Predicting a moving target’s next move
HIV owes much of its devastating success to its envelope protein Env, which interacts with receptors on a host cell’s surface to allow viral entry. Because it is the only HIV protein directly exposed to the environment, Env is a major target of the anti-HIV immune response, but it evolves so rapidly that Env-targeting antibodies usually become ineffective within weeks. Much effort has thus been directed towards understanding the process of Env evolution, with the goal of developing broadly effective antibodies that are more difficult for HIV to escape.
As part of this effort, multiple projects in the Bloom Laboratory (Basic Sciences Division) have explored the evolutionary space available to Env. For example, graduate student Hugh Haddox, now a post-doc at UW, quantified the inherent mutational tolerance of each site in Env by measuring the effect of every possible amino acid substitution on viral replication in cell culture. Using a related approach, Adam Dingens, a graduate student in the Bloom and Overbaugh labs, studied how mutations in Env affect its ability to escape detection by a specific antibody.
While experiments like these provide valuable insight into the evolution of a single Env protein, a major question in the field is whether the results from a given study are relevant to Env variants isolated from different strains HIV. To address this question, Hugh and Adam teamed up to compare the mutational tolerances at each site for two evolutionarily distinct Env proteins. Their work was recently published in eLife. “This project examined how the evolutionary space available to HIV changes as the virus evolves. It was a very impressive collaboration” says Dr. Jesse Bloom.
Adam and Hugh chose to compare Envs that differ at 115 of 836 sites, both of which were isolated by the Overbaugh lab from infants infected through mother-to-child transmission. For each Env variant, they performed a technique called deep mutational scanning (DMS) by generating a library of viruses containing random single amino acid substitutions in the env gene and using high-throughput sequencing to determine which variants are able to infect human cells and replicate (see Figure). If a particular mutation reduces Env functionality, it will be statistically underrepresented in the population of viruses isolated after infection compared to its starting frequency in the initial library.
Most substitutions had similar effects on the function of both Env proteins, allowing the authors to determine which regions are highly tolerant of mutations and which are functionally constrained. In addition, both Env variants exhibited similar amino acid preferences at most positions. “Thus, although Env homologs can be highly divergent at the sequence level, this divergence does not necessarily imply divergence in evolutionary potential” explains Hugh. About 30 sites did show a shift in preference between the different Envs; these generally preferred a specific amino acid in one Env and tolerated many amino acids in the other.
Some mutations were tolerated differently by the two Env homologs, indicating that the mutational space available to a given virus can be context-dependent. The authors expected sites with contrasting tolerances to be located in diverged parts of the protein, but this was not the case. “Instead, it appears that changes in mutational effects at a given site are often due to long-range interactions between that site and other sites that differ in sequence. Interestingly, many of the sites with altered mutational effects are located in regions of Env that are highly conformationally dynamic. Thus, it is possible that the dynamic nature of these regions could help propagate such long-range interactions,” Hugh hypothesized.
Importantly, the DMS experiment was carried out under laboratory conditions, in the absence of an immune response to the virus. This means that the results only identify mutations that affect Env function, not those that allow immune escape. Hugh and Adam used this to their advantage by combining the amino acid preference data from both Envs into a generalized model of the functional constraints on Env’s evolution. Using this model, they asked which sites appear to be evolving faster or slower in nature compared to what is expected given the functional constraints measured in the lab. This approach successfully pinpointed sites already known to be under strong immune selection and identified additional sites that may play a role in immune evasion by HIV.
The Bloom lab’s finding that diverged Env proteins exhibit similar amino acid preferences at most sites suggests that mapping of mutational tolerances for a small number of variants may be sufficient to approximate the evolutionary space available to all Envs, with the caveat that shifts in amino acid preference over time can generate context dependence. Thus, amino acid preference maps generated from DMS data can inform future efforts to analyze sequence-structure-function relationships during evolution. Further exploration of how long-range intra-protein interactions and immune pressures constrain viral evolution may help researchers anticipate HIV’s next move.
Haddox HK, Dingens AS, Hilton SK, Overbaugh J and Bloom JD. 2018. Mapping mutational effects along the evolutionary landscape of HIV envelope. eLife. 28:460-473
This work was supported by the National Institutes of Health, National Science Foundation, Howard Hughes Medical Institute, Simons Foundation and Collaboration for AIDS Vaccine Discovery.
For the Media