OUP user menu

Genetic and antigenic diversity of human rotaviruses: potential impact on the success of candidate vaccines

Enzo A. Palombo
DOI: http://dx.doi.org/10.1111/j.1574-6968.1999.tb08819.x 1-8 First published online: 1 December 1999


Licensing of the first human rotavirus vaccine raises the hope of a reduction in the burden of paediatric diarrheal disease. However, the less than optimal performance of this vaccine in trials carried out in developing countries indicates that improvements in vaccine design are necessary. Analysis of the genetic and antigenic diversity of rotavirus isolates collected from various geographical locations suggests that future vaccine formulations may need to include a broader spectrum of strains. This may increase vaccine efficacy by providing comprehensive coverage against circulating viruses. Continued surveillance and genetic analysis of the rotavirus population prior to and after the introduction of routine vaccination will reveal if the diversity of this pathogen will impact on the success of vaccine programs.

  • Rotavirus
  • Genetic variation
  • Vaccine development
  • Viral antigen

1 Introduction

Group A rotaviruses are recognized as the major cause of severe dehydrating gastroenteritis in infants and young children worldwide [1]. The high annual mortality in developing countries and the high morbidity in developed countries justify the introduction of effective mass vaccination programs. The first human rotavirus vaccine (a human-animal tetravalent reassortant virus mixture) was licensed in the USA in 1998 [2]. Evaluation of the long-term efficacy of this and future vaccines requires a greater understanding of the molecular epidemiology of the wild-type rotavirus population. Hence, epidemiological studies aimed at assessing the potential for genetic and associated antigenic change of the rotavirus population over time are a vital part of further vaccine development.

The rotavirus particle contains the viral genome (11 segments of double-stranded RNA) surrounded by three capsid layers consisting of a core, an inner capsid and an outer capsid [3]. Antigenic sites used for rotavirus classification are located on the immunogenic proteins VP4 (encoded by gene segment 4), VP6 (gene segment 6) and VP7 (gene segment 7, 8 or 9). VP6, the inner capsid protein, determines the group and subgroup specificity. The majority of human disease results from infection with Group A rotaviruses of either subgroup I or II specificity. VP7, the major outer capsid protein, is a glycosylated protein of 34–38 kDa and is the major target for neutralizing antibodies. There are 14 VP7 serotypes (G-types); serotypes G1, G2, G3 and G4 are the clinically and epidemiologically most important serotypes worldwide [4]. Sequence analysis of VP7 has identified regions of the protein that are conserved within serotypes but divergent between serotypes. Three of these, regions A (aa 87–101), B (aa 142–152) and C (aa 208–221), correlate with the major epitopes mapped by studies of neutralizing monoclonal antibody (NMAb) escape mutants [1,2]. VP4, the minor outer capsid protein, is an 84–88 kDa protein which is cleaved by trypsin into the subunits VP5* (60 kDa) and VP8* (28 kDa). VP4 is classified into P-types, including 13 serotypes and 19 genotypes (determined by hybridization and sequence analysis). However, the majority of clinically important strains found in humans belong to the common serotypes P1A and 1B, corresponding to genotypes P[8] and P[4].

2 Human rotavirus vaccine strategies

The goal of an effective rotavirus vaccine is to duplicate or exceed the protection induced by natural infection, the details of which are yet to be fully elucidated. Primary infection results in the production of both homotypic and heterotypic neutralizing antibodies, primarily directed against VP4 [5]. Subsequent infections result in a broadening of the cross-reactive VP7 neutralizing antibody response [5]; however, epitopes involved in eliciting the broadened VP7 response have not been defined. A number of human vaccine strategies have been developed including: (1) the use of bovine and rhesus viruses in what has been termed the ‘Jennerian’ vaccine approach; (2) bovine-human virus reassortant vaccines; (3) rhesus-human virus reassortant vaccines; (4) attenuated human virus vaccines; (5) virus-like particles composed of baculovirus-expressed proteins; and (6) DNA vaccines. Approaches (2) and (3) were deemed necessary after it was realized that serotype-specific immunity against each of the epidemiologically important human serotypes would be required for maximum protection.

