Evidence of Vaccine-related Reassortment of Rotavirus, Brazil, 2008–2010

Analysis of 27 rotavirus strains from vaccinated and unvaccinated children revealed reassortment events in 3 strains: a gene derived from a vaccine; a gene acquired from a circulating strain; and reassortment between circulating strains. Data suggest that the widespread use of this monovalent rotavirus vaccine may introduce vaccine genes into circulating human rotaviruses or vice versa.

Analysis of 27 rotavirus strains from vaccinated and unvaccinated children revealed reassortment events in 3 strains: a gene derived from a vaccine; a gene acquired from a circulating strain; and reassortment between circulating strains. Data suggest that the widespread use of this monovalent rotavirus vaccine may introduce vaccine genes into circulating human rotaviruses or vice versa. G roup A rotaviruses (RVAs) are a frequent cause of diarrhea in children. The RVA genome consists of 11 dsRNA segments that encode 6 structural (VP1-VP4, VP6, VP7) and 6 non-structural (NSP1-NSP6) proteins (1). The basis of a new classification system is phylogenetic analysis of 11 RVA genome segments, although binary classification is still used; genotyping is based on the coding genes for VP7 (G) and VP4 (P) (2).
Vaccination is considered effective in reducing the consequences of RVA. Two vaccines, Rotarix (Glaxo SmithKline, Brentford, UK) and RotaTeq (Merck & Co., Whitehouse Station, NJ, USA), are licensed in several countries. Both vaccines demonstrated broad protection against the most common RVA genotypes (3).
In Brazil, Rotarix, a monovalent attenuated human rotavirus vaccine for infants 6-24 weeks of age, was introduced in the National Immunization Programs in March, 2006. The vaccine is delivered in 2 doses, >4 weeks apart. In 2009, when vaccine coverage achieved 85.9%, reduction in hospitalization associated with diarrhea (17%) and in related mortality rates (22%) was observed (4). This study aimed to assess phylogenetic relationships between G1P [8] RVA obtained from vaccinated and unvaccinated children hospitalized for acute gastroenteritis.
Strain ES15221-08, detected in an unvaccinated child, is genetically distinct and differs in origin from other G1P [8] circulating strains (online Technical Appendix Figures 1, 2). The VP1 gene (648 bp; online Technical Appendix Figure  2, panel A) in this sample showed 100% nt identity with the Rotarix strain. Strain MA19006-10 was genetically similar to the strain; however, the NSP5 segment (online Technical Appendix Figure 1, panel E) was closely related to a G1P [8] strain from Australia (JF490152). These 2 strains appear to have been generated by reassortment with this vaccine strain.
Strain MA19030-10, detected in a child after 1 vaccine dose, was closely related to the MA group, but the NSP2 segment differed in origin from other MA samples (online Technical Appendix Figure 1, panel B) because it clustered with other Wa-like strains. The NSP3 gene belonged to genotype 3 because it was 99.2% (nt) similar to the AU-1 prototype strain (DQ490535.1). These results suggested reassortment events between Wa-like and AU-1 like co-circulating strains.
NSP2 and NSP3 genes from sample MA19030-10 and NSP5 genes from sample MA19006-10 differed from each other and from those of their respective clusters. Samples MA19006-10 and MA19030-10 are genetically distinct and might have distinct evolutionary histories. Studies that included this genogrouping system were performed to prove the existence of inter-genogroup reassortment between human RVA genogroups or human and animal genogroups. The existence and effectiveness of heterogeneous genome constellations remains unclear, probably because it is caused by mechanisms that create protein sets that work better when kept together (8).
Phylogenetic analysis of the VP8* (aa 1-247) portion of VP4 encoding gene showed circulating strains (online Technical Appendix Figure 2, panel D) clustered into 2 lineages (P[8]-3 and P[8]-1). The alignment of the deduced aa sequences showed potential trypsin cleavage sites at arginine 240 and 246 conserved in all samples. All circulating strains contained 91.3%-100% identical aa residues to the Rotarix strain in VP8* antigenic epitopes; no changes were observed on epitopes 8-2 and 8-4.
Two VP7 lineages (G1-I and G1-II) were identified. Samples closely related to the Rotarix strain belonged to G1-II; remaining strains belonged to G1-I. Comparing the regions defined as antigenic epitopes (7-1 and 7-2) for VP7 protein, we found >2 epitopes (aa 94, 123, 148, 217) were not conserved among circulating strains compared with the Rotarix strain. The following amino acids were conserved in the non-VP7/non-VP4 segments of either the Rotarix strain or other strains analyzed: cysteine residues involved in disulfide bonds at positions 6, 8, 85, and 285 in NSP2 protein; the leucine, isoleucine, aspartic acid, methionine and glutamine at positions 275, 281, 288, and 295 in the NSP3 protein; the N-linked glycosylation sites at positions 8 and 18, and cysteine residues at positions 63 and 71 in NSP4 protein; and serine residues at positions 153, 155, 163, and 165 in NSP5 protein.

Conclusions
This report characterizes the complete genome of G1P [8]  sequences clustered consistently with the region of sample collection. Three strains circulating in Brazil were closely related to those in Rotarix; 1 of the 3 was 100% identical to Rotarix, likely representing shedding of this vaccine. Sequence analysis confirmed the presence of the Rotarix VP1-derived segment in 1 sample (ES15221-08), indicating an unreported reassortment event between the vaccine and a community strain. The backbone of MA19030-10 sample was Wa-like, but the NSP3 segment exhibited a T3 genotype that was described for the AU-1 genogroup. This finding suggests that this strain derived its NSP3 gene from an AU-1-like strain through reassortment.
Changes in antigenic regions of VP4 and VP7 proteins have been associated with mutated RVA strains that spread (9,10). Comparison of the aa sequences of VP7 and VP8* of strains circulating in Brazil and the monovalent vaccine strain demonstrated that VP7 and VP8* of the circulating strains showed similar antigenic regions to those of the vaccine. No differences between strains from vaccinated and unvaccinated children were observed.
Considering the segmented RVA genome and that the Rotarix vaccine is an attenuated RVA human strain, it is expected that reassortants will arise and circulate among humans. The effects of such events are not known. This study described strains that originated from reassortment events between the Rotarix vaccine strain and strains detected in vaccinated and unvaccinated children. Improvement of RVA surveillance programs that include full genome sequencing analysis will strengthen the understanding of how vaccines will affect the RVAs circulating among humans, and how those events could affect the use of live vaccines, the frequency of RVA intra-and intergenogroup reassortment events under natural conditions, and the stability of RVA generated by such events.