Genetic diversity of human sapovirus across the Americas

Abstract

Background:

Sapoviruses are responsible for sporadic and epidemic acute gastroenteritis worldwide. Sapovirus typing protocols have a success rate as low as 43% and relatively few complete sapovirus genome sequences are available to improve current typing protocols.

Objective/study design:

To increase the number of complete sapovirus genomes to better understand the molecular epidemiology of human sapovirus and to improve the success rate of current sapovirus typing methods, we used deep metagenomics shotgun sequencing to obtain the complete genomes of 68 sapovirus samples from four different countries across the Americas (Guatemala, Nicaragua, Peru and the US).

Results:

VP1 genotyping showed that all sapovirus sequences could be grouped in the four established genogroups (GI (n = 13), GII (n = 30), GIV (n = 23), GV (n = 2)) that infect humans. They include the near-complete genome of a GI.6 virus and a recently reported novel GII.8 virus. Sequences of the complete RNA-dependent RNA polymerase gene could be grouped into three major genetic clusters or polymerase (P) types (GI.P, GII.P and GV.P) with all GIV viruses harboring a GII polymerase. One (GII.P-GII.4) of the new 68 sequences was a recombinant virus with the hotspot between the NS7 and VP1 regions.

Conclusions:

Analyses of this expanded database of near-complete sapovirus sequences showed several mismatches in the genotyping primers, suggesting opportunities to revisit and update current sapovirus typing methods.

1. Background

Since the introduction of rotavirus vaccines, norovirus has become the leading cause of medically-attended acute gastroenteritis (AGE) in many countries including the US [1,2]. Of the other viruses associated with AGE, sapoviruses have increasingly been detected in endemic and epidemic AGE [3–6]. Using real-time RT-PCR, studies from low-, middle-, and high-income countries have shown that the prevalence of sapovirus in children < 5 years of age ranges from 3.3 to 17% [7–13]. Although earlier reports described sapovirus infection as one with less severe clinical symptoms than norovirus and rotavirus [14], more recent studies have shown that infections with sapovirus can result in hospitalizations and severe dehydration [9,15]. Sapoviruses were first detected by electron microscopy in children of an infant home with acute gastroenteritis in October 1977 in Japan. The name sapovirus refers to the well-studied prototype strain, Sapporo virus, from another AGE outbreak in the Sapporo prefecture in Japan in 1982. Initially classified as typical caliciviruses or Sapporo-like viruses by electron microscopy, sequencing of the complete genome showed that these non-enveloped 30–35 nm viruses belong to a separate genus Sapovirus in the family Caliciviridae [16]. Sapoviruses have a positive-sense, single stranded RNA genome of 7.3–7.5 kb in length which contains two open reading frames (ORFs). Cleavage of the ORF1 polyprotein by the virus-encoded 3C-like cysteine proteinase yields the mature non-structural (NS) proteins (NS1–NS7) including the RNA-dependent RNA polymerase (RdRp or NS7) as well as the major capsid protein VP1, whereas ORF2 encodes for a minor capsid protein VP2 [17]. The antigenic epitopes are in the hypervariable region of VP1, which is the most diverse region of the genome, likely in the predicted P2 domain [18]. Based on complete VP1 nucleotide sequences, sapoviruses can be divided into up to 19 genogroups (GI–GXIX) [19] of which viruses from 4 (GI, GII, GIV and GV) infect humans while viruses in the other genogroups have been detected in swine (GIII and GV-GXI), sea lions (GV), mink (GXII), dogs (GXIII), bats (GXIV, GXVI-GXIX) and rats (GXV). The human sapoviruses can be further classified into 17 genotypes [17], and an additional proposed GII.8 genotype [20]. GI and GII sapoviruses are the most frequently detected viruses in recent years. Although GIV viruses are relatively rare, in some studies they accounted for up to one third of the cases [21]. They have been detected worldwide in stool samples from young children as early as 1992 in Pakistan [22]. GV viruses were first reported in stool samples from children in Argentina in 1995 [23] but in most studies viruses of this genogroup have been rarely reported [24]. GI.6 and GII.3 viruses became predominant in Japan, but these genotypes gradually disappeared in subsequent years [25]. Several recombinant sapovirus strains have been documented [26] and, like for noroviruses, such strains may have an altered virulence possibly leading to an increased disease burden [27]. Since 2006, most research groups have used the same protocols for the detection [28] and genotyping [29] of human sapoviruses. The real-time RT-PCR assay described by Oka et al. [28] is able to detect viruses from all 4 genogroups. However, reported genotyping success rates for sapovirus range from 43% to 100% [30,31] indicating the current hemi-nested PCR assays do not detect all circulating sapovirus strains.

