Molecular Characterisation of a Rare Reassortant Porcine-Like G5P[6] Rotavirus Strain Detected in an Unvaccinated Child in Kasama, Zambia

A human-porcine reassortant strain, RVA/Human-wt/ZMB/UFS-NGS-MRC-DPRU4723/2014/G5P[6], was identified in a sample collected in 2014 from an unvaccinated 12 month old male hospitalised for gastroenteritis in Zambia. We sequenced and characterised the complete genome of this strain which presented the constellation: G5-P[6]-I1-R1-C1-M1-A8-N1-T1-E1-H1. The genotype A8 is often observed in porcine strains. Phylogenetic analyses showed that VP6, VP7, NSP2, NSP4, and NSP5 genes were closely related to cognate gene sequences of porcine strains (e.g., RVA/Pig-wt/CHN/DZ-2/2013/G5P[X] for VP7) from the NCBI database, while VP1, VP3, VP4, and NSP3 were closely related to porcine-like human strains (e.g., RVA/Human-wt/CHN/E931/2008/G4P[6] for VP1, and VP3). On the other hand, the origin of the VP2 was not clear from our analyses, as it was not only close to both porcine (e.g., RVA/Pig-tc/CHN/SWU-1C/2018/G9P[13]) and porcine-like human strains (e.g., RVA/Human-wt/LKA/R1207/2009/G4P[6]) but also to three human strains (e.g., RVA/Human-wt/USA/1476/1974/G1P[8]). The VP7 gene was located in lineage II that comprised only porcine strains, which suggests the occurrence of independent porcine-to-human reassortment events. The study strain may have collectively been derived through interspecies transmission, or through reassortment event(s) involving strains of porcine and porcine-like human origin. The results of this study underline the importance of whole-genome characterisation of rotavirus strains and provide insights into interspecies transmissions from porcine to humans.


Introduction
Group A rotaviruses (RVA), of the family Reoviridae, are the number one viral pathogens causing severe diarrhoea in children below five years of age [1]. In 2016, an estimated 128,000 deaths in children below five years were due to RVA infections, 90% of which occurred in developing countries [2,3]. Similarly, RVA are the primary cause of acute gastroenteritis in new-born piglets [4].
The findings that show animals can serve as potential reservoirs for genetically diverse rotavirus strains that can be passed on to humans have elicited a large amount of interest and topics for further research [21]. Several novel and rare animal-like or animal-human reassortant rotavirus strains have been identified globally [22][23][24][25][26][27][28]. The detection of animal strains in humans is presumed to be as a result of zoonotic transmission, along with reassortment, which contributes to the diversity of circulating RVA [4,29,30]. Inter-and intragenogroup reassortment may occur when multiple RVA simultaneously infect a host. This is attributed to the segmented nature of the rotavirus genome [1,31]. It is, therefore, necessary to continuously carry out the monitoring of animal RVA and the role they play in contributing to the diversity of circulating RVA in humans.
The G5, one of the most common porcine genotypes, has sporadically been identified in human populations in Brazil (G5P[X]), Cameroon (G5P [7] and G5P[8]), Argentina (G5P[8]), and the United Kingdom(G5P[X]) [32][33][34][35][36]. The P [6] is presumed to be of porcine origin. They have also been identified in human populations [37][38][39][40]. The first human G5P [6] strain, LL36755, was detected in a child who had acute gastroenteritis in China in 2007 [41]. Other G5P [6] strains were detected in Vietnam, Taiwan, Bulgaria, Japan, and Thailand [37,[42][43][44][45]. To date, the whole-genome of only two human G5P [6] strains-Bulgarian BG620 (nt sequences unavailable in the DDBJ, EMBL, and GenBank data libraries as of 13 August 2020) and Japanese Ryukyu-1120 (full open reading frame, available in GenBank)-have been analysed [45,46]. Diarrhoea is a burden for the Zambian healthcare system, with about 33% of the extreme cases being attributable to RVA [47][48][49]. In an attempt to generate disease burden attributable to rotavirus diarrhoea in children, the Zambian Ministry of Health, with support from WHO, launched rotavirus surveillance at the University Teaching Hospital (UTH) in 2006 [50,51]. Surveillance data generated provided evidence of the burden of rotavirus diarrhoea that supported the introduction of the rotavirus vaccine, Rotarix ® , as a pilot project in Lusaka, Zambia in 2012, and was later rolled out nationwide in November 2013 [50]. According to the estimates reported by the World Health Organization (WHO) and the United Nations International Children's Emergency Fund (WHO/UNICEF), rotavirus vaccine coverage in Zambia has been consistently high for the last six years, increasing from 73% in 2014 to 90% in 2019 [52]. Over this period, a sustained and significant reduction in rotavirus-associated hospitalisations and mortality was observed in children under 5 years [51].
The African Rotavirus Surveillance Network, coordinated by the World Health Organization Regional Office for Africa (WHO/AFRO), is actively monitoring the diversity and distribution of RVA genotypes in children hospitalised with acute diarrhoea [53]. Initially, the network was established with four countries in 2006, and expanded to 29 countries by the end of 2016 [54,55]. The Diarrhoeal Pathogens Research Unit at Sefako Makgatho University in Pretoria (South Africa) and the Noguchi Memorial Institute for Medical Research in Accra (Ghana) are the two WHO Rotavirus Regional Reference Laboratories (RRLs) for the network that conducts monitoring of rotavirus epidemiology in Africa [55]. The WHO/AFRO is currently supporting the University of the Free State-Next Generation Sequencing (UFS-NGS) unit to undertake rotavirus surveillance of rotavirus strains that circulated in Zambia between 2013 and 2016 at the whole-genome level. A G5P [6] strain, UFS-NGS-MRC-DPRU4723, was identified among these strains and was analysed so as to elucidate its origin and evolution. The sample was collected in 2014 from an unvaccinated 12 month old male hospitalised for gastroenteritis at Arthur Davison Children's Hospital in Ndola, Zambia.

