INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Rotaviruses are the most common etiologic agents of acute viral gastroenteritis in young children throughout the world, and a worldwide effort is under way to design an effective vaccination strategy. In the United States, rotavirus infection is primarily a problem of morbidity and associated health care costs (48, 52), whereas in developing countries, mortality is high, with >870,000 deaths per year attributed to rotavirus (39). Live attenuated rotavirus vaccine candidates consisting of human-animal (simian or bovine) rotavirus reassortants were tested in children but showed variable effectiveness in different settings (20, 74). In recent trials, these vaccines provided approximately 70% effectiveness against severe diarrhea (16, 43, 44, 61, 69, 75, 76).

Rotaviruses belong to the Reoviridae family and are composed of three protein layers surrounding 11 segments of double-stranded RNA (32). The innermost layer is composed of VP1, VP2, VP3, and the genome, the middle layer is composed of VP6; and the outer layer is composed of the glycoprotein VP7 and spikes of VP4 dimers (32, 64, 70). VP4 and VP7 possess distinct antigenic activities, defining P serotypes and G serotypes, respectively. VP4 and VP7 independently elicit antibodies capable of neutralizing rotavirus infectivity and inducing protective immunity (32).

Rotavirus genes encoding the rotavirus structural proteins VP2, VP6, VP4, and VP7 have been cloned in baculovirus, and the recombinant rotavirus proteins have been coexpressed in the baculovirus expression system (26, 31, 47). Stable virus-like particles (VLPs) self-assemble following expression of VP2 alone (47). Coexpression of VP2 and VP6 alone or with VP4 results in the production of double-layered 2/6- or 2/4/6-VLPs, respectively (26, 47). Coexpression of VP2, VP6, and VP7, with or without VP4, results in triple-layered 2/6/7- or 2/4/6/7-VLPs (26). All VLPs maintained the structural and functional characteristics of native particles (26, 63, 65, 69), including binding to and internalization of 2/4/6/7-VLPs into MA104 cells (26, 28).

Models of rotavirus infection, without disease, were developed in rabbits (14, 17-19, 38, 73) and in adult mice (58, 79, 80) to monitor the development of active serum and mucosal immunity as well as protection from infection following a live rotavirus challenge. We demonstrated that parenteral vaccination with live or inactivated rotavirus induces active immunity and protection in the rabbit model (19). Preliminary results for rabbits showed that VLPs administered parenterally in Freund's adjuvants and aluminum phosphate (AlP) were immunogenic and induced active protection from homologous serotype G3 oral rotavirus challenge (21, 23). Here, we report the immunogenicity and protective efficacy of parenterally administered VLPs of different compositions (2/6, 2/6/7, and 2/4/6/7) in rabbits, using different adjuvants: Freund's, AlP, and QS-21. QS-21 has the advantage that it may be licensed for use in humans (46) and has been tested with VLPs in mice (23, 24, 42, 50).

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Cells and viruses. The lapine rotavirus ALA (P[14], G3) strain (11), used for rabbit rotavirus challenge inoculations or enzyme-linked immunosorbent assays (ELISAs) and fluorescent-focus neutralization assays (FFNAs), was passaged in the presence of trypsin 10 times in fetal rhesus monkey kidney MA104 cells or was passaged in MA104 cells and plaque purified three times, respectively, as described previously (17). The simian rotavirus SA11 clone 3 (SA11 Cl3; P[2], G3) (5, 30, 51), bovine rotavirus B223 (P8[11], G10), and SA11 Cl3 × B233 rotavirus reassortant viruses R-N33 (P8[11], G3) and R-A32 (P[2], G10) used in FFNAs were propagated in MA104 cells with trypsin and plaque purified two or three times as described elsewhere (29). Original derivation and characterization of the SA11 Cl3 × B223 rotavirus reassortant viruses R-N33 and R-A32 were described previously (66). R-N33 (P8[11], G3) contains genome segments 1 and 4 from B223 and the remainder of its genes from SA11 Cl3, and R-A32 (P[2], G10) contains genome segments 2 and 4 from SA11 Cl3 and all other genome segments from B223 (66, 67).