The most extensively evaluated vaccine consists of an oral rhesus rotavirus reassortant tetravalent formulation (RRV-TV) that combines VP7 serotype G1, G2, G3 and G4 specificities [2]. This vaccine, however, lacks human VP4 specificity. Although RRV-TV showed high efficacy (61–100%) against severe disease in trials conducted in developed countries (USA and Finland), trials in developing countries have yielded equivocal results. One trial in Venezuela demonstrated high efficacy (88%), but efficacy rates against severe disease ranging from 0–46% in trials conducted in Brazil and Peru [2] indicate that improvements in human vaccine design are necessary. The RRV-TV vaccine has not been evaluated in Australasia or Africa, regions where serotypes not covered by this and other candidate vaccines are circulating (see below). Current vaccine efficacy could be undermined by genetic and antigenic instability of the wild-type rotavirus population, so a better understanding of the population genetics of rotavirus will assist in the development of effective vaccine formulations and strategies. Although the serotypes of strains circulating during RRV-TV vaccine trials were determined, little information about the genetic and antigenic diversity of these strains has been sought. RRV-TV has been licensed in the USA as a live, oral, three-dose vaccine, despite these uncertainties.

Recent developments have suggested a possible association between immunization with the RRV-TV vaccine and an increased rate of intussusception in children receiving 1–3 doses of the vaccine [6]. This has led to the decision to postpone immunization with this vaccine. Should a causal association be established, this may discourage the continued development of other live, oral vaccine candidates, even though trials with attenuated human virus and bovine-human reassortant virus based vaccines are underway. This, in turn, may stimulate the development of alternative vaccines and novel delivery methods (e.g. DNA vaccines, non-replicating particles, intra-nasal administration).

3 Molecular epidemiology of rotavirus field isolates

A number of molecular epidemiological studies have examined the extent of genetic diversity of wild-type rotavirus populations. Such studies have been facilitated by the advances in nucleic acid based techniques, in particular, the development and application of methods that allow the production of gene-specific cDNA, by RT-PCR, from RNA isolated from faecal samples. Sequencing of PCR-derived cDNA, either directly or after cloning, has enabled clinical samples to be rapidly and readily investigated. The majority of studies have focussed on VP7, which is the major immunogenic target of rotavirus vaccines, and on serotypes G1 or P1A[8], which are the most prevalent worldwide [4] (Fig. 1). Recent analysis of rotavirus gene sequences has yielded information about the degree of temporal diversity of clinical samples, the phylogenetic relationships between isolates and the association between genetic and antigenic diversity.

Figure 1

Global distribution of rotavirus isolates according to G- and P-types. ‘Others’ includes isolates that were not typable. Adapted from [2].

3.1 Outer capsid protein, VP7

Preliminary comparisons of VP7 amino acid sequences from standard strains indicated that a high level of conservation (>91% amino acid identity) existed within viruses of the same serotype, while inter-typic variation was more extensive (71–86% amino acid identity) [1]. This has generally been supported from genetic studies of field isolates where nucleotide and amino acid identities ranged from 92–100% (Table 1). However, two studies from different geographical locations suggested that the degree of intra-typic variation was greater, i.e. up to 11% amino acid divergence for G1 [10] and up to 19% nucleotide and 22% amino acid divergence for G3 [14] (Table 1). The G1 isolates were classified into distinct genetic and antigenic types (see below), while the high degree of divergence in G3 isolates suggested that this serotype comprised a number of subtypes. Interestingly, the geographical origins (Chinese and Japanese) of the G3 isolates correlated with the different genetic types suggesting restricted circulation of these strains and minimal exchange of viruses between these locations.

