Main

Smallpox, caused by Variola virus, was eradicated in 1977 by widespread vaccination with VACV, but without an adequate understanding of the immune responses necessary for protection1. Today there is concern that smallpox might re-emerge due to bioterrorism. The vaccines used previously have an imperfect safety record; thus, a safer smallpox vaccine is needed. With no means of testing new vaccines against human smallpox, however, their licensing will rely, in part, on comparative studies using the conventional vaccines as a benchmark.

Antibody responses induced after vaccination with VACV contribute to protective immunity against Variola virus1, and memory T-cell and B-cell responses remain more than 50 years after vaccination2,3. It is unclear, however, if this residual immunity protects against smallpox. Studies in animal models show that antibodies have the most important role in protection against orthopoxviruses4,5,6,7, whereas CD8+ T-cell responses prevent mortality and mediate late recovery4,6,7. These results suggest that smallpox vaccines are similar to many other human vaccines in that neutralizing antibodies are important for protection.

There are two antigenically distinct infectious forms of VACV, the IMV and the EEV (ref. 8). IMV is responsible for host-to-host spread, whereas EEV, containing a second lipid envelope and several virus glycoproteins, is better adapted to spread infection within the host. It was shown recently that IMV enters by fusion with the plasma membrane9, whereas EEV sheds its outer membrane by a unique, ligand-dependent, nonfusogenic mechanism that helps EEV to evade antibodies to IMV (ref. 10). In vivo studies in rabbits and mice showed that EEV-specific antibodies protected better against lethal challenge than did IMV antibodies11,12. These findings indicate that assays monitoring the humoral response to smallpox vaccination should evaluate both EEV-specific and IMV-specific antibodies11,12. EEV surface glycoproteins B5 and A33 are targets for protective antibodies in animal models13,14,15,16,17, but only B5 is a target for EEV-neutralizing antibodies. In spite of their importance, EEV-specific antibody responses have not been studied in humans. This is partly because the importance of EEV was not recognized until near the end of the smallpox eradication campaign, and partly because of the technical difficulty in working with EEV (ref. 8). IMV-neutralizing antibodies are directed against proteins A27, L1, H3, D8 and A17, and immunization with A27 (refs. 18,19), L1 (refs. 14,17), H3 (refs. 19,20) and D8 (ref. 19) conferred protection in mice and/or macaques. Antibody responses against specific IMV surface proteins, however, have not been studied clinically.

The objective of our study was to quantify the antibody responses against several VACV surface antigens that are probable targets for antibodies contributing to protection against smallpox. The correlations between antigen-specific antibody titers and virus neutralization against both IMV and EEV were studied. Here we provide a benchmark against which the potency of new vaccines and the level of immunity existing in those immunized previously can be compared.

Results

Antibody responses against EEV and IMV surface proteins

Previous analyses of human immune responses to smallpox vaccination considered EEV-specific antibodies only rarely, in spite of their importance for protection against disease. Here, we measured antibodies against three different EEV glycoproteins (B5, A33 and A56) by ELISA, and EEV-neutralizing antibodies by plaque reduction neutralization (PRN). In addition, antibodies against IMV were monitored by (i) ELISA using three IMV surface proteins (A27, H3 and L1) and virus-infected cell lysates (VACV), (ii) PRN using wild-type virus and (iii) enhanced green fluorescent protein (EGFP) reduction neutralization (ERN) using EGFP-tagged virus.

We used a panel of 60 serum samples from unvaccinated teenagers as a negative control group to calculate cutoff titers defining seropositivity (B5 1:75, A33 1:93, A56 1:111, A27 1:114, H3 1:120, L1 1:36, VACV 1:364). These cutoff titers were defined as three times the geometric mean titer obtained for the unvaccinated cohort and generated 100% specificity, except for A33 (93%), A56 (95%), H3 (95%) and A27 (93%). Serum samples from our study participants with antibody levels below the cutoff titers were considered seronegative and given an arbitrary value of one-half of that titer to allow calculation of geometric mean titers and to determine effective seroconversion. Seroconversion for VACV-naive vaccinees or effective boosting for revaccinees was defined as a four-fold rise in titer from that on day 0.

