Vaccination with live vaccinia virus-based vaccines (e.g., Dryvax) elicits long-lasting and robust protection against orthopoxvirus infection and targets both the EV and MV forms of the virus (8
). However, there is little information regarding the actual correlates of the protective immune response against smallpox infection (2
). Previous studies of the protective humoral immune response against smallpox as represented by VIG, an antiserum collected from Dryvax vaccinees and used to treat vaccination adverse events (26
), have identified numerous virus antigen targets recognized by antibodies in VIG. In addition to numerous MV proteins recognized by VIG serum antibodies, the EV-specific proteins A33R, B5R, and A56R were detected in Western blot analysis by antibodies in VIG (4
). Recent studies with DNA or protein candidate vaccines have reinforced the importance of an immune response against EV proteins, as well as MV envelope proteins, to provide the best protection against virus infection (17
In order to quantify the specific antibody response directed against the EV proteins, we developed a set of EV-specific ELISAs based on rEV antigens corresponding to the natural A33R, A34R, A56R, and B5R ORFs. While our studies were in progress, other laboratories reported the development of EV-specific ELISAs using truncated forms of A33R and B5R proteins produced in insect cells or truncated forms of A33R, B5R, and A56R produced in Chinese hamster ovary cells (17
). Although these truncated EV antigens were found to be viable ELISA substrates in quantifying the EV-specific antibody responses, we elected to use an SFV expression system to produce full-length EV antigens in BHK-21 cells, where mammalian-specific posttranslation modification may be conferred, and subsequently immunoaffinity purified the proteins by virtue of the C-terminal epitope tag (31
). Natural EV proteins are transmembrane proteins that acquire numerous mammalian-specific posttranslational modifications that may influence their correct folding. Protein expression was verified by Western blot analysis using primary antibodies directed against the epitope tag. Western blot analysis using anti-Dryvax serum as the primary antibody recognized prominent protein bands of the expected sizes for the corresponding EV A33R, A56R, and B5R. The predominant EV protein species recognized by Western blot analysis with the epitope tag antibody or anti-vaccinia virus sera appeared as numerous diffuse bands after SDS-PAGE under reducing and nonreducing conditions. These multiply reactive EV protein species probably represent unmodified, multimeric, posttranslationally modified EV proteins and cleavage products from EV protein precursors from rSFV-infected cell lysates, as previously described for these proteins (24
). The response to the A33R and A56R protein species treated under nonreducing conditions in the Western blot analysis using the anti-Dryvax serum was much stronger than that observed under reducing conditions (Fig. ). Serum antibodies reacted predominantly with the high-molecular-mass forms of A33R and A56R protein species. The effect was less pronounced for B5R. These results suggest that a strong antibody response is elicited against epitopes present only in the oligomeric forms of A33R and A56R. The purified A56R and B5R preparations also contained small-molecular-mass protein fragments of less than 14 kDa that were only detected using the epitope tag antibody, suggesting that these viral peptide fragments at the C-terminal end were probably not very antigenic in vivo. No protein bands corresponding to A34R were detected after Western blot analysis using serum from mice immunized with Dryvax or the SFV recombinant expressing A34R. Western blot analysis using the epitope tag antibody demonstrated that the A34R protein made by the SFV vector is abundant and migrates with the expected molecular mass (Fig. ). Although we cannot formally rule out the possibility that recombinant A34R is lacking posttranslational modifications that affect its antigenicity, the most likely explanation is that A34R is not very antigenic, since VIG does not detect A34R when used as the primary antibody in Western blot analysis (25
). Moreover, these results are consistent with data showing that antibodies raised against A34R did not neutralize EV (20
To optimize the utility of each of the four EV proteins as an immobilized antigen on an ELISA plate, we tested the ability of each EV protein to quantify antibody titers using the epitope tag antibody as a primary ELISA antibody and a panel of vaccinia virus antisera. The results from these ELISA experiments recapitulated the results seen for the Western blot experiments in that antibodies to all four proteins were detected using the epitope tag as the primary antibody in an ELISA, while none of the tested vaccinia virus hyperimmune sera detected A34R. The specificities of the A33R, A56R, and B5R ELISAs were demonstrated by using the murine antisera produced by using vA5Lint (a recombinant WR strain vaccinia virus that lacks the A56R gene) or UV-inactivated MV particles. While Western blot analysis demonstrated that anti-vA5Lint immune serum detected A33R and B5R but not A56R, the A56R ELISA also detected a trace amount of activity against A56R in the anti-vA5Lint immune serum. Since vA5Lint was constructed from the pVOTE vectors (60
