Novel porcine-like human G26P[19] rotavirus identified in hospitalized paediatric diarrhoea patients in Ho Chi Minh City, Vietnam

During a hospital-based diarrhoeal disease study conducted in Ho Chi Minh City, Vietnam from 2009 to 2010, we identified four symptomatic children infected with G26P[19] rotavirus (RV) – an atypical variant that has not previously been reported in human gastroenteritis. To determine the genetic structure and investigate the origin of this G26P[19] strain, the whole genome of a representative example was characterized, revealing a novel genome constellation: G26–P[19]–I5–R1–C1–M1–A8–N1–T1–E1–H1. The genome segments were most closely related to porcine (VP7, VP4, VP6 and NSP1) and Wa-like porcine RVs (VP1–3 and NSP2–5). We proposed that this G26P[19] strain was the product of zoonotic transmission coupled with one or more reassortment events occurring in human and/or animal reservoirs. The identification of such strains has potential implications for vaccine efficacy in south-east Asia, and outlines the utility of whole-genome sequencing for studying RV diversity and zoonotic potential during disease surveillance.

We conducted a prospective multicentre hospital-based study of acute diarrhoea in children during 2009-2010, Vietnam. This study was conducted according to the principles expressed in the Declaration of Helsinki, and was approved by the institutional ethical review boards of all participating hospitals and the Oxford Tropical Research Ethics Committee (My et al., 2013). Stool samples from symptomatic and asymptomatic children were collected and screened, as described previously, for multiple aetiological agents, including parasites by microscopy, bacteria (Shigella, Salmonella, Campylobacter, Yersinia, Pleisiomonas and Aeromonas) by microbiological culture and viruses (norovirus and RV) by reverse transcription (RT)-PCR amplification (My et al., 2013). For RV identification and genotyping, RT-PCR was performed on total dsRNA from stool samples to detect the RV capsid genes, VP7 and VP4, followed by direct sequencing of the PCR amplicons for confirmation and genotyping (My et al., 2011). G and P types were determined via the group A RV genotyping tool, RotaC v2.0 (http://rotac.regatools.be), according to the guidelines of the Rotavirus Classification Working Group (Maes et al., 2009).
Over the study period (from May 2009 to April 2010), RV was detected in the stool samples of 664 of 1419 (46.8 %) children with acute diarrhoea, of which 590 (88.4 %) patients were infected with RV only, and none of the other diarrhoeal pathogens screened for were identified. Sequence analysis of the RV-positive samples revealed a diverse array of genotypes, including a novel G26P[19] genotype in four children with symptomatic disease. The ages of the four patients infected with a G26P[19] strain ranged from 14 to 24 months (Table S1, available in the online Supplementary Material), slightly higher than the median age of RV infection caused by other genotypes (13 months; interquartile range: 9-19 months). The geographical and temporal distribution of the four G26P [19] infections suggested that these cases were not associated epidemiologically, making it difficult to determine whether some or all of these cases resulted from a human-to-human transmission chain or represented multiple jumps from non-human hosts to humans (Table S1). We found none of the other screened diarrhoeal pathogens in the stool samples of these four children and therefore assumed that G26P[19] RV was the causative agent of diarrhoeal episodes observed in all cases. These G26P[19] RV infections were detected at locations across HCMC at different time points throughout enrolment, which suggested that additional undetected G26P[19] infections may have occurred in the broader population during this period. As this study was hospital-based, milder G26P[19] infections in the community may have gone unreported or presented to alternative healthcare facilities in the city. Furthermore, as sampling was conducted over a period of 1 year, we were unable to determine whether this atypical genotype continued to circulate locally after the period of investigation.
In accordance with recent changes in RV classification (Matthijnssens et al., 2011) and to investigate the potential origins of the strain, we characterized all 11 gene segments of a representative G26P[19] strain to determine its genome constellation. Genome sequencing was performed on the strain RVA/Human-wt/VNM/30378/2009/G26P[19], using conventional Sanger sequencing after a series of combinatorial RT-PCR amplifications with an array of internal primers (Table S2) using a previously described approach (Cowley et al., 2013). The coding region for the 11 gene segments was determined and the resulting genome constellation was determined to be G26-P[19]-I5-R1-C1-M1-A8-N1-T1-E1-H1 (GenBank accession numbers HG513045-HG513055) -a novel combination of genome segments that has not been reported previously. The genome constellation of the sequenced strain was compared to other sequences of selected animal and human strains in GenBank, including representative strains containing G26 and P[19] segments (Tables 1 and S3).
To investigate segment-specific evolutionary relationships for the sequenced G26P[19] strain, maximum-likelihood phylogenies were inferred for each of the 11 gene nucleotide sequences and related reference RV sequences obtained from GenBank. Multiple alignments of nucleotide sequences and amino acid sequences were performed using MEGA5 employing CLUSTAL W algorithms, and were visually checked and realigned using Se-AL (v2.0) (http://tree.bio. ed.ac.uk/software/seal/). Only complete gene sequences were utilized in phylogenetic analyses, with the exception of VP4 [which included all Vietnamese P[19] sequences from this study regardless of sequence length (minimum sequence length: 619 nt) and RVA/Human-wt/THA/ 57vp7w/200X/G3P [19]] and VP7 [which included all Vietnamese G26 sequences from this study (minimum sequence length: 821 nt) and any available G26 sequences available in GenBank regardless of length]. Phylogenetic trees were inferred for each gene segment using aligned nucleotide sequences, employing a maximum-likelihood method in PhyML v3.0 (Guindon et al., 2010), using the HKY85 model of substitution (which was determined to be the best fit model for all gene segments using MODELTEST; Posada & Crandall, 1998) with 1000 bootstrap replicates. The resulting phylogenetic trees were visualized and edited in FigTree (v1.4) (http://tree.bio.ed.ac.uk/software/figtree/) and genetic distances (p-uncorrected) were estimated using HyPhy v2.0 (http://www.datam0nk3y.org/hyphy/ doku.php) (Pond et al., 2005).
