the ABO blood group, the Lewis phenotype and the secretor status) [6] since NoVs recognize HBGAs in humans [5]. The expression of the α-1,2-fucosyltransferase (FucTII) enzyme, encoded by the FUT2 gene, determines the secretor status of an individual [10]. Secretor-positive individuals have A, B, H, Leb, and/or Ley antigens. In contrast, secretor-negative or non-secretor individuals have an inactive FUT2, leading to the lack of A, B, H, Leb and/or Ley antigens. However, they may be Lea and Lex positive as products of FUT3 [11, 12]. It has been previously shown that human susceptibility to NoV is not completely determined by the secretor status, since secretor negative individuals can also be infected by NoVs [15, 16]. Here, we report large-scale outbreaks in a university town in China during the period when the GII.4 Sydney 2012 variant emerged and spread globally. Mechanistic studies focusing on HBGA binding capability and host susceptibility, as well as antigenic features, indicated that the GII.4 Sydney 2012 variant in China had gained a wider spectrum of HBGA binding properties compared to previous GII.4 variants. Subjects, materials, and methods Ethics statement The research received approval from the Institutional Review Board (IRB) at the China CDC, ensuring the protection of human subjects. Written informed consent was obtained from symptomatic patients and asymptomatic individuals for the collection of various samples, including stool specimens and saliva samples. Outbreak investigation and sample collection Between January 8 and April 3, 2013, outbreaks of acute gastroenteritis occurred in the Guangzhou Higher Education Mega Center (university town), situated on an island in Guangzhou city, Guangdong Province, southern China. This area includes approximately 170,000 teachers and students from 10 universities and one middle school, as described previously [20, 21]. A NoV outbreak was defined as more than five acute gastroenteritis cases occurring within three days after exposure in a common setting, with more than two samples (whole fecal, rectal swab, or vomitus) laboratory-confirmed as NoV. Probable cases were defined as individuals experiencing symptoms of vomiting (≥ 2 episodes) and/or diarrhea (≥ 3 episodes) within one day, who were students, teachers, or other staff between January 8 and April 3, 2013. Potential cases were identified through outpatient logs from the university town hospital, outpatient department logs of the school, the academic affairs department, and the island health station. Emergency monitoring systems were established for outpatient departments on each campus, with specialized medical personnel assigned to report cases daily. Additionally, a unified epidemiological questionnaire for acute gastroenteritis cases was used to gather basic information, clinical characteristics, and epidemiological contact history. A case-control study was conducted to identify transmission risk factors. A total of 386 stool or anal swab samples from gastroenteritis cases were collected, including 90 from ZD, 42 from HS, 93 from HG, 57 from GG, 25 from GDFZ and 79 from GD—abbreviations representing the five universities and one middle school involved in the outbreak. Fifty-four saliva samples were also collected from symptomatic patients in one classroom and one cafeteria. rRT-PCR, Genotyping and genome amplification All clinical samples (stool, anal swabs, and vomitus) were collected in sterile containers, transported to the laboratory at 4 °C within 4 h of collection, and stored at -80 °C until nucleic acid extraction and analysis. For stool samples requiring suspension preparation, samples were processed immediately after thawing without repeated freeze-thaw cycles. For stool samples, a 10% stool suspension was prepared by mixing 0.1 g of stool with 1.0 ml of phosphate-buffered saline (pH 7.2). For swab or vomitus samples, nucleic acid was extracted directly. Commercial NoV real-time RT-PCR kits (BioPerfectus Technology Co., http://www.s-sbio.net) were used for NoV detection. Two NoV fragments from positive samples, located in the polymerase and capsid genes, were amplified using primer pairs targeting highly conserved regions of these proteins [19]. Positive samples were dual-typed using conventional RT-PCR, targeting a 579 bp product for GI viruses or a 570 bp product for GII viruses, which includes the 3’-end of the polymerase region and the 5’-end of the major capsid gene. The complete genome sequence of strain GZ3135 was determined, including the full-length VP1 region (1623 bp) and the full-length ORF1 region (5100 bp). The genome was amplified using primers designed as described previously [22]. All outbreaks were detected as negative for E. coli, Salmonella, Shigella, Campylobacter, Y. enterocolittica, rotavirus, adenovirus, and astrovirus. DNA sequencing, phylogenetic analysis and Homology model To analyze the DNA sequencing of NoV PCR products and the genome PCR products of strain GZ3135, we utilized the Big-Dye terminator cycle sequencing kit and the ABI Prism 310 Genetic Analyzer from Applied Biosystems Inc., Foster City, CA. The obtained NoV sequences were examined using CLUSTAL X (Version 1.83) and subjected to phylogenetic analysis using MEGA version 4.1. The