Rotavirus A strains obtained from children with acute gastroenteritis in Mozambique, 2012-2013: G and P genotypes and phylogenetic analysis of VP7 and partial VP4 genes

In Mozambique rotavirus (RV) was shown to be the greatest cause of acute diarrhoea in infants from 0 to 11 months, and in 2015, national rotavirus vaccination was introduced. As with other developing countries, there is very limited active strain characterisation. Rotavirus positive clinical specimens, collected between 2012 and 2013, have now provided information on the genotypes circulating in southern Mozambique prior to vaccine introduction. Genotypes G2 (32.4%), G12 (28.0%), P[4] (41.4%) and P[6] (22.9%) (n = 157) strains were commonly detected with G2P[4] (42.3%) RVs being predominant, specifically during 2013. Phylogenetic evaluation of the VP7 and VP8* encoding genes showed, for the majority of the Mozambican strains, that they clustered with other African strains based on genotype. RVA/Human-wt/MOZ/0153/2013/G2P[4], RVA/Human-wt/MOZ/0308/2012/G2P[4] and RVA/Human-wt/MOZ/0288/2012/G12P[8] formed separate clusters from the other Mozambican strains with similar genotypes, suggesting possible reassortment. Amino acid substitutions in selected epitope regions also supported phylogenetic clustering. As expected, the VP7 and VP8* genes from the Mozambican strains differed from both the RotaTeq® (SC2-9) G2P[5] and Rotarix® (A41CB052A) G1P[8] genes. This study provides information on the genetic diversity of rotavirus strains prior to vaccine introduction and generates baseline data for future monitoring of any changes in rotavirus strains in response to vaccine pressure. Electronic supplementary material The online version of this article (doi:10.1007/s00705-017-3575-y) contains supplementary material, which is available to authorized users.


Introduction
Rotaviruses are one of the leading causes of severe-dehydrating diarrhoea in infants and young children. The number of global deaths due to rotavirus infection in children under the age of five was estimated to be 215 000 in 2013; of these deaths, 56% occurred in Sub-Saharan Africa [1]. Rotaviruses are taxonomically classified within a genus of the Reoviridae family and contain an 11-segment doublestranded RNA (dsRNA) genome. The dsRNA segments encode six structural (VP1-VP4, VP6 and VP7) and six nonstructural (NSP1-NSP6) proteins. The structural viral proteins (VPs) are assembled in three concentric layers enclosing the genomic segments, the viral RNA-dependent RNA polymerase (VP1) and the viral capping enzyme (VP3). The three capsid layers consist of 60 dimers of the inner capsid protein, VP2, 260 trimers of the middle layer protein, VP6, and 780 monomers of the glycosylated VP7 protein. Spike

Rotavirus strains
Between February 2012 and September 2013, a cross-sectional study was conducted at two sites in Mozambique; Mavalane General Hospital (MGH) in Maputo and Manhiça District Hospital (MDH) in the Manhiça district (Supplementary Material 1). Faecal samples were collected from children under-five-years old that had been admitted and hospitalized with acute diarrhoea (duration no more than 7 days); defined by three or more looser-than-normal stool passages or watery diarrhoea in the 24 hours prior to the hospital visit. A total of 163 rotavirus positive specimens were detected by EIA (Oxoid, UK) and kept for further analysis [14].

RNA extraction and RT-PCR genotyping
RNA extraction was performed from 10% faecal sample suspensions in distilled water using the QIAamp Viral RNA Kit (Qiagen, USA), following the manufacturer's instructions. Total RNA was eluted in 60 µL AVE buffer. The extracted RNA was amplified in a reverse-transcription polymerase chain reaction (RT-PCR) with AMV reverse transcriptase (Promega, USA) and Taq DNA polymerase (Promega, USA). The reactions targeted the full VP7 encoding gene (sBeg9/End9; 1062 bp) and the partial VP4 encoding gene for amplification (VP8*; Con3/Con2; 876 bp) [15,16] as described previously.

Nucleotide sequencing
Nucleotide sequencing of selected VP7 and VP8* encoding genes was performed using the dideoxynucleotide chain termination method (Supplementary Material 2). Specifically, AMV reverse transcriptase (Thermo Scientific) and KAPA HiFi polymerase (Kapa Biosystems) were used in RT-PCR reactions to amplify the VP7 (sBEG/End9) [15] and VP8* (Con3/Con2) [16] encoding genes. PCR amplicons were purified using the NucleoSpin® PCR clean-up and Gel extraction kit (Macherey-Nagel, Germany), according to the manufacturer's instructions. The nucleotide sequences of these amplicons were determined using the same forward and reverse primers used for amplicon generation and the BigDye terminator v.3.1 kit (Applied Biosystems), again according to the manufacturer's instructions. Sequencing products were analysed on an ABI 3130 Genetic Analyzer (Applied Biosystems) and the resulting electropherograms were edited in CLC Bio (Qiagen).

