Changes in norovirus genotype diversity in gastroenteritis outbreaks in Alberta, Canada: 2012–2018

Gastroenteritis outbreak investigations in Alberta were managed by public health officials in collaboration with the Provincial Laboratory for Public Health (ProvLab) [8]. Stool samples collected between July 2012 and February 2018 during outbreak investigations were tested at ProvLab for NoV genogroup I and genogroup II using a real time RT-PCR assay [14]. A NoV- confirmed outbreak was defined as ≥2 epidemiologically linked cases with gastroenteritis and at least one sample tested positive for NoV. Data was analyzed from July to June of the following year considering the winter seasonality of NoV.

Outbreak settings were classified into 6 different groups: 1) community long-term care, hospital long-term care, supportive living, and group homes facilities; 2) hospital acute care; 3) food establishments, catering events, food shops, community functions, conferences and hotels; 4) day care centers; 5) other types of group residences, e.g. camp, dormitory; and 6) other settings including community shelters, community services, household, schools and cruise ships. Groups 1 and 2 were classified as healthcare-related settings, whereas groups 3 to 6 were classified as non-healthcare-related.

For each NoV-positive outbreak, one NoV positive sample was selected for genotyping. Briefly, the nucleic acid extract from each stool sample was subjected to reverse transcription (RT) with random primers and the resulting cDNA was PCR-amplified. Samples collected up to February 2017, were amplified in region C (ORF2) using primer pair G2SKF/G2SKR for genogroup II strains or primer pair G1SKF/G1SKR for genogroup I strains [15]. Samples with emergent or unassigned genotypes based on region C sequence analysis were further genotyped targeting the 3’end of the polymerase gene using primers LV4282-99F [16] and COG2R [17]. Samples collected from March 2017 onwards, were genotyped using a dual polymerase-capsid genotyping protocol based on a single PCR amplicon obtained with primer pair MON432/G1SKR for genogroup I strains and primer pair MON431/G2SKR for genogroup II strains [18]. All PCR products were subjected to Sanger sequencing and genotypes were assigned using the Norovirus Genotyping tool [19]. A large majority of strains from outbreaks occurring between July 2012 and June 2015 were left uncharacterized at ORF1; retrospective characterization was not attempted based on observations that 69% of norovirus GII outbreaks in Alberta were caused by a single ORF2 genotype, GII.4 Sydney, and reports from North America [18] and diverse countries from different continents [20] suggest that GII.Pe/GII.4 Sydney was the major strain in circulation worldwide during that time frame.

Phylogenetic analyses of GII.17, GII.P16/GII.4 Sydney, GII.P4 New Orleans/GII.4 Sydney and GII.P16/GII.2 sequences obtained in our study were performed with MEGA 6.06 [21]. Maximum likelihood trees were constructed using the substitution model producing the lowest Bayesian Information Criterion scores, as calculated by the software. For all trees, branch significance was estimated based on 1000 bootstrap replicates.

The capsid P domain of two GII.P16/GII.4 Sydney outbreak strains (AB-2016-26 and AB-2016-190) were amplified by RT-PCR using forward primer ACGCGGATCCTCAAGAACTAAACCATTCTCTGTCC and reverse primer ATAAGAATGCGGCCGCTTAGCAAAAGCAATCGCCACGGCAATCGCATACTGCACGTCTACGCCCCGTTCC and cloned into pGEX-4 T-1 vector (GST Gene fusion System, GE Healthcare Life Sciences) between the Bam HI and Not I sites. An RGD4C tag (CDCRGDCFC) was linked to the C terminus of the P domain for P particle formation [22]. The recombinant P domain protein was expressed in E. coli (BL21, DE3) with induction by 0.25 mM isopropyl-β-D-thiogalactopyranoside (IPTG) at room temperature (~ 21 °C) overnight as described elsewhere [22]. Purification of the glutathione S-transferase (GST)-P domain fusion protein was performed using resin of Glutathione Sepharose 4 Fast Flow (GE Healthcare Life Sciences) according to the manufacturer’s instruction. GST was removed from the target proteins by thrombin (GE Healthcare Life Sciences) cleavage either on beads or in solution (phosphate buffer saline, PBS, pH 7.4).