View this table:
Table 1

Temporal sequence diversity of human rotavirus structural (VP) and non-structural (NSP) genes and proteins determined from studies of field isolates

Gene/proteinAntigenic/genetic typeGeographical locationTime period of virus isolation (years)Nucleotide identity (%)Amino acid identity (%)Reference
VP6Subgroup IAustralia8n.d.100[17]
NSP4Group IIAustralia9n.d.>97[17]
Group IPhilippines/India/Thailand492−10092−100[19]
Group IIChina/Japan19>85n.d.[16]
Group IIUSA/Brazil/Uganda/India/Thailand1291−10086−100[19]
  • In most cases, sequence comparisons with standard virus strains were carried out in these studies. Only the comparisons relating to the clinical isolates under investigation are included here.

  • n.d., not determined.

Analysis of a large collection of sequences from samples collected over an extended time period in China, Japan and Pakistan suggested the existence of genetic subtypes of G1 [7]. This was supported by phylogenetic analysis of G1 VP7 sequences which revealed four major global lineages of this protein [9,20]. The global distribution of lineages indicated that some locations consisted of predominantly one lineage while others had a number of co-circulating genetic types. The presence of different G1 lineages is dynamic as demonstrated by the replacement of one type by another over a 6-year period in Australia [10]. Comparisons of the sequences of the neutralization epitope regions of VP7 revealed differences within regions of conserved amino acid sequences in different lineages; this finding suggests these should exhibit antigenic differences. Consistent with this, a study of immune responses following vaccination with RRV-TV found that serum neutralizing antibody responses stimulated by RRV-TV were significantly higher against a standard strain belonging to the same lineage as the G1 vaccine component compared to the responses against antigenically distinct G1 strains isolated from vaccine ‘failures’ who experienced diarrheal episodes [9]. The strains isolated from the vaccine ‘failures’ were subsequently found to belong to a different lineage. However, similar immune responses were found in subjects who were vaccine ‘successes’. Nonetheless, studies using NMAbs support the antigenic diversity of G1 lineages (see below). The impact of intra-serotypic diversity of circulating strains on the success of vaccines is yet to be determined. RRV-TV vaccine trials were remarkably successful in Finland, a location where all four G1 lineages were present over a 5-year period [20]. The significance of this success will remain unclear until the genetic composition of circulating strains during trials is determined.

The existence of genetic subtypes of serotype G2 VP7 was first suggested by analysis of isolates from Asia [11]. This was supported by a wider comparison of serotype G2 VP7 sequences from isolates collected in various geographical locations (Australia, China, Japan, Pakistan, Venezuela and USA) which suggested the existence of distinct genetic lineages [12]. Although showing overall conservation in VP7 sequence, serotype G2 isolates collected during separate epidemic seasons in Taiwan exhibited particular amino acid changes that could be analyzed phylogenetically [13]. This suggests the potential for continued generation of antigenic diversity. This study also concluded that some Taiwanese strains may have been imported from other countries. Specifically, the Taiwanese strains contained VP7 genes that were similar to those from atypical G2 viruses isolated in Australia at the same time. This demonstrates the potential for rapid global spread of new rotavirus strains.

Studies of G1 and G2 isolates demonstrated that the VP7 genes incorporated into the RRV-TV vaccine were genetically distinct from the majority of global isolates investigated. The G2 VP7 vaccine component (derived from the strain DS1 isolated in the United States in 1976) exhibits sequence variation in at least one of the three neutralization epitope regions when compared with other G2 viruses [12]. Significantly, a candidate bovine-human virus reassortant virus vaccine has incorporated the same human VP7 genes as RRV-TV into its design [21]. Future vaccination strategies may require that the selection of strains or antigens included in vaccine formulations consider the extent of variation in virus populations.