Preparation of rotavirus SA11 Cl3 virus. Live or formalin-inactivated SA11 Cl3-infected cell lysates were prepared as described elsewhere (19). Following formalin inactivation, SA11 Cl3 preparations were characterized by ELISA as described previously (26) with a panel of VP4- and VP7-specific neutralizing and nonneutralizing monoclonal antibodies (MAbs) to confirm that virus inactivation did not result in loss of epitope reactivity. SA11 Cl3 triple-layered particles (TLPs) were purified by CsCl density gradient centrifugation (26). Purified SA11 Cl3 TLPs were inactivated with 4'-aminomethyltrioxalin-hydrochloride (psoralen; Lee Biomolecular Research Laboratories, Inc., San Diego, Calif.) (37). Before and after psoralen inactivation, SA11 Cl3 TLPs were (i) examined by electron microscopy (EM) to confirm their integrity; (ii) analyzed by staining of polyacrylamide gels with silver nitrate following sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis (PAGE) to determine purity; and (iii) following SDS-PAGE, analyzed by Western blotting (immunoblotting) using a mouse hyperimmune serum prepared against triple-layered SA11 Cl3 virus to determine protein composition as described previously (5, 26). Psoralen-inactivated CsCl-purified SA11 Cl3 TLPs were not analyzed by ELISA with the panel of MAbs for epitope reactivity since psoralen inactivation of rotavirus results in antigenically intact particles (37). To determine whether virus was completely inactivated, both formalin- and psoralen-treated viruses were passaged three times in cell culture; no cytopathic effect or plaques were seen, indicating that inactivation was complete. Complete virus inactivation was more readily obtained with psoralen than with formalin; therefore, in the later studies with CsCl-purified virus, psoralen inactivation was used.

Production and characterization of VLPs. Sf9 cells were coinfected with the baculovirus recombinants as described previously (26) to produce 2/6, (G3) 2/6/7, and (P[2], G3) 2/4/6/7-VLPs. Therefore, 2/6/7- or 2/4/6/7-VLPs contained a VP7 of serotype G3 specificity, while 2/4/6/7-VLPs contained a VP4 of a P[2] type. VLPs were purified by isopycnic CsCl gradient centrifugation and characterized essentially as described previously (26). Briefly, VLP composition, purity, and integrity were confirmed by Western blotting (using a mouse hyperimmune serum prepared against triple-layered SA11 Cl3 virus as described elsewhere [5, 26]), staining of polyacrylamide gels with silver nitrate, and EM, respectively. The G3 serotype of the 2/6/7- and 2/4/6/7-VLPs was confirmed by immunoelectron microscopy and a MAb-based ELISA as described previously (26). The total protein concentration was determined by the Bio-Rad (Hercules, Calif.) protein assay, using bovine immunoglobulin G (IgG) as the standard, and endotoxin levels were quantitated by the Limulus amebocyte lysate test (<0.05 endotoxin unit [EU]/dose); Associates of Cape Cod, Inc., Woods Hole, Mass.).

Preparation of rotavirus or VLP vaccines. Vaccine preparations were either (i) adsorbed to AlP (kindly supplied by Wyeth Lederle Vaccines and Pedriatics, West Henrietta, N.Y.) as described previously (19); (ii) mixed 1:1 with Freund's adjuvants (complete for the first dose and incomplete for the second dose); or (iii) mixed with 20 µg (unless otherwise specified) of QS-21 adjuvant (kindly supplied by Wyeth Lederle Vaccines and Pediatrics), a purified triterpene glycoside (saponin) of aquillaic acid derivative from the bark of the South American Molina tree Quillaja saponaria (45, 46).

Animals. The 3- to 6-month-old rotavirus antibody- and specific-pathogen-free New Zealand White rabbits of either sex used in this study were either from our own rabbit breeding colony (14, 17-19) or from a commercial source (Charles River Laboratories, Inc., Wilmington, Mass.). During the immunization phase of the study, rabbits were housed in open cages in two rooms in a BL2 containment facility. During the rotavirus challenge, rabbits were housed individually in negative-pressure isolator units.

Animal inoculations and sample collection. Approximately 1 to 2 months prior to rotavirus or VLP inoculations, all rabbits, except those parenterally immunized with 10 or 20 µg of 2/4/6/7-VLPs in Freund's adjuvants or absorbed to AlP, were vaccinated twice intramuscularly with a Clostridium spiroforme toxoid (developed and kindly supplied by R. Carman, TechLabs, Inc., Blacksburg, Va.), to prevent disease due to endemic C. spiroforme and its associated toxin (3, 19, 60).

Detection of rotavirus in fecal samples by ELISA. The presence of rotavirus antigen in fecal samples was determined 0 to 14 dpc by ELISA as described previously (13, 14, 17, 19, 25). A sample was considered positive if the optical density (OD) value of the virus well minus the OD value of the control well was 0.1 and this OD value was at least 2 standard deviations greater than the OD values of the negative control samples (18). Virus antigen shedding curves (OD versus dpc) for each animal were plotted, and the area under the shedding curve for each animal was calculated. Percent reduction in virus shedding compared to control animals was calculated for each animal by comparing the area under the curve for each individual animal to the mean area under the curves of virus shedding in the control group within each experiment. Percent reduction in virus antigen shedding was then calculated for each vaccine group by determining the mean percent reduction of virus antigen shedding for each vaccine group.

