Genetic characterization of G12P[6] and G12P[8] rotavirus strains in six African countries

Background: G12 rotaviruses were rst observed in sub-Saharan Africa in 2004 and since then have continued to emerge and spread across the continent and are reported as a signicant human rotavirus genotype in several African countries, both prior to and after rotavirus vaccine introduction. This study investigated the genetic variability of 15 G12 rotavirus strains with either P[6] or P[8] identied between 2010 and 2014 from Ethiopia, Kenya, Rwanda, Tanzania, Togo and Zambia. Methods: The investigation was carried out by comparing VP7 and partial VP4 sequences of the African G12P[6] and G12P[8] strains with the available GenBank sequences and mapping the recognized neutralization epitopes of these strains. Results: The ndings suggested that the VP7 and VP4 genes of the G12 strains circulating in African countries are homologous at the nucleotide and amino acid level, irrespective of country of origin and year of detection, although there was a unique clustering of the Ethiopian strains. The study strains shared a common ancestry with G12 strains circulating globally. Neutralization epitope mapping revealed that rotavirus VP4 P[8] genes associated with G12 had amino acids similar to those reported globally including the vaccines RotaTeq® P[8] and Rotarix®. Conclusions: It is unlikely that widespread vaccine use has driven the molecular evolution and sustainability of G12 strains in Africa. Furthermore, it is too early post vaccine introduction to indicate any effect of vaccine-induced pressure on maintaining the stability of these strains in circulation. Continuous monitoring of rotavirus genotypes is recommended to assess the long-term impact of rotavirus vaccination on the dynamic nature of rotavirus evolution on the continent. diversity of the circulating G12 strains in Africa is still limited, we sought to investigate the genetic variability of the two recognized neutralization antigens, VP7 and VP4, of G12 strains from across the continent. Thus, this study investigated the genetic variability of the gene segments 4 (encoding VP4) and 9 (encoding VP7) of G12 strains identied in several African countries and mapped the neutralization epitopes in an effort to provide insights on the evolutionary mechanisms and possible origins of the G12 strains in Africa. VP4 sequences were mapped with other globally circulating P[6] reference strains. Maximum likelihood phylogenetic trees constructed using MEGA 6.0 1000 bootstraps. 3 Parameter The African G12P[8] and G12P[6] rotavirus strains belonging to lineage III circulating in these countries are not unique and are the same as the globally circulating rotavirus G12 strains and there is no evidence of molecular evolutionary pressure from widespread vaccine use. The antigenic epitopes display limited diversity to each other and other global strains, including to the two rotavirus vaccines (RotaTeq and Rotarix), indicating that this is unlikely to be associated with sustained circulation over time in Africa. However, as new rotavirus vaccines which do not carry the common human rotavirus VP4 genotypes, (RotaVac and RotaSIIL from India) are introduced in selected African countries, it will be imperative to continue genotypic surveillance to identify and monitor emerging strains.


Introduction
Diarrhoeal disease is a major cause of death in infants and young children below the age of ve and rotavirus is the most signi cant pathogen associated with that mortality [1]. Rotavirus is associated with 122,232 − 215,757 Under 5 deaths annually [2]. Furthermore, it has been estimated that diarrhoeal diseases are signi cantly more severe in immunocompromised children, especially those infected with HIV which is relevant to sub-Saharan Africa [3]. Recent estimates showed that the introduction of rotavirus vaccines globally has resulted into 59% and 36% relative reduction of rotavirus hospitalizations and all cause acute gastroenteritis, respectively [4]. Before the introduction of rotavirus vaccines in many African countries, it was estimated that almost 40% of all diarrhoeal cases on the continent were due to rotaviruses [3]. The introduction of rotavirus vaccines into 29 sub-Saharan African countries before 2016, resulted in a reduction of approximately 21,000 deaths and 135,000 hospitalizations in 2016 alone [5], highlighting the major impact that rotavirus vaccines have had on rotavirus diarrhoea.
The diarrheal stool samples were collected as a routine diagnostic clinical specimen when the parents brought their child to a health facility for clinical management, requiring no written informed consent. As part of the WHO-coordinated rotavirus surveillance network, the archived rotavirus-positive specimens, were anonymized and utilized for strain characterization under a Technical Service agreement and a Materials Transfer Agreement to the WHO AFRO Regional Reference Laboratory based at Sefako Makgatho Health Services University. The WHO Research Ethics Review Committee granted an exemption activity, noting that the procedures involved in the study are part of routine hospital-based rotavirus surveillance.