phylogenetic tree was constructed using the neighbour-joining method (distance calculation by the Kimura-2-parameter correction; pairwise deletion) implemented in the MEGA 4.1 program, with results validated by 1000 bootstrap replicates. For the comparative modeling of the GII.4 Sydney 2012 variant (GZ3135), the GII.4 DenHaag variant (BJ55019), and the GII.4 95/96US variant (SZ9711), we used the x-ray crystal structure of the P domain of the GII.4 NoV strain 2012 NSW in complex with the B-trisaccharide [23] (accession number 4 × 7D in the Protein Data Bank [PDB; https://www.rcsb.org/structure/4X7D]). These models were generated using the Homology Modeling Database SWISS-MODEL (http://swissmodel.expasy.org). HBGA Phenotyping of Saliva samples To determine the HBGA phenotypes of the saliva samples, we employed EIA assays along with monoclonal antibodies (MAbs) specific to H1, H2, A, B, Lea, Leb, Lex, and Ley antigens, using a previously described method [24, 25]. Saliva samples were diluted at 1:1000 in PBS and then coated onto microtiter plates (Dynex Immulon) overnight at 4 °C. After blocking with 5% milk, MAbs specific to Lewis and ABH antigens were added. MAbs anti-H1, -H2, -Lex, -Ley, -type A, and -type B were used at a dilution of 1:100; anti-Lea at 1:300; and anti-Leb at 1:200. After incubation for 1 h at 37 °C, HRP-conjugated goat anti-mouse IgG or IgM antibodies were added. The cut-off for a positive signal was OD 450 = 0.1. An individual was classified as a secretor if any of the secretor antigens (A, B, H1, Ley,or Leb) were positive. Non-secretor individuals were identified when Lex or Lea was positive while other antigens were negative. Secondary antibody controls without the primary antibody were performed in pre-experiments to exclude the possible binding of the secondary antibody directly to the saliva samples. All controls were negative. Amplification of the capsid protein-encoding gene for in vitro expression For the amplification of the P domain-encoding cDNA sequences of the GII.4 Sydney 2012 variant GZ3135. we conducted PCR using a sense primer (GCACGGATCCTCAAGAACTAAACCATTCTCTGTCCC) and an antisense primer (GCATGCGGCCGCTTAGCAAAAGCAATCGCCACGGCAATCGCATAGTGCACGTCTACGCCCCG). This resulted in the P domain-encoding cDNA with coding sequences for a cysteine-containing tag (CDCRGDCFC) at the 3’ end, potentially facilitating the formation of P particles [26, 27]. The resulting PCR product was then cloned into the expression vector pGEX-4T-1 (GE Healthcare Life Sciences) between the BamHI and NotI sites. Subsequently, recombinant P protein expression in E.coli was conducted as previously described [26]. The P proteins of the GII.4 DenHaag 2006b variant BJ55019 and GII.4 95/96 US SZ9711 were prepared in previous studies [28]. Western blotting was conducted to detect the P protein expression of strains GZ3135, BJ55019, and 95/96 US SZ9711. HBGA binding assay The binding assays, primarily performed as described in previous studies [24, 29], involved using synthetic oligosaccharides and saliva. To prepare the saliva samples, they were boiled and then diluted 1000-fold. Plates were coated with synthetic oligosaccharides at a concentration of 2 µg/ml. The binding assays using synthetic oligosaccharides employed a panel of 9 PAA-biotin-conjugated HBGAs (H type 3 disaccharide; type A (tri), type B (tri), Led (H type 1), H type 2, Lea and Lex trisaccharides; and Leb and Ley tetrasaccharides), as previously reported [24, 29, 30]. The saliva samples and synthetic oligosaccharides were immobilized on 96-well microtiter plates and allowed to bind overnight at 4 °C. The plates were then blocked with 5% nonfat dry milk before the addition of P proteins (5 µg/ml). The detection antibodies were rabbit antisera obtained from immunization with the SZ9711 P particle, diluted at 1:2000 [24, 29]. The secondary antibody was used at a dilution of 1:3000 [28]. Two controls were performed in pre-experiments: in both P particle-saliva and P particle-oligosaccharide binding assays, the P particles were omitted to exclude the possible binding of the P particle-detecting antibody directly to the saliva samples or the oligosaccharides. Both control results were negative. Statistical analysis Categorical data are described using frequencies and percentages. The comparison of categorical data across different groups is conducted using chi-squared tests or Fisher’s exact test, depending on whether the expected frequencies are less than 1. Odds ratios (ORs) and 95% confidence intervals (CIs) for categorical variables are calculated using a two-tailed approach. A significant difference is identified at a p-value of less than 0.05 with a two-tailed test. Results Description of the NoV outbreaks From January 8 to April 3, 2013, multiple outbreaks of acute gastroenteritis occurred in the Guangzhou Higher Education Mega Center, also known as the university town, located on an island in Guangzhou city, Guangdong province, southern China. This university town includes approximately 170,000 students and teachers associated with 10 universities and one middle school. During these outbreak events, 652 individuals associated with 5 universities (ZD, HS, HG, GD and GG) and 1 middle school (GDFZ) reported illness to the Guangzhou