Nucleotide Sequencing and Identity of the Strain
Illumina ® MiSeq sequencing exhibited a phred score of Q30 and collectively yielded 98.8 Mbs of data for this specific sample. The whole genome of RVA/Human-wt/ ZMB/UFS-NGS-MRC-DPRU4723/2014/G5P [6] was 18272 bps in size. The length and ORF of the 11 gene segments as determined by nucleotide sequencing are shown in Table 1. A BLASTn search was performed, and it appeared to exhibit maximum sequence identities of 95.7%-98.0% with porcine and human porcine-like strains (Table 1). Based on the whole genome classification system, RVA/Human-wt/ZMB/UFS-NGS-MRC-DPRU4723/2014/G5P [6] exhibited a G5-P[6]-I1-R1-C1-M1-A8-N1-T1-E1-H1 genotype constellation ( Table 2). The genetic constellation of the study strain was compared to those of other G5 and non-G5 strains retrieved from the GenBank ( Table 2).

Sequence and Phylogenetic Analysis
To investigate the potential origin of RVA/Human-wt/ZMB/UFS-NGS-MRC-DPRU4723/2014/G5P [6], phylogenetic trees were constructed for each of the 11 gene segments along with cognate gene sequences of RVA strains obtained from the GenBank.
Blue shading indicates the gene segments with genotypes identical to those of UFS-NGS-MRC-DPRU4723. Bold font indicates genotypes associated with porcine strains. "−" indicates that no sequence data were available in GenBank/EMBL/DDBJ data banks. * Genotype assignment based on reports by [37] (strain 03-98sP50) and (strain BG260) [46]. To date, the nucleotide accession numbers for the 11 gene segments of strains 03-98sP50 and BG260 are not available in the GenBank, EMBL, or DDBJ data banks.

Sequence and Phylogenetic Analysis of the VP4 Gene
The VP4 gene of RVA/Human-wt/ZMB/UFS-NGS-MRC-DPRU4723/2014/G5P [6] was phylogenetically compared to the already established five lineages (I-V) of genotype P [6] [65] (Figure 2). The P [6] gene of the study strain clustered into lineage V, which consisted of porcine and putative human porcine-like strains detected in parts of Europe and one African strain. A similarity analysis of the P [6] gene of the study strain with strains obtained from GenBank showed that the Zambian G5P [6] exhibited the highest sequence identity of 98.1% (98.3%) with a porcine-like human strain RVA/Human-wt/COD/KisB332/2008/G4P [6] from the Democratic Republic of Congo (Supplementary data 1). All the African strains clustered into a separate lineage, lineage I, with sequence identities of 85.7%-86.8% (92.5%-93.9%) (Supplementary data 1).

