Neutralizing antibody responses in HIV dual infection: lessons for vaccine design

Sheward, Daniel James

Neutralizing antibody responses in HIV dual infection: lessons for vaccine design

Doctoral Thesis

2022

Abstract

The development of a safe, effective prophylactic HIV vaccine remains a major global health priority. Stabilized, soluble trimers that mimic the native functional HIV trimer have been developed that elicit strain-specific neutralizing HIV antibodies in animal models, and are currently being evaluated in several human clinical trials. Identifying whether multiple immunogens could be administered to facilitate the broadening of responses represents a pivotal challenge. In this thesis, we characterized the antibody response in individuals infected with multiple HIV strains to inform the development of polyvalent and sequential HIV vaccine regimens. We found that conventional approaches to detect HIV co- and superinfection are confounded by recombination. Therefore, we developed an automated, Bayesian approach to detect superinfection explicitly accounting for recombination. Using simulated and real sequence data, we demonstrated that this approach is sensitive, highly specific, and robust to recombination. Furthermore, analyzing previously published sequence datasets, we identified cases of superinfection that previously went undetected, indicating that superinfection occurs more frequently than previously estimated. We characterized the development of antibodies in five superinfected individuals identified in the CAPRISA 002 acute infection cohort. Specifically, we evaluated whether superinfection re-engaged cross-reactive memory B cells, promoting the development of cross-neutralizing antibodies. By comparing the breadth of the neutralizing antibody response in superinfected individuals to those that typically develop in singly infected individuals, we showed that HIV superinfection was not sufficient to broaden responses. By characterizing the kinetics and specificity of autologous neutralizing antibody responses, we show that responses to the superinfecting viruses failed to efficiently recruit neutralizing memory B cells. Instead, the secondary infection elicited strain-specific, de novo responses. This occurred even though the superinfecting viruses were relatively closely related (from the same subtype). To determine whether the co-exposure to diverse Env antigens favours the development of cross-neutralizing antibodies better than sequential exposure, we characterized the development of neutralizing antibodies in HIV co-infected individuals where several divergent viruses were transmitted prior to seroconversion. We identified three cases of co-infection that encompassed immunological exposure to: (i) two diverse, unlinked Envs, (ii) two related Envs with diversity uniformly distributed over the trimer, and (iii) two diverse but recombined Envs such that clusters of high homology were preserved in the presence of high diversity elsewhere. We found that, like superinfection, co-infection was not sufficient to broaden neutralizing antibody responses. Co-exposure to two HIV Env antigens did not necessarily produce additive or cross-neutralizing antibody responses, and in some cases was subject to immunological interference. This was most evident in the case of co-infection with two related Envs where diversity was uniformly distributed across the Env trimer; in this case neutralizing antibody responses to one variant arose to the near exclusion of responses to the other. However, in the case of co-exposure to diverse Envs but where the trimer apex was conserved in both variants through recombination, potent neutralization of both variants was evident. This was the co-infected participant who developed the broadest neutralizing antibody response, and we show that cross-neutralization was mediated, in part, by trimer apextargeting neutralizing antibodies. In conclusion, we find that HIV superinfection fails to efficiently recruit neutralizing memory B cells and, at best, results in additive nAb responses rather than a synergistic effect leading to cross-neutralization; a distinction that is highly relevant for vaccine design. While sequential immunizations with heterologous Env immunogens may be able to improve the potency of elicited responses, alone, they are unlikely to promote the development of bnAbs. Our observations from cases of co-infection suggests that cocktails of divergent stabilized Env trimers are unlikely to drive the development of cross-neutralizing antibodies, and may be subject to interference. However, the rational design of more similar immunogen cocktails where conserved epitopes are preserved across immunogens may be able to facilitate neutralizing antibodies to these targets, as seen in one individual. Thus, the use of related, stabilized Env trimers with diversity introduced in key regions together with strategies to reduce the immunogenicity of immunodominant, strain-specific epitopes may represent one path to a cross-neutralizing antibody response to multiple Envs within a cocktail.