Detection of infectious virus from fecal samples by FFA. The titers of infectious virus from selected fecal samples collected from 17 rabbits following ALA rotavirus challenge were measured by fluorescent-focus assay (FFA) (10, 12). The 10% fecal suspensions were serially diluted (10-fold), and each dilution was assayed in duplicate on MA104 cell monolayers. Infectivity titers were expressed as fluorescent focus units (FFU) per milliliter. Due to toxicity, the lowest dilution tested for infectivity titers was 1:10. When fluorescent foci in 1/10 dilutions could not be visualized by fluorescence microscopy, the samples were considered negative, and a value of 5 FFU/ml was arbitrarily assigned.

Detection of antibody responses by Western blotting, ELISA, and FFNA. Western blots to examine rotavirus-specific antibodies were performed by a procedure modified from those used by Burns et al. (4) and Crawford et al. (26). Briefly, SDS-PAGE was performed as described above, using SA11 Cl3-infected cell lysate (3.5 × 10⁸ PFU/ml) as the antigen. Following immunoblotting, the nitrocellulose membranes were blocked for 30 min with 5% BLOTTO and were then cut into strips prior to overnight incubation at room temperature with the different primary antibodies diluted in 0.5% BLOTTO. Monoclonal or polyclonal antibodies used as controls were (i) purified anti-VP4 MAbs 3D8, 9E3, and 5E4 (4) at a final concentration of 2 µg/ml (used conjointly), (ii) polyclonal sera specific to VP6 (guinea pig) (33) and to VP7 (rabbit) (62) at dilutions of 1:600 and 1:1,000, respectively, and (iii) anti-SA11 polyclonal mouse serum at a dilution of 1:7,500. Pre- and postchallenge (0 and 28 dpc, respectively) serum samples from all rabbits were diluted 1:100 for Western blot analyses. The strips were then washed three times with 0.5% BLOTTO and incubated for 2 h with a 1:3,000 dilution of either horseradish peroxidase-conjugated goat anti-rabbit total immunoglobulin (IgM, IgG, and IgA) antibody (Cappel Research Products, Durham, N.C.), anti-mouse total immunoglobulin (IgM, IgG, and IgA) antibody (Southern Biotechnology Associates, Inc., Birmingham, Ala.), or anti-guinea pig IgG antibody (Cappel Research Products) as appropriate. The strips were washed and developed with 3,3',5,5'-tetramethylbenzidine-peroxidase membrane enhancer (Kirkegaard & Perry Laboratories, Inc., Gaithersburg, Md.) according to the manufacturer's directions.

Statistical analysis. Statistical analyses were performed with SPSS version 7.5 for Windows (SPSS, Inc., Chicago, Ill.). Percent reductions in shedding were compared by the Mann-Whitney U test. Antibody titers before and after virus challenge within a group were compared by the Wilcoxon signed ranks test, and titers between groups were compared by using the Kruskal-Wallis test followed by the Mann-Whitney U test. Trend analysis was performed by linear regression, and correlation coefficients were calculated by Pearson's correlation coefficients.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Preparation and characterization of VLPs used for inoculation of rabbits. Coinfection of Sf9 cells with the respective baculovirus recombinants resulted in the formation of 2/6-, 2/6/7-, and 2/4/6/7-VLPs. 2/6-VLPs resembled double-layered particles and 2/6/7- and 2/4/6/7-VLPs resembled rotavirus TLPs (data not shown). EM analysis revealed that the structural integrity of all VLP types was high (90% [data not shown]). The total protein concentration of each VLP type was determined by the Bio-Rad protein assay. The endotoxin levels in the VLP preparations used for rabbit inoculations were quantitated at 0.037 EU/10 µg, 0.016 EU/20 µg, and 0.015 to 0.05 EU/50 µg. The purity and protein composition of all VLPs used in this study were confirmed by electrophoretic analysis and staining of gels with silver nitrate and by Western blot analysis (data not shown).

Immune response and protection from challenge of rabbits parenterally immunized with 10 or 20 µg of P[2], G3 2/4/6/7-VLPs in Freund's adjuvants or absorbed to AlP. Two doses of live or formalin-inactivated simian rotavirus SA11 was previously shown to induce protection in rabbits (19). We next examined whether VLPs would afford the same protection and whether the adjuvant used would influence the immunogenicity or protective efficacy of the VLPs. In an initial experiment, rabbits were immunized twice parenterally (0 and 21 dpv) with either 10 or 20 µg of 2/4/6/7-VLPs either in Freund's adjuvants (n = 5) or absorbed to AlP (n = 5). Control animals were immunized with either live or formalin-inactivated SA11-infected cell lysates (1.4 × 10⁷ PFU) either in Freund's adjuvants (n = 3) or absorbed to AlP (n = 2). Negative control rabbits were immunized with MA104 cell lysates in either adjuvant.