Reassortment Analysis
The concatenated whole genome alignment of RVA/Human-wt/ZMB/UFS-NGS-MRC-DPRU4723/ 2014/G5P [6], together with the Japanese G5P [6] strain and selected Chinese porcine-like human P [6] strains, was visualised ( Figure 4). The whole genome of the Zambian G5P [6] strain demonstrated a relatively high degree of conservation with the Japanese G5P [6] strain and the two Chinese G4P [6] strains. With the exception of VP7 and VP4, the genome of the Chinese strain E931 exhibited the overall highest genomic conservation to the study strain. With the exception of VP7, VP3, and NSP1 genes, the Chinese strain GX54 shared a highly conserved genome with the study strain. The Japanese strain Ryukyu-1120 demonstrated a highly similar genome to the study strain for seven of the 11 genes, the exceptions being VP1, VP3, VP6, and VP7. The results of this analysis confirmed the genetic similarity between RVA/Human-wt/ZMB/UFS-NGS-MRC-DPRU4723/2014/G5P [6] and Asian (Chinese) porcine-like human strains, hence suggesting that the Zambian G5P [6] strain may have been derived via reassortment events.

Reassortment Analysis
The concatenated whole genome alignment of RVA/Human-wt/ZMB/UFS-NGS-MRC-DPRU4723/2014/G5P [6], together with the Japanese G5P [6] strain and selected Chinese porcine-like human P [6] strains, was visualised ( Figure 4). The whole genome of the Zambian G5P [6] strain demonstrated a relatively high degree of conservation with the Japanese G5P [6] strain and the two Chinese G4P [6] strains. With the exception of VP7 and VP4, the genome of the Chinese strain E931 exhibited the overall highest genomic conservation to the study strain. With the exception of VP7, VP3, and NSP1 genes, the Chinese strain GX54 shared a highly conserved genome with the study strain. The Japanese strain Ryukyu-1120 demonstrated a highly similar genome to the study strain for seven of the 11 genes, the exceptions being VP1, VP3, VP6, and VP7. The results of this analysis confirmed the genetic similarity between RVA/Human-wt/ZMB/UFS-NGS-MRC-DPRU4723/2014/G5P [6] and Asian (Chinese) porcine-like human strains, hence suggesting that the Zambian G5P [6] strain may have been derived via reassortment events.