2. Objective

With protocols for sequencing of the complete genome of enteric viruses directly from stool samples becoming more widely available [32], the aim of our study was to increase the number of complete sapovirus genomes especially for those genotypes for which currently no or only few complete genomic sequences are available in public databases. A larger sequence database will help to develop more broadly-reactive PCR assays with overall improved genotyping success to better understand the molecular epidemiology of human sapoviruses.

3. Study design

3.1. Fecal specimens

Sapovirus positive stool specimens used in this study were obtained from outbreaks or sporadic cases of AGE and collected between 2010 and 2016. CDC’s Internal Program for Research Determination deemed that this study is categorized as public health non-research and that human subject regulations did not apply. Specimens representing rare or uncommon genotypes, or strains for which no complete genomes were available in public databases were selected for whole genome sequencing. In addition, specimens that could not be amplified using the hemi-nested PCR assay were included [29]. All specimens, except four samples collected during an outbreak in a long-term care facility in 2016 in the United States, were collected from children under 5 years of age with sporadic AGE. Sapovirus positive specimens in the US were obtained from two sites (Nashville, Oakland) participating in New Vaccine and Surveillance Network [33] and from an all-age active surveillance study for medically–attended acute gastroenteritis in Oregon. The Peruvian specimens were obtained from children hospitalized at the Instituto Nacional de Salud del Niño in Lima, Peru. The specimens from Nicaragua were from a population-based study in 2010 and had been tested for Shigella, Salmonella, E.coli, Campylobacter, Cryptosporidium, rotavirus, adenovirus, and norovirus [7]. The specimens from Guatemala had been collected as part of an AGE health facility-based surveillance in two Guatemalan departments (Santa Rosa and Quetzaltenango) from April 2010 to February 2016. Table 1 summarizes the number of complete sapovirus genomes (i.e., sequences containing the 5′ UTR, the complete ORF1 and ORF2 and the 3′UTR regions) in public databases (as of April 1st, 2018) and the complete genomes obtained in this study.

Table 1.

Distribution of near-complete sapovirus genomes sequenced in this study compared to complete genomes available in NCBI.

Genotype	This study	NCBIa
	n	n
GI.1	9	10
GI.2	3	3
GI.3	0	1
GI.4	0	1
GI.5	0	1
GI.6	1	1
GI.7	0	1
GII.1	10	2
GII.2	4	1
GII.3	6	3
GII.4	1	1
GII.5	5	1
GII.6	0	1
GII.7	0	1
GII.8b	4	4
GIV.1	23	5
GV.1	2	2
GV.2	0	2
total	68	41

^aNCBI = National Center of Biotechnology Information, complete genomes available on April 1st 2018.

^bProposed new genotype.

3.2. RNA extraction, sapovirus detection by RT-qPCR and VP1 genotyping

Clarified 10% stool suspensions were prepared in phosphate-buffered saline and centrifuged at 10,000 × g for 10 min. Viral nucleic acid was extracted using the MagMAX total nucleic acid isolation Kit (Thermo Fisher Scientific, Carlsbad, CA). Sapovirus RNA was detected by TaqMan-based quantitative RT-PCR as described previously [28]. VP1 subtyping was performed by hemi nested RT-PCR followed by Sanger sequencing of the 2nd round PCR positive products [29].

