Polypeptide composition of rotavirus empty capsids and their possible use as a subunit vaccine

Development and characterization of a plant-derived rotavirus-like particle vaccine

BACKGROUND

Virus-like particles (VLPs) are unable to replicate in the recipient but stimulate the immune system through recognition of repetitive subunits. Parenterally delivered rotavirus-VLP (Ro-VLP) vaccine could have the potential to overcome the weaknesses of licensed oral live-attenuated rotavirus vaccines, namely, low efficacy in low-income and high mortality settings and a potential risk of intussusception.

METHODS

A monovalent Ro-VLP composed of viral protein (VP) 7, VP6 and VP2 of G1 genotype specificity was produced in Nicotiana benthamiana using Agrobacterium tumefaciens infiltration-based transient recombinant expression system. Plants expressing recombinant G1 Ro-VLP were harvested, then the resultant biomass was processed through a series of clarification and purification steps including standard extraction, filtration, ultrafiltration and chromatography. The purified G1 Ro-VLP was subsequently examined for its immunogenicity and toxicological profile using animal models.

RESULTS

G1 Ro-VLP had a purity of ≥90% and was structurally similar to triple-layered rotavirus particles as determined by cryogenic transmission electron microscopy. Two doses of aluminum hydroxide-adjuvanted G1 Ro-VLP (1 μg, 5 μg or 30 μg), administered intramuscularly, elicited a robust homotypic neutralizing antibody response in rats. Also, rabbits administered G1 Ro-VLP (10 μg or 30 μg) four times intramuscularly with aluminum hydroxide adjuvant did not show any significant toxicity.

CONCLUSIONS

Plant-derived Ro-VLP composed of VP7, VP6 and VP2 structural proteins would be a plausible alternative to live-attenuated oral rotavirus vaccines currently distributed worldwide.

Chimaeric virus-like particles derived from consensus genome sequences of human rotavirus strains co-circulating in Africa

Rotavirus virus-like particles (RV-VLPs) are potential alternative non-live vaccine candidates due to their high immunogenicity. They mimic the natural conformation of native viral proteins but cannot replicate because they do not contain genomic material which makes them safe. To date, most RV-VLPs have been derived from cell culture adapted strains or common G1 and G3 rotaviruses that have been circulating in communities for some time. In this study, chimaeric RV-VLPs were generated from the consensus sequences of African rotaviruses (G2, G8, G9 or G12 strains associated with either P[4], P[6] or P[8] genotypes) characterised directly from human stool samples without prior adaptation of the wild type strains to cell culture. Codon-optimised sequences for insect cell expression of genome segments 2 (VP2), 4 (VP4), 6 (VP6) and 9 (VP7) were cloned into a modified pFASTBAC vector, which allowed simultaneous expression of up to four genes using the Bac-to-Bac Baculovirus Expression System (BEVS; Invitrogen). Several combinations of the genome segments originating from different field strains were cloned to produce double-layered RV-VLPs (dRV-VLP; VP2/6), triple-layered RV-VLPs (tRV-VLP; VP2/6/7 or VP2/6/7/4) and chimaeric tRV-VLPs. The RV-VLPs were produced by infecting Spodoptera frugiperda 9 and Trichoplusia ni cells with recombinant baculoviruses using multi-cistronic, dual co-infection and stepwise-infection expression strategies. The size and morphology of the RV-VLPs, as determined by transmission electron microscopy, revealed successful production of RV-VLPs. The novel approach of producing tRV-VLPs, by using the consensus insect cell codon-optimised nucleotide sequence derived from dsRNA extracted directly from clinical specimens, should speed-up vaccine research and development by by-passing the need to adapt rotaviruses to cell culture. Other problems associated with cell culture adaptation, such as possible changes in epitopes, can also be circumvented. Thus, it is now possible to generate tRV-VLPs for evaluation as non-live vaccine candidates for any human or animal field rotavirus strain.