CDC. The affected individuals were aged 12 to 57 years, with a male to female ratio of 1.18:1 (353 males to 299 females). Among the patients, 629 were students, accounting for 96.5% (629/652) with an incidence rate of 0.6% (629/10376), followed by 15 cafeteria chefs, 5 teachers, 1 medical staff, 1 hotel employee, and 1 off-campus staff member. The main clinical symptoms included diarrhea (89.0%, 580/652), vomiting (41.3%, 269/652), and fever (11.7%, 76/652). All cases had mild symptoms, with no severe cases or deaths. Real-time RT-PCR tests revealed that 114 of 386 stool and anal swab samples were NoV-positive. Among these, 19 samples were successfully genotyped by conventional RT-PCR and sequencing, and all 19 were identified as the GII.4 Sydney[P31] variant (previously called GII.4 Sydney-GII.Pe). The 19 genotyped samples were distributed across the outbreak period as follows: 5 samples from the first wave (January 8–18, all from ZD campus) and 14 samples from the second wave (March 4-April 3, including 4 from HS, 5 from HG, 3 from GD, and 2 from GG campus). Epidemic curve analysis revealed distinct transmission phases across the affected campuses (Fig. 1). For ZD, HS, and HG campuses, the outbreaks exhibited a biphasic pattern: an initial sharp increase in cases within 24–48 h characteristic of a point-source foodborne exposure, followed by a prolonged tail of secondary cases consistent with person-to-person transmission. In ZD campus, the first 48 h accounted for 78% (110/141) of total cases, supporting the identification of restaurant A as the point source. Subsequently, cases continued at a lower rate (1–3 cases/day) for an additional 9 days, indicating secondary transmission among close contacts. Similarly, in HS and HG campuses, foodborne transmission (linked to infected food handlers and contaminated sandwiches, respectively) triggered the initial case clusters, with sustained transmission through person-to-person contact in dormitories and classrooms during the later stages. In contrast, GD, GG, and GDFZ campuses showed more gradual epidemic curves with multiple peaks over extended periods, suggesting primarily person-to-person transmission without a clear point-source exposure. The first outbreak began on January 8th at the ZD campus and lasted for 11 days, during which 141 cases were reported, peaking on January 8–9 (Fig. 1). This outbreak was evidently associated with foodborne transmission. A case-control study (52 symptomatic individuals in the case group and 67 asymptomatic individuals in the control group) revealed that restaurant A was the primary outbreak setting for early cases (OR = 3.46, 95% CI: 1.07–11.16). The infection risk was particularly high for lunch on January 7 (OR = 4.34, 95% CI: 1.18–17.37), and individuals who consumed the cold shredded chicken set meal had a significantly higher infection risk (OR = 17.82, 95% CI 4.46–78.17) [20, 21]. Fig. 1The date distributions of the NoV outbreaks caused by a novel Sydney variant occurring in the university town of Guangzhou between January 8 and April 2. The outbreaks were associated with the campuses of 5 universities (ZD, HS, HG, GG and GD) and 1 middle school (GDFZ) The date distributions of the NoV outbreaks caused by a novel Sydney variant occurring in the university town of Guangzhou between January 8 and April 2. The outbreaks were associated with the campuses of 5 universities (ZD, HS, HG, GG and GD) and 1 middle school (GDFZ) The second wave of NoV outbreaks started on March 4, coinciding with the beginning of the new semester after winter vacation, and lasted for 31 days across multiple campuses, peaking on March 18–20 (Fig. 1). The cases from ZD in the north of the town spread to HS and HG in the south, and GD in the southwest corner in early March, continuing to spread southward to GDFZ and GG in mid-March after the winter vacation (Fig. 2) [31]. It was noted that eleven symptomatic chefs continued to work while symptomatic at Restaurant B, posing a risk of transmission through contaminated food during the early stages of the HS outbreak. Additionally, fifteen symptomatic students potentially contaminated sandwiches, which they had prepared themselves to celebrate Women’s Day on March 8th during the early stages of the HG outbreak. Of the six campuses, three (ZD, HS, and HG) potentially experienced foodborne transmission in the early stages and contact transmission in the later stages (Table 1). Fig. 2The progress of the NoV outbreaks caused by the novel GII.4 Sydney variant shown in geographical distribution during different periods between January 8 and April 2. Stripes represent public areas. Different colors represent the number of gastroenteritis cases that occurred on each campus (white: no case; light gray: a few cases; dark gray: increasing cases; black: the most cases) The progress of the NoV outbreaks caused by the novel GII.4 Sydney variant shown in geographical distribution during different periods between January 8 and April 2. Stripes represent public areas. Different colors represent the number of gastroenteritis cases that occurred on each campus (white: no case; light gray: a few cases; dark gray: increasing cases; black: the most cases) Table 1Summary of the outbreaks caused by the