) that retain a portion of the A56R gene, the low A56R activity seen in the ELISA is likely directed against a truncated product of the A56R ORF. Western blot analysis using the anti-MV antiserum did not detect A33R and weakly detected B5R. Unexpectedly, however, the anti-MV antiserum contained sufficient antibody activity to recognize the A56R protein on a Western blot and to be quantified in the A56R-specific ELISA. In addition, the A56R and B5R ELISAs measured considerable A56R antibody, as well as detectable B5R antibody, in the anti-MV serum. Since there was no detectable live virus in the UV-inactivated MV sample used to produce the MV hyperimmune serum (limit of detection, <100 PFU/ml), it is unlikely that replicating vaccinia virus was responsible for the generation of A56R and B5R antibodies in this antiserum. Although it is possible that residual A56R and B5R protein was present in the UV-inactivated MV preparation, it seems more likely that, since inactivated virus is still capable of infecting cells, these proteins were synthesized following limited early gene expression. All three EV gene products appear to be transcribed at both early and late times postinfection (13
). The relative immune responses generated by the UV-inactivated particles can be explained by the observation that the product of the A56R gene partially localizes on the cell surface (33
), whereas the B5R protein sequesters within the Golgi apparatus of the infected cell (33
). The absence of an A33R antibody response in the anti-MV serum can be explained by the observation that the expression of A33R is predominantly late (44
). The sum of the experiments with the panel of hyperimmune antisera and the antisera from mice immunized with a single dose of 106
Dryvax demonstrated that the SFV-recombinant A33R, A56R, and B5R proteins were suitable antigens for use in sensitive and quantitative EV-ELISAs, particularly in the mouse model of vaccination.
Since previous reports have shown that numerous vaccinia virus proteins are recognized by VIG, including the EV proteins A33R, A56R, and B5R (9
), we also evaluated the EV-specific ELISAs in a limited number of experiments for their utility in quantifying EV antibody titers in human sera. While the ELISAs for A33R, A56R, and B5R demonstrated antibody titers approximately 250-fold-higher than those in the control human serum, there were considerable background amounts using the conditions established for the murine vaccination experiments. Interestingly, the A34R-specific ELISA detected virtually no antibody above the control human serum background, consistent with previous results using Western blot analysis. Nevertheless, the results suggested that similar sensitive quantitative EV-ELISAs could be developed for human vaccine trials using the rSFV EV reagents.
We also conducted a series of EV neutralization assays in order to determine if the humoral response measured by the EV-specific ELISAs correlated with biological activity against the EV form of the vaccinia virus. As demonstrated in the comet inhibition assay, both the mouse and human anti-Dryvax immune sera were able to inhibit new infections due to EV by blocking the spread of virus from the infected cells. Similar inhibitory activity was observed with the hyperimmune serum raised against vA5Lint, which is only missing the A56R protein, but no inhibitory activity was detectable using the hyperimmune serum raised against purified UV-inactivated MV particles, which is greatly reduced in anti-A33R and anti-B5R antibodies but contains relatively high anti-A56R titers. The results suggested that A56R antibodies are not important for EV-neutralizing activity. Although our data do not discriminate between A33R-specific and B5R-specific antibodies in EV neutralization, it has recently been reported that polyclonal antibodies and MAbs directed against A33R or B5R possess EV neutralization activity (1
) and that the response against B5R is thought to constitute the primary EV neutralization activity of VIG (4
). While the comet inhibition assay provides a reliable and simple test to detect anti-EV antibody activity, it is not quantitative or particularly sensitive. In our hands, for example, we were unable to reliably measure comet inhibition using antisera from mice immunized with a single dose of 106
Dryvax, thus limiting the assay's usefulness for comparing EV responses to candidate vaccines. Similarly, the A33R, A56R, and B5R monospecific antisera generated by rSFV vector immunization were unable to limit satellite plaque formation in a comet inhibition assay at any dilution (data not shown).
One of the major goals of the present study was to develop sensitive assays for quantifying the EV antibody response to candidate smallpox vaccines such as MVA (50
). We have recently compared MVA to Dryvax in its ability to protect against a lethal poxvirus challenge in a small animal model (40
). The results of these studies indicated that an effective humoral response elicited by MVA or Dryvax immunization plays an important role in protection, but the role of individual antigens in protective immunity is still largely unknown. The results from the experiments in the present study demonstrated that the EV-ELISAs are useful for quantifying the antibody response to individual EV antigens and can distinguish the responses to different vaccines and vaccination protocols. For example, the A33R and B5R responses following a two-dose vaccination with MVA (108
PFU) at 0 and 6 weeks were significantly higher than with a single dose of Dryvax (106
PFU). As specific EV vaccine markers are identified and correlated with protection, these quantitative assays will be increasingly valuable for assessing their efficacies in preclinical animal studies.