Upon screening 387 serum samples, we observed similar profiles for antibodies against B5, A33, A56, A27, H3 and VACV, although titers against VACV were higher than those against the single purified proteins (Fig. 1). Only very low, or no, antibody response was detected against L1. This result was not due to the use of misfolded antigen, because the recombinant protein used to crystallize L1 (ref. 21) gave equivalent titers (data not shown).

Figure 1: Antibody responses in humans (n = 92) after smallpox vaccination.
figure 1

Antibody end-point titers against EEV (B5, A33, A56) and IMV (A27, H3, L1) surface proteins and virus-infected cell lysates (VACV) were detected at different times after vaccination by ELISA. IMV-specific EGFP reduction neutralization (IMV ERN) titers are also depicted. Geometric means, 95% confidence interval of previously unvaccinated (red bars) and vaccinated (blue bars) participants, median values of whole population (black bars), cutoff titers for seropositivity (dashed black line) and representative titer of commercial VIG (dotted green line) are indicated. Significant differences between groups of vaccinated and unvaccinated subjects are shown as *P < 0.0167, **P < 0.003 and ***P < 0.0003.

Full size image

The B5-, A33-, A56-, A27-, H3- and VACV-specific ELISA titers correlated well with EEV-specific (EEV PRN) and IMV-specific (IMV ERN, IMV PRN) neutralization titers (ND50) (Supplementary Table 1 online). Among the EEV antigens, we observed the highest correlation (r = 0.92) between the B5-specific titers and the EEV-specific ND50 values. This high correlation did not, however, prove a cause-effect relationship between B5-specific and EEV-neutralizing antibodies. Therefore, we carried out EEV-specific PRN assays using sera (n = 8) to which purified recombinant B5, A33 or A56 glycoprotein was added (Fig. 2a). The EEV-neutralizing activity of day 21 sera was abrogated by addition of 1 μg or 10 μg of B5 protein and was reduced by >50% with 0.1 μg (P < 0.001), whereas there was no decrease in EEV neutralization after addition of A33 or A56 proteins. Sera from primary or secondary vaccinees gave the same result and addition of recombinant proteins had little or no effect on EEV-specific infectivity in the absence of human serum (data not shown). These data show that B5 is the only target for EEV-neutralizing antibodies induced after smallpox vaccination. Comparison of the B5-specific ELISA and the EEV PRN data showed that EEV-specific ND50 titers of 1:50 or 1:100 corresponded to B5-specific ELISA titers of 1:1,500 or 1:3,800, respectively.

Figure 2: Mean residual neutralizing activity in human sera (n = 8) after antibody depletion.
figure 2

To deplete sera from specific antibodies against B5, A33, A56, H3 or L1, a volume of serum containing one EEV-specific or IMV-specific ND50, respectively, was incubated with different amounts of proteins for 1 h before the assay. The remaining neutralizing activity was determined in duplicate and compared with untreated serum. (a) EEV-neutralizing activity in PRN after addition of indicated amounts of purified B5, A33 or A56 glycoproteins. (b) IMV-neutralizing activity in PRN after addition of various amounts of purified A27, H3 or L1 proteins or ultraviolet-inactivated IMV (4 × 107 (open bars), 4 × 106 (gray bars) and 4 × 105 (filled bars)). Residual EEV- and IMV-specific activities of day-21 serum samples from 8 vaccinees (3 previously unvaccinated, 4 previously vaccinated and 1 unknown) are shown.