3.3. Full genome sequencing, de novo assembly, and U₅₀ N₅₀ metrics calculation

Viral metagenomics were performed according to a previously published protocol [34–36]. Briefly, virus particles were filtered using 0.45 μm centrifugal filters (Millipore, Billerica, MA), followed by a nuclease treatment consisting of a cocktail of RNase A and TURBO^™ DNase (Thermo Fisher Scientific, Carlsbad, CA), Baseline-ZERO^™ (Epicentre, Madison, WI), and Benzonase (Sigma-Aldrich, St. Louis, MO). Viral nucleic acids were extracted using QIAamp^® Viral RNA Mini Kit (QIAGEN, Hilden, Germany). Complementary DNA (cDNA) synthesis and random amplification were performed by sequence-independent single primer amplification (SISPA) [35–37]. PCR products were purified using Agencourt^® AMPure^® XP beads (Beckman Coulter, Brea, CA). An approximate 200-bp fragment library was constructed using the Nextera^® XT DNA Library Preparation Kit (Illumina, San Diego, CA). The Nextera^® product was purified using Agencourt^® AMPure^® XP beads (Beckman Coulter, Brea, CA) and quality of the purified library was assessed on an Agilent HS D1000 ScreenTape System (Agilent Technologies, Santa Clara, CA). Library concentration for pooling was determined with a KAPA Library Quantification Kit for Illumina^® platforms (Roche, Wilmington, MA). Samples were sequenced on an Illumina MiSeq using MiSeq Reagent Kits v2 (250-cycle paired-end). Full-length sapovirus genomic sequences were generated using a custom bioinformatics pipeline as described previously [38,39]. Briefly, sequences were trimmed/filtered to remove adapters and low quality bases, sequences shorter than 50 nt, and human (host) sequences identified through mapping of reads to the human reference genome hg19 using bowtie2 [40]. Sapovirus were first assembled using SPAdes [41], a de novo assembler, followed by reference mapping and gene annotation using Geneious version 9.1.4 [Biomatters] [42] to verify assembled sequences. For all de novo assembled sequences the new metric U50 was calculated which is circumventing the limitations of N50 by identifying unique, target-specific contigs using a reference sequence as a baseline [43].

3.4. Phylogenetic and genome similarity analyses

Sequences alignment was performed with MUSCLE [44] and phylogenetic trees were constructed using the maximum-likelihood method with 100 bootstraps replications to assess phylogenetic robustness using MEGA version 6 [45]. Using the Sequence Demarcation tool [46] we aligned every unique pair of sequences and calculated the sequence pairwise identity among sequences from this study and published references. Possible recombination events were analyzed using SimPlot [47].

4. Results

Of the 108 sapovirus positive samples that were tested by deep metagenomics shotgun sequencing, we successfully obtained near-complete genome sequences for 68 (63%) strains of which 44 had UG50% values of over 99% indicating that they were successfully obtained during de novo assembly by the bioinformatics pipeline [43] (Table 2). Of the remaining 40 samples, 26 sequences were incomplete and/or had gaps and/or had too few reads by NGS to be confidently assembled and assigned. The other 14 samples had a complete VP1 sequence which allowed genotyping, but since no complete genomes was obtained, they were not further analyzed in this study. All 68 (near-)complete genomes belonged to the four established sapovirus genogroups (GI, GII, GIV and GV) and included 13 GI, 30 GII, 23 GIV and 2 GV sequences were typed based on VP1 sequences. They included 12 sequences from rare sapovirus genotypes, GI.6 [n = 1], GII.3 [n = 6], GII.4 [n = 1] and GII.5 [n = 4]. In addition, based on a pairwise distance cut-off value of ≤0.169 to distinguish different genotypes and ≤0.488 to distinguish different genogroups [17], four sequences could be classified into a recently reported new genotype, GII.8 (Figs. 1 and 2B) [20]. These sequences shared only 75% nucleotide identity with its closest neighbor, GII.7 viruses.