GII.4 Sydney variant in the university townOutbreak campusStart timeEnding timeIll No.Incidence rate%, (ill No./total No.)Other infected individuals outside of students (No.)Risk factorsSaliva collection (symptomatic/Asymptomatic No.)ZD8-Jan18-Jan1410.78 (141/18057)Teacher (2) and food operator (1)Hand Shredded Chicken Set MealNot collectedHS4-Mar25-Mar680.42 (68/16300)Teacher (1) and food operators (11)Infected food operators8/10HG7-Mar27-Mar1200.68 (120/17520)Teacher (1) and food operator (5)Sandwiches and Wraps7/5GD10-Mar3-Apr1610.56 (161/28764)Not involvedNot involved9/15GG13-Mar31-Mar1020.31 (102/32546)Teacher (1)Not involvedNot collectedGDFZ13-Mar27-Mar602.61 (60/2300)Teacher (1)Not involvedNot collectedTotal6520.56 (652/115469)24/30 Table 1Summary of the outbreaks caused by the GII.4 Sydney variant in the university townOutbreak campusStart timeEnding timeIll No.Incidence rate%, (ill No./total No.)Other infected individuals outside of students (No.)Risk factorsSaliva collection (symptomatic/Asymptomatic No.)ZD8-Jan18-Jan1410.78 (141/18057)Teacher (2) and food operator (1)Hand Shredded Chicken Set MealNot collectedHS4-Mar25-Mar680.42 (68/16300)Teacher (1) and food operators (11)Infected food operators8/10HG7-Mar27-Mar1200.68 (120/17520)Teacher (1) and food operator (5)Sandwiches and Wraps7/5GD10-Mar3-Apr1610.56 (161/28764)Not involvedNot involved9/15GG13-Mar31-Mar1020.31 (102/32546)Teacher (1)Not involvedNot collectedGDFZ13-Mar27-Mar602.61 (60/2300)Teacher (1)Not involvedNot collectedTotal6520.56 (652/115469)24/30 Summary of the outbreaks caused by the GII.4 Sydney variant in the university town Genome sequence analysis of GII.4 Sydney [P31] variant that cause these outbreaks Based on the genomic sequences, the NoV isolated from these outbreaks was identified as a GII.4 Sydney variant-like virus, named GZ3135. The full-length capsid gene (1623 bp) of the GZ3135 strain exhibited a nucleotide identity of 98.8% with that of the GII.4 Sydney prototype isolated in 2012 (JX459908). It also shared 96.4% identity with strain SZ9711 (GII.4 95/96US) and 96.6% with strain BJ55019 (GII.4 2006b DenHaag) from China, which were studied previously [28]. Phylogenetic analysis of the full-length ORF2 showed that strain GZ3135 clustered in the same branch as the prototype Sydney GII.4 strain NSW0514 (GenBank No. JX459908) detected globally (Fig. 3A). Similarly, ORF1 typing indicated that strain GZ3135 could be typed as GII.P31 (previously called GII.Pe), consistent with the prototype Sydney GII.4 strain NSW0514 (Fig. 3B). Based on X-ray crystallographic study findings [32], the HBGA binding pocket is composed of four fundamental components within the P domain: loop-1 (residues 372–378), loop-2 (residues 390–396), loop-3 (residues 441–445), and a β-sheet (residues 341–346) (Fig. S1). Sequence alignments showed that the major residues of the β-sheet and loop-3 remain conserved among all clusters, while residues in loop-1 and loop-2 experienced epochal evolution (Fig. S1). For the GII.4 Sydney variants, residue changes occurred at positions 372, 373, and 393 compared to previous GII.4 variants. Compared to the prototype Sydney 2012, GZ3135 had two unique residue mutations: R373H and G393S (Fig. 4A). Conformation differences at residue 393 were observed in the GII.4 Sydney strain GZ3135 compared to BJ55019 (GII.4 DenHaag 2006b) and SZ9711 (GII.4 95/96US) (numbered based on GII.4 VA387) (Fig. 4 B and C). Amino acid alignment indicated that these GII.4 strains reveal an epochal evolution. Compared to previous GII.4 variants, GII.4 Sydney variants, including the GII.4 Sydney prototype and strain GZ3135 from the present study, exhibited 11 residue mutations in Epitope A (T294, E368, D372, R/H373), Epitope C (T340, A377, and N378), Epitope E (T413), and Epitope G (D357, A359, and R364). Additionally, three residue differences (R373H, G393S, and D310N) were noted between GZ3135 and the GII.4 Sydney prototype (Fig. 4). Fig. 3Phylogenetic analyses of the GII.4 Sydney variant (GZ3135) in comparison with other GII.4 variants (SZ9711 and BJ55019) isolated in China with reference strains from GenBank based on the VP1 (A) and RdRp (B) sequences, respectively Fig. 4Evolutionary changes of the HBGA-binding interface and putative antigenic epitopes on the P domains of GII.4 VP1s. (A) The antigenic residues comparison of the GII.4 Sydney variant (GZ3135) and other GII.4 variants (SZ9711 and BJ55019) isolated in China with the representative GII.4 variants from 1987 to 2012 in major capsid protein (P2 region). Major changes of HBGA binding sites in loop 1 and loop 2 are marked with asterisks (*) and pound symbols (#), respectively. Variant-specific color coding indicates the residue substitutions that occurred compared with the previous lineage where substitutions were retained in subsequent variant. The residue positions of A-H epitopes and the binding pocket (loop-1, loop-2, loop-3 and a β-sheet) are shown in sequence alignments of P proteins in Fig S1. Based on anterior and top views, there are conformational differences in residue 393 among the GII.4 Sydney strain GZ3135, BJ55019 (2006b), and SZ9711 (95/96US) (B and C) Phylogenetic analyses of the GII.4 Sydney variant (GZ3135) in comparison with other GII.4 variants (SZ9711 