View larger version (17K): [in this window] [in a new window]   FIG. 1.   Serum and intestinal antibody responses following two parenteral immunizations (0 and 21 dpv) of rabbits with formalin-inactivated SA11 rotavirus, 10 or 20 µg of G3 2/4/6/7-VLPs, or MA104 cell lysate in either Freund's adjuvants or AlP. Total (IgM, IgG, and IgA) serum (A), intestinal IgA (B), and intestinal IgG (C) antirotavirus antibodies were measured by ELISA before (56 dpv) and after (84 dpv) oral challenge with lapine ALA rotavirus. Formulation of vaccine and adjuvant are indicated along the x axis. Antibody titers were measured for individual rabbits, and results are plotted as the GMTs of the groups. Error bars represent 1 standard error of the mean. For titers of <50 and <5, 25 and 2.5, respectively, were used to calculate the GMT. A GMT of 25 or 2.5, for serum or intestinal antibodies, respectively, was considered negative. A significant (P < 0.05, Wilcoxon signed ranks test) increase in rotavirus antibody GMTs following challenge is indicated by an asterisk.

View larger version (16K): [in this window] [in a new window]   FIG. 2.   Protection from lapine ALA rotavirus challenge in rabbits immunized twice parenterally with formalin-inactivated SA11 rotavirus, 10 or 20 µg of G3 2/4/6/7-VLPs, or MA104 cell lysate in either Freund's adjuvant or AlP. Virus antigen shedding curves were plotted for individual rabbits, and the mean area under the curve was calculated for each group. Results are plotted as percent reduction in shedding for each individual animal (), as well as the mean for each vaccine group (---), calculated by comparing the mean area under the curve for each vaccine group to the mean area under the curve for the control group. The number of rabbits with overlapping levels of protective efficacy is indicated next to the dots (n = 2). Formulation of vaccine, adjuvant, and number of rabbits per group are indicated along the x axis.

Immune response and protection from challenge of rabbits parenterally immunized twice with 50 µg of P[2], G3 2/4/6/7- or G3 2/6/7-VLPs in Freund's adjuvants or QS-21. We next examined whether increasing the dose of 2/4/6/7-VLPs would increase protective efficacy from ALA rotavirus challenge (10³ ID₅₀). Since the response with AlP was weaker than with Freund's adjuvants, we also examined whether QS-21, a saponin adjuvant, would be able to induce higher antibody responses and protection rates than AlP. In addition, to investigate the contribution of the outer capsid protein VP4, to induction of protective immunity, we compared the immunogenicity and protective efficacies of 50 µg of 2/4/6/7- or 2/6/7-VLPs in Freund's adjuvants (n = 5 and n = 9, respectively) and QS-21 (n = 10 and n = 10, respectively). Control rabbits were either mock vaccinated with TNC buffer or vaccinated with CsCl-purified live or psoralen-inactivated SA11 Cl3 virus in QS-21 only, in the same way as controls in Freund's adjuvants were performed previously.

View larger version (30K): [in this window] [in a new window]   FIG. 3.   Serum and intestinal antibody responses following two parenteral immunizations (0 and 21 or 28 dpv) of rabbits with formalin-inactivated SA11 rotavirus, MA104 cell lysate, or 50 µg of G3 2/4/6/7-, G3 2/6/7-, or 2/6-VLPs in Freund's adjuvants (A to C) and with 50 µg of live/psoralen-inactivated SA11 rotavirus, TNC buffer, or 50 µg of G3 2/4/6/7-, G3 2/6/7-, or 2/6-VLPs in QS-21 (D to F). Total (IgM, IgG, and IgA) serum (A and D), intestinal IgG (B and E), and intestinal IgA (C and F) antirotavirus antibodies were measured by ELISA before (56 or 69 dpv) and after (84 or 97 dpv) oral challenge with lapine ALA rotavirus. Antibody titers were measured for individual rabbits, and results are plotted as GMTs of the groups. The GMT of intestinal IgG antibody titers for both 0 and 28 dpc of the rabbits vaccinated with 2/6-VLPs in Freund's adjuvants is 368. Therefore, for this group, only one box appears on the graph (B). Error bars represent 1 standard error of the mean. For titers of <50 and <5, 25 and 2.5, respectively, were used to calculate the GMT. A GMT of 25 or 2.5, for serum or intestinal antibodies, respectively, was considered negative. A significant (P < 0.05, Wilcoxon signed ranks test) increase in rotavirus antibody GMTs following challenge is indicated by an asterisk.

View larger version (11K): [in this window] [in a new window]   FIG. 4.   Protection from lapine ALA rotavirus challenge in rabbits immunized twice parenterally with formalin-inactivated SA11 rotavirus, MA104 cell lysate, or 50 µg of G3 2/4/6/7-, G3 2/6/7-, or 2/6-VLPs in Freund's adjuvant (A) or with 50 µg of live/psoralen-inactivated SA11 rotavirus, TNC buffer, or 50 µg of G3 2/4/6/7-, G3 2/6/7-, or 2/6-VLPs in QS-21 (B). Virus antigen shedding curves were plotted for individual rabbits, and the mean area under the curve calculated for each group. Results are plotted as percent reduction in shedding for each individual animal (), as well as the mean for each vaccine group (---), calculated by comparing the mean area under the curve for each vaccine group to the mean area under the curve for the control group. The number of rabbits with overlapping levels of protective efficacy is indicated next to the dots (n = 2 or 3); formulation of vaccine and number of rabbits per group are indicated along the x axis.