Table 2.

Near-complete genomes of 68 sapovirus strains generated in this study.

Accession Number	Sample name	Ct	Genotype	Sequence length	U50a	N50	UG50	NG50	UG50%
MG012441	GI.2_Oakland_6371	25	GI.2	7465	4100	235	4100	6023	54.92%
MG012442	GI.2_Oakland_3023	19	GI.2	7457	7457	1473	7457	7496	99.99%
MG012399	GI.1_Portland_3639	24	GI.1	7430	7420	505	7420	9503	99.87%
MG012425	GIV.1_Portland_3587	23	GIV	7359	5520	836	5520	5520	75.01%
MG012428	GIV.1_Portland_3602	25	GIV	7361	1938	826	1938	6497	26.13%
MG012424	GIV.1_Portland_3629	21	GIV	7420	7399	548	7399	7399	99.69%
MG012459	GIV.1_Portland_3631	25	GIV	7414	7390	779	7390	7392	99.68%
MG012423	GIV.1_Portland_3632	28	GIV	7358	4690	1059	4690	4778	63.68%
MG012427	GIV.1_Portland_3633	25	GIV	7383	7339	1024	7339	7352	99.15%
MG012426	GIV.1_Portland_3636	25	GIV	7329	7311	865	7311	7311	99.75%
MG012457	GIV.1_Portland_3641	26	GIV	7420	7420	522	7420	7420	100.00%
MG012458	GIV.1_Portland_3643	23	GIV	7412	7412	829	7412	7420	100.00%
MG012422	GIV.1_Portland_3644	24	GIV	7387	7350	1077	7350	7350	99.50%
MG012456	GIV.1_Portland_3649	23	GIV	7413	7409	747	7409	7409	99.82%
MG012455	GIV.1_Portland_3651	21	GIV	7426	7404	1073	7404	7430	99.66%
MG012397	GI.1_Nashville_9427	20	GI.1	7356	7356	491	7356	7394	100.00%
MG012400	GI.1_Nashville_9491	22	GI.1	7346	6398	526	6398	8745	87.10%
MG012435	GI.1_Nashville_9691-1	20	GI.1	7425	7421	800	7421	7520	99.95%
MG012437	GI.1_Nashville_9432	19	GI.1	7388	7378	728	7378	7411	99.86%
MG012438	GI.1_Nashville_9478	19	GI.1	7428	655	540	0	6151	0.00%
MG012440	GI.2_Nashville_9411	22	GI.2	7449	7449	769	7449	10174	99.99%
MG012443	GI.6_Nashville_9367	17	GI.6	7430	7427	575	7427	10982	99.96%
MG012444	GII.1_Nashville_9343	15	GII.1	7483	2115	192	2115	2253	28.26%
MG012445	GII.1_Nashville_9346	15	GII.1	7394	379	269	379	1856	5.13%
MG012402	GII.1_Nashville_9360	14	GII.1	7469	7469	2852	7469	7592	100.00%
MG012403	GII.1_Nashville_9422	20	GII.1	7464	7464	517	7464	7492	100.00%
MG012404	GII.1_Nashville_9353	18	GII.1	7363	7363	385	7363	7410	100.00%
MG012405	GII.1_Nashville_9511	21	GII.1	7417	7416	619	7416	7457	99.99%
MG012406	GII.1_Nashville_9510	26	GII.1	7463	885	403	885	5949	11.86%
MG012407	GII.1_Nashville_9525	18	GII.1	7448	7446	1086	7446	16746	99.97%
MG012408	GII.1_Nashville_9691-2	20	GII.1	7148	4031	800	4031	7520	56.39%
MG012409	GII.1_Nashville_9385	14	GII.1	7388	7386	7553	7386	7553	99.97%
MG012410	GII.2_Nashville_9521	22	GII.2	7418	4360	560	4360	5162	58.78%