and BJ55019) isolated in China with reference strains from GenBank based on the VP1 (A) and RdRp (B) sequences, respectively Evolutionary changes of the HBGA-binding interface and putative antigenic epitopes on the P domains of GII.4 VP1s. (A) The antigenic residues comparison of the GII.4 Sydney variant (GZ3135) and other GII.4 variants (SZ9711 and BJ55019) isolated in China with the representative GII.4 variants from 1987 to 2012 in major capsid protein (P2 region). Major changes of HBGA binding sites in loop 1 and loop 2 are marked with asterisks (*) and pound symbols (#), respectively. Variant-specific color coding indicates the residue substitutions that occurred compared with the previous lineage where substitutions were retained in subsequent variant. The residue positions of A-H epitopes and the binding pocket (loop-1, loop-2, loop-3 and a β-sheet) are shown in sequence alignments of P proteins in Fig S1. Based on anterior and top views, there are conformational differences in residue 393 among the GII.4 Sydney strain GZ3135, BJ55019 (2006b), and SZ9711 (95/96US) (B and C) HBGA phenotypes and Symptomatic Infection Three acute gastroenteritis (AGE) outbreaks caused by the GII.4 Sydney [P31] variant were identified and further studied (Table 1). To determine the HBGA phenotypes, saliva samples were collected from both symptomatic and asymptomatic individuals and analyzed using enzyme immunoassays (EIAs) (Table 2). The results demonstrated that this GII.4 Sydney variant infected individuals with blood types A, B, H (O), and Lewis b/y-positive phenotypes. Interestingly, one secretor-negative individual was found in the symptomatic group. This individual (female, aged 20 years) was Lewis a-positive (Lea) and Lewis x-positive (Lex), with no detectable A, B, H, Leb, or Ley antigens, consistent with the non-secretor phenotype (FUT2 inactive, FUT3 active). Stool samples from this individual tested positive for NoV GII.4 Sydney[P31] by rRT-PCR, confirming symptomatic infection. Table 2Distribution of HBGA phenotypes among individuals with or without symptoms of acute gastroenteritisHBGAsSymptomaticAsymptomaticTotalNo. (%), n =24No. (%), n =30No. (%), n = 54x² P OR(95% CI) ABO phenotypes1: O9 (37%)19 (63%)28 (51%)3.5640.0590.347 (1.114~1.055)A8 (33%)4(13%)12 (22%)3.0680.0793.250 (0.841~12.564)B3 (12%)6 (20%)9 (16%%)0.5490.4620.571 (1.27~2.573)A/B4 (16%)1(3%)5 (9%)2.8230.0915.800 (0.603~55.813) Lewis phenotypes: Leb17 (70%)23(76%)40 (74%)0.2360.6270.739 (0.218~2.506)Ley19 (79%)20(66%)39 (72%)1.0380.3081.900 (0.548~6.509)Lea17 (70%)14 (46%)31(57%)3.1850.0742.776 (0.892~8.639)Lex7 (29%)5 (16%)12 (22%)1.2050.2722.059 (0.560~7.574) Secretor status: Positive23 (95%)25 (83%)48 (88%)2.1090.1464.600 (0.499~42.373)Negative1 (4%)5 (16%)6 (11%)2.1090.1462.17 (0.024~2.007)Binding to P proteinof strain GZ313520 (83%)22(73%)0.7710.381.818 (0.474~6.974) Table 2Distribution of HBGA phenotypes among individuals with or without symptoms of acute gastroenteritisHBGAsSymptomaticAsymptomaticTotalNo. (%), n =24No. (%), n =30No. (%), n = 54x² P OR(95% CI) ABO phenotypes1: O9 (37%)19 (63%)28 (51%)3.5640.0590.347 (1.114~1.055)A8 (33%)4(13%)12 (22%)3.0680.0793.250 (0.841~12.564)B3 (12%)6 (20%)9 (16%%)0.5490.4620.571 (1.27~2.573)A/B4 (16%)1(3%)5 (9%)2.8230.0915.800 (0.603~55.813) Lewis phenotypes: Leb17 (70%)23(76%)40 (74%)0.2360.6270.739 (0.218~2.506)Ley19 (79%)20(66%)39 (72%)1.0380.3081.900 (0.548~6.509)Lea17 (70%)14 (46%)31(57%)3.1850.0742.776 (0.892~8.639)Lex7 (29%)5 (16%)12 (22%)1.2050.2722.059 (0.560~7.574) Secretor status: Positive23 (95%)25 (83%)48 (88%)2.1090.1464.600 (0.499~42.373)Negative1 (4%)5 (16%)6 (11%)2.1090.1462.17 (0.024~2.007)Binding to P proteinof strain GZ313520 (83%)22(73%)0.7710.381.818 (0.474~6.974) Distribution of HBGA phenotypes among individuals with or without symptoms of acute gastroenteritis Binding to P protein of strain GZ3135 HBGA-binding profile of the GII.4 Sydney variant The HBGA-binding P domains of three GII.4 variants [GZ3135 (Sydney 2012), BJ55019 (DenHaag 2006b), and SZ9711 (95/96 US)] were expressed for P particle production (Fig. S2) [26]. All three P proteins bound to secretor-positive saliva samples from individuals with O, A, B and AB blood types, showing strong binding signals. Interestingly, the P protein of the. GII.4 Sydney 2012 variant (GZ3135) also bound to two secretor-negative saliva samples, one from a symptomatic individual and the other from an asymptomatic individual (Fig. 5 and Fig. S3B). The symptomatic individual was confirmed to be Lewis a-positive and Lewis x-positive (Lea/Lex), consistent with the binding of GZ3135 to synthetic Lea and Lex oligosaccharides (Fig. 5B). The GZ3135 P domain also bound to one of five secretor-negative saliva samples from a total of 113 phenotype-determined saliva samples collected in our previous study (Fig. S3D) [33]. The HBGA (A, B, H1, H2, Lea, Leb, Lex or Ley) binding signals of the nonsecretor saliva samples from symptomatic (S) and asymptomatic (C) individuals of this study (Fig. S3A) and the previous study (Fig. S3C) were shown. The binding patterns of SZ9711 and BJ55019 were determined using the 113 phenotype-determined saliva samples from the previous study [33]. In that study, three other P proteins, including 91 (Camberwell), JB03040042 (Hunter), and JB03050156 (Sakai), had similar binding patterns to SZ9711 and BJ55019 but exhibited weaker binding signals [34]. Previously, there had been reports on the characterization of carbohydrate binding patterns in six VLPs derived from different GII.4 variants, including GII.4–1987 (Camberwell), GII.4–1997 (95/96US), and GII.4–2002 (Farmington Hills), GII.4–2002a (two replacements at site P226S and A395T compared with GII.4–2002), GII.4–2004 (Hunter), and GII.4–2005 (Sakai), spanning various years. These patterns include the FUT2-dependent HBGA binding pattern (observed in GII.4–1987, GII.4–1997, and GII.4–2002), the binding pattern involving non-secretor antigens (Lewis a and Lewis x), and FUT2-dependent A antigen (seen in GII.4–2002). Additionally, there were GII.4–2004 and GII.4–2005 variants that did not display any carbohydrate binding pattern [2]. In this study, the binding pattern of strain SZ9711 isolated in 1997 was consistent with the previously reported GII.4–1997 variant, showing that strain SZ9711 utilized known FUT2-dependent HBGAs and bound efficiently to Ley, as well as A and B antigens. Strain BJ55019, isolated in 2006 (GII.4 DenHaag 2006b), also belonged to the FUT2-dependent A antigen binding pattern. Compared to the binding pattern of strain SZ9711, Strain BJ55019 bound to Ley, A and B antigens, as well as H2. Strain GZ3135, isolated in 2013 (GII.4 Sydney 2012), not only bound to FUT2-independent products (H2, A, B, Leb and Ley) but also to non-secretor antigens (Lea and Lex). Discussion In March 2012, the novel GII.4 variant Sydney emerged in Australia, raising concerns about a possible increase in NoV epidemics in the following winter in the northern hemisphere [18]. Indeed, the newly emerged GII.4 Sydney variant caused increased NoV activity in many countries globally during the 2012/13 season [35, 36]. In China, the GII.4 Sydney variant caused both outbreaks and sporadic cases in different areas during the same period [37, 38]. The outbreaks described in this study occurred during the period when the emerging Sydney variant was sweeping across the world. This study reports the large-scale outbreaks of gastroenteritis caused by an emerging GII.4 Sydney variant [P31] (previously called GII.Pe-GII.4 Sydney 2012), which resulted in sustained transmission involving six campuses in the university town of Guangzhou, China. The strong spread capacity of this novel GII.4 NoV variant made the outbreaks difficult to control, lasting several months. The interval of winter vacation did not stop the epidemic. Fig. 5Binding of GII.4 P domain proteins to saliva samples and synthetic oligosaccharides. (A) Saliva-based HBGA-binding assays were performed using recombinant P domain proteins of strain GZ3135 (red), BJ55019 (blue), and SZ9711 (green). The ABO phenotypes of saliva samples are indicated as A, B, AB, O secretor (O+) and O nonsecretor (O-). (B) Binding of P domain proteins of GII.4 strains, GZ3135 (red), BJ55019 (blue), and SZ9711 (green) to a panel of synthetic oligosaccharides representing 8 different HBGAs (H1, H2, A, B, Lea, Leb, Lex and Ley) (X-axis). All oligosaccharides were linked to polyacrylamide (PAA)-biotin as backbones. All data represent the average values of triplicate experiments. The dashed box indicates positive binding for nonsecretor oligosaccharides (Lex and Lea) Binding of GII.4 P domain proteins to saliva samples and synthetic oligosaccharides. (A) Saliva-based HBGA-binding assays were performed using recombinant P domain proteins of strain GZ3135 (red), BJ55019 (blue), and SZ9711 (green). The ABO phenotypes of saliva samples are indicated as A, B, AB, O secretor (O+) and O nonsecretor (O-). (B) Binding of P domain proteins of GII.4 strains, GZ3135 (red), BJ55019 (blue), and SZ9711 (green) to a panel of synthetic oligosaccharides representing 8 different HBGAs (H1, H2, A, B, Lea, Leb, Lex and Ley) (X-axis). All oligosaccharides were linked to polyacrylamide (PAA)-biotin as backbones. All data represent the average values of triplicate experiments. The dashed box indicates positive binding for nonsecretor oligosaccharides (Lex and Lea) Although the two waves of outbreaks (January 8–18 and March 4-April 3) were separated by the winter vacation period (approximately mid-January to late February), epidemiological and molecular evidence supports their relatedness. First, the second wave outbreaks occurred immediately after students returned from winter vacation, suggesting that the virus may have been reintroduced to the campus through asymptomatic carriers or contaminated environments. Second, all 19 genotyped samples from both waves were identified as the GII.4 Sydney[P31] variant with high sequence similarity (99.5–100% nucleotide identity in the VP1 region), indicating that the same viral strain was responsible for both waves. The winter vacation interval likely interrupted active transmission but did not eliminate the virus from the environment or the population, leading to resurgent outbreaks at the beginning of the new semester. At the beginning of the new spring semester, the second wave of the outbreaks involved five campuses. The outbreaks that occurred in three schools were found to be associated with foodborne