Full size image

A27- and H3-specific ELISA titers correlated better with IMV-specific than with EEV-specific neutralization assays (Supplementary Table 1). To assess the contribution of specific IMV antigens as targets for neutralization, IMV-specific PRN assays were performed with sera containing increasing amounts of purified recombinant A27, H3 and L1 protein (Fig. 2b). IMV-neutralizing activity was reduced by 8% (P = 0.003), 13% (P < 0.001) or 20% (P < 0.001) by adding 0.1 μg, 1 μg or 10 μg of A27, respectively. Similarly, adding increasing amounts of H3 reduced neutralization by 6% (P = 0.04), 11% (P = 0.01) or 9% (P = 0.01). In contrast, the addition of L1 had no effect. Combining various amounts of A27 and H3 (up to 50 μg) reduced IMV neutralization by 20–42% (P < 0.001; data not shown). IMV neutralization was abrogated, however, by adding the equivalent of 4 × 107 plaque-forming units (PFU) of purified, ultraviolet-inactivated IMV and reduced by 62% (P < 0.001) or 20% (P < 0.001) by 4 × 106 or 4 × 105 PFU, respectively (Fig. 2b). This shows that, in contrast to EEV neutralization, IMV-neutralizing antibodies target multiple surface proteins, including A27 and H3. Although L1 can be a target for IMV-neutralizing antibodies14,17, only very low levels are induced after smallpox vaccination in humans.

Impact of vaccination history on antibody responses

A comparison of antibody titers with time since vaccination showed that antibody levels decreased significantly between day 21 and 6 months after vaccination (Fig. 1). However, beyond 10 years after vaccination (for revaccinees), no significant (P > 0.05) decrease in EEV-specific or IMV-specific antibody titers was detected (Supplementary Fig. 1 online). The inability to detect a long-term decline in a cross-sectional study might result from the wide range of antibody levels observed in vaccinated individuals or the inherent stability of antibody titers after the initial decline.

The 92 vaccinees were divided into three groups according to their vaccination history: unvaccinated (n = 18), vaccinated (n = 64, including 6 who had been vaccinated more than once) and vaccination status unknown (n = 10). Analysis of the day 0 samples, however, revealed that 4 of 18 subjects with no previous reported vaccination were seropositive against all antigens tested in ELISA (except for L1), and two of these contained IMV-neutralizing antibodies by ERN and PRN assays. To compare the immune responses deriving from primary or secondary vaccination accurately, we regarded these four subjects as individuals with unknown vaccination status.

Among those vaccinated previously, 70–77% were seropositive for antibodies to B5, A33, A56, A27 and VACV at day 0 (Supplementary Fig. 2 online). All those vaccinated more than once previously were seropositive against all antigens tested (except L1). Those with an unknown vaccination history displayed high seropositivity levels at day 0 indicating that they were mostly previously vaccinated people.

We defined benchmark ELISA antibody titers for each antigen in each vaccination group (Table 1). Titers at days 0 and 7 were higher in vaccinated than in unvaccinated people. We observed no significant differences for all the antigens tested between the group of previously vaccinated participants and those of unknown vaccination history. Notably, at day 0, those vaccinated more than once previously displayed higher B5-specific (P = 0.003) and A33-specific (P = 0.048) antibody levels than those with only one previous vaccination, whereas multiple vaccinations had no effect on IMV-specific titers, that is, A27, H3, L1 or VACV.

Table 1 Comparative serological findings by vaccination history
Full size table

In general, there was little increase in antibody levels by day 7 after vaccination (Fig. 1) irrespective of the subjects' vaccination history (Fig. 1 and Table 1) or prevaccination antibody levels (only six individuals exhibited augmented antibody levels against one or two antigens tested in ELISA; Supplementary Figs. 3 and 4 online). Among the individuals vaccinated previously, 78–92% showed an increase by day 21 against B5, A27 or VACV (Supplementary Fig. 4). Similarly, by day 21, 85–100% of the vaccinia-naive individuals seroconverted against B5, A27 or VACV. In the revaccinees, there was no correlation (P > 0.05) between an effective boosting response against B5 (r = 0.09), A33 (r = 0.11), A56 (r = 0.09), A27 (r = 0.08), H3 (r = 0.10), L1 (r = 0.06) or VACV (r = 0.08) and the time since their last vaccination. The level of preexisting antibodies at day 0, however, correlated (P < 0.01) inversely with the respective boosting against B5 (r = –0.27), A33 (r = –0.45), A56 (r = –0.64), A27 (r = –0.47) and VACV (r = –0.48) at day 21, so that vaccinees with existing high antibody levels were less likely to show boosting. At day 21, we found considerably lower (P < 0.001) B5-, H3- and VACV-specific antibody titers in unvaccinated people compared with those vaccinated or individuals with unknown history (Fig. 1 and Table 1). In addition, B5-specific titers were higher in those individuals vaccinated more than once compared with those vaccinated only once (P = 0.030). Antibody titers against A33, A56, A27 and L1 at day 21 were unaffected by previous vaccinations.