Immune response and protection from challenge of rabbits parenterally immunized twice with 50 µg of 2/6-VLPs in Freund's adjuvants or QS-21. We recently showed that 2/6-VLPs administered parenterally, orally, or intranasally to mice provide protection (42 to 100%) from homologous EC_wt murine rotavirus challenge (2, 33, 42, 58, 59). To determine if either of the outer capsid proteins, VP7 or VP4, is required to achieve protection from infection in the rabbit model, 50 µg of 2/6-VLPs in Freund's adjuvants (n = 5) and QS-21 (n = 5) was administered parenterally to rabbits.

Immune response and protection from challenge of rabbits parenterally immunized twice with different concentrations of 2/6-VLPs and QS-21. Experiments in mice showed that parenteral administration of 2/6-VLPs in QS-21 induced high levels of protection from challenge (2, 33, 42). However, 2/6-VLPs in QS-21 administered parenterally to rabbits failed to induce high levels of protection, and 2/6-VLPs in Freund's adjuvants afforded high levels of protection in only two of five rabbits (Fig. 4). To determine if either an increase of VLP or QS-21 dose would afford higher levels of protection in rabbits, 200 µg of 2/6-VLPs in 20 µg of QS-21 (n = 5) or 50 µg of 2/6-VLPs in 50 µg of QS-21 (n = 5) was administered parenterally to rabbits. Increases in the dose of either 2/6-VLPs (200 µg) or QS-21 (50 µg) failed to increase the levels of total serum antibodies (P  0.072), intestinal IgG antibodies (P  0.228), or mean protective efficacy (5 and 18%, respectively) (P  0.141) compared to 50 µg of 2/6-VLPs in 20 µg of QS-21 (data not shown). As observed with lower doses of VLPs or QS-21, the total serum and intestinal IgG antibody levels or mean protective efficacy with either the increased dose of VLPs or adjuvant were significantly lower (P  0.048) than those induced by 2/6-VLPs in Freund's adjuvants (data not shown).

Analyses of the immune response to individual rotavirus proteins and correlation of protective efficacy and immune responses detected by Western blotting and FFNA. Table 1 summarizes the rabbit immune responses (as measured by ELISA and FFNA), percent protective efficacy, and the corresponding statistical analyses among all vaccine groups. Immunization with virus or with the inclusion of VP4 in 2/4/6/7-VLP vaccines resulted in significantly higher (P < 0.05, Mann-Whitney U test) mean protective efficacy than in the 2/6/7-VLP vaccines. Therefore, we investigated whether rabbits vaccinated with 50 µg of live or psoralen-inactivated SA11 rotavirus or 2/4/6/7-, 2/6/7-, or 2/6-VLPs in Freund's adjuvants or QS-21 developed VP4-specific immune responses.

View larger version (116K): [in this window] [in a new window]   FIG. 5.   Serum immune responses to individual rotavirus proteins before (56 or 69 dpv) or after (84 or 97 dpv) oral challenge with lapine ALA rotavirus. Representative results are shown for rabbits vaccinated parenterally either with 2/4/6/7-VLPs in Freund's adjuvants (F) or with 2/4/6/7-, 2/6/7-, or 2/6-VLPs, psoralen-inactivated SA11, or TNC buffer in QS-21 (Q). Proteins in a SA11-infected cell lysate were separated by SDS-PAGE, transferred to nitrocellulose, and incubated with MAbs directed to VP4, VP6, or VP7, with a hyperimmune serum to SA11, or with paired sera from rabbits immunized with the indicated vaccine formulations. Percent protection from infection for individual rabbits is indicated below each lane. Locations of the viral proteins are indicated on the left.