MG012411	GII.2_Nashville_9331	20	GII.2	7375	7374	505	7374	7494	99.99%
MG012412	GII.2_Nashville_9393	18	GII.2	7429	7427	655	7427	7512	99.97%
MG012413	GII.2_Nashville_9795	22	GII.2	7372	7370	1386	7370	7887	99.97%
MG012415	GII.3_Nashville_9517	27	GII.3	7452	485	451	0	2343	0.00%
MG012418	GII.3_Nashville_9513	16	GII.3	7464	7464	418	7464	7544	100.00%
MG012419	GII.3_Nashville_9354	20	GII.3	7439	7424	523	7424	7434	99.80%
MG012447	GII.5_Nashville_9349	23	GII.5	7383	6187	533	6187	6238	83.80%
MG012449	GII.5_Nashville_9387	20	GII.5	7432	7432	1373	7432	7451	100.00%
MG012430	GIV.1_Nashville_9327	22	GIV	7348	7348	660	7348	7374	100.00%
MG012433	GV.1_Nashville_9424	26	GV.1	7455	4929	573	4929	5399	66.14%
MG012434	GV.1_Nashville_9492	25	GV.1	7499	7498	1045	7498	7503	99.99%
MG012421	GIV.1_Nashville_9489	20	GIV	7415	7415	1275	7415	7568	100.00%
MG012453	GII.8_Monterey_6283	ND	GII.8	7404	7403	714	7403	7426	99.99%
MG012452	GII.8_Monterey_6284	ND	GII.8	7447	7446	842	7446	7829	99.99%
MG012398	GI.1_Lima_1845	NA	GI.1	7311	7311	474	7311	7311	100.00%
MG012436	GI.1_Lima_1870	NA	GI.1	7424	7421	681	7421	7456	99.96%
MG012439	GI.1_Lima_1863	NA	GI.1	7429	774	254	774	5139	10.42%
MG012417	GII.3_Lima_1856	NA	GII.3	7454	7454	508	7454	7454	100.00%
MG012446	GII.4_Lima_1873	NA	GII.4	7398	253	254	252	3209	3.41%
MG012451	GII.5_Lima_1868	NA	GII.5	7434	355	224	355	1958	4.78%
MG674584	GII.8_Lima_1859	NA	GII.8	7450	321	221	321	1150	4.31%
MG674583	GII.8_Lima_1867	NA	GII.8	7300	4593	473	4593	4610	62.91%
MG012420	GIV.1_Lima_1858	NA	GIV	7418	7298	545	7298	7298	98.38%
MG012429	GIV.1_Lima_1861	NA	GIV	7389	7389	514	7389	7414	100.00%
MG012431	GIV.1_Lima_1869	NA	GIV	7402	7398	462	7398	7402	99.99%
MG012432	GIV.1_Lima_1871	NA	GIV	7387	924	462	276	6039	3.74%
MG012461	GIV.1_Lima_1862	NA	GIV	7437	7437	750	7437	7439	100.00%
MG012462	GIV.1_Lima_1864	NA	GIV	7415	254	340	245	1280	3.30%
MG012463	GIV.1_Lima_1865	NA	GIV	7437	7434	644	7434	7524	99.96%
MG012414	GII.3_Leon_6701	19	GII.3	7453	3135	246	2165	7415	29.04%
MG012416	GII.3_Leon_6708	20	GII.3	7454	7452	783	7452	8083	99.97%
MG012454	GIV.1_Leon_1751	22	GIV	7401	7399	600	7399	7469	99.97%
MG012448	GII.5_Santa_Rosa_3812	21	GII.5	7447	7447	850	7447	7574	100.00%
MG012450	GII.5_Santa_Rosa_3693	21	GII.5	7448	499	250	434	4756	5.83%
MG012460	GIV.1_Quetzaltenago_3711	25	GIV	7436	7425	406	7425	7577	99.85%

ND: not detected; NA: not available.