Discussion
The detection of genotype G5 in humans, which is typical for pigs, is possibly due to interspecies transmission [35,45]. In Zambia, as with many countries in Africa, humans and farm animals live in proximity. The interaction between humans and animals could be the primary cause for zoonotic transmission, which could result in genetic reassortments and perhaps other mechanisms of genetic diversity, ultimately leading to the introduction and spread of animal genotypes into human populations [69].
In this study, an analysis was conducted on a sample collected from a child admitted to a paediatric ward presenting with clinical symptoms (vomiting, diarrhoea, and fever) that are usually present during typical rotavirus infection. This raises the question whether such animal-derived strains are capable of mutating and effectively spreading within/across human populations as in the case of established typical Wa-like and DS-1-like genotype constellations, with the same magnitude of rotavirus disease severity. Furthermore, taking into consideration that the G5 and P [6] genotypes are not included in the currently available vaccines, the probability for such strains to have the potential to spread more swiftly from human to human may have implications for the effectiveness of current rotavirus vaccine candidates that are in use in African countries.
A phylogenetic analysis of RVA/Human-wt/ZMB/UFS-NGS-MRC-DPRU4723/2014/G5P [6] showed that this strain was a possible reassortant, as it was closely related to both porcine and porcine-like human strains, predominantly from Asia, than to typical human RVA strains. The VP6, VP7, NSP2, NSP4, and NSP5 segments of this strain showed a close similarity to porcine strains. Although the remaining gene segments (VP1, VP3, VP4, and NSP3) were closely related to human strains, all of these were porcine-like human strains [26,56,[58][59][60]70]. With a genotype 1 (Wa-like) backbone, this finding is consistent with the hypothesis that human Wa-like strains and porcine strains have a common ancestor [5]. However, the origin of the VP2 gene of the study strain was not very definitive, as it was not only close to porcine and porcine-like human strains but also to three human strains (DC1476, DC582, and DC1127). Phylogenetically, the clusters of these three strains were shown to be distinctive from the genes of contemporary, wild-type human strains [71]. Notably, the VP7 gene of RVA/Human-wt/ZMB/UFS-NGS-MRC-DPRU4723/2014/G5P [6] was located in lineage II, which comprised only porcine strains, hence implying the possibility of porcine-to-human interspecies transmission [63]. Phylogenetic analysis of porcine and human P [6] strains indicated that both porcine and human P [6] strains were present in P [6] lineages I, III, and V, hence showing that human P [6] strains might have separately emerged from at least three porcine-to-human transmissions [65]. This finding supports the Zambian G5P [6] strain, as the VP4 gene clustered and shared high nucleotide and amino acid identities with lineage V of P [6] porcine and porcine-like human strains. The NSP1 gene was most similar to porcine-like human strains. However, it was revealed to have the porcine genotype A8. Taking this together, it is likely that RVA/Human-wt/ZMB/UFS-NGS-MRC-DPRU4723/2014/G5P [6] originated by zoonotic transmission, coupled with reassortment events.
Several amino acid changes were identified in the nine variable regions when the VP7 gene of the study strain was compared to other G5 strains within each of the three lineages [64]. Additionally, the previously described conserved N-glycosylation site at residues 69-71 within the variable region 4 (VR-4) was found to be conserved in all the G5 strains used in this analysis [64,72]. Four major antigenic regions have been described for the VP7 protein in rotaviruses (A, B, C and F) [73,74]. Marked differences in the antigenic regions of RVA/Human-wt/ZMB/UFS-NGS-MRC-DPRU4723/2014/G5P [6] were seen when it was compared to other globally circulating G5 strains. Usually, antigenic regions A and C are said to be conserved within serotypes [75]. However, multiple substitutions were observed in these regions when comparing the Zambian G5 strain to other G5 strains globally.
The amino acid sequence for the VP4 gene was 775 amino acids long and displayed amino acid identity values ranging from 91.0% to 98.3% with the reference P [6] strains. Considering it has been established that strains with amino acid identities greater than 89% belong to the same P genotype [76], our findings show that RVA/Human-wt/ZMB/UFS-NGS-MRC-DPRU4723/2014/G5P [6] belongs to the genotype P [6]. The analysis of the amino acid sequences showed that the hypervariable region (amino acid 71-208) which houses the variable region 3 (VR-3) contained most of the substitutions. Furthermore, the potential trypsin cleavage sites [68] were conserved in all the P [6] strains. Several amino acid substitutions were observed among the lineage I P [6] strains. The presence of several amino acid changes in the VP4 gene of this strain compared to other circulating P [6] strains globally is in agreement with the hypothesis that the P [6] gene has been introduced to humans via independent reassortment events [40,65,77].
Rotaviruses are genetically diverse in nature and are host-species specific, suggesting that host species barriers and restrictions exist. However, rotaviruses of animal origin may cross the host species barrier and may acquire human rotavirus gene segments, which enables the viruses to efficiently spread across human populations [4]. In this regard, G5 rotavirus strains have sporadically been documented in Latin America, Asia, Europe, and Africa [33][34][35][36][37]41,45,46]. Porcine P [6] strains seem to pose a lesser species barrier to humans [20]. Even though the relationship between porcine and human rotaviruses has already been established [5], whole genome analysis in this study presented the possible occurrence of interspecies transmission and reassortment between human and porcine rotaviruses.

Sample Collection
The sample was collected in 2014 from an unvaccinated 12 month old male at Arthur Davidson Children's Hospital (ADCH) in Ndola, a rotavirus surveillance sentinel site. The child had travelled with parents from Kasama, a town in the Northern Province of Zambia which is approximately 760 km away from Ndola, Zambia. This child was admitted to a paediatric ward at ADCH, with gastroenteritis of four days duration and a history of fever. Frequency of vomiting and diarrhoea was three episodes and two episodes, respectively, in the previous 24 h. The level of dehydration was assessed as mild and the child received an oral rehydration solution and was discharged after a few days. The stool sample was screened using the enzyme immunoassay (EIA) technique for the presence of RVA antigen in the Virology laboratory in Lusaka. It was randomly picked and sent to the Diarrhoeal Pathogens Research Unit (DPRU), a World Health Organization Rotavirus Regional Reference Laboratory (WHO-RRL) in Pretoria, South Africa, as part of the WHO/AFRO annual rotavirus surveillance. Conventional genotyping was carried out at DPRU. Thereafter, the sample was shipped to the UFS-NGS unit for sequencing and whole-genome analysis.

