Phylogenetic analysis of the viral proteins VP4/VP7 of circulating human rotavirus strains in China from 2016 to 2019 and comparison of their antigenic epitopes with those of vaccine strains

Group A rotaviruses (RVAs) are the most common etiological agents of severe acute diarrhea among children under 5 years old worldwide. At present, two live-attenuated RVA vaccines, LLR (G10P[15]) and RotaTeq (G1–G4, G6 P[8], P[5]), have been introduced to mainland China. Although RVA vaccines can provide homotypic and partially heterotypic protection against several strains, it is necessary to explore the genetic and antigenic variations between circulating RVAs and vaccine strains. In this study, we sequenced viral protein VP7 and VP4 outer capsid proteins of 50 RVA strains circulating in China from 2016 to 2019. The VP7 and VP4 sequences of almost all strains showed high homology to those of previously reported human strains and vaccine strains of the same genotype. However, in the presumed antigenic epitopes of the VP7 and VP4, multiple amino acid variations were found, regardless of the G and P genotypes of these strains. Moreover, all circulating G3 RVA strains in China potentially possess an extra N-linked glycosylation site compared with the G3 strain of RotaTeq. The potential N-linked glycosylation site at residues 69–71 was found in all G9 strains in China but not in the G9 strain of the Rotavac or Rotasill vaccine. These variations in antigenic sites might result in the selection of strains that escape the RVA neutralizing-antibody pressure imposed by vaccines. Furthermore, the G4 and P[6] genotypes in this study showed high homology to those of porcine strains, indicating the transmission of G4 and P[6] genotypes from pigs to humans in China. More genetic surveillance with antigenic evaluation in prevalent RVAs is necessary for developing and implementing rotavirus vaccines in China.


Introduction
Group A rotaviruses (RVAs) were first detected in the 1970s and are the most common etiological agents of severe acute gastroenteritis among children under 5 years old worldwide (Banyai et al., 2018;Raju et al., 2019).Rotavirus gastroenteritis (RVGE) caused an estimated 128,500 deaths in 2016, mostly in low-income countries, and rotavirus vaccines have reduced deaths by more than 280,000 (Troeger et al., 2018).In Asia, rotavirus accounts for 37.5% of hospitalized gastroenteritis cases (Troeger et al., 2018).In addition, treatment of RVGE incurs the greatest cost (US$365 million annually) in China, followed by Japan and India (Kawai et al., 2012).Vaccination is the most effective means of preventing RVGE.The first (RotaTeq andRotarix, approved in 2008 and2009, respectively) and subsequent (Rotavac and Rotasill, approved in 2018) rotavirus vaccines were prequalified by the World Health Organization (World Health Organization, 2021;Sadiq et al., 2022) and have contributed to the rapid decline in deaths caused by RVAs.The global rates of diarrhea among hospitalized children under 5 years old before the introduction of rotavirus vaccines in 2013 and after the introduction of rotavirus vaccines in 2019 were 38% and 23%, respectively (Sadiq et al., 2018;World Health Organization, 2021).
VP7 and VP4 are outer capsid proteins of RVA, and their genes are the most important antigenic genes for the major neutralizing antigenic epitopes in RVA vaccines (Estes and Cohen, 1989).However, the segmented genomic complexity with evolution of the VP7 and VP4 genes has reduced the efficacy of RVA vaccines (Estes and Cohen, 1989).We performed phylogenetic analysis of the VP4 and VP7 of circulating RVA strains in China from 2016 to 2019 and identified potentially important antigenic disparities compared with these proteins of vaccine strains, to facilitate the development and introduction of RVA vaccines in China.

Nucleic acid extraction
Stool samples (100 mg) were suspended in 1 ml of Hanks' Balanced Salt Solution (HBSS) solution, homogenized by vortex mixing, and centrifuged.Viral RNA was extracted from stool samples and purified using the QIAamp viral RNA Mini Kit (Qiagen, Hilden Germany) according to the manufacturer's instructions.

Amplification of VP7 and VP4
Extracted viral double-stranded RNA was used as the template for Multiplex RT-PCR using specific VP7 and VP4 consensus primer pairs (Matthijnssens et al., 2006).The full CDS region of VP7 (978 base pairs) and partial CDS region of VP4 (2088 base pairs) were amplified.RT-PCR was performed using the Qiagen One-Step RT-PCR kit (Qiagen, Hilden Germany) according to the manufacturer's instructions.RT-PCR was also performed with an independent RT step using the SuperScript ™ III Reverse Transcription Kit (Invitrogen California, USA).Taq polymerase was activated for 3 min at 94°C, followed by 35 cycles of amplification (30 s at 94°C, 30 s at 55°C, and 60 s at 72°C), with a final extension for 10 min at 72°C.

