Norovirus Infections in Symptomatic and Asymptomatic Food Handlers in Japan^▿

Specimens.

Fifty-five outbreaks and 35 sporadic cases of norovirus-associated gastroenteritis occurred at food-catering settings during the two winter periods between 10 November 2005 and 9 December 2006 (Fig. 1). Outbreaks were defined as having both (i) two or more food handlers with symptoms of gastroenteritis, i.e., nausea, vomiting, stomachache, diarrhea, or fever, and (ii) two or more specimens that were positive for norovirus by real-time RT-PCR. Sporadic cases were defined as having (i) only one symptomatic food handler and/or (ii) only one specimen positive for norovirus by real-time RT-PCR (even if two or more food handlers were symptomatic). Outbreak settings included nursing care centers, fast food establishments, hospitals, school canteens, hotels, restaurants, university cafeterias, and a kindergarten (see Tables 1 and 2). Stool specimens were collected from both symptomatic and asymptomatic employees. In total, 2,376 of approximately 7,000 specimens were examined for norovirus by real-time RT-PCR.

Virus detection.

A 10% (wt/vol) stool suspension was prepared with sterilized MilliQ water and centrifuged at 10,000 × g for 10 min. The QIAamp viral RNA minivacuum protocol (Qiagen, Hilden, Germany) was used to extract RNA from 140 μl of the clarified supernatant according to the manufacturer's instructions. Briefly, cDNA synthesis was carried out with 10 μl of the RNA in 20 μl of the reaction mixture containing 50 pmol random hexamer (Takara, Tokyo, Japan), 1× Superscript III reverse transcriptase buffer (Invitrogen, Carlsbad, CA), 10 mM dithiothreitol (Invitrogen), 0.4 mM each deoxynucleoside triphosphate (Roche, Mannheim, Germany), 1 U RNase inhibitor (Toyobo, Tokyo, Japan), and 10 U Superscript reverse transcriptase III (Invitrogen). RT was performed at 37°C for 15 min, followed by 50°C for 1 h. Real-time RT-PCR was performed as previously described, and the cutoff for positive norovirus specimens was set at >10 copies per well (13).

Sequencing and phylogenetic analysis.

Conventional RT-PCR was carried out to sequence the real-time RT-PCR-positive specimens. Briefly, for norovirus GI PCR, primers G1SKF and G1SKR were used, and for norovirus GII PCR, primers G2SKF and G2SKR were used (17). RT-PCR products were excised from the gel and purified with the QIAquick gel extraction kit (Qiagen, Germany). Nucleotide sequences were prepared with the BigDye Terminator cycle sequencing kit (version 3.1) and determined with the ABI 3130 sequencer (ABI, Boston, MA). Nucleotide sequences were aligned with Clustal X, and the distances were calculated by Kimura's two-parameter method. Phylogenetic trees with bootstrap analysis from 1,000 replicas were generated by the neighbor-joining method as described previously (14).

Nucleotide sequence accession numbers.

The accession numbers for sequences determined in this study are GenBank EF630426 to EF630534.

Specimens and screening for norovirus by real-time RT-PCR.

Fifty-five outbreaks and 35 sporadic cases of norovirus-associated gastroenteritis in Japan were examined between 10 November 2005 and 9 December 2006 (Tables 1 and 2; Fig. 1). Most outbreaks occurred at nursing care centers (25 of 55), followed by hospitals (12 of 55), cafeterias (5 of 55), fast food establishments (7 of 55), schools (3 of 55), hotels (2 of 55), and restaurants (1 of 55). Sporadic cases were also found at a number of these settings (Table 2). In total, 2,376 stool specimens were collected from both symptomatic and asymptomatic food handlers at different food-catering settings. These specimens were screened by real-time RT-PCR, and norovirus was detected in 449 of 2,376 (19%) specimens. Real-time RT-PCR can distinguish between norovirus GI and GII sequences, and both GI and GII sequences were detected. Twenty-six of 2,376 (1%) stool specimens were positive for norovirus GI sequences, and 423 of 2,376 (18%) were positive for norovirus GII sequences. Noroviruses were detected in specimens from both symptomatic and asymptomatic food handlers (see Tables 1 and 2).