transmission, and two of them may be attributed to sick food handlers. In the United States, 42% of foodborne gastroenteritis outbreaks from 2009 to 2010 were caused by NoV, while 53% of foodborne NoV outbreaks were associated with contaminated food through food handlers [39]. In this study, infected food handlers or contaminated foods were also the main risk factors in the early stage of outbreaks, suggesting that foodborne transmission played an important role in the rapid transmission. While most predominant GII.4 variants circulated for approximately 4 years or more, the GII.4 Sydney variant predominated for more than 10 years, similar to the 95/96 US variant [40]. A recent study analyzed 1400 capsid sequences from GII.4 Sydney and identified changes in 4 (A, C, D, and E) of the 6 major antigenic sites, where the major mutations (10/12) for strain GZ3135 and the prototype Sydney strain also occurred in this study [40]. Of the four most immunodominant antigenic sites, Site A could account for intra-variant antigenic diversification of the Sydney_2012 variant, with mutation occurring in residues 297, 372, and 373. Residues 297 and 372 co-evolved and changed over time [40]. Of these sites, residues 297 and 372 were the same, but residue 373 differed between strain GZ3135 (373 H) and the Sydney prototype (373R). Overall, despite circulating for over a decade, the GII.4 Sydney variant presents minimal antigenic diversification, providing novel insights into diversification [40]. While these highly variable amino acids on the capsid surface constitute important epitopes, they also altered HBGA binding affinity and specificity by constituting the binding pocket with changed features [40]. In this study, the four unique residue mutations for GII.4 Sydney variants occurred in Loop-1 (372, 373, 377, and 378). Previous studies have shown that Asp374 in loop-1 is conserved and plays a crucial role in recognizing HBGAs. However, other residues in loop-1 are highly variable and may contribute to changes in HBGA binding affinity and specificity of GII.4 variants. In this study, residues 373 and 393 of strain GZ3135 differ from those in the Sydney variant prototype. Previous research has discussed how changes in these residues correlate with the affinity of HBGAs [2, 30, 31]. Lindesmith et al. demonstrated that residues 393 to 395 are important not only for HBGA interactions but also for antigenicity in GII.4 NoV [2]. The rearrangements of loop-2, particularly residues 391, 392, and 393, are believed to be associated with binding, contributing to the enhanced NoV-HBGA binding affinity during the evolution of GII.4 NoV strains [2, 30]. The removal of the threonine residue at position 395 in the Hunter variant (post-2002 variant) did not affect binding to the H antigen, but binding to A and B antigens was severely hampered. Interestingly, the mutant with residue 395 deleted gained the capacity to bind to Lewis x and sialyl-Lewis x antigens compared to the wild-type VLP, indicating that amino acid residues outside the HBGA binding site can modify NoV binding properties [31, 34]. Liang et al. revealed that a mutation from Asn to Ser at this residue resulted in the side chain changing from pointing out of the binding pocket to pointing inward, bringing it significantly closer to the attachment factor HBGAs [30]. This conformational alteration potentially accounts for the substantially enhanced binding affinity between NoV-HBGA and the Sydney variant [30]. Residue 393 is a Ser in GZ3135, whereas it is a Gly in the GII.4 Sydney variant prototype. Additionally, the conformation differences of residue 393 among the GII.4 Sydney strain GZ3135, BJ55019 (GII.4 DenHaag 2006b), and SZ9711 (GII.4 95/96US) are shown in Fig. 4 (B and C) (numbered based on GII.4 VA387). Therefore, the change from Gly to Ser at residue 393 might be related to the capacity of GZ3135 to bind to Lewis X. The role of HBGAs in the evolution of GII.4 NoVs has been widely acknowledged [10, 26, 27]. Consequently, understanding the evolution of GII.4 NoVs requires studying NoV epidemiology and host susceptibility. In this study, we examine outbreaks caused by the GII.4 Sydney variant and observe that both secretors and a non-secretor displayed susceptibility to this variant. One female non-secretor was symptomatically infected, with the virus found in her stool. The symptomatic infection observed in a Lea/Lex non-secretor individual provides direct evidence that the GII.4 Sydney variant can overcome the traditional genetic resistance barrier associated with FUT2 inactivation. This observation is consistent with our in vitro binding data showing that the GZ3135 P protein binds to both Lea and Lex oligosaccharides, as well as to saliva samples from non-secretors. The Lewis-positive status of this individual (Lea and Lex) explains the susceptibility, as the virus has evolved to recognize these FUT3-dependent antigens as alternative attachment factors. These findings support the hypothesis that the emergence and success of the GII.4 Sydney variant may be partially attributed to its expanded HBGA binding spectrum, enabling infection of previously resistant populations. Our previous study also observed that four non-secretors were infected in one outbreak