Phylogenetic analysis of VP7 was performed on all available global G26 sequences and additional sequences from the three G26P[19] strains identified in this study (GenBank accession numbers HG513056-HG513058).
The VP7 sequences of the G26 strains fell within a single distinct lineage (Fig. 1) (Miyazaki et al., 2011) and RVA/Pig-wt/CHN/Z650/2007/G26P[X] (GenBank accession number KC292205; unpublished). Calculations of p-uncorrected genetic distance also showed the highest similarity between the VP7 sequence of our strain and that of the 57vp7w strain (97.1 %; 736 bp), followed by the corresponding segments of TJ4-1 (93.2 %; 987 bp) and Z650 (92.8 %; 1061 bp) strains (Tables 1 and S3). All of the VP7 segments from the G26 strains identified in human hosts fell within a sublineage of the G26 genotype that was closely related to those isolated from porcine hosts (Fig. 1). Although these analyses were limited by the paucity of available full-length RV genome sequences from human and animal species, phylogenetic clustering suggested that viruses of this novel genotype may have moved between porcine and human hosts on at least one occasion, followed by limited onward transmission in humans (Fig. 1 (Okada et al., 2000). Cumulative evidence from genetic investigations and the sustained detection of P [19] in humans and pigs support the theory that P[19] may have originated in pigs, suggesting that P [19] was introduced into human RV strains via human-porcine RV reassortment Ghosh et al., 2012;Krishnan et al., 1994;Maneekarn et al., 2006;Mukherjee et al., 2011;Nguyen et al., 2008;Okada et al., 2000;Theamboonlers et al., 2008;Urasawa et al., 1992;Varghese et al., 2004;Wu et al., 2011;Zade et al., 2009 (Table 1), and clustered with other full-length porcine and porcine-like human P[19] sequences in the VP4 phylogeny (Fig. 1). A RV strain incorporating a P[19] VP4 segment was reported previously in a symptomatic RV patient in HCMC in 2003 (Nguyen et al., 2008); the resulting P[19] sequence for this RV strain (RVA/ Human-wt/VNM/VN375/2003/GXP [19]) was only 558 bp in length, but exhibited 95.7 % nucleotide sequence identity to our more recently sampled viruses. These findings raise a question as to whether RV strains containing P[19] represent an established but rarely detected variant circulating in humans in Vietnam or are a porcine-maintained genotype that has undergone repeated jumps into the human population.  The remaining genes encoding the structural proteins (VP6 and VP1-3) exhibited the highest nucleotide sequence identity to the corresponding genes of porcine strains or porcine-human reassortants. The VP6 full-length sequence of our G26P[19] exhibited a sequence identity of 91.1 % to the cognate genes of the porcine strain RVA/Pig-wt/ARG/ CN86/1988/GXP[X] (Table 1), and clustered phylogenetically with other porcine, porcine-like human and porcinelike non-human strains (Fig. S1). Additionally, the VP1-3 sequences of our G26P[19] strain clustered phylogenetically with the cognate segments of porcine strains and (porcinelike) human strains (Fig. S1), and demonstrated the highest sequence similarity to porcine and porcine-like human strains (Tables 1 and S3). Similarly, further genetic distance determination and phylogenetic analyses of the nonstructural genes of this novel G26P[19] strain provided additional evidence of associations between this virus and the corresponding segments of porcine and porcinehuman reassortant strains (Tables 1 and S3, Fig. S2).  (Theamboonlers et al., 2008), no other G26P [19] has been detected in subsequent RV studies in Thailand or in any continual RV surveillance programmes elsewhere. The detection of such similar viruses in two countries in South-east Asia raises questions as to the prevalence of this strain in human and animal RV infections, and the ability of this rarely detected strain to infect and transmit between humans. This warrants further epidemiological and biological investigation.
Given the potential for mixed RV infections, atypical strain circulation, and the close proximity of people and domesticated animals in parts of Asia, the characterization of genetic reassortment and interspecies transmission is key to understanding the evolutionary and ecological dynamics of RV (Cook et al., 2004;Santos & Hoshino, 2005). Here, we have identified and described the genome of a novel G26P[19] RV in children with diarrhoea in southern Vietnam, confirming the circulation of potentially zoonotic RV strains alongside common human RV strains causing human disease. There is a lack of whole RV genome sequences available from this region of Asia (from both humans and animals), which limits our understanding of the epidemiology and zoonotic potential of this important human pathogen. Additionally, the efficacy of currently available vaccines [Rotarix (GlaxoSmithKline) and RotaTeq (Merck)] against novel and exotic RV genotypes has not, as yet, been determined. Continual RV strain surveillance, combined with whole-genome sequencing from both human and animal populations, is essential for understanding the epidemiological relevance of novel RV strains and assessing their zoonotic potential. Such investigations are essential for determining the potential impact of vaccination in urban and rural areas of Asia with alternative circulating organisms and atypical RV epidemiology.