Viral dsRNA Extraction
The viral double-stranded RNA (dsRNA) was extracted from human stool suspensions using a previously described method with modifications [78]. Approximately 100 mg stool was suspended in 200 µL phosphate-buffered saline (PBS) solution (Sigma-Aldrich ® , St Louis, MO, United States). The faecal suspension was mixed with 900 µL TRI Reagent ® LS (Molecular Research Centre, Cincinnati, OH, United States) and homogenized for five minutes. A 300 µL volume of chloroform (Sigma-Aldrich ® , St Louis, MO, United States) was used to achieve phase separation, which was followed by centrifugation (Eppendorf microcentrifuge 5427 R, Germany) at 17,319× g for 20 min at 4 • C. The supernatant was precipitated using 700 µL ice-cold isopropanol (Sigma-Aldrich ® , United States) and centrifuged (Eppendorf microcentrifuge 5427 R, Germany) at 17,319× g for 30 min at 4 • C. The supernatant was discarded, and the tubes were air-dried for 5 min, followed by the precipitation of single-stranded RNA (ssRNA) using 30 µL 8 M lithium chloride (Sigma, St Louis, MO, United States) at 4 • C for 16 h. The dsRNA was purified using the MinElute gel extraction kit (Qiagen, Hilden, Germany). RNA integrity was determined by electrophoresis on 1% TBE agarose gel stained with ethidium bromide (Sigma-Aldrich ® , St Louis, MO, United States), which was visualised on a G: Box UV transilluminator (Syngene, Cambridge, United Kingdom).

cDNA Synthesis and Purification
cDNA synthesis was carried out using the Maxima H Minus Double-stranded cDNA kit (Thermo Fisher Scientific, Waltham, MA, United States) according to the manufacturer's instructions with minor modifications captured at the UFS-NGS SOP, whereby the dsRNA was denatured at 95 • C for 5 min. First strand synthesis was carried out for two hours at 50 • C. Random hexamer primer was employed for cDNA synthesis. The cDNA was purified using the MSB ® Spin PCRapace purification kit (Stratec, Invitek Molecular, Berlin, Germany).

DNA Library Preparation and Illumina ® MiSeq Sequencing
DNA libraries for Illumina ® sequencing were prepared using the Nextera ® XT DNA library preparation kit (Illumina, San Diego, CA, United States) according to the manufacturer's instructions. Briefly, DNA was tagmented at 55 • C for five minutes followed by ligation to Illumina ® sequencing index 1 and index 2 adapters by PCR amplification. Size selection and clean-up of the DNA libraries was performed using Agencourt AMPure XP beads (Beckman Coulter, South Kraemer Boulevard Brea, CA, United States). The quantity of DNA was determined on the Qubit 2.0 fluorimeter (Invitrogen, Carlsbad, CA, United States), and a quality check of the libraries was performed on a Bioanalyzer 2100 (Agilent Technologies, Santa Clara, CA, United States). After this, sequencing was performed on an Illumina ® MiSeq sequencer (Illumina, San Diego, CA, United States) using a MiSeq reagent kit v3 for 600 cycles (2 × 300 bp paired reads) with a 10% PhiX DNA control spike-in.

Assignment of Genotypes
The genotypes of each of the 11 rotavirus genome segments were determined using the online Virus Pathogen Resource (ViPR).

Phylogenetic Analysis
Gene-specific multiple sequence alignments were made using the MAFFT plugin implemented in Geneious Prime ® 2019.2.1 and the MUSCLE algorithm embedded in MEGA 6.06 (for the VP2 and NSP1 segments) [80,81]. Once aligned, the DNA Model Test program in MEGA 6.06 was used to identify the optimal evolutionary model for each genome segment [82]. Using an Akaike information criterion (corrected) (AICc), the following models were found to best fit the data: HKY+G+I (VP1), GTR+G+I (VP2, VP3, and VP4), T92+G (VP6, NSP1, NSP2, NSP3, NSP4, and NSP5), and T92+G+I (VP7). Maximum likelihood trees were constructed using the optimal models in MEGA version 6.06 [82,83] with 1000 bootstrap replicates to estimate branch support [84]. The shared nucleotide and amino acid sequence identities among strains were calculated for each gene using the p-distance algorithm in MEGA 6.06. Analysis and visualization of the aligned concatenated whole genomes was performed on the mVISTA online platform [85].

Conclusions
In summary, RVA/Human-wt/ZMB/UFS-NGS-MRC-DPRU4723/2014/G5P [6] was a reassortant possessing gene segment of porcine and porcine-like human origin, and was closest to Asian strains. It is presumed that pigs play a crucial part as a source for new or newly-evolved emerging human rotaviruses. This highlights the need for continuous large-scale surveillance and whole genome analysis of circulating porcine and human rotaviruses. Furthermore, it was imperative to examine the prevalence of G5P [6] strains in Zambia. Eventually, this should result in a greater understanding of the genes that determine the transmission between hosts successfully as well as to gain insights on complex reassortment patterns between porcine and human rotaviruses.