Triple layered rotavirus VLP production: Kinetics of vector replication, mRNA stability and recombinant protein production

Rotavirus infection causes diarrhoeal disease in infants, killing more than half million children each year. Virus-like particles (VLP) seem to be excellent vaccine candidates, since they are cheaper to produce than attenuated viral vaccines and safer, as they do not contain genetic material. The present work focus on a triple layered particle composed by three rotavirus structural proteins: VP2, VP6 and VP7, produced in an insect cell/baculovirus expressing system. Two strategies were evaluated for 2/6/7 VLP production: co-infection with three monocistronic baculovirus vectors or single-infection with a tricistronic multi-gene baculovirus vector; these strategies were followed at different levels: baculovirus DNA replication kinetics, mRNA stability, protein production and VLP formation. This study highlights some of the reasons why the tricistronic baculovirus strategy is more efficient for production of triple layered rotavirus 2/6/7 VLP than monocistronic co-infection, in particular: (i) the tricistronic vector presents higher DNA replication rates than the monocistronic vectors, (ii) the mRNA stability is invariant for all mRNAs corresponding to VP2, VP6 and VP7 and (iii) the tricistronic baculovirus strategy produces an excess of VP7 over VP6 when compared to the VP7/VP6 stoichiometric ratio in the native rotavirus. Although the co-infection strategy leads to protein production akin to the rotavirus VP7/VP6 stoichiometric ratio, the tricistronic vector strategy yields higher amounts of rotavirus-like particles.

Cloning of complete genome sets of six dsRNA viruses using an improved cloning method for large dsRNA genes

Cloning full-length large (>3 kb) dsRNA genome segments from small amounts of dsRNA has thus far remained problematic. Here, a single-primer amplification sequence-independent dsRNA cloning procedure was perfected for large genes and tailored for routine use to clone complete genome sets or individual genes. Nine complete viral genome sets were amplified by PCR, namely those of two human rotaviruses, two African horsesickness viruses (AHSV), two equine encephalosis viruses (EEV), one bluetongue virus (BTV), one reovirus and bacteriophage Phi12. Of these amplified genomes, six complete genome sets were cloned for viruses with genes ranging in size from 0.8 to 6.8 kb. Rotavirus dsRNA was extracted directly from stool samples. Co-expressed EEV VP3 and VP7 assembled into core-like particles that have typical orbivirus capsomeres. This work presents the first EEV sequence data and establishes that EEV genes have the same conserved termini (5' GUU and UAC 3') and coding assignment as AHSV and BTV. To clone complete genome sets, one-tube reactions were developed for oligo-ligation, cDNA synthesis and PCR amplification. The method is simple and efficient compared to other methods. Complete genomes can be cloned from as little as 1 ng dsRNA and a considerably reduced number of PCR cycles (22-30 cycles compared to 30-35 of other methods). This progress with cloning large dsRNA genes is important for recombinant vaccine development and determination of the role of terminal sequences for replication and gene expression.

Effect of MOI ratio on the composition and yield of chimeric infectious bursal disease virus-like particles by baculovirus co-infection: deterministic predictions and experimental results

Virus-like particles (VLPs) are empty particles consisting of virus capsid proteins that closely resemble native virus but are devoid of the native viral nucleic acids and therefore have attracted significant attention as noninfectious vaccines. A recombinant baculovirus, vIBD-7, which encodes the structural proteins (VP2, VP3, and VP4) of infectious bursal disease virus (IBDV), produces native IBD VLPs in infected Spodoptera frugiperda insect cells. Another baculovirus, vEDLH-22, encodes VP2 that is fused with a histidine affinity-tag (VP2H) at the C-terminus. By co-infection with these two baculoviruses, hybrid VLPs with histidine tags were formed and purified by immobilized metal affinity chromatography (Hu et al., 1999). Also, we demonstrated that varying the MOI ratio of these infecting viruses altered the extent of VP2H incorporated into the particles. A dynamic mathematical model that described baculovirus infection and VLP synthesis (Hu and Bentley, 2000) was adapted here for co-infection and validated by immunofluorescence labeling. It was shown to predict the VLP composition as a dynamic function of MOI. A constraint in the VP2H content incorporated into the particles was predicted and shown by experiments. Also, the MOI ratio of both infecting viruses was shown to be the major factor influencing the composition of the hybrid particles and an important factor in determining the overall yield. ELISA results confirmed that VP2H was exhibited to a varied extent on the outer surface of the particles. This model provides insight on the use of virus co-infection in virus-mediated recombinant protein expression systems and aids in the optimization of chimeric VLP synthesis.