Sanger sequencing
Amplicons were sequenced using the dideoxynucleotide termination Sanger sequencing method with ABI 3500XL sequencer. A region of VP7 and VP4 was sequenced using reverse and forward primers used for RT-PCR. The sequence chromatograms were edited using chromasPro version 1.49 beta resulting into 961 bp (VP7) and 876 bp (VP4) fragments (www.technelysium. com.au/chromas.html).

Sequence analysis
Sequencing data was then compared with available rotavirus sequences in the GenBank using the NCBI-BLAST software (www.ncbi.nlm.nih.gov/BLAST/ ,USA). MEGA 6.0 was used to align the sequences with G12 strains retrieved from GenBank, by muscle codon [39]. To expand the analysis, VP7 and VP4 sequences available in the GenBank isolated from other African countries were included. Dot conservation plots were constructed using BioEdit sequence alignment editor [40] mapping the variable and antigenic regions within VP7 of the study strains with G12 reference strains belonging to the four G12 lineages (I-IV).
Simultaneously, P [8] VP4 sequences of the study strains were mapped with P [8] of both the Rotarix® and RotaTeq® vaccine strains and other recent circulating strains. While the P [6] VP4 sequences were mapped with other globally circulating P [6] reference strains. Maximum likelihood phylogenetic trees were constructed using MEGA 6.0 with 1000 bootstraps. General time reversible (GTR), Tamura 3 Parameter (T92) and Hasegawa-Kishino-Yano (HKY) were selected as the best evolutionary models for the constructing of G12, P [8] and P [6] phylogenetic trees in MEGA 6.0, respectively. Either the gamma (G) or gamma invariable (G + I) evolutionary distribution were applied for each phylogenetic tree. Nucleotide and amino acid distance homology matrix was constructed using the p-distance algorithm in MEGA 6.0 [39].

Accession numbers
The partial VP7 and VP4 sequences have been made available on the NCBI GenBank database (Accession numbers: MK059426 -MK059453)

VP7 genotype analysis
The nucleotide and amino acid sequences of 15 G12 rotavirus strains collected during 2010-2014 rotavirus seasons across Africa, were analysed and compared with the strains from GenBank database. High nucleotide (97-99%) and amino acid (98-100%) sequence similarity was observed amongst the VP7 gene sequence of the G12 study strains as well as between the study strains and the circulating global human G12 strains. However, strain MRC-DPRU6219 from Rwanda shared 95-98% nucleotide and 97-98% amino acids identity with the other 14 study strains (Supplementary Table 2).
Phylogenetic analysis showed that the African rotavirus genotype G12 VP7 sequences clustered within lineage III, and sub-lineage III A-C (labelled for the purpose of discussion in this study) (Fig. 1). The two study strains, MRC-DPRU4540 (Tanzania) and MRC-DPRU2118 (Togo) in sub-lineage IIIA were closely related to globally circulating non-African G12 strains. G12 strains included in this study from Ethiopia (MRC-DPRU2268, MRC-DPRU857, MRC-DPRU2273, MRC-DPRU5683, MRC-DPRU4959 and MRC-DPRU4165) formed a monophyletic cluster within sub-lineage IIIB with reference strains from Nepal and Belgium. The study strains isolated from Kenya and Zambia clustered in sub-lineage IIIC closely related to other African G12 strains and interestingly most of these African strains all shared a G12P [6] genotypic constellation. Within the same IIIC sub-lineage a single Rwandan strain (MRC-DPRU6219) isolated in 2014 seemed distinct and clustered closer to strains isolated from Mozambique and India.
Sequence analysis within the nine variable regions (VR) and four antigenic regions (AR A-C,F) of VP7 of the study strains were considerably conserved when compared to the rst reported African G12 strain from South Africa (SA4958JHB) as well as representative strains belonging to the four lineages. The comparison of the amino acid showed differences mostly within the variable regions of the gene compared to the antigenic regions which carry the recognized antigen-speci c epitopes (Table 2). Although the study strains were homologous to the rst reported South African strain, at certain positions the study strains shared amino acids similar to the prototype (L26, lineage I) and porcine (RU172, lineage IV) G12 strains.
Interestingly, in antigenic region A, strains from Ethiopia and Zambia had an N100S amino acid substitution. An alignment of G12 lineage III strains circulating globally identi es this substitution commonly in strains from Nepal, Italy and Belgium but not from other African strains (data not shown). Furthermore, notable amino acid substitutions A125S and V142I differentiated lineage III isolates from those in lineages I, II and IV. This substitution was seen in all global G12 strains belonging to lineage III.

