Funded Grants

Measuring the predictability of evolution with adaptive immunity

A basic question in biology is, can we predict evolution? Although evolutionary theory can retrospectively explain many cases of adaptation and genetic drift, predicting the trajectories of specific populations remains difficult. Is prediction merely limited by incomplete knowledge of selective pressures and functional constraints, or is evolution so strongly affected by chance that prediction is effectively impossible? The answers to these questions determine the inferences we draw from current populations, the type of data we collect, and the kinds of outcomes we plan for.

The problem of predicting evolution underlies a central mystery of adaptive immunity. When encountering a pathogen, B cells begin mutating rapidly and undergo intense selection for improved pathogen binding. B cells with good binding later secrete high-affinity antibodies, which help clear infection. B cells can bind any number of sites, such that some cells bind sites that disrupt essential activities, while others do less damage. Which sites B cells evolve to recognize thus affects protection against current and future pathogens. B cells that neutralize pathogens are more protective, and B cells that target evolutionarily conserved sites may confer broader immunity. Actual B cell repertoires vary among individuals in ways that cannot be explained simply by individuals’ unique histories of infection. Some sites are recognized by a few people and other sites by nearly everyone. To immunologists, this is the problem of immunodominance.

The aim of the proposed work is to investigate the fitness landscape on which B cells evolve to a common pathogen, which will reveal the role of chance in evolution and show how immune repertoires are shaped by infection. B cells provide a unique context for studying evolution on adaptive landscapes: they evolve very quickly by point mutations, their fitnesses can be easily defined and measured as binding affinities to proteins, they are abundant, ancestral forms can be reconstructed, and their population structure naturally provides several forms of experimental replication. An investigation of B cell evolution could thus yield the most comprehensive survey to date of an adaptive landscape, allowing the roles of chance, accessibility, and convergence in evolution to be evaluated statistically. Moreover, this adaptive landscape is of great importance to pathogen evolution and public health. Adaptive immunity shapes the diversity of many pathogens, including influenza and HIV. Tremendous effort has been dedicated to studying how broadly neutralizing antibodies to such pathogens may be induced. Such antibodies naturally arise in a subset of the infected population, and vaccines that attempt to boost their numbers are in development. However, it is not known how extensively antibody repertoires can be shaped or if there are consequences of homogenizing herd immunity for viral evolution. We will pursue these related problems via theory, computation, and experiment.