Glycosphingolipid binding specificities of rotavirus: identification of a sialic acid-binding epitope

The glycosphingolipid binding specificities of neuraminidase-sensitive (simian SA11 and bovine NCDV) and neuraminidase-insensitive (bovine UK) rotavirus strains were investigated using the thin-layer chromatogram binding assay. Both triple-layered and double-layered viral particles of SA11, NCDV, and UK bound to nonacid glycosphingolipids, including gangliotetraosylceramide (GA1; also called asialo-GM1) and gangliotriaosylceramide (GA2; also called asialo-GM2). Binding to gangliosides was observed with triple-layered particles but not with double-layered particles. The neuraminidase-sensitive and neuraminidase-insensitive rotavirus strains showed distinct ganglioside binding specificities. All three strains bound to sialylneolactotetraosylceramide and GM2 and GD1a gangliosides. However, NeuAc-GM3 and the GM1 ganglioside were recognized by rotavirus strain UK but not by strains SA11 and NCDV. Conversely, NeuGc-GM3 was bound by rotaviruses SA11 and NCDV but not by rotavirus UK. Thus, neuraminidase-sensitive strains bind to external sialic acid residues in gangliosides, while neuraminidase-insensitive strains recognize gangliosides with internal sialic acids, which are resistant to neuraminidase treatment. By testing a panel of gangliosides with triple-layered particles of SA11 and NCDV, the terminal sequence sialyl-galactose (NeuGc/NeuAcalpha3-Galbeta) was identified as the minimal structural element required for the binding of these strains. The binding of triple-layered particles of SA11 and NCDV to NeuGc-GM3, but not to NeuAc-GM3, suggested that the sequence NeuGcalpha3Galbeta is preferred to NeuAcalpha3Galbeta. Further dissection of this binding epitope showed that the carboxyl group and glycerol side chain of sialic acid played an important role in the binding of such triple-layered particles.

Chimeric infectious bursal disease virus-like particles expressed in insect cells and purified by immobilized metal affinity chromatography

Chimeric virus-like particles (VLPs) of infectious bursal disease virus (IBDV) were produced by coinfecting Spodoptera frugiperda (Sf-9) insect cells with two recombinant baculoviruses, vIBD-7 and vEDLH-22. vIBD-7 encodes VP2, VP3, and VP4 of the IBDV structural proteins. vEDLH-22 encodes VP2 with five histidine residues at the carboxy-terminus (VP2H). Coinfection produced hybrid VLPs composed of VP2, VP2H, and VP3. The additional histidine residues on VP2H enabled the efficient purification of VLPs based on immobilized metal affinity chromatography (IMAC). These results demonstrated that the VLPs formed are comprised of chimeric subunits with attached affinity ligands, and further, that sufficient His5 ligand was available for binding to the IMAC metal-chelating resin. Additionally, these novel particles were fully characterized for antigenicity by a series of monoclonal antibodies, and appeared identical to the two wild-type IBDV strains contributing subunits to the chimeric VLP. IMAC purification provides a promising low-cost and simple scheme to purify VLPs as vaccines.

Biochemical and immunologic comparison of virus-like particles for a rotavirus subunit vaccine

A parenterally administered rotavirus vaccine composed of virus-like particles (VLPs) is being evaluated for human use. VLPs composed of bovine VP6 and simian VP7 (SA11, G3) proteins (6/7-VLPs) or of bovine VP2, bovine VP6, and simian VP7 (SA11, G3) proteins (2/6/7-VLPs) were synthesized and purified from Sf9 insect cells co-infected with recombinant baculoviruses. 6/7- and 2/6/7-VLP administered parenterally (i.m.) in mice had comparable immunogenicity, but the 2/6/7-VLPs were more homogeneous and stable. The inclusion of the VP2 capsid contributed to particle formation and stability. The adjuvant QS-21 significantly enhanced the immunogenicity of 2/6/7-VLPs over A10H or saline alone. Equivalent serum neutralizing antibody responses were induced over the range of 1-15 microg/dose of 2/6/7-VLPs administered with the range of 5-20 microg/dose of QS-21. The immunogenicity of 2/6/7-VLPs and inactivated SA11 virus were comparable. 2/6/7-VLPs are a promising candidate for a parenterally delivered rotavirus subunit vaccine.