View larger version (30K): [in this window] [in a new window]   FIG. 6.   Correlation of higher protection rates with presence of serum N-Abs to VP4. (A) Serum N-Ab responses following two parenteral immunizations (0 and 28 dpv) of rabbits with 50 µg of P[2], G3 2/4/6/7-, G3 2/6/7-, or 2/6-VLPs in Freund's adjuvants or with 50 µg of live or psoralen-inactivated SA11, TNC buffer, or 50 µg of P[2], G3 2/4/6/7-, G3 2/6/7-, or 2/6-VLPs in QS-21 were measured by FFNA against rotavirus strains SA11 (P[2], G3) and B223 (P8[11], G10) and against reassortant virus strains R-N33 (P8[11], G3) and R-A32 (P[2], G10) prior to (56 or 69 dpv) oral challenge with lapine ALA rotavirus. Serum N-Ab titers were measured for individual rabbits, and results are plotted as the GMTs of the groups. Error bars represent 1 standard error of the mean. For titers of <100, 50 was used to calculate the GMT. A GMT of 50 and its log were considered negative. A significant difference (P  0.021, Mann-Whitney U test) in VP4-specific N-Abs induced by 2/4/6/7-VLPs vaccines measured by comparing serum N-Abs measured against B223 and R-A32 is indicated by an asterisk. (B) Scatterplot of the log of the SA11 VP4-specific-NAb titers and the percent reduction of virus antigen shedding for individual rabbits (n = 58). Results were graphed to indicate whether rabbits were immunized with both VP4 and VP7 (; SA11 or 2/4/6/7-VLPs in either adjuvant), absence of VP4 but presence of VP7 (; 2/6/7-VLPs in either adjuvant), or absence of both VP4 and VP7 (; 2/6-VLPs in either adjuvant or TNC buffer in QS-21). Correlation coefficient was calculated by Pearson's correlation coefficient.

Detection of infectious virus in fecal samples collected 0 to 10 dpc by FFA. To determine if protective efficacy measured by percent reduction in virus antigen shedding by ELISA predicts the amount of infectious virus shed, the infectivity titers of the ALA virus shed following challenge were measured by FFA (Fig. 7). For comparison, a subset of rabbits (n = 17) were chosen to cover the whole range of protection, independent of vaccine formulation. To establish a correlation, Pearson's correlation coefficient was calculated by plotting the areas under the curve for the virus antigen shedding curves and the virus titer curves. A significant (P < 0.001, r = 0.979) correlation coefficient was found (Fig. 7A).

View larger version (16K): [in this window] [in a new window]   FIG. 7.   Correlation between reduction of virus antigen shedding as measured by ELISA and reduction of infectious virus titers as measured by FFA independently of vaccine formulation administered to rabbits. (A) Scatterplot of the areas under the curve for virus antigen shedding (as measured by ELISA) versus the log of infectious virus titers (as measured by FFA). Analysis was performed on a subset of rabbits (n = 17) vaccinated parenterally with buffer, SA11 virus, or different VLP formulations. Correlation coefficient was calculated by Pearson's correlation coefficient. Line depicts linear regression analysis [f(x) = 0.247X  0.969]. Boxes indicate clusters of rabbits according to percent protection ranges. (B) Daily mean infectious (0 to 10 dpc) ALA rotavirus titers measured by FFA were determined for rabbits grouped based on protection, independent of vaccine formulation administered to rabbits. Based on the scatterplot, rabbits were grouped based on 0% (), 20 to 65% (), 68 to 98% (), and 100% () protection as measured by ELISA. For infectious titers of <10, 5 was used to calculate the mean infectious virus titer. An infectious titer of 5 was considered negative.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

We examined the immunogenicity and protective efficacy of different formulations of parenterally administered candidate rotavirus VLP subunit vaccines in the rabbit model of rotavirus infection. Previously, we showed that parenteral vaccination with live or inactivated rotavirus induces complete active protective immunity in rabbits against virus challenge. This protection correlated with the presence of rotavirus-specific IgG, but not IgA, in the intestine (19). In this study, we showed that VLPs administered parenterally are immunogenic and can induce active protective immunity in rabbits. Vaccinated rabbits developed high levels of total (IgM, IgG, and IgA) rotavirus-specific antibody in the serum. Rabbits vaccinated with VLPs or with live or inactivated SA11 virus developed antirotavirus intestinal antibody of the IgG, but not IgA, isotype. Protective efficacy varied with both VLP formulation and adjuvant. With all adjuvants tested, 2/4/6/7-VLPs consistently induced the highest levels of protection from homotypic (G3) ALA rotavirus challenge.

There was no significant difference in the mean protective efficacy afforded by live or formalin-inactivated SA11 in either Freund's adjuvants (87%) or AlP (76%) or purified live or psoralen-inactivated SA11 in QS-21 (83%) (P  0.462). Although the results are similar to our previous work comparing live or inactivated SA11 virus administered in Freund's adjuvants and AlP (19), in the present study, unlike the previous study, total protection from rotavirus challenge was not achieved in rabbits vaccinated with SA11 in either adjuvant. The live and formalin-inactivated SA11 vaccine preparations were the same vaccine preparations used previously (19), but antigen ELISA results indicated that limited degradation of the immunogen occurred in this study (data not shown). Therefore, the lower protective efficacy in this study is likely due to immunization with a lower dose of antigen.