caused by the DenHaag 2006b variant. However, the failure of P protein production prevented us from determining the HBGA-binding profile of the outbreak virus [41]. In this study, the HBGA binding assay showed that the GII.4 Sydney variant GZ3135 bound to non-secretor saliva samples, as well as Lex and Lea oligosaccharides, representing the non-secretor HBGA types. A similar observation was reported previously, where Carlsson and co-workers demonstrated that saliva from a secretor-negative patient bound to the VLP derived from GII.4 Valencia/2004/Es, which caused an outbreak [41]. Another interesting finding was from an outbreak caused by the GII.4 New Orleans 2009 variant, where a stool sample from an asymptomatic secretor-negative individual tested positive for NoV by RT-PCR. This observation was confirmed by the detection of a high anti-NoV IgA titer in the saliva of this non-secretor individual, similar to the titers in secretor-positive patients [42]. A very similar DenHaag variant of GII.4 NoV, Cairo1 (EU876892), revealed binding ability to both secretor and non-secretor saliva samples, as well as to Lex and sialyl Lex oligosaccharides [31]. Liang et al. also showed that three GII.4 Sydney variants exhibited high binding affinities with Lea and Lex, while previous GII.4 variants showed no or lower binding affinities to these oligosaccharides [30]. These data imply that the emergence of novel variants is influenced by both herd immunity and new HBGA binding properties [2, 12]. In conclusion, this study described large-scale outbreaks that affected several university campuses in China during the emergence and global spread of the GII.4 Sydney 2012 variant. The ability to infect both secretor and non-secretor populations, the binding capability to both secretor and non-secretor HBGAs, and adaptive changes in key antigenic epitopes may have contributed to the success of strains belonging to the GII.4 Sydney variant. Our data support the hypothesis that host susceptibility and antigenic variation play critical roles in the emergence and evolution of viruses within the GII.4 lineage. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1. Fig S1. Sequence alignments between the partial VP1 proteins of the GII.4 variants isolated in China (GZ3135, BJ55019, and SZ9711) and representative GII.4 variants circulating from 1987 to 2013. The residues in loops or sheets involved in the HBGA binding pocket, i.e., loop-1, loop-2, loop-3 and a β-sheet, are marked by red dashed boxes. The potential key epitopes (A-H) are shaded in different colors. The gray color indicates the residue substitutions that occurred compared with the previous lineage in the HBGA binding pocket and the GII.4 potential epitopes. Fig S2. SDS-PAGE analysis of the P domain proteins (~35 kDa) of three GII.4 variants, GZ3135 (lane 1), SZ9711 (lane 2), and BJ55019 (lane 3). The gel was stained with Brilliant Blue (G-250). The lane M represents protein standards with indicated molecular weights. The arrow indicates the target P proteins. Fig S3. Nonsecretor status determination of individuals by histo-blood group antigen (HBGA) phenotyping and binding of three GII.4 P domain proteins (GZ3135, BJ55019, and SZ9711) to saliva samples with nonsecretor phenotype. The HBGA (A, B, Lea, Leb, Lex or Ley) signals of the nonsecretor saliva samples from symptomatic (labeled as S## ) and asymptomatic (labeled as C##) individuals in this study (A) and in a previous study (C) were shown. The signals of Lea and Lex are marked in red. The binding signals of GII.4 P domain proteins (GZ3135, BJ55019, and SZ9711) to saliva samples with nonsecretor phenotype in this study (B) and in a previous study (D) are shown. Supplementary Material 1. Fig S1. Sequence alignments between the partial VP1 proteins of the GII.4 variants isolated in China (GZ3135, BJ55019, and SZ9711) and representative GII.4 variants circulating from 1987 to 2013. The residues in loops or sheets involved in the HBGA binding pocket, i.e., loop-1, loop-2, loop-3 and a β-sheet, are marked by red dashed boxes. The potential key epitopes (A-H) are shaded in different colors. The gray color indicates the residue substitutions that occurred compared with the previous lineage in the HBGA binding pocket and the GII.4 potential epitopes. Fig S2. SDS-PAGE analysis of the P domain proteins (~35 kDa) of three GII.4 variants, GZ3135 (lane 1), SZ9711 (lane 2), and BJ55019 (lane 3). The gel was stained with Brilliant Blue (G-250). The lane M represents protein standards with indicated molecular weights. The arrow indicates the target P proteins. Fig S3. Nonsecretor status determination of individuals by histo-blood group antigen (HBGA) phenotyping and binding of three GII.4 P domain proteins (GZ3135, BJ55019, and SZ9711) to saliva samples with nonsecretor phenotype. The HBGA (A, B, Lea, Leb, Lex or Ley) signals of the nonsecretor saliva samples from symptomatic (labeled as S## ) and asymptomatic (labeled as C##) individuals in this study (A) and in a previous study (C) were shown. The signals of Lea and Lex are marked in red. The binding signals of GII.4 P domain proteins (GZ3135, BJ55019, and SZ9711) to saliva samples with nonsecretor phenotype in this study (B) and in a previous study (D) are shown.