Antigenic subtypes (monotypes) of rotavirus VP7 G-types were first recognized by differences in the reactivity patterns to G1-specific NMAbs displayed by standard G1 viruses [22]. The genetic basis for these monotypes was defined by analysis of VP7 sequences, which indicated that the presence of particular amino acids located within neutralization epitope regions of VP7 correlated with monotype designation [23]. The combination of genetic and antigenic analysis (using NMAbs) showed that two of the monotypes (G1a and G1b) correlated with two of the global VP7 lineages (lineages II and I, respectively) described above [10]. A particular amino acid substitution at residue 94 in neutralization epitope region A correlated absolutely with monotype and lineage specificity. Monotypes of serotypes G2 and G4 have also been defined [24]. For serotype G4, temporal sequence changes in neutralization epitope regions A and C were associated with the emergence of different monotypes [15].

3.2 Other viral proteins

Studies of VP4 diversity, though few in number and not all in agreement, have suggested that variability within P-types exists (Table 1). One study of P1A[8] isolates from Australia indicated a high degree of conservation over a 17-year period [15]. In contrast, a 2-year study from the United States demonstrated remarkably greater variation in isolates of the same P-type [9]. Although Kirkwood et al. [17] showed limited variation in P2[6] isolates collected over a 10-year period (Table 1), the existence of antigenically distinct phylogenetic lineages of P2[6] has recently been suggested [25]. Although lineages within P1A[8] have also been described [20,26], antigenic differences between these have not been defined. Clearly, further studies of temporal VP4 diversity are required, especially if incorporation of human VP4 into next generation vaccines is deemed worthwhile.

The single investigation of VP6 diversity suggested that this protein is well conserved within a subgroup [17]. In contrast, the observation of marked diversity in the gene encoding VP1 (the viral RNA-dependent RNA polymerase [3]) suggested the existence of phylogenetically distinct genetic types within and between serotypes [13]. Recently, genetic subtypes of the non-structural protein NSP4 have been described [19]. Variation is apparent within (Table 1) and between these types.

4 Emerging serotypes of human rotavirus

Diversity in serotype distribution of human rotaviruses is well documented [4]. Global serotyping studies have indicated that over 83% of strains causing diarhhoea in children belong to G-types 1–4 and P-types 1A[8] and 1B[4] (Fig. 1). However, strains of less common or unusual serotypes which also cause diarrhea have been described, and constitute 14% of global isolates (Fig. 1). In particular, recent surveys carried out in Brazil, India and Bangladesh have demonstrated the emergence of serotypes G5 and G9 that are not covered by the RRV-TV vaccine formulation. Given the limited cross-protection among G serotypes induced by monovalent vaccines, this raises the concern that vaccines directed only at the common serotypes G1–4 will not prove efficacious against these less common serotypes. Serotype G9 rotaviruses are known to be present in the United States. Human infection by serotype G6, G8 and G10 viruses has also been described and these G-types may constitute the major serotype in some locations [27]. Some of these serotypes are more commonly found in animals, so their emergence in humans is seen as evidence for inter-species transmission of rotaviruses. Unusual combinations of VP7 and VP4 (whose genes can segregate independently) suggest that reassortment between common and less common serotypes or between human and animal viruses enhances the diversity of circulating strains [4].

5 Atypical VP7 types

There have been several reports of human rotaviruses that exhibit such unusual genetic and antigenic features that they are not able to be readily classified into defined serotypes. Strain M3014 was isolated in Australia in 1995 and was characterized by a VP7 gene that showed significant identity to a prototype G4 strain [28]. However, this strain was not reactive with a G4-specific NMAb. Moreover, the sequence of the neutralization epitope region A exhibited maximum identity with that of a G9 virus (14/15 amino acids) yet the virus was also non-reactive with G9-specific NMAbs. It was proposed that the VP7 of M3014 was derived by sequential mutation of a G4-like progenitor gene, resulting in a protein with novel antigenic properties. Whether this VP7 represents a new G-type in unknown.

A Brazilian strain with dual G5-G11 serotype specificity, IAL-28, was shown by sequence analysis to carry a VP7 protein that resembled both G5 and G11 prototype strains [29]. Phylogenetic analysis suggested that IAL-28 was a common ancestor of G5 and G11 viruses, both of which are commonly found in pigs. IAL-28 resembled the porcine virus MDR-13 which exhibited dual G5-G3 specificity [30] and may indicate the existence of an evolutionary link between porcine and human rotavirus VP7.