^aThe performance of a de novo assembly method is measured by N50. U₅₀ is a metric that identifies unique, target-specific non-overlapping contigs by using a reference genome as baseline, aiming at circumventing limitations that are inherent to the N₅₀ metric.

Fig. 1.

Pairwise sequence comparison of novel sapovirus strains (GII.8 Monterey6283, GII.8 Monterey6284, GII.8 Lima1859, and GII.8 Lima1867) with sapovirus GII.1-GII.7reference sequences. In the color-coded pairwise identity matrix each cell includes the percentage identity among two sequences (horizontally to the left and vertically at the bottom). The colored key indicates the correspondence between pairwise identities and the colors displayed in the matrix.

Fig. 2.

Phylogenetic analysis of 99 sapovirus RdRp (A) and 70 VP1 (B) sequences using Maximum Likelihood method. The tree was inferred by using the Maximum Likelihood method based on the Tamura-Nei model. The trees with the highest log likelihood are shown, on the basis of nucleotide sequence of the complete RdRp (A) and VP1 (B) genes. The trees are drawn to scale with branch lengths measured in the number of substitutions per site. Bootstrap values (100 replicates) are shown next to each branch. Branches have been compressed for clarity and only the GII branch is partially expanded to show how the four samples of GII.8 branch out from the closest GII.7 sapovirus.

Phylogenetic analyses of the complete RdRp gene (approximately 1550 nucleotides) [17] of 99 sapovirus strains, including 68 from this study, showed that they can be grouped into 3 main clusters or polymerase types (P-types: GI.P, GII.P and GV.P) which are distinguished by having less than 43% nucleotide identity (Fig. 2A). GI.P includes all GI viruses based on VP1, GII.P includes all GII and GIV viruses including the new GII.8 and GV.P includes all GV viruses.

The non-structural (NS) protein sequences in ORF1, contained previously recognized conserved motifs in the first five amino acids of NS1 (MASKP) and around the RdRp-VP1 junction region [NS7-VP1] cleavage site (FEME/G, the slash indicates the putative cleavage site by viral protease NS6) in all strains [48]. The rest of non-structural proteins biological functions are not yet experimentally determined, but NS3 and NS5 have typical calicivirus NTPase and VPg motifs, respectively. Conserved amino acids motifs including (G(A/P)PGIGKT) in NS3, (KGKTK and DDEYDE) in NS5, (GDCG) in NS6, (WKGL, KDEL, DYSKDST, GLPSG and YGDD) in NS7 and (PPG and GWS) in VP1 were, with some minor variations, present in all 68 sapovirus sequences obtained in this study. These minor variations were found in the NS3 and NS5 genes. In NS3, the GAPGIGKT motif was present only in GI.1 sequences whereas all other strains, irrespective of genogroup, had the GPPGIGKT motif. In NS5, all sequences had the common KGKTK motif while GI.2 viruses had KGKSK and the novel GII.8 strains had a KGKNK motif.

We identified an RdRp-VP1 recombinant (GII.P-GII.4_Lima_1873) (Fig. 3) with the recombination break point at the RdRp-VP1 junction. No other recombinant viruses were identified throughout ORF1.

Fig. 3.

Evidence of recombination for sapovirus GII.4 Lima1873 compared to previous sequenced GII.1 and GII.4 genomes. Sapovirus genomes were analyzed by SimPlot and similarity scores using a 200 nt sliding window are plotted. Three reference strains (A, B and C) were analyzed against the query sequence GII.4_Lima1873. Similarity scores of all three references (two GII.1 and one GII.4) were almost > 85% in the entire NS region, with a notable drop in similarity scores over the capsid region and ORF2 (VP2). The query sequence most closely related to previously described sapovirus recombinant strain with a GII.P polymerase and GII.4 capsid (KP067444). Red boxes indicate the approximate location of the hemi nested genotyping primers [29].