The difference in virus inactivation procedures probably did not influence the immune responses because (i) we did not observe any significant difference in either immunogenicity or protective efficacy between live, formalin-inactivated, and psoralen-inactivated SA11 virus, (ii) we had previously shown (19) that following formalin treatment of virus, the reactivity levels of MAbs for several viral epitopes remain unchanged on formalin-inactivated virus compared to live virus, and (iii) likewise, others have shown that psoralen-inactivated virus retains epitope reactivity to the same panel of MAbs (37). Therefore, the effect of the differences in inactivation procedures on the immune responses elicited was not a major concern.

Live or inactivated SA11 rotavirus elicited the highest antibody responses and levels of protection in all three adjuvants tested. The high levels of antibody were not due to replication of the SA11 rotavirus administered intramuscularly to rabbits because prior to challenge none of the SA11-vaccinated rabbits developed serum antibodies to nonstructural proteins (Fig. 5 and data not shown), and fecal samples collected 2 to 14 dpv from rabbits vaccinated intramuscularly with live or formalin-inactivated SA11 rotavirus did not contain rotavirus antigen, indicating that virus replication did not occur (19). It is probable that the higher responses induced by SA11 compared to VLPs were due to differences in the immunizing dose. The doses of SA11 and VLPs administered to rabbits were based on total protein concentration determined by the Bio-Rad protein assay. Other studies have determined that if vaccine dose is based on the amount of VP7 in VLP and SA11 preparations, then SA11 has to be administered at a lower total protein concentration to yield equivalent doses of VP7 (50). Therefore, SA11 may have provided better responses in all adjuvants in this study compared to VLPs because of slightly higher doses, not differences in inherent properties between SA11 and VLPs.

Since the rabbit is an infection, rather than a disease, model, we determined protective efficacy of the VLPs by calculating percent reduction in antigen virus shedding measured between vaccinated and mock-vaccinated rabbits. Prevention of virus antigen shedding may be a more stringent measure of protection than protection from clinical disease, since detectable levels of virus antigen shedding can occur in the absence of clinical signs (21). Based on the strong correlation (r = 0.979, P < 0.001, Pearson's correlation coefficient) between the amount of virus antigen and infectious virus shedding, calculation of percent protection from infection reflects both the amplitude and the duration of both infectious virus and virus antigen shedding. The significant reduction of infectious virus shedding in vaccinated rabbits that had 68% protection from virus antigen shedding indicates that substantially less virus replication occurred in the intestines of vaccinated rabbits. A decrease in the amount of virus shedding could lead to a considerable decrease in the amount of infectious virus disseminated among the population. Although VLPs administered parenterally can effect a significant reduction in rotavirus shedding, it is difficult to predict how the reduced shedding relates to protection from any rotavirus diarrhea or severe rotavirus diarrhea because comparable measurements of these parameters have not been reported for naturally infected children or for any disease animal model.

To assess the immunogenicity and protective efficacy of different VLP formulations in the rabbit model, Freund's adjuvants, AlP, and QS-21 were tested as adjuvants. Only AlP is licensed for use in humans. Although our results indicate that VLPs can induce homotypic protection from challenge by the parenteral route, our data combined with data obtained for mice (42, 50) indicate that AlP and aluminum hydroxide are not potent adjuvants for use in conjunction with VLPs. Nevertheless, in previous rabbit experiments (14), the antibody titers induced by SA11 virus in AlP was sufficient to provide protection. It is probable that higher doses of VLPs in AlP will be needed to induce complete protective immunity. QS-21 has been shown to be a powerful immunopotentiator with VLPs in mice (42, 50) and has been tested with soluble protein antigens in preclinical studies in humans (45, 46, 56). Although Freund's adjuvant proved to be superior to QS-21 in our study, QS-21 may prove to be a better adjuvant than AlP.

Studies with children and animals have both supported and refuted protection based on the development of N-Abs (1, 15, 16, 40, 49, 53, 57, 77, 78, 81, 82). Recently, we showed that mucosally administered 2/6-VLPs, parenterally administered 2/6-VLPs, or inactivated double-layered murine EDIM rotavirus with QS-21 induced protection from rotavirus challenge in mice (2, 33, 55, 58, 59), indicating that neither VP7 nor VP4 is absolutely required for protection. However, parenteral immunization of rabbits with 2/6/7- or 2/6-VLPs with QS-21 resulted in levels of protection lower than those achieved with the 2/4/6/7-VLPs. Increases in the dose of either 2/6-VLPs or QS-21 failed to improve the protective efficacy. The presence of neutralizing VP4 antibodies in VLP- or SA11-vaccinated rabbits correlated (P < 0.001, r = 0.55, Pearson's correlation coefficient) with enhanced protection rates. Therefore, it appears that the presence of antibodies to VP4 plays an important role in protection of rabbits from infection. Although an absolute requirement for VP7 is not supported by our results with rabbits, VP7 may be needed to stabilize VP4 or to achieve proper expression of the epitopes on VP4 required for the induction of protective antibodies (26). It remains unclear whether N-Abs are required to induce high levels of protection; however, if N-Abs are needed for protection, then a limited number of VLP formulations may be sufficient to produce a broadly protective VLP subunit vaccine (27). A parenteral 2/4/6/7-VLP vaccine as opposed to an inactivated rotavirus vaccine will have the advantages of safety (i.e., it will be nonreplicating), no residual infectious virus or chemicals following inactivation, antigen stability (>6 years at 4°C), purity, and ability to alter protein content as needed (26, 27).

