Comprehensively mapping the effects of mutations to influenza virus

Abstract

Influenza virus is a rapidly evolving threat to public health. Influenza evolves through point mutations, so knowing the effects of all amino-acid mutations on the ability of the virus to withstand various selective pressures reveals the evolutionary paths accessible to the virus. There are approximately 10,000 different amino-acid mutations that can be made to the average influenza gene, but accurately predicting the effects of any one mutation is difficult. However, new methods leveraging the latest technologies in mutagenesis and high-throughput DNA sequencing have made it possible to measure the effects of all possible mutations to an influenza gene. This approach involves introducing all codon mutations to an influenza gene, reconstituting mutant virus libraries carrying these mutations and the corresponding protein variants, imposing selective pressure on the mutant virus libraries, and using accurate deep sequencing methods to quantify the frequencies of all mutations before and after selection. The effect of each mutation can then be computed from the change in mutation frequency during selection, where beneficial mutations will increase in frequency and deleterious mutations decrease in frequency. Here I describe several applications of this approach to comprehensively measure the effects of mutations within two influenza genes. First, I examine the extent that mutational effects shift during the course of protein evolution by measuring mutational effects to two homologs of influenza nucleoprotein separated by over thirty years of evolution. Although there are a few protein sites with strong shifts in which amino acids are preferred, the effects of mutations at most sites are conserved across these homologs. The mutational effects measured in these two human influenza nucleoprotein homologs accurately describe the evolution of more distant influenza viruses infecting pigs, horses, and birds, demonstrating the feasibility of using measurements on one virus strain to model the evolution of more distantly related strains. Next, I describe technical improvements to the process of generating and selecting comprehensive mutant virus libraries of influenza hemagglutinin that yield more accurate and reproducible measurements of mutational effects than previously possible. Finally, I extend this approach to comprehensively map all mutations in hemagglutinin that enable the virus to escape from neutralizing antibodies. These results reveal the striking mutation-level idiosyncrasy of antibody escape: at most epitope sites only a subset of mutations confer escape to a given antibody, and similar antibodies targeting the same antigenic site elicit distinct profiles of escape mutations. Collectively, these studies enhance our understanding of influenza evolution and immune evasion and expand our ability to comprehensively map the evolutionary potential of viruses.