Quantifying how the mutational tolerance of HIV's envelope protein shapes its evolution

Abstract

HIV's most rapidly evolving proteins is its envelope protein (Env). This rapid evolution is driven by continuous selection to evade immunity within HIV-infected hosts. However, as Env evolves, it is also under functional constraint to perform essential functions in the viral lifecycle, including receptor binding and membrane fusion. Since both of the above forces strongly shape Env's evolution, their effects have been difficult to disentangle from one another. As a result, our understanding of these forces is far from complete. A central goal of my graduate research has been to experimentally measure the functional constraint on Env in the lab in the absence of external immune selection. There are ~10,000 single amino-acid mutations to a protein of Env's length (=19 * ~850). Using a high-throughput technique called deep mutational scanning, I measured the effects of each of these mutations to Env in context of viral replication in cell culture. The results provide an in-depth profile of Env's ability to tolerate each of the 20 amino acids at each site in the protein. Using these data, I examined Env's mutational tolerance variable loops, which rapidly evolve to evade antibodies. It is possible that these loops have a high tolerance for mutations, and that that is one reason they so readily evolve. However, I did not find statistical support that these loops are more tolerant of mutations than other parts of the protein, suggesting that their variability in nature may mainly be due to high levels of diversifying pressure from antibodies. I also examined epitopes of broadly neutralizing antibodies targeting the CD4 binding site. These epitopes that are highly conserved in nature and are targets in vaccine design. A common assumption is that this conservation is due to high functional constraint at these sites. Indeed, I found that they were less tolerant of mutations than other parts of Env, providing rigorous support for a long-standing hypothesis, and suggesting that these epitopes may have a diminished evolutionary capacity to evade antibodies relative to other sites in the protein, which would make them more vulnerable to immune targeting. Another central goal of my thesis has been to compare Env's mutational tolerance among divergent strains. The same mutation (e.g., A12N) can have different effects in two related proteins due to epistasis (e.g., A12N may only be tolerated in one homolog, but not the other). However, the extent that mutational effects to Env differ between divergent HIV strains is largely unknown. To address this knowledge gap, I repeated the deep mutational-scanning experiment of two Env homologs that have 85% amino-acid identity. The results allowed me to compare each homolog's ability to tolerate each of the 20 amino acids at 616 homologous sites. I found that at a small fraction of sites, the amino acids tolerated in one homolog were largely distinct from the amino acids tolerated in the other homolog. However, only a few sites showed such extreme differences; most sites had changes in mutational tolerance that were only small-to-intermediate in effect size. Thus, these results indicate that Env's mutational tolerance is still substantially conserved between homologs. Overall, my graduate research has increased our knowledge of how Env's underlying mutational tolerance shapes the evolution of antibody epitopes, providing experimental support for the assumption that conserved epitopes targeted in vaccine design are indeed less tolerant of mutations than the rest of the protein, and may thus be less likely to evade an immune response. This work also provides a comprehensive measure of differences in mutational effects across Env, finding that mutational effects are largely conserved between divergent homologs. More broadly, this work was also the first time that deep mutational scanning had been used to comprehensively measure mutational effects to an HIV protein in context of viral replication. In the future, this technique could be adapted to study any phenotype that is selectable in the lab (e.g., antibody escape).