The quantitative and qualitative nature of the antibody response required to protect against initial HIV-1 infection has been difficult to study because of the limited animal models for HIV-1. The recent construction of SHIV isolates based on primary HIV-1 envelope genes allows the evaluation of anti-envelope antibodies in a nonhuman primate model system. We studied three well-characterized antibodies, alone and in combination, to correlate in vitro neutralizing activity with protection. HIVIG, 2F5, and 2G12 were chosen because their anti-HIV-1 neutralizing activities have been well described (30
) and because they were available in sufficient quantities to perform passive-transfer studies. Since our in vitro data showed that double and triple combinations of these antibodies were substantially more potent than any single antibody (30
), these data could be compared to the protective effect in vivo.
The passive-transfer data showed that sterile protection against intravenous challenge with SHIV-89.6PD could be achieved by preexisting antibody but that a high level of neutralizing antibodies was required. Even with the triple-combination HIVIG/2F5/2G12, sterile protection was achieved in only three of six animals. Thus, the level and potency of infused antibody appeared to be below the threshold for complete protection. This corresponded to our in vitro neutralization data for these monkeys; the neutralizing activity in plasma at the time of challenge was greater than 99% but not consistently 100%. The association between in vitro neutralization and in vivo effect was further strengthened by the data showing that sterile protection was achieved only in monkeys infused with the triple-antibody combination (which was most potent in vitro) and that the double combination (2F5/2G12) was more effective than any single antibody. Despite this association of neutralizing activity and sterile protection, extrapolation of the absolute amount of neutralizing antibody in serum required to protect against human HIV-1 infection should be approached with caution. SHIV-89.6PD is highly pathogenic, causing AIDS in 12 to 14 weeks, and it was administered by the intravenous route at a relatively high inoculum. Thus, the level of antibody required to protect in this model may not be the same as that required to protect against a low-dose mucosal exposure to HIV-1 in humans. Additionally, there are several antibody-mediated mechanisms that may play a role in protection. Our experiments sought only to correlate virus neutralization, measured against activated PBMC targets, with in vivo protection.
It is important to note that antibody levels far below those required for sterile protection had a substantial effect on the viral load in plasma, the CD4+
cell count, and the clinical outcome. In contrast to IVIG controls, animals that received HIVIG or MAb 2G12 alone did not suffer complete CD4+
cell loss, were able to mount an anti-SHIV antibody response, and had little or no evidence of lymphoid-cell depletion or opportunistic infections at necropsy. Our unpublished data and data published by others (28
) suggest that such animals can often control SHIV-infection through 1 to 2 years of follow up. The fact that the concentrations of HIVIG and 2G12 in serum at the time of challenge were only in the IC90
range against the challenge virus (and well below the IC99
level) suggests that moderate levels of preexisting neutralizing antibody could be an important component of protective immunity. Compared to MAb 2G12, the lack of protective effect by MAb 2F5 (Fig. ) was unexpected because its in vitro neutralizing activity was substantially greater than that of 2G12. At the time of challenge, 2F5 concentrations in serum were 10 times the IC90
while 2G12 concentrations were approximately equal to the IC90
(Table ). Thus, in this case, the magnitude of preexisting neutralizing activity did not correlate with the level of partial protection. One possible explanation was the shorter half-life of 2F5 than of 2G12 (5 days versus 13 days). While the almost threefold shorter half-life of 2F5 does not appear to completely account for the lack of protective effect by 2F5, it is likely that a sustained neutralizing-antibody level contributes to protection by reducing early plasma viremia. In future experiments, we could address the role of antibody half-life by reinfusing antibody shortly after virus challenge. Also of note, the half-life of 2F5 was further shortened (to 2.75 days) when administered as part of the triple combination HIVIG/2F5/2G12. This may have been due to some combination of antibody agglutination and/or increased clearance, but the specific cause is not clear. In vitro experiments did not reveal any inhibitory effect on the ELISA measurement of 2F5 concentration when the three antibodies were combined. This effect was not seen with 2G12 (i.e., the pharmacokinetics were similar when it was administered alone and in combination). A comparison of viral loads in SHIV-infected monkeys in the HIVIG/2F5/2G12 and 2F5/2G12 groups reveals lower viral loads in the double-antibody group. This further supports the hypothesis that antibody half-life influences the overall protective effect. Given the potent beneficial effect of the double-MAb combination on the CD4+
cell count, viral load, and clinical disease, future studies could titrate down the infusion dose to investigate the lowest level of neutralizing antibodies that can attenuate the SHIV-89.6PD disease course.
In summary, our data from the SHIV-macaque model show a general correlation between in vitro neutralization and protection. Sterile protection was achieved only with the triple antibody combination, which demonstrated the most potent virus neutralization in vitro. Similarly, the double-antibody combination, which was highly synergistic in vitro, was substantially more effective in vivo than either MAb alone. While no individual antibody provided sterile protection, there were beneficial effects on viral load, CD4+
cell count, and clinical outcome. The data suggest that a vaccine that induces neutralizing antibody could have a protective effect against HIV-1 infection or disease. Since current HIV-1 vaccine candidates do not elicit the high levels of neutralizing antibody that were achieved in our passive-transfer experiments (32
), one interpretation of the data is that these vaccines will fail to protect against HIV-1. However, in the setting of active immunization and natural mucosal exposure to HIV-1, numerous factors could influence protection. Our passive-transfer experiments do not account for the role of cellular responses, and active immunization should elicit amnestic responses with raising rather than falling antibody levels. This latter point is of particular interest in light of the partial protective effect afforded by HIVIG and MAb 2G12, despite modest in vitro neutralizing activity. Since the SHIV-macaque animal model appears to be useful for the study of anti-envelope antibody responses, future passive-transfer experiments evaluating intravenous and intravaginal SHIV challenge should provide further data on the role of antibody in protection against HIV-1.