Passive immunity to bovine rotavirus in newborn calves fed colostrum supplements from cows immunized with recombinant SA11 rotavirus core-like particle (CLP) or virus-like particle (VLP) vaccines

Heterotypic passive immunity to IND (P/5/G6) bovine rotavirus (BRV) was evaluated. Three groups of calves (n = 5 per group) were fed 1% pooled colostrum supplements (birth to 7 days of age) from BRV seropositive cows vaccinated with recombinant SA11(P/2/G3) rotavirus-like particles (VLPs), recombinant SA11 rotavirus core-like particles (CLPs), or inactivated SA11 rotavirus (SA11). Control calves (n = 5 per group) received either pooled colostrum from unvaccinated (BRV field exposure seropositive) control cows, or no colostrum. IgG1 antibody titers to IND BRV for the pooled colostrum were: 1,048,576 (VLP); 1,048,576 (CLP); 262,144 (SA11); and 16,384 (control colostrum). Elevated titers of BRV neutralizing (VN) antibodies were present in VLP colostrum (98,000), and SA11 colostrum (25,000), but not in CLP colostrum (1400), compared to colostrum from nonvaccinates (2081). Calves were orally inoculated with virulent IND BRV at 2 days of age and challenged at post-inoculation day (PID) 21. Calves were monitored daily for diarrhea and faecal BRV shedding through PID 10 and post-challenge day (PCD) 10. After colostrum feeding, the IgG1 antibody titers were highest in serum and faeces of calves fed VLP and CLP colostrum, but VN and IgA antibodies were highest in calves fed VLP colostrum. After BRV inoculation, calves fed colostrum from vaccinated cows had significantly fewer days of BRV-associated diarrhea and BRV shedding than control calves. All calves fed VLP colostrum were protected from diarrhea after BRV inoculation; two calves shed BRV. In the CLP colostrum group, one calf developed BRV-associated diarrhea and all calves shed virus. In the SA11 colostrum group, three calves developed BRV-associated diarrhea and four calves shed virus. BRV-associated diarrhea and shedding occurred in 9 of 10 control calves. Active IgM antibody responses occurred in faeces and/or serum of most calves after BRV inoculation. However, the highest active antibody responses (IgM and IgG1 in serum, and IgM, IgG1 or IgA in faeces) after BRV inoculation were in calves fed control or no colostrum, in association with clinical diarrhea in most of these calves. After challenge at PID 21, BRV-associated diarrhea and shedding were of short duration or absent, in all groups. These results demonstrate the efficacy of colostrum from VLP vaccinated cows to provide heterologous, passive protection against BRV diarrhea and shedding in calves. In comparison, calves fed CLP or SA11 colostrum were only partially protected against BRV diarrhea or shedding.

Three-dimensional structural analysis of recombinant rotavirus-like particles with intact and amino-terminal-deleted VP2: implications for the architecture of the VP2 capsid layer

Rotaviruses are the leading cause of severe infantile gastroenteritis worldwide. These viruses are large, complex icosahedral particles consisting of three concentric capsid layers enclosing a genome of eleven segments of double-stranded RNA (dsRNA). The amino terminus of the innermost capsid protein VP2 possesses a nonspecific single-stranded RNA and dsRNA binding activity, and the amino terminus is also essential for the incorporation of the polymerase enzyme VP1 and guanylyltransferase VP3 into the core of the virion. Biochemical and structural studies have suggested that VP2, and especially the amino terminus, appears to act as a scaffold for proper assembly of the components of the viral core. To locate the amino terminus of VP2 within the core, we have used electron cryomicroscopy and image reconstruction to determine the three-dimensional structures of recombinant virus-like particles that contain either full-length or amino-terminal-deleted forms of VP2 coexpressed with the intermediate capsid protein VP6. A comparison of these structures indicates two significant changes along the inner surface of VP2 in the structure lacking the amino terminus: a loss of mass adjacent to the fivefold axes and a redistribution of mass along the fivefold axes. Examination of the VP2 layer suggests that the proteins are arranged as dimers of 120 quasi-equivalent molecules, with each dimer extending between neighboring fivefold axes. Our results indicate that the amino termini of both quasi-equivalent VP2 molecules are located near the icosahedral vertices.