By 6 months and 1 year after vaccination, in all groups, VACV-specific antibodies decreased to 27–37% and 25–27% of their day-21 titers, respectively (Supplementary Fig. 5 online). Also, at 1 year, 93–100% of the participants vaccinated previously were still seropositive for B5 and VACV, whereas only 67–83% of primary vaccinees remained seropositive (Supplementary Fig. 2). In addition, we detected two- to six-fold higher antibody levels against B5, H3 and VACV in those vaccinated previously compared with unvaccinated individuals (Fig. 1 and Table 1). Similar observations were made when comparing the group with unknown history with the unvaccinated group. No differences were found against any antigen tested between the vaccinated participants and those with unknown history. In participants with more than one previous vaccination, B5-specific antibodies remained significantly higher after 1 year than in the subjects with one former immunization (P = 0.034).

Discussion

This study analyzed the antibody responses in humans after vaccination with VACV strain Lister, the International Smallpox Vaccine Reference strain, and provides a benchmark against which residual immunity in today's population and the immunogenicity of new smallpox vaccines may be compared. Data presented show that both primary vaccinees and revaccinees elicit antibody responses within 21 d of vaccination with the Lister vaccine, confirming effective 'takes' (ref. 22). Antibody responses were measured by ELISA using virus-infected cell lysates, three different EEV antigens and three different IMV antigens. Strong responses were detected against B5, A33, A56, A27, H3 and VACV. L1 induces neutralizing and protective antibodies in animal models14,17, but we found only low levels of L1-specific antibodies after smallpox vaccination.

Hitherto, the antibody responses to EEV in humans have largely been ignored, although antibodies to EEV antigens are important in protecting against poxviruses11,12,13,18. Here, we showed that VACV strain Lister induces antibodies to the EEV proteins B5, A33 and A56 in humans. Comparative analysis showed that the levels of antibodies to B5 induced in humans correlated closely with the magnitude of EEV-specific neutralization in vitro; therefore, our B5-specific ELISA predicts EEV-neutralizing activity accurately. B5 was known to be a major target of EEV-neutralizing antibodies in vaccinia immune globulin (VIG)16. Our results confirm this and also show that B5 is the only target of EEV-neutralizing antibodies after smallpox vaccination. This is important for the development of new smallpox vaccines and is particularly relevant to the attenuated VACV strain LC16m8, which was derived from the Lister strain and has a disrupted B5R gene. A previous study (ref. 23) showed that a virus lacking gene B5R, which was derived from LC16m8, induced levels of protection in mice similar to those of the NYCBH strain (Dryvax) and reported that B5 is nonessential for protective immunity. Given, however, that (i) antibodies are critical for vaccine-mediated protection against secondary poxvirus infections4,5,7, (ii) EEV-specific immunity correlates with increased protection11,12,13, (iii) combinations of subunit vaccines inducing antibodies neutralizing both EEV and IMV confer the best protection in animal models17,18, (iv) B5 is the only reported target for EEV-neutralizing antibody after live immunization in humans and animals, and (v) the B5R gene is present in all strains of variola virus that have been sequenced, we believe it is unwise to exclude B5 from any future smallpox vaccine. A study reporting discrepancies in antibody profiles after VACV immunization in mice, macaques and humans indicated that B5-specific antibodies cannot always be found in human VIG (ref. 24). These results, however, were obtained with B5 that was expressed in Escherichia coli and might not represent authentic protein from mammalian cells. Notably, immunization of rabbits with B5 antigen from E. coli did not induce EEV-neutralizing antibody15.

