Metagenomic Surveillance of Viral Gastroenteritis in a Public Health Setting

Abstract

Norovirus is the primary cause of viral gastroenteritis (GE). To investigate norovirus epidemiology, there is a need for whole-genome sequencing and reference sets consisting of complete genomes. To investigate the potential of shotgun metagenomic sequencing on the Illumina platform for whole-genome sequencing, 71 reverse transcriptase quantitative PCR (RT-qPCR) norovirus positive-feces (threshold cycle [C_T], <30) samples from norovirus surveillance within The Netherlands were subjected to metagenomic sequencing. Data were analyzed through an in-house next-generation sequencing (NGS) analysis workflow. Additionally, we assessed the potential of metagenomic sequencing for the surveillance of off-target viruses that are of importance for public health, e.g., sapovirus, rotavirus A, enterovirus, parechovirus, aichivirus, adenovirus, and bocaparvovirus. A total of 60 complete and 10 partial norovirus genomes were generated, representing 7 genogroup I capsid genotypes and 12 genogroup II capsid genotypes. In addition to the norovirus genomes, the metagenomic approach yielded partial or complete genomes of other viruses for 39% of samples from children and 6.7% of samples from adults, including adenovirus 41 (N = 1); aichivirus 1 (N = 1); coxsackievirus A2 (N = 2), A4 (N = 2), A5 (N = 1), and A16 (N = 1); bocaparvovirus 1 (N = 1) and 3 (N = 1); human parechovirus 1 (N = 2) and 3 (N = 1); Rotavirus A (N = 1); and a sapovirus GI.7 (N = 1). The sapovirus GI.7 was initially not detected through RT-qPCR and warranted an update of the primer and probe set. Metagenomic sequencing on the Illumina platform robustly determines complete norovirus genomes and may be used to broaden gastroenteritis surveillance by capturing off-target enteric viruses. IMPORTANCE Viral gastroenteritis results in significant morbidity and mortality in vulnerable individuals and is primarily caused by norovirus. To investigate norovirus epidemiology, there is a need for whole-genome sequencing and reference sets consisting of full genomes. Using surveillance samples sent to the Dutch National Institute for Public Health and the Environment (RIVM), we compared metagenomics against conventional techniques, such as RT-qPCR and Sanger-sequencing, with norovirus as the target pathogen. We determined that metagenomics is a robust method to generate complete norovirus genomes, in parallel to many off-target pathogenic enteric virus genomes, thereby broadening our surveillance efforts. Moreover, we detected a sapovirus that was not detected by our validated gastroenteritis RT-qPCR panel, which exemplifies the strength of metagenomics. Our study shows that metagenomics can be used for public health gastroenteritis surveillance, the generation of reference-sets for molecular epidemiology, and how it compares to current surveillance strategies.

Keywords: enteric viruses; gastroenteritis; next-generation sequencing; norovirus; public health; surveillance; surveillance studies; transmissible gastroenteritis virus.

FIG 1

Patient age per detected virus species or norovirus genotype. The left light-gray rectangle shows young children (<5 years old), and the right one shows the elderly (>65 years old). The histogram depicts the total number of detected viruses of public health relevance per age group, with bins of 2.5 years. *, Norovirus GII.17[P25] (N = 1) and GII.4[P4] (N = 1) were present in the same sample and could not be plotted due to unknown patient age. N, total number of times a virus genome was observed.

FIG 2

Percentage of genome completeness of norovirus and off-target GE viruses versus C_T-values. Genome completeness was determined for a minimum 3× depth of coverage.

FIG 3

Example depth of coverage plot of norovirus (NoV) GII strains in sample R02-13. Two strains were present in this sample, a GII.17[P17] norovirus strain (top) with high depth of coverage (DoC) and a GII.4[P31] norovirus strain (bottom) with lower DoC, denoted “major” and “minor,” respectively. The characteristic block-like high coverage peaks near the ORF1/2 junction of the minority strain are due to homology between the strains in these regions. The aligned reads of the majority strain adversely affect the consensus sequence of the minority strain, necessitating manual correction. A full overview of all norovirus GI, GII, and off-target viruses with public health relevance is shown in Fig. S1 to S4. The coverage is shown as a 1 + log₁₀ transformation with a maximum 999× depth of coverage.

FIG 4

Maximum likelihood tree of 61 GII strains. (A) Maximum likelihood trees were inferred for ORF1 based on 5,166 nucleotides using the GTR best-fit model. (B) ORF2 based on 1,689 nucleotides using the TIM2 best-fit model. Reference strains from Chhabra et al. (11) are shown in red; clusters are annotated with numbered arrows (Table 1). The scale bar represents nucleotide substitutions per site, and selected bootstrap values greater than 70 are shown. For samples containing multiple strains, the lowest depth of coverage strain is denoted with “minor” in the tip label (e.g., “R01-04-minor”). All GII.4 and GII.P4 strains were Sydney and New Orleans variants, respectively.

FIG 5

Maximum likelihood tree of 9 GI strains. (A and B) Maximum likelihood trees were inferred for (A) ORF1 based on 5,412 nucleotides using the TIM2 best-fit model and (B) ORF2 based on 1,665 nucleotides using the TIM2 best-fit model. Genotype reference strains from Chhabra et al. (11) are shown in red, while additional references from NCBI GenBank are shown in blue. The scale bar represents nucleotide substitutions per site, and selected bootstrap values greater than 70 are shown. In sample R04-15, two strains were detected, with the strain of lower coverage denoted as “minor” in the tip label.