VP4 genotype analysis
The partial VP8* gene sequence of VP4 (876 bp) was also analysed for the 15 study strains and compared to sequences in GenBank. Of the sequences analysed, both P[8] (n = 8) and P [6] (n = 7) strains were included. Sequence comparison showed that the G12 P [8] strains shared 98-100% amino acid and 98-99% nucleotide similarity with most African P [8] strains available in the GenBank database. Also, they shared similar percentage homology with each other (Supplementary Table 3). The P [8] rotavirus strains clustered in lineage III distantly from the Rotarix and RotaTeq P [8] vaccine components, which clustered in lineage I and lineage II, respectively (Fig. 2). Four of the ve strains from Ethiopia formed their own monophyletic cluster as was also seen with their VP7 sequences. While MRC-DPRU5683 (Ethiopia), MRC-DPRU2118 (Togo), MRC-DPRU6219 (Rwanda) and MRC-DPRU4540 (Tanzania) dispersed differently on the phylogenetic tree, clustering closer to strains from Hungary, USA, Australia and India, respectively.
The comparison of the eight strains bearing VP4 P [8] genotype with the VP4 P [8] gene included in the two vaccines, Rotarix and RotaTeq, and other strains representing different P [8] lineages, revealed various amino acid substitutions within the VP8* neutralizing antigenic epitopes. Within VP8* there are four de ned neutralization epitopes, designated 8 − 1 to 8 − 4 ( Table 3). As shown in Table 3, the study strains had similar amino acids with the VP4 of RotaTeq® at positions 125 and 131. At positions 150 and 113, both the vaccine VP4 components were the same as the study strains (except for the monophyletic strains from Ethiopia). The study strains as well as a reference strain belonging in lineage III, had Glycine amino acid instead of Aspartic acid (RotaTeq) or Asparagine (Rotarix) in position 195.
Similarly, the P[6] study strains shared 98-100% amino acid and 98-99% nucleotide similarity amongst themselves and with P [6] sequences available in the GenBank database. However, strain MRC-DPRU857 from Ethiopia shared lesser nucleotide and amino acid similarity with the study strains, 96-97% and 97-98% respectively (Supplementary Table 4). The P [6] study strains clustered in lineage Ia with other global strains and tended to cluster more closely with other strains African strains (Fig. 3). Amino acid conservation plot of the study strains with other P [6] strains representing the four lineages show that the study strains are conserved within lineage I into which the study strains cluster (Table 4).