Although EEV-neutralizing antibodies are directed against a single antigen (B5), IMV-neutralizing antibodies recognize several proteins. Antibody-depletion studies confirmed that neutralizing antibodies recognize both A27 and H3. The fact that addition of A27 or H3 reduced the IMV-neutralizing titer only modestly, whereas inactivated IMV virions blocked all IMV-neutralizing activity of human serum, suggests there are other IMV targets (for example, D8 and A17). Consistent with data presented here, H3-specific antibodies with IMV-neutralizing activity have been described in human VIG (ref. 20), whereas L1-specific neutralizing antibodies were not detected25. The median antibody levels against specific antigens observed 1 year after vaccination were very similar to the titers detected in commercial VIG (Fig. 1).

Humoral immunity after smallpox vaccination seems to be long-lived. Residual antibodies against VACV, EEV glycoproteins B5, A33 and A56, and the IMV proteins A27 and H3 were detected in the day 0 sera of health care workers vaccinated more than 50 years ago. Cross-sectional studies2,3,26 described long-term persistence of VACV-specific ELISA or IMV-neutralizing antibodies and of EEV-neutralizing antibodies26. This report, however, is the first analysis of antibody responses against EEV surface glycoproteins B5 and A33 that are important for antibody-mediated protection against disease. In a cross-sectional study, we corroborated these findings and showed that B5-specific antibodies remain in 48% of the Italian population more than 25 years after the end of the smallpox eradication campaign27. We found higher titers of B5-specific and A33-specific antibodies in vaccinees with multiple previous vaccinations, indicating that repeated vaccination provides long-lasting, increased humoral immunity against EEV antigens. A previous study drew similar conclusions from their cross-sectional study26. In this regard it is noteworthy that the World Health Organization recommended revaccination every three years for those in areas where Variola virus was endemic1. The benefit of multiple vaccinations is also shown by analysis of the day 21 sera. The group having received two former vaccinations had the highest amount of antibodies to B5, and the lowest titer was found in the VACV-naive recipients. Similarly, increased IMV-specific antibodies are induced and maintained in people with multiple vaccinations3. Unfortunately, it is difficult to correlate the outcome of multiple vaccinations with protection.

It was reported that immune responses to vaccination in VACV-primed individuals occur more rapidly than in primary vaccinees28,29. We detected no significant increase by ELISA or PRN, however, 7 d after immunization, irrespective of any previous vaccination. The precise knowledge of the kinetics of antibody induction is valuable for postexposure vaccination against smallpox. In our study, the time of peak antibody response cannot be deduced because of the limited number of time points analyzed.

Our data show that B5 is the only target for neutralization on EEV, whereas IMV has multiple targets for neutralizing antibody; multiple vaccinations improve long-lived B5-specific and thus EEV-neutralizing antibody responses for protection; and more than one vaccination induces increased A27-specific and H3-specific, that is, IMV-neutralizing antibody responses. The data also define neutralizing titers against IMV and EEV following vaccination with an effective smallpox vaccine and define ELISA antibody titers against multiple IMV and EEV surface proteins induced by the Lister vaccine. These immune responses provide a benchmark against which the level of existing immunity in a population and the immunogenicity of new smallpox vaccines can be compared.

Methods

Study design, vaccine and subjects.

In December 2002, the UK government decided to immunize ∼350 health care workers against smallpox using the VACV strain Lister/Elstree vaccine (Swiss Serum Institute, Berne, Switzerland, batch number 84430 and 84431) to constitute regional groups that could respond to a potential smallpox emergency. Participants in this study were healthy volunteers and gave signed, informed consent. Ethical approval was given by the Thames Valley Multi-centre Research Ethics Committee. Exclusion criteria were defined and vaccinations were performed following the guidelines of the Joint Committee on Vaccination and Immunization and Committee on Safety of Medicines, the Department of Health. Vaccination, clinical follow-up of participants and blood collection was carried out at regional Occupational Health Departments in the UK (ref. 22). For reasons of confidentiality, serum samples were tested without knowledge of sex or age of vaccinees. In all, 92 vaccinees volunteered to donate blood, including 18 individuals with no previous recorded smallpox immunization, 58 with one previous vaccination, 6 with two vaccinations and 10 subjects whose vaccination status was unknown.

Antigen production.