Chakladar and Chakrabarti [31] have described a novel VP7 in a virus, ID45/2, isolated in India. This VP7 exhibited only 67–78% nucleotide sequence identity to standard strains, with most amino acid differences clustered at the N-terminus of the protein. Nucleotide sequence changes caused a shift in the position of the two initiation codons and the retention of only one of the two hydrophobic signal sequences of VP7. Homology of neutralization epitope regions A and C with a standard G1 virus suggested that ID45/2 represented a new subtype of G1. However, this has not been confirmed by antigenic characterization of this strain. Of interest was the demonstration that viruses possessing an ID45/2-like VP7 represented a majority (67%) of Indian strains tested by hybridization analysis.

How are rotaviruses like M3014, IAL-28 and ID45/2 derived? Suzuki et al. [32] investigated two strains (CH55 and CHW17) that were immunologically classified as G3 but clustered with G1 viruses by phylogenetic analysis. Mathematical modelling suggested that both strains possessed cross-serotypic mosaic VP7 genes derived from intragenic recombination between G1 and G3 sequences. The crossover sites were almost the same for both strains, suggesting that the recombination events may have occurred in a common ancestor of CH55 and CHW17. Intragenic recombination leading to the exchange of antigenic regions was proposed as a mechanism by which rotaviruses could produce new variants which are capable of escaping the immune response elicited by vaccines. However, the mechanisms of recombination in rotaviruses are difficult to explain. Although mixed rotavirus infections have been reported, rotavirus RNA synthesis occurs within particles and free RNA is not found in infected cells [3]. This reduces the likelihood of homologous recombination or template switching between different viral RNA species during replication. The alternative mechanism proposed for the derivation of the M3014 VP7 (see above) may be more feasible.

6 Conclusions

This review has highlighted the extent of genetic and antigenic diversity in rotavirus field isolates belonging to serotypes covered by the recently available vaccine. Although comparisons have been made between the genetic composition of wild-type virus populations and the RRV-TV formulation, the implications of genetic and antigenic diversity for vaccine success are relevant to other vaccine strategies. In particular, variation is evident in VP7, the viral protein targeted by current vaccine strategies, and likely to be the target of other vaccines. Together with the emergence of novel serotypes, this may necessitate the inclusion of a greater variety of strains into future formulations. The selection of strains may need to take into consideration the generation of new immunological variants in this pathogen. Equally important, vaccine formulations may need to be tailored to cover the diversity of strains within particular geographical locations. Immune selection caused by wide-scale immunization with rotavirus vaccines may result in the emergence of escape mutants, as has occurred with hepatitis B [33]. Ultimately, the effect that diversity of the viral population will have on the success of human rotavirus vaccines, and the role that mass vaccination will have on rotavirus evolution, must be determined from surveillance and epidemiological studies of clinical isolates before and after the introduction of vaccination programs. Complicating factors such as nutritional status, age at infection, size of inoculum of infectious agent and interference by other enteropathogens may also impact on vaccine success in developing countries which suffer the greatest burden of rotavirus disease.


Studies in our laboratory have been supported by the National Health and Medical Research Council of Australia and the Royal Children's Hospital Research Institute. I thank Professor Ruth Bishop for support and critical comments on the manuscript.


  1. [1].
  2. [2].
  3. [3].
  4. [4].
  5. [5].
  6. [6].
  7. [7].
  8. [8].
  9. [9].
  10. [10].
  11. [11].
  12. [12].
  13. [13].
  14. [14].
  15. [15].
  16. [16].
  17. [17].
  18. [18].
  19. [19].
  20. [20].
  21. [21].
  22. [22].
  23. [23].
  24. [24].
  25. [25].
  26. [26].
  27. [27].
  28. [28].
  29. [29].
  30. [30].
  31. [31].
  32. [32].
  33. [33].
View Abstract