Integrating measurements of immune escape and in vitro replication with