Sequence analysis
The nucleotide sequences of VP7 and VP4 were analyzed, and consensus alignments were conducted using ClustalW (Larkin et al., 2007).Maximum-likelihood phylogenetic trees were drawn in MEGA 7.0 software based on the GTR+G+I model and calculated by the bootstrap method using 1,000 replicates.All positions with less than 95% site coverage were eliminated.Amino acid sequence similarity was calculated using MegAlign (DNA Star), and potential N-linked glycosylation sites were screened using NetNGlyc1.0Server (https://services.healthtech.dtu.dk/service.php?NetNGlyc-1.0).For antigenic characterization, sequences were aligned using BioEdit, and structural diagrams were generated using Pymol 2.5.The VP7 and VP4 sequences were submitted to GenBank under accession numbers OM920725-920812 (sequenced in this study) or downloaded (OM037825-038088).
The circulating G3 strains were included in G3 lineages I, IV, and V and clustered with strains from Belgium, the USA, Russia, Japan, and G3 strains isolated previously in China (Figure 1C).RotaTeq G3 had higher amino acid homology (96.6%-97.3%)with G3 lineage I than with G3 lineage IV (93.9%) or G3 lineage V (91.6%-92.7%)(Table 2).G4 is rare in China; the one VP4 sequence sequenced in this study showed 84.41%-85.01%homology to that of other human G4 strains reported in China and 96.84% and 95.87% identities with that of HLJKD and HLJ strains isolated from porcine stool samples in China (Figure 1D).This G4 strain had low pairwise homology (87.3%) to RotaTeq G4 (Table 2).
Seventeen G9 strains were included in G9 lineage VI-e and three in G9 lineage III-d.The 20 G9 strains clustered with strains from Belgium, Russia, and China (Figure 1E).They showed 91.6%-93.6%identity with the G9 strain of Rotavac (lineage II) and 93.3-94.3%identity with the G9 strain of Rotasill (lineage I) (Table 2).
Compared with RotaTeq G3, the G3 strains showed five or six amino acid differences in the VP7 antigenic epitopes.Indeed, their VP7 antigenic epitopes contained mutations at residues 238 (K238D/N) and 242 (D242A/N), both in the 7-1b epitope (Figure 3A).Interestingly, each of these G3 strains contained K238N or D145N, which provides a potential N-linked glycosylation site that does not exist at the corresponding position in RotaTeq G3 (Ben Hadj Zeller et al., 2012;Fredj et al., 2013).KN238 or D145N was surrounded by other residues of epitope regions (Figure 3B).This glycosylation may have a profound impact on the antigenicity of this epitope.