Isotype-specific antibody responses to rotavirus and virus proteins in cows inoculated with subunit vaccines composed of recombinant SA11 rotavirus core-like particles (CLP) or virus-like particles (VLP)

The isotype antibody responses to bovine IND P5, G6 and simian SA11 P2, G3 rotavirus and SA11 rotavirus proteins (VP4, VP6 and VP7) in serum, colostrum and milk were analysed by ELISA in three groups of vaccinated cows and nonvaccinated controls. Pregnant cows were vaccinated intramuscularly and intramammarily with recombinant baculovirus-expressed SA11 rotavirus VLP (triple-layered virus-like particles containing rotavirus VP2, VP4, VP6 and VP7); CLP (double-layered core-like particles containing rotavirus VP2 and VP6); or inactivated SA11 rotavirus, respectively. Rotavirus antigen titers were highest (30-200-fold) in ELISA in the VLP vaccine compared to the inactivated SA11 vaccine. The IgG1, IgG2 and IgM geometric mean antibody titers (GMT) to rotavirus (titers to bovine rotavirus vs SA11 rotavirus did not differ significantly for any isotype or group) and the IgG2 GMT to VP6 in serum at calving in the vaccinated groups were significantly (P < 0.05) higher than in the control group. In colostrum, IgG1 and IgA rotavirus antibody titers were significantly elevated for VLP (IgG1 GMT 832225; IgA GMT 16384), CLP (IgG1 GMT 660561; IgA GMT 10321) and SA11 (IgG1 GMT 131072; IgA GMT 1448) vaccinated cows compared to control cows (IgG1 GMT 11585; IgA GMT 45). The IgG1 and IgA GMT to rotavirus were significantly elevated (6-100-fold) in milk of VLP and CLP vaccinated cows compared to SA11 vaccinated or control cows. The isotype antibody responses to VP6 in serum, colostrum and milk paralleled the responses to rotavirus, but titers were approximately 2-10-fold lower. Only cows vaccinated with VLP had significantly enhanced serum, colostral and milk antibody titers to rotavirus VP4 and VP7. These results demonstrate that rotavirus antibody titers in serum, colostrum and milk are significantly enhanced by use of non-infectious VLP, CLP and inactivated SA11 rotavirus vaccines, but the VLP or CLP vaccines induced the highest antibody responses, corresponding to their higher rotavirus antigen titers measured by ELISA.

Rotavirus vaccines and vaccination potential

The development of a successful rotavirus vaccine is a complex problem. Our review of rotavirus vaccine development shows that many challenges remain, and priorities for future studies need to be established. For example, the evaluation of administration of a vaccine with OPV or breast milk might receive less emphasis until a vaccine is made that shows clear efficacy against all virus serotypes. Samples remaining from previous trials should be analyzed to determine epitope-specific serum and coproantibody responses to clarify why only some trials were successful. Detailed evaluation of the antigenic properties of the viruses circulating and causing illness in vaccinated children also should be performed for comparisons with the vaccine strains. In future trials, sample collection should include monitoring for asymptomatic infections and cellular immune responses should be analyzed. The diversity of rotavirus serotype distribution must be monitored before, during, and after a trial in the study population and placebo recipients must be matched carefully to vaccine recipients. Epidemiologic and molecular studies should be expanded to document, or disprove, the possibility of animal to human rotavirus transmission, because, if this occurs, vaccine protection may be more difficult in those areas of the world where cohabitation with animals occurs. We also need to have an accurate assessment of the rate of protection that follows natural infections. Is it realistic to try to achieve 90% protective efficacy with a vaccine if natural infections with these enteric pathogens only provide 60% or 70% protection? Subunit vaccines should be considered to be part of vaccine strategies, especially if maternal antibody interferes with the take of live vaccines. The constraints on development of new vaccines are not likely to come from molecular biology. The challenge remains whether the biology and immunology of rotavirus infections can be understood and exploited to permit effective vaccination. Recent advances in developing small animal models for evaluation of vaccine efficacy should facilitate future vaccine development and understanding of the protective immune response(s) (Ward et al. 1990b; Conner et al. 1993).

Development of candidate rotavirus vaccines

Candidate rotavirus vaccines tested to date have been developed using a 'Jennerian' approach. Strains of bovine and simian rotaviruses that are naturally attenuated for humans have been assessed and found to confer immunity that is serotype specific in a varying proportion of recipients. The spectrum of protection has been widened by developing reassortants in which the bovine or simian gene coding for VP7 (the major outer capsid protein) has been replaced by the corresponding gene from human VP7 types 1, 2, 3 or 4. Once the protective antigen(s) are identified it may be possible to develop subunit vaccines that eliminate side effects sometimes observed with live vaccine candidates.