The saliva-based binding assays were performed as previously described [13]. Briefly, boiled human saliva with known HBGA phenotypes collected from Cincinnati, OH, United States, were diluted 1000-fold and used to coat 96-well microtiter plates (Dynex Immulon; Dynatech, Franklin, MA). After blocking with 5% non-fat milk in PBS, different concentrations of P-domain protein (15, 7.5, 3.75 ng/μl) were added to the wells. The bound P proteins were detected using a guinea pig anti-NoV antiserum (1:3000), followed by the addition of HRP-conjugated goat anti-guinea pig IgG. The HRP activity was then measured with TMB kit (Kierkegaard & Perry Laboratories, Gaithersburg, MD) and the OD450 values were read with an ELISA spectrum reader (Tecan, Durham, NC).

A total of 1572 gastroenteritis outbreak investigations were performed in Alberta between July 1st 2012 and February 30th 2018, of which 859 (54.6%) had specimens submitted to the ProvLab for laboratory testing. Norovirus was identified in 530 (61.7%) of all tested outbreaks. The monthly distribution of NoV-positive outbreaks peaked in the winter months (Fig. 1), however, spring peaks with higher activity than that of winter months occurred in March 2014 and May 2016. Compared to historical data, the annual numbers of NoV outbreaks between July 2012 to June 2017 were lower than those observed in the previous 5 years (July 2007 to June 2012 vs. July 2012 to June 2017, p = 0.0489, one-tailed t-test).

Genogroup II strains were responsible for 440 out of 530 (83.0%) of laboratory confirmed NoV outbreaks (Table 1), while genogroup I and mixed genogroup I and genogroup II strains were responsible for 83 (15.7%) and 7 (1.3%) NoV outbreaks, respectively. Twenty ORF2-based genotypes were identified throughout the study period: GI.1, GI.2, GI.3, GI.4, GI.5, GI.6, GI.7, GI.9, GII.1, GII.2, GII.3, GII.4, GII.5, GII.6, GII.7, GII.8, GII.13, GII.14, GII.16 and GII.17 (Table 2). Overall, GII.4 was the most common genotype and was responsible for at least 319 (60.2%) out of 530 NoV-positive outbreaks. Strains carrying an ORF2 of variant Sydney were accountable for the majority of GII.4 outbreaks (297/319, 93.1%). GII.4 variants Den Haag and New Orleans, which caused pandemics in 2006 and 2009, respectively, disappeared after June 2014. Genotypes GI.6 and GI.7 had an increased presence during July 2012 to June 2013 whereas GI.3 was the predominant GI strain from July 2015 to June 2016. Four novel NoV strains emerged during the last three annual periods of study (Fig. 2): 1) GII.17; 2) GII.P16/GII.4 Sydney, a recombinant strain carrying a polymerase gene of GII.16 and a GII.4 Sydney capsid gene; 3) GII.P16/GII.2, a recombinant strain carrying a GII.16 polymerase gene and a GII.2 capsid gene and 4) GII.P4 New Orleans/GII.4 Sydney, a recombinant strain with a GII.4 New Orleans polymerase gene and a GII.4 Sydney capsid gene.

The first GII.17 outbreak in Alberta occurred in September 2014. Phylogenetic analysis of complete capsid sequences identified all GII.17 strains as part of the Kawasaki cluster (Additional file 1: Figure S1). Representatives of two sub-clusters of GII.17 Kawasaki previously defined by Chan et al. [23] were identified: strains from 2014 grouped into the Kawasaki323-like sub-cluster (0.7–1% nucleotide and 0.7–0.9% amino acid p-distances with AB983218), and strains from 2015 and 2016 grouped into the Kawasaki308-like sub-cluster (0.4–0.9% nucleotide and 0.2–0.6% amino acid p-distances with LC037415). GII.17 was responsible for 4.3, 11.5 and 25.6 (%) of all NoV outbreaks during the periods of July 2014–June 2015, July 2015–June 2016 and July 2016–June 2017, respectively. No GII.17 outbreaks were observed after April 2017.