Discussion
Live-attenuated RVA vaccines have been introduced in more than 100 countries, contributing to herd immunity (Zeller et al., 2012;Ben Hadj Fredj et al., 2013).In countries in which an RVA vaccine was included in the national immunization plan (NIP), the number of hospitalizations due to rotavirus infection has decreased markedly (Cates et al., 2022).However, the RVA vaccine was more effective against severe rotavirus disease in high-income countries (80%-90%) than in low-to-middleincome countries (40%-70%) (Cates et al., 2022).Rotavirus vaccination does not affect the natural turnover of rotavirus strains.However, the fact that v were no longer predominant after RotaTeq and Rotarix introduction (Patel et al., 2011), vaccination is important for surveillance of circulating rotavirus strains.VP7 and VP4 are the most important rotavirus neutralization antigens, and analysis of their antigenic differences between circulating and vaccine strains is warranted.
In 2001, the predominant rotavirus strains shifted from G1 to G3 in China; after 2012, G3 was replaced by G9.The predominant P genotype has been P [8] for the past 20 years.In fact, from January 2016 to December 2019, 5147 samples were genotyped by PCR and sequencing.G9P [8] was most predominant, followed by G3P [8], G2P[4], G1P [8], G2P[4], and G4P [6] (unpublished data from the National Viral Diarrhea Surveillance Network of China).On the basis of preliminary phylogenetic analyses, we selected 50 strains representative of the circulating rotavirus genotypes and Phylogenetic analysis of the VP4 protein of circulating and vaccine RVA strains (RotaTeq P [8], P[5], Rotarix P [8], LLR).Maximum-likelihood trees (A-C) were constructed based on the partial VP7 CDS region sequences (2088 base pairs).A GTR+G+I nucleotide substitution model was used to construct the phylogenetic tree.The pigeon RVA strain WVL21015-FL was used as the outgroup.Chinese strains are marked by red dots and vaccine strains by black triangles.Bootstrap values (1,000 replicates) of 70% are shown.
investigated the extent of their prevalence (years and regions) in mainland China.The two most widely used vaccines, Rotarix and RotaTeq, were included in the phylogenetic and neutralization epitope analyses of the VP7 and VP4 genes.Furthermore, Rotasill and Rotavac (G9 genotype) were included in the analysis of VP7 of G9.LLR (G10P[15] genotype), a completely attenuated strain from sheep, obtained marketing approval in 2001 in China.Because it has marked sequence differences in VP7 and VP4 compared with prevalent human strains, it was not analyzed further.
G1 RVA strains are generally of the P [8] genotype; therefore, both RotaTeq and Rotarix (which have G1 and P [8]) can provide homotypic protection against G1and P [8] strains.However, the VP7 epitopes of the prevalent G1 RVA strains in this study had N94S, which is associated with immune escape (Taniguchi et al., 1988).G2 RVA strains, which are uncommon in China, are generally associated with the P[4] genotypes.Compared with G2 of RotaTeq, only minor mutations were found in antigenic epitopes in the 7-1a and 7-1b regions of G2 strains.Compared with other G genotype strains, G3 strains did not possess more mutations but rather additional glycosylation site mutations, N238 (7-1b) and N145 (7-2), which are absent from G3 vaccine strains.Combined with previous reports and VP7 structural models, the immunogenicity of the 7-1a epitope could also be affected by glycosylation of residue N238 (Ben Hadj Fredj et al., 2013).K238N reduced neutralization of animal RVA strains (Ciarlet et al., 1994;Ciarlet et al., 1997).Therefore, this glycosylation site mutation may explain the antigenicity difference compared with G3 vaccine strains.(A) Alignment of the antigenic epitopes in VP7 of the RVA strains circulating in China with those in Rotarix, RotaTeq, and other vaccines.Red box, the sites involved in neutralization escape.(B) Surface representation of the VP7 trimer (PDB 3FMG).Antigenic epitopes are in lime green (7-1a), yellow (7-1b), and cyan (7-2).Red, the surface-exposed residues that differ between circulating and vaccine RVA strains.
Analysis of the VP7 gene of G9 revealed that three later strains (GX19-1013, SZ18-2197, and SC19-1063) in G9 lineage III-d clustered with earlier Chinese strains (E1545), and other G9 lineage VI-e strains clustered with newly reported Chinese strains in GenBank.Similar findings were reported by a study in Jiangsu Province, China (Wang et al., 2015).However, in Africa and Belgium, G9 strains independently clustered in G9 lineage VI or III (Zeller et al., 2012;Xu et al., 2018;Manouana et al., 2021).Moreover, the 20 G9 strains in this study had potential Nlinked glycosylation sites (positions 69-71), which are absent from VP7 of Rotasill and Rotavac.In addition, these glycosylation sites are rare in field RVA strains.Glycosylation determines viral immunogenicity by modulating virus receptor binding or masking antigenic sites, as in SARS-CoV-2 (Petrovic et al., 2021).Further studies should address whether these glycosylation sites alter the antigenicity of VP7 (Zeller et al., 2012;Harastani et al., 2020) The G4 and P [6] genotypes in this study displayed marked intra-genotypic variety and high homology to porcine strains, indicating spillover of G4 and P [6] from pigs to humans in China.Indeed, P [6] is a widespread zoonotic RVA genotype in developing countries (Malasao et al., 2018).Consequently, P [6] is likely to spread between human and animals as a result of poor sanitation and constant contact with livestock.No approved RVA vaccine contains a P [6] or P[4] genotype strain.Partial heterotypic protection against P[4] and P [6] RVA is provided by the approved live-attenuated vaccines.However, it is difficult to compare the antigenic epitopes of P [6] or P[4] genotypes with those of vaccine strains.
P [8] is the most common rotavirus P genotype worldwide.Rotateq, Rotarix, and Rotavac with the P [8] genotype provide homotypic protection against P [8] strains.Consistent with reports from other countries, P [8] strains in China are mainly of P [8] lineage III (Zeller et al., 2012;Sadiq and Bostan, 2020).Interestingly, strain SD18-0121, the sole exemplar of P [8] lineage IV, showed low amino acid sequence similarity with the P [8] of vaccine strains, as is evident in the structure of VP4.VP8* is the location of most antigen mutations between vaccine and circulating strains.However, the P [8] RVA strains in this study had few amino acid changes in their VP8* epitopes.Four strains (NM16-1059, GX16-1119, HEB16-1059, and SD18-0121) showed amino acid mutations in VP5*.These amino acid differences in antigenic epitopes warrant further investigation of their effects on vaccine efficacy.
RVA vaccination has significantly reduced the burden of RVA disease among children worldwide, especially in developing countries.RotaTeq and LLR are approved in mainland China but are not included in the NIP.A study in six provinces of China in 2018 showed a low rate of LLR rotavirus vaccination and late vaccination age, which may be improved by addition of an RVA vaccine to the NIP.Compared with the vaccine strains, the amino acid differences in the VP7 and VP4 antigenic epitopes of Chinese strains may reduce vaccine effectiveness, necessitating further research on these epitopes.In addition, efforts to accelerate the development of new RVA vaccines, including reverse genetics, are warranted.
FIGURE 3 FIGURE 4 (A) Alignment of the antigenic epitopes in VP4 of RVA strains circulating in China with those in Rotarix, RotaTeq, and other vaccines.Red box, the site involved in neutralization escape.(B) Surface representation of the VP8* monomer (PDB 1KQR).Upper and lower images are of the front and rear, respectively, of VP8*.Antigenic epitopes are shown in chartreuse (8-1), yellow-orange (8-2), and violet (8-3).Red surfaceexposed residues indicate differences between circulating and vaccine RVA strains.

TABLE 1
VP7 and VP4 lineage, date of isolation, and patient information for the Chinese rotavirus strains analyzed in this study and vaccine strains.

TABLE 2
Distance matrix for VP7 and VP4 based on amino acid identities.