Data analyses
Sequences were analysed using the Nucleotide Basic Local Alignment Search Tool (BLASTn) and genotypes were confirmed with the online database Virus Pathogen Database and Analysis Resource (ViPR) [19]. The nucleotide sequences of the Mozambican RVA strains have been submitted to GenBank with accession numbers KY315699-KY315722 being assigned to the VP8* and KY426071-KY426094 to the VP7 encoding sequences, respectively. Phylogenetic analyses were carried out using MEGA 7.0.14 [20]. A MUSCLE alignment was performed to align the Mozambican sequences with relevant nucleotide sequences obtained from GenBank. Phylogenetic trees were built with the Maximum Likelihood method using 1000 bootstraps. The Tamura 3 correction parameter was determined as the best substitution model [21] for the VP7 encoding sequences and the Hasegawa-Kishino-Yano model [22] for the VP8* encoding sequences. Amino acid sequences were aligned in Clustal Omega and epitopes were identified as described by Aoki and co-workers [23]. Amino acid sequences of the VP7 Mozambican strains were compared to those of the vaccine strains, RotaTeq ® (SC2-9) G2P [5] (VP7: GU565068) and Rotarix ® (A41CB052A) G1P [8] (VP7: JN849114) whereas VP8* Mozambican sequences were compared to RotaTeq ® (WI79-4) G6P [8] (VP4: GU565044) and Rotarix ® (A41CB052A) G1P [8] (VP4: JN849113).
A total of 24 G and 23 P genotypes were sequenced and the sequencing results correlated with the genotyping Table 1 Frequency of G/P genotype combinations detected in Manhiça (rural) and Mavalane (urban) area X refers to strains that were non-typeable for G or P, NT refers to strains not typed for both G and P. Gmix and Pmix represents samples with more than one G or P type detected 1  could not be confirmed as P [6].
Since the G2 genotype was missing for several samples (by genotyping PCR), the primer binding site of the G2 forward genotyping primer (aCT2) was analysed (Supplementary Material 3). Although differences were observed between the last three nucleotides of the aCT2 primer and the G2 sequences of Mozambican strains, most Mozambican G2 strains could be genotyped using the aCT2 primer ( [6] as P [8] was also investigated by comparing the primer binding regions for the respective genotypes (Supplementary Material 3). Again, it was not clear why the P [6] primer (3T-1) failed to detect the P[6] genotype using the genotyping PCR. Fifteen of the 18 base pairs in the P [8] genotyping primer (1T-1D) did not match the sequence of RVA/Human-wt/MOZ/0042/2012/GXP [6] and it is therefore unclear why the RVA/Human-wt/MOZ/0042/2012/ GXP [6] strain was incorrectly genotyped as P [8].

Discussion
The present study reports the genotyping and molecular characterisation of rotavirus strains from southern Mozambique during 2012-2013. Globally, the most common G and P rotavirus strains are G1-4, G9 and G12 as well as P [4], P [6] and P [8] [25]. In our study, G2P [4] was the prevalent genotype, accounting for 42.3% of strains detected in 2013. This differed from RVA genotypes detected in 2011 in Chókwè, a site situated closely to the area reported in this study. The G2P [4] genotype has been reported worldwide [26] and was also reported to be the predominant type (47.0%) circulating in South African in 2013 [27]. However, it is important to highlight that Mozambique had not yet introduced rotavirus vaccines during the study period while South Africa introduced Rotarix ® into their Expanded Programme on Immunisation in 2009.
We also detected uncommon genotypes such as G8P [4] and G8P [8], albeit at low rates (4-5.0%). The G8 genotype is known to cause infection in cattle and has been responsible for rotavirus infection in humans, mainly in African countries [28,29]. Recently it was reported that G8P [6] was the predominant type in São Tome Principe in 2011, being responsible for 71.1% of rotavirus cases [30]. Other genotypes detected in low frequencies were G1P [8] and G9P [8]. Genotype G1P [8] strains were the second most common genotype detected in the Chókwè study [13]. The G9 genotype was not observed at high frequencies in either the current study or the study carried out in Chókwè.
In our study, we observed a difference in distribution of rotavirus strains between the two years. In 2012, G12P [6] was predominant and more rotavirus strain diversity was noted. Genotype G2P [4] was prevalent in 2013 and in less diversity genotypes was observed compared to 2012. Variation in circulating rotavirus genotypes can occur yearly [26], with the change due to the natural variability of rotavirus strains over time. It has been suggested that rotavirus vaccines may influence the distribution of genotypes [31]. A study carried out in Belgium reported a higher prevalence of G2P [4] in vaccinated rotavirus gastroenteritis patients than in unvaccinated rotavirus gastroenteritis patients [32]. In Mozambique, the prevalence of G2P[4] will require continuous monitoring, especially because of the introduction of rotavirus vaccines in 2015.
Geographical differences in genotype distribution within the same country or region can occur [25]. We also observed differences in the geographical distribution of genotypes. In Mavalane, an urban area, G12P[6] (28.6%) was the most prevalent, while in contrast to Manhiça, a rural area, G2P [4] (39.4%) was predominant.
Analyses of primer binding sites did not shed light on the discrepancies between PCR-genotyping results and nucleotide sequencing. In our study, mixed genotypes were observed in 11.7% (13/111) of the samples typed for G and P. This is similar to the 12-14% mixed genotypes reported by the African Rotavirus Surveillance Network [26].
The phylogenetic analyses provided some evidence on the widespread circulation of rotavirus strains in Mozambique, with similar strains detected in Manhiça, Mavalane and Chókwè (Supplementary Material 1). The Chókwè G12 and P [6] strains phylogenetically clustered with strains detected in the current study indicating that these strains circulated in southern Mozambique from 2011 to 2013 (Fig. 3A, B). Interestingly, no G8 strains were identified in the Chókwè study during 2011 [13].