Discussion
This study analysed circulating G12P [6] and G12P [8] rotaviruses from several African countries during the period 2010-2014 and prior to widespread use of rotavirus vaccines on the continent. Genotype G12 strains, which emerged approximately two decades ago, have been reported to be the cause of severe dehydrating diarrhoea in vaccinated children in several countries, particularly in Latin America which started vaccination about six years prior to sub-Saharan Africa [41][42][43]. However, if one looks at a temporal association of the emergence of the G12 strains, it is associated with the global spread of these strains, rather than causally associated with wide-spread vaccine use. Nevertheless, with the introduction of rotavirus vaccine in 2012-2014 in many of the African countries included in this study, the opportunity existed to conduct an analysis of circulating G12P [6] and G12P [8] strains in several countries, just prior to and as vaccines were introduced and to evaluate whether these strains might become predominant due to evading the vaccine. Five of the six studied countries had introduced the Rotarix vaccine. The exception is Rwanda which uses RotaTeq vaccine.
Clearly, G12 strains do not share the VP7 G-speci city with vaccine strains; however, both licenced rotavirus vaccines (RotaTeq and Rotarix) have demonstrated clinical protection against heterotypic strains, including G12 strains. For instance, the phase III Rotarix® clinical trial conducted in Malawi and South Africa showed cross protection against diverse rotavirus strains, including G12 with vaccine e cacy of 51.5% [33]. Similar results were observed with the RotaTeq vaccine study in three African countries [34]. However, rotavirus vaccines have also been shown to exercise protection via the immune responses to the VP4 neutralization antigens [44], and the VP4 P [8] is shared between both vaccines and a proportion of the G12 strains evaluated, those with G12P [8]. Thus, understanding the genetic variability of both the VP4 and VP7 genes of the circulating G12 rotavirus strains should provide insights into the evolutionary relationships and potential biological advantages of these strains in Africa.
Phylogenetic analysis of G12 rotavirus strains globally, shows segregation of the strains into four lineages (I -IV). Lineage I is the prototype strain L26 identi ed in 1987 and which was not apparently biologically competitive in humans and did not spread; lineage II is the G12P [9] strains from Asia which appear to be a unique class of natural reassortants with a VP4 P [9]; and lineage IV includes the only porcine strain (G12P [7]) [19,45,46]. Lineage III strains, on the other hand, are the mostly contemporary G12 strains detected since the mid-2000's and which are now globally prevalent in most continents. This analysis con rms that the genotype G12 strains circulating in these six sub-Saharan African countries (Ethiopia, Kenya, Rwanda, Tanzania, Togo and Zambia) clustered in lineage III with strains circulating all over the world, showing the dominance and biological competitiveness of these strains, which have persisted over the last two decades, in most continents [47][48][49].
Evidence of genetic variation was observed amongst the four G12 lineages in this study. Amino acid substitution S25N (VR2), N87S (antigenic region A) and A213T (antigenic region C) in lineages II & III segregate the prototype lineage I detected in 1987 and the porcine lineage IV. The lineages were further characterised by the amino acid substitutions A125S in VR 6 and V142I in antigenic region B detected only in the current circulating lineage III strains. The latter change from Valine to Isoleucine, where the amino acids share similar chemical properties, might not impose a conformational change to the VP7 protein. However, the A125S substitution, in which Alanine acquired a hydroxyl group to change to Serine over the period of early 2000s to late 2000s could in uence the capsid structure. The mechanism of rotaviruses mutating to advance epidemiological spread was observed with recent G2 rotavirus strains belonging to lineage IVa that spread globally. All these strains exhibited an amino acid substitution D96N which seemed to confer survival advantage to these lineage IVa G2 rotavirus strains [50]. It needs to be investigated further whether the A125S amino acid substitution observed in lineage III G12 strains has contributed to its competitiveness and spread. The amino acid substitutions and phylogenetic clustering of the study strains away from the porcine lineage IV, indicates that they are not genetically related although animal-human rotavirus transmission is often reported in the African continent.
Amino acids changes within the antigenic regions of VP7 can result in alteration to the antigenicity of the virus and potentially enhance immunity [51]. It has been shown that the antibodies targeting neutralization epitopes stabilize the capsid and prevent uncoating of the virus which is required for viral replication [52]. The observed amino acid substitutions in the antigenic region A-C within VP7 of different G12 lineages might not substantially explain the increased detection of G12 strains which have evolved naturally but might support the continual detection of more competitive G12 strains belonging to lineage III globally.
Zeller and colleagues proposed that differences in the neutralizing epitopes in VP4 could undermine the vaccines effectiveness [51]. If the vaccine e cacy is mediated through the VP4 antigen, then considering these mutations may provide further insight. The study strains had similar amino acids in most of the antigenic epitopes to the VP4 P [8] gene of RotaTeq, with some differences to Rotarix, which is the preferred vaccine in most African countries. The major amino acid substitution is in position 131, in which Rotarix had a Serine and RotaTeq and study strains had an Arginine. Arginine is a positively charged amino acid able to interact with negatively charged particles, and such change from a non-essential to essential amino acid in the VP4 rotavirus protein involved in virus entry might cause conformational changes leading to neutralizing antibody escape. It is therefore not possible to draw conclusions that the prevalence of G12 strains was affected by vaccine introduction. The amino acid changes in the neutralization epitopes of VP7 and VP4 may cause conformational changes that provide selective advantages of the new strain and establish continued infection within the population, particularly with high vaccine induced antibody levels and should be monitored prospectively. Possibly, assessing the G12 strains that have emerged in Latin America and Africa at different stages after rotavirus vaccine introduction might shed light on the evolutionary pressure exerted by the vaccines.

Conclusions
The study ndings suggest that the novel G12 strains circulating in African countries are homologous at the nucleotide and amino acid level, irrespective of geographical distribution and year of detection. The African G12P [8] and G12P [6] rotavirus strains belonging to lineage III circulating in these countries are not unique and are the same as the globally circulating rotavirus G12 strains and there is no evidence of molecular evolutionary pressure from widespread vaccine use. The antigenic epitopes display limited diversity to each other and other global strains, including to the two rotavirus vaccines (RotaTeq and Rotarix), indicating that this is unlikely to be associated with sustained circulation over time in Africa. However, as new rotavirus vaccines which do not carry the common human rotavirus VP4 genotypes, (RotaVac and RotaSIIL from India) are introduced in selected African countries, it will be imperative to continue genotypic surveillance to identify and monitor emerging strains.

Declarations
Ethics Approval and consent to participate

Consent for publication Not applicable
The partial VP7 and VP4 sequences have been made available on the NCBI GenBank database (Accession numbers: MK059426 -MK059453)