Several GII.4 strains causing outbreaks in early 2016 could not be classified at the variant level by the genotyping tool based solely on region C sequences. BLAST analysis with additional sequence data from the 3’end of ORF1 (~ 740 nt) revealed a high nucleotide identity (99%) with Kawasaki194-like sequences (GenBank accession number LC175468), previously characterized as GII.P16/GII.4 Sydney recombinant strains [24]. Phylogenetic trees of partial ORF1 and partial ORF2 sequences of this recombinant are shown on Fig. 3. The complete capsid sequences of four GII.P16/GII.4 Sydney recombinants collected between 2016 and 2018, AlbertaEI84/2016, AlbertaEI277/2016, AlbertaEI487/2017 and AlbertaEI231/2018, had 3.3–3.9% nucleotide and 1.5–2.2% amino acid p-distances with the GII.Pe/GII.4 Sydney 2012 reference strain NSW0514/AU/2012 (accession number JX459908) (Additional file 2: Figure S2). Compared with GII.Pe/GII.4 Sydney, the GII.P16/GII.4 Sydney strains did not present distinctive mutations at the P2 domain, however, all GII.P16/GII.4 Sydney sequences in our data set shared two isoleucines at residues 119 and 145, both located at the S domain (Additional file 3: Figure S3). The first GII.P16/GII.4 Sydney outbreak in Alberta occurred in February 2016. Overall, the GII.P16/GII.4 Sydney recombinant was responsible for 21 (27.0%) of 78 NoV outbreaks from July 2015–June 2016, 3 (7.7%) out of 39 NoV outbreaks from July 2016–June 2017 and 44 (62.9%) out of 70 NoV outbreaks from July 2017–February 2018.

In order to investigate the HBGA binding properties of the novel GII.P16/GII.4 Sydney recombinant, we performed binding assays using p-particles of two representative strains, AB-2016-26 and AB-2016-190 and saliva of individuals with known HBGA phenotypes. Both strains demonstrated ability to bind saliva of A, B and O secretors and did not bind saliva of non-secretors (Additional file 4: Figure S4). AB-2016-190 showed stronger binding to secretors than AB-2016-26, but both strains showed no binding to non-secretors, similar to the binding pattern of Syd9-2B, a GII.Pe/GII.4 Sydney strain.

Based on retrospective ORF1-genotyping analysis of GII.2 outbreaks back to November 2011, we identified that the first GII.P16/GII.2 outbreak in Alberta occurred in August 2016. In BLAST analysis, the GII.P16/GII.2 outbreak strain shared highest identity (99%) with Volvic-E15317, a strain collected in France in late 2016 (Fig. 4). GII.P16/GII.2 was the predominant NoV strain between July 2016–June 2017 (13 of 39 NoV outbreaks, 33.3%).

The recombinant strain GII.P4 New Orleans/GII.4 Sydney was identified during the period July 2016–June 2017 (Fig. 3). The strain was responsible for four out of five GII.4 outbreaks from that period and was not observed after June 2017.

Between July 2012 and February 2018, the majority of outbreaks occurred in community long-term care, hospital long-term care, supportive living, and group homes facilities (63/83, 75.9% of all NoV GI and 346/440, 78.6% of all NoV GII outbreaks, respectively) and hospital acute care (5/83, 6% of GI and 53/440,12% of GII outbreaks) (Additional file 5: Figure S5). A larger proportion of GI outbreaks occurred in non-health care settings compared to GII (15/83, 18.1% vs. 41/440, 9.3%, p = 0.0180, Chi-square test).