Based on previous immunization studies done with SFV-based vaccines (5
), we suspected that these rSFV expression vectors could be used to generate monospecific antisera against each of the EV antigens. Following rSFV vector immunization, a strong serum antibody response against each of the expressed EV proteins except for A34R was observed. This response was greater than the corresponding specific-antibody titers elicited by two doses of Dryvax (106
PFU). The absence of A34R-specific antibodies following rSFV-A34R immunization is consistent with other results that suggest that the A34R protein does not induce a substantial antibody response. The monospecific antisera elicited by the rSFV-A33R and -A56R vectors reacted strongly with protein bands in Western blot analysis derived from vaccinia virus infected-cell lysates that corresponded to A33R and A56R. In the immunofluorescence studies, all three monospecific antisera specifically stained Dryvax-infected cells, albeit the staining by the B5R-monospecific antiserum appeared less intense than the staining by the A33R- and A56R-monospecific antisera. Thus, the rSFV vectors expressing EV proteins are capable of generating polyclonal EV-monospecific antisera that can be used for dissecting the EV immune response.
Recent studies have reported that the induction of antibody responses against A33R and B5R antigens is important in the protective humoral response (7
). An antibody response against A33R correlates with protection (17
); an anti-B5R antibody response is primarily responsible for the EV neutralization activity in VIG (4
), and B5R antibodies are able to partially protect mice in passive immunization and challenge studies (7
). These results not only suggest the value of antigen-specific assays to quantify the response to EV antigens following vaccination, they also suggest that the rSFV-A33R and rSFV-B5R vectors, or the antibodies produced by them, might confer protection against a poxvirus infection in animal models. Although we have not explored this concept in depth, the animals used to produce the monospecific EV antisera were challenged intranasally with 25 times the 50% lethal dose of WR strain vaccinia virus (data not shown). Mice inoculated with rSFV-A33R exhibited partial protection (two of five survived) against the lethal intranasal challenge with pathogenic vaccinia virus WR, even though the prechallenge A33R-specific antiserum from these mice contained no comet inhibition activity. This partial protection afforded by the rSFV-A33R inoculations is consistent with previous reports. In contrast, all mice inoculated with rSFV-A56R and rSFV-B5R particles were susceptible to the vaccinia virus WR challenge dose. The induction of an A56R-specific response was not expected to be critical for protection, given that it has been reported that one-third of EV particles lack A56R antigens (27
) and that the hyperimmune antiserum (anti-A5Lint) that was deficient in A56R antibodies exhibited potent comet inhibition activity. At the present time, we do not have enough data to speculate on the reason for the inability of the rSFV-B5R-immunized animals to resist challenge. Further immunization experiments are planned with the rSFV-EV vectors, both singly and in combination. Nevertheless, the preliminary results point out the usefulness of these vectors in dissecting the protective immune response to vaccinia virus EV proteins.
In summary, the evaluation of new-generation smallpox vaccines will require the use of animal models as surrogate markers for vaccine efficacy (49
). Although the correlates for protection against smallpox are unknown, investigations of DNA-based and protein subunit vaccines have shown the requirement for both MV and EV components for maximum efficacy (17
), and assays will be needed to accurately quantify the EV humoral response after vaccination. The assays described here using the A33R, A56R, or B5R proteins were sensitive, specific, and reproducible. The proteins appeared to be bona fide copies of the vaccinia virus EV proteins, as antisera derived from vaccinia virus-based immunizations recognized the rSFV-expressed EV proteins in Western blotting, and the proteins were able to adopt the oligomeric forms observed by natural vaccinia virus EV proteins under nonreducing conditions. The EV-specific assays allowed a quantitative comparison of the EV antibody responses following Dryvax and MVA vaccination in a mouse model; similar quantitative comparisons of other candidate smallpox vaccines can be used to correlate protection against lethal challenge with EV-directed antibody response. The rSFV-EV expression vectors will also be valuable as tools for dissecting the protective immune response to vaccination, as vectors for generating polyclonal monospecific antisera, and possibly as experimental vaccines.