Genotyping and phylogenetic analysis of norovirus.

To confirm the positive real-time RT-PCR results and determine the genotypes, we reamplified and sequenced the partial capsid gene. We simplified the phylogenetic trees to include only unique sequences (italicized in Table 1, specimen number column); that is, when two or more sequences from the same outbreak had 100% nucleotide similarity, we named a single consensus sequence, and if a sequence had one or more nucleotide mismatches with others in the same outbreak, we gave the sequence a distinct name. All 26 norovirus GI-positive specimens (from 25 outbreak cases and 1 sporadic case) were sequenced, and these clustered into four distinct genotypes: GI/3, GI/4, GI/8, and GI/14 (Fig. 2). One or more norovirus GII-positive specimens in each setting were confirmed by RT-PCR and sequenced. The GII sequences belonged to 12 genotypes, including one new genotype: GII/1, GII/2, GII/3, GII/4, GII/5, GII/6, GII/7, GII/8, GII/9, GII/10, GII/14, and GII/New (Fig. 3).

Molecular epidemiology of outbreaks.

Norovirus GI sequences were detected in 5 of 55 (9%) outbreaks (Table 1). We found 14 of 20 (70%) symptomatic food handlers and 11 of 59 (19%) asymptomatic food handlers positive for norovirus GI strains. Sequence analysis showed that each of the five norovirus GI-associated outbreaks was caused by a single norovirus GI strain; that is, two or more sequences from the same outbreak shared 100% nucleotide similarity. Norovirus GII sequences were detected in 50 of 55 (91%) outbreaks. We found 267 of 364 (73%) symptomatic food handlers and 122 of 1,786 (7%) asymptomatic food handlers positive for norovirus GII strains. In 37 of 50 norovirus GII-associated gastroenteritis outbreaks, a single norovirus GII strain was assumed to be responsible, since two or more sequences from the same outbreak shared 100% nucleotide similarity. In the remainder of the GII-associated outbreaks (13 of 50 outbreaks; settings 3, 8, 9, 10, 17, 23, 34, 38, 48, 49, 51, 62, and 68), several norovirus GII sequences were detected; that is, there were one or more nucleotide changes or different genotypes (Table 1). For example, in the norovirus GII-associated setting 48 (Table 2), we detected three norovirus genotypes (GII/3, GII/5, and GII/6) in stool specimens collected from different symptomatic food handlers in the same hotel between 10 and 16 January 2006, whereas for setting 8, a nursing care center, we detected different GII/4 sequences on 25 November 2006 (specimens 401 and 402) (Fig. 3). Interestingly, symptomatic and asymptomatic food handlers were positive by real-time RT-PCR in 9 of these 13 mixed norovirus GII outbreaks (settings 3, 9, 10, 17, 23, 34, 49, 51, and 68). For example, in the norovirus GII-associated setting 34 (Table 1), we detected three norovirus genotypes (GII/6, GII/4, and GII/2); the former two sequences (specimens 52 and 53) were detected in specimens from symptomatic food handlers, and the latter sequence (specimen 237) was detected in a specimen from an asymptomatic food handler.

GII/4 sequences were predominant, accounting for 20 of 55 (36%) outbreaks, excluding the outbreaks with multiple norovirus genotypes (Fig. 4). A considerable number of outbreaks (7 of 55) were also caused by strains belonging to GII/3, followed by GII/5, which caused 3 outbreaks. Interestingly, in 12 of 13 mixed-GII-genotype outbreaks (settings 3, 8, 9, 10, 17, 23, 34, 38, 49, 51, 62, and 68), we always detected a norovirus GII/4 sequence (Table 1).