We obtained purified, soluble forms of recombinant B5, A33 and A56 glycoproteins expressed in Chinese hamster ovary cells and Western Reserve (WR)-infected cell lysates (VACV) for ELISA as described12. We cloned the full-length A27L gene into pET28a (Novagen) and the first 555 nucleotides of the L1R gene (corresponding to amino acids 1–185) into pET24a (Novagen). Both antigens contained a carboxy-terminal 6xHis tag and were expressed in E. coli strain BL21(DE3). Soluble A27 was purified by affinity chromatography. L1 inclusion bodies were denatured, the protein refolded and purified as described21. The VACV strain WR-derived antigens are very highly conserved among other vaccine strains and orthopoxviruses (Supplementary Table 2 online).

ELISA.

We determined EEV glycoprotein-, IMV protein- and VACV-specific serum IgG titers by ELISA as described12. Briefly, we coated plates with 100 ng/well of purified B5, A33, A56, A27, H3, L1 or bovine serum albumin (negative control) or ultraviolet-inactivated (105 PFU) VACV-cell lysate. After blocking, we incubated the plates for 90 min with two-fold serial serum dilutions and then subsequently incubated them with goat antibody to human IgG (γ-chain specific) alkaline phosphatase conjugate (1:1,000, Sigma-Aldrich). We added P-nitrophenyl phosphate substrate (Sigma-Aldrich) and measured the optical density after 60 min. IgG end-point titers were defined as the reciprocal serum dilutions giving twice the average optical density values obtained with bovine serum albumin. We used a control serum from an individual vaccinated multiple times (titers: B5 1:1,201, A33 1:256, A56 1:516, A27 1:987, H3 1:360, L1 1:81, VACV 1:10,240) to normalize end-point titers between plates and assays.

Neutralization assays.

We determined IMV-specific and EEV-specific neutralizing activity using PRN assays12. We also determined IMV neutralization by detection of reduction of an EGFP-tagged virus using flow cytometry. The ND50 values obtained in neutralization assays were defined as the reciprocal of the dilution giving a 50% reduction of plaque number or EGFP expression. We used a control serum from a vaccinated individual (titers: IMV ERN 1:454, IMV PRN 1:579, EEV PRN 1:47) to normalize ND50 values. For PRN, we used IMV (VACV strain WR) purified from sucrose gradients or freshly prepared EEV containing monoclonal antibody 2D5 (1:1,000)12. In the IMV-specific EGFP reduction neutralization assay (IMV ERN), we measured antibody titers using sucrose-gradient–purified IMV from VACV strain vA5L-EGFP-N (ref. 30). We prepared two-fold serial serum dilutions in duplicate, added vA5L-EGFP-N (10,000 PFU) and incubated for 1 h at 37 °C. Subsequently, we added 2 × 105 HeLa cells, and plates were left on ice for 30 min and then incubated at 37 °C for 15 h. We collected cells, fixed with 1% paraformaldehyde, gated for forward/side scatter and for EGFP expression, and acquired 12,000 events (FACS Calibur, BD).

Statistical analysis.

We carried out correlation analysis between data sets from ELISAs (n = 387), IMV ERN (n = 206), IMV PRN (n = 81) and EEV PRN (n = 21) using a nonparametric, two-tailed Spearman correlation test (SPSS 12.0, SPSS). The Spearman coefficient r and the corresponding P value are indicated. We also performed correlation analysis between ELISA titers at day 0 and the number of years since vaccination for the individuals whose previous vaccination dates were known (n = 59). We also carried out correlation analysis between the seroconversion rate and the time since previous vaccination (n = 55) and the titer at day 0 (n = 86), respectively. We compared ELISA and IMV ERN titers obtained in groups of different vaccination history using a nonparametric, unpaired, two-tailed Mann-Whitney U-test. Corresponding P values are indicated. We applied a Bonferroni correction (3 tests, P < 0.0167) for multiple comparisons.

URLs.

Department of Health, UK: http://www.dh.gov.uk. Exclusion criteria: http://www.dh.gov.uk/assetRoot/04/08/39/64/04083964.pdf. Poxvirus Bioinformatics Resource Center: http://www.poxvirus.org/.

Note: Supplementary information is available on the Nature Medicine website.