Monoclonal anti-idiotype induces antibodies against bovine Q17 rotavirus

This study describes, for the first time, the production and use of an "internal-image" anti-idiotypic monoclonal antibody (MAb) to elicit a rotavirus-specific antibody response. An immunoglobulin G2a MAb, designated RQ31 (MAb1), specific for the outer capsid protein VP4 of bovine Q17 rotavirus and capable of neutralizing viral infection in vitro was used to generate an anti-idiotypic MAb (MAb2). This MAb2, designated RQA2, was selected by enzyme-linked immunosorbent assay (ELISA) using F(ab')2 fragments of RQ31. RQA2 (MAb2) inhibited the binding of RQ31 (MAb1) to the virus but had no effect on the binding of other rotavirus-specific MAbs. The MAb2 also inhibited virus neutralization mediated by MAb1 in a dose-dependent fashion. Naive guinea pigs immunized with the MAb2 produced anti-anti-idiotypic antibodies (Ab3) that reacted with bovine Q17 rotavirus in an ELISA and neutralized rotavirus infection in vitro. The Ab3 response was characterized as MAb1-like because the Ab3 recognizes only the Q17 and neonatal calf diarrhea virus rotavirus strains in ELISA, as did RQ31 (MAb1). The Ab3 response also possessed two other characteristics of RQ31: the abilities to bind the 1.36 (double-capsid) but not the 1.38 (single-capsid) purified rotavirus fraction in ELISA and to immunoprecipitate the VP4 rotavirus protein.

Immunization with baculovirus-expressed recombinant rotavirus proteins VP1, VP4, VP6, and VP7 induces CD8+ T lymphocytes that mediate clearance of chronic rotavirus infection in SCID mice

Clearance of chronic murine rotavirus infection in SCID mice can be demonstrated by adoptive transfer of immune CD8+ T lymphocytes from histocompatible donor mice immunized with a murine homotypic rotavirus (T. Dharakul, L. Rott, and H.B. Greenberg, J. Virol 64:4375-4382, 1990). The present study focuses on the protein specificity and heterotypic nature of cell-mediated clearance of chronic murine rotavirus infection in SCID mice. Heterotypic cell-mediated clearance was demonstrated in SCID mice infected with EDIM (murine) rotavirus after adoptive transfer of CD8+ T lymphocytes from BALB/c mice that were immunized with a variety of heterologous (nonmurine) rotaviruses including Wa (human, serotype 1), SA11 and RRV (simian, serotype 3), and NCDV and RF (bovine, serotype 6). This finding indicates the serotypic independence of T-cell-mediated rotavirus clearance. To further identify the rotavirus proteins that are capable of generating CD8+ T cells that mediate virus clearance, donor mice were immunized with SF-9 cells infected with a baculovirus recombinant expressing one of the following rotavirus proteins: VP1, VP2, NS53 (from RF), VP4, VP7, NS35 (from RRV), VP6, and NS28 (from SA11). SCID mice stopped shedding rotavirus after receiving CD8+ T cells from mice immunized with VP1, VP4, VP6, and VP7 but not with VP2, NS53, NS35, NS28, or wild-type baculovirus. These results suggest that heterotypic cell-mediated clearance of rotavirus in SCID mice is mediated by three of the major rotavirus structural proteins and by a putative polymerase protein.

Expression of rotavirus VP2 produces empty corelike particles

The complete VP2 gene of bovine rotavirus strain RF has been inserted into the baculovirus transfer vector pVL941 under the control of the polyhedrin promoter. Cotransfection of Spodoptera frugiperda 9 cells with wild-type baculovirus DNA and transfer vector DNA led to the formation of recombinant baculoviruses which contain bovine rotavirus gene 2. Infection of S. frugiperda cells with this recombinant virus resulted in the production of a protein similar in size and antigenic properties to the authentic rotavirus VP2. The protein binds double-stranded RNA and DNA in an overlay protein blot assay. Expressed VP2 assembles in the cytoplasm of infected cells in corelike particles 45 nm in diameter. These corelike particles were purified by sucrose gradient centrifugation and found to be devoid of nucleic acid. Coexpression of VP2 and VP6 from heterologous rotavirus strains (bovine and simian) resulted in the formation of single-shelled particles. These results definitively show the existence of an innermost protein shell in rotavirus which is formed independently of other rotavirus proteins. These results have implications for schemes of rotavirus morphogenesis.