Molecular epidemiology of sporadic cases.

Among 35 sporadic cases, norovirus GI sequences were detected in 1 case (3%), which was caused by a GI/4 strain (Table 2). Norovirus GII sequences were detected in 34 of 35 (97%) sporadic cases. The majority of the GII sporadic cases were caused by GII/4 strains (16 of 35), followed by GII/3 (9 of 35). Interestingly, for sporadic cases, only food handlers with symptoms of gastroenteritis were positive for norovirus; 35 of 49 (71%) were positive for GII strains, and 1 of 1 (100%) was positive for GI strains.

Genetic analysis.

The partial capsid sequence was used to describe the genetic diversity of the norovirus sequences. All sequences detected in this study closely matched other, published sequences (Fig. 2 and 3). The GII/4 sequences shared >95% nucleotide similarity, although there was noticeable subclustering within the GII/4 genotype (Fig. 3). Norovirus GII/3 also appeared to have subclusters, while the GII/3 sequences shared >96% nucleotide similarity (Fig. 3). Of interest was the detection of a novel GII genotype (GII/New; sequence 299). As shown in Fig. 5, the amino acid start sequence for the capsid was MRM, whereas all other norovirus GII sequences had MKM. To further investigate this finding, we sequenced the entire capsid gene. Only one other full-length capsid sequence (norovirus strain NLVJ23; accession number GenBank AY130762) closely matched the norovirus 299 sequence, with 99.3% amino acid identity over the entire capsid gene. Interestingly, norovirus NLVJ23 also had the unusual amino acid start sequence of MRM for the capsid. The N-terminal regions of the capsids of these two sequences were quite unlike that of any other norovirus GII sequence. In addition, both norovirus 299 and NLVJ23 had an amino acid insertion at the 10th residue, whereas no other norovirus GII sequences had any insertions or deletions within the first 149 amino acids of the capsid gene (Fig. 5).

Analysis of norovirus loads.

The real-time RT-PCR results were used to analyze the viral loads by genotype and genogroup (Fig. 6). All positive sequences (not just the consensus sequences shown in the trees) were used for this analysis. The numbers of positive specimens examined for each genotype were as follows: 6 for GI/3, 6 for GI/4, 3 for GI/8, 4 for GI/14, 6 for GII/1, 7 for GII/2, 35 for GII/3, 96 for GII/4, 13 for GII/5, 11 for GII/6, 2 for GII/7, 3 for GII/8, 1 for GII/9, 3 for GII/10, 2 for GII/14, and 1 for GII/new (Fig. 6). Overall, norovirus GII had a slightly higher mean viral load (1 log unit higher) than norovirus GI (3.81 × 10⁸ versus 2.79 × 10⁷ copies/g of stool) (Fig. 6). The highest viral load for norovirus GI was 2.02 × 10⁸ copies/g of stool for GI/4, and the highest viral load for GII was 7.96 × 10⁹ copies/g of stool for GII/4 (data not shown).

A comparison between the mean viral loads of symptomatic and asymptomatic individuals found that GI-infected symptomatic individuals had a slightly higher mean viral load than GI-infected asymptomatic individuals (4.43 × 10⁷ versus 3.79 × 10⁶ copies/g of stool), whereas GII-infected symptomatic and asymptomatic individuals had similar mean viral loads (3.31 × 10⁸ versus 5.53 × 10⁸ copies/g of stool). Our results also showed that GI/4-infected symptomatic individuals had a slightly higher mean viral load than GI/4-infected asymptomatic individuals (6.73 × 10⁷ versus 6.34 × 10⁶ copies/g of stool), whereas GII/4-infected symptomatic individuals had a mean viral load similar to that of GII/4-infected asymptomatic individuals (2.17 × 10⁸ versus 6.58 × 10⁸ copies/g of stool). Comparisons between the other genotypes were difficult to perform because the asymptomatic individuals were not infected with those genotypes (data not shown).