Computational Models of Adaptive Immunity to Influenza

PROJECT SUMMARY The adaptive immune response underpins protection against influenza virus infection, and strengthening the response is the goal of vaccination. However, we lack a good understanding of what factors determine a strong adaptive immune response to vaccination, particularly in individuals with prior exposure to influenza virus. The goals of the proposed study are to describe conditions under which strong, durable, and broadly protective adaptive immune responses to influenza arise and to use mechanistic and statistical models to understand precisely how these responses change over time, especially in response to vaccination. We will begin by identifying adults with “good” (strong, durable, and broad) and less good (transient, narrow, and/or weak) B and T cell responses in a randomized, placebo-controlled trial of repeat influenza vaccination. Participants will have received the recombinant influenza vaccine Flublok one to four years in a row, while a control group will not be vaccinated until year five. In the first phase of the proposed study, we will investigate participants’ antibody neutralization landscapes and B and T cell repertoires and ask how infections, vaccinations, and individual- specific factors shape the strength, durability, and breadth of the response, focusing on neutralizing antibodies and T cell phenotypes that we and others have identified as correlates of protection. Our statistical and mechanistic models will assess how exposures shape the evolution of B and T cell clones and the relationship between B and T cell repertoires and the specificity of the serum neutralizing antibodies. We will also investigate how B and T cell phenotypes, including their transcriptomes, relate to repertoire dynamics and the qualities of serum antibody and T cell responses. We will re-assess correlates of protection from infection. These foundational models will test many mechanistic hypotheses and involve exploration, including the use of machine learning for prediction. Consequently, in the next phase of the study, we will use a distinct observational cohort that includes vaccinated and unvaccinated participants from a wider age range to cross-validate our leading models and hypotheses, including correlates of protection. This approach should increase the robustness of the conclusions and deliver a set of statistical and mechanistic approaches that describe conditions leading to different responses to vaccination, predict B and T cell repertoire changes in response to influenza exposures, estimate multidimensional correlates of protection, and infer the dynamics of the adaptive immune response to influenza in individuals over time to understand in detail how variable vaccine responses arise. Our data and software will be findable, accessible, interoperable, and reusable (FAIR) to maximize their utility for other researchers investigating the evolution of adaptive immunity to influenza and other pathogens.