We identified several mismatches between the widely used genotyping primers published a decade ago [29] and 134 VP1 sequences including 68 sequences from this study. Several GII viruses, including the new GII.8, and GIV viruses had multiple mismatches in the primers used in the hemi-nested typing PCR (Fig. 4).

Fig. 4.

Alignment of 68 near complete genome sequences obtained in this study and the primers used for hemi-nested PCR [29]. Mismatches between individual strains and the primer binding regions are highlighted in yellow.

^aNucleotide positions based on Manchester strain (Accesion number X86560).

4.1. Nucleotide sequence accession numbers

The sapovirus nucleotide sequences determined in this study have been deposited in GenBank under the accession numbers MG012397-MG012400; MG012402-MG012463 and MG674583- MG674584.

5. Discussion

To better understand the magnitude of genetic variability of human sapoviruses and to increase the available sequences to improve the success rate of current genotyping protocols, we sequenced (near-) complete genomes of 68 sapovirus strains from four different countries including several rare genotypes. We used the complete VP1 nucleotide sequences to classify strains into established sapovirus genogroups and genotypes [17,23]. Samples from an AGE outbreak in a long-term care facility in the US and from two hospitalized children in Peru could be typed as GII.8, a new genotype which was recently reported [20,49]. Due to the limited number of samples from adults (2 samples from an outbreak in a long term care facility), we cannot conclude if adults are infected by different genotypes then young children which is an important research area to be addressed in future studies. A major contribution of the current study is the addition of 68 near-complete genome sequences to the current 41 complete sapovirus genomes in GenBank. Two types of recombination events have been described including inter-genogroup and intra-genogroup [17]. In the current study one intra-genogroup recombinant (GII.1/GII.4) event was observed among the 68 complete sapovirus genomes analyzed, similar to previously described recombinant viruses in Vietnam [50] and in the Philippines [26]. Other intra-genogroup recombinations have also been also described in sapovirus GI (GI.1/GI.8) in Japan [51] consistently with the breakpoint located between the RdRp and the VP1 genes. Among caliciviruses, RdRp is the most conserved region of the genome and coinfection with multiple sapovirus strains may lead to the emergence of recombinant strains [52,53]. Nevertheless, the frequency of recombination observed in sapovirus is lower than in viruses in the closely related genus Norovirus, in which this phenomenon occurs frequently [27]. This can be partially explained because of the relatively few number of RdRp sequences available for sapovirus compared to norovirus. Thus, the large number of sequences provided in this study allow for a better assessment of the frequency of recombination among sapoviruses.

The heminested PCR for typing of sapoviruses was also designed in 2006 based on the limited number of sequences available at that time. Genotypes such as GII.3, and GII.8 have up to 5 mismatches with the SV-R2 reverse primer but design of more broadly-reactive typing primers is compromised by the large genetic variability of GII strains.

In summary, we obtained full and near-complete sapovirus genome sequences of 68 human stool samples from four different countries in the Americas using deep metagenomics shotgun sequencing. Further optimizations of the metagenomics shotgun protocol may be needed to consistently obtain complete genomes from samples with a lower viral load either by using virus-specific RNA baits to enrich sapovirus nucleic acids prior to sequencing or sequence-specific amplification of complete sapovirus genomes. We showed that the current oligonucleotide primers used for genotyping of human sapoviruses [29] have mismatches that are less likely to be successful for amplifying GII sapoviruses, but optimization of amplification conditions of the current protocol may increase the success rate (data not shown). Continued surveillance of sapovirus from different geographic regions using improved detection and typing protocols will help to better determine the disease burden and genotype diversity of these, until recently, underappreciated viral gastrointestinal infections.