Taken together, our results using VLPs and QS-21 in rabbits suggest that both VP4 and VP7 neutralization antigens may be required in a parenteral immunogen to induce protection from rotavirus challenge. Nevertheless, the correlation is not absolute because individual rabbits immunized with 2/6/7- or 2/6-VLPs in Freund's adjuvants achieved high levels of protection (87 to 100%). The widely variable protective efficacies obtained with the 2/6-VLPs in Freund's adjuvants may be due to variation in the immunodominance of epitopes or immunological regulatory mechanisms of the outbred rabbit, which may be important factors influencing the outcome of the immune response.

The dichotomy in results obtained in mice and rabbits may be influenced (i) by differences in the P types of the mouse (P[16]) and rabbit (P[14]) virus strains used for challenge relative to the immunizing P type (P[2]) or (ii) by inherent differences in antigen uptake, processing, and epitope recognition between rabbits and mice. Following oral infection of mice with a heterologous virus, the majority of the intestinal antibody-secreting cells were directed toward VP2 and VP6, and only about 1% of the total virus-specific response was directed to VP4 (71). It would be interesting to perform similar experiments in rabbits. Ultimately, clinical testing of VLPs in the gnotobiotic piglet rotavirus disease model and humans will be required to determine whether results from the rabbit or mouse model will be more predictive of protection against disease in children. Results from studies of children naturally infected with human rotaviruses or bovine-human virus reassortants have suggested that VP4 may be the immunodominant neutralization antigen in a homotypic response (9, 81), while data from another study of children suggested that VP7 is the immunodominant neutralization antigen (78). Currently, virus reassortants with different VP4-VP7 combinations are being applied to study immune responses to each neutralization antigen after primary rotavirus infections and reinfections in humans (16, 36). In our study, we were unable to measure the VP4-specific immune response to ALA, the challenge strain, but it will be of interest to determine if N-Abs to ALA VP4 (P[14]) were also induced in 2/4/6/7-VLP- or SA11-vaccinated rabbits. Alternatively, the use of VLPs with ALA VP4 and VP7 proteins may increase the protection of 2/4/6/7-VLPs in the rabbit model because of the effects of protein interactions on immunogenicity and epitope presentations (7, 8).

The difference in protective efficacy afforded by the different VLP formulations in QS-21 did not result from variable immunogenicity of the different VLPs since significant differences in total serum or intestinal IgG antibody titers were not observed. However, protection in VLP-vaccinated rabbits correlated with the presence of rotavirus-specific intestinal IgG (P = 0.001, r = 0.495; Pearson's correlation coefficient). Our studies in mice also indicated that VLPs administered by parenteral or mucosal routes preferentially induced intestinal IgG with limited induction of intestinal IgA and that protection in mice immunized intranasally correlated with serum and intestinal IgG (51, 52). In contrast, our data for rabbits suggest that the presence of VP4-specific intestinal IgG may be required to induce high levels of protection since protective efficacy of 2/6/7- or 2/6-VLPs in QS-21 was poor. Studies of the immune responses to individual proteins in the intestine prior to virus challenge may be useful to determine if the differences in protective efficacy correlate with the presence of protein-specific intestinal antibodies in the rabbit (41). However, new sensitive assays that can detect native epitopes need to be developed for such studies.

Although intestinal antibody responses are expected to play a major role in protection, recent data in gene-knockout mice indicate that CD8⁺ T cells can mediate total short-term and partial long-term protection from rotavirus challenge (34, 35, 59). We do not yet know if VLPs can induce CD8⁺ T cells; therefore, we cannot exclude the possibility that CD8⁺ T cells also play a role in protection following immunization of rabbits with VLPs.

Protection studies in rabbits and mice suggest that VLPs may provide a safe and efficacious alternative to live oral rotavirus vaccines. A parenteral vaccine might overcome two of the problems observed with oral live rotavirus vaccines: (i) interference by maternal antibodies and (ii) poor replication of the vaccine virus. Because a parenteral vaccine would not require replication of the rotavirus vaccine strains, formulation of the vaccine might be easier since differences in immunogenicity based on growth properties of individual viruses would not come into play. However, the need for multivalent parenteral vaccines to induce broad immunity against the various human rotavirus serotypes will need to be examined. The results of our studies support the further development of rotavirus subunit vaccines as well as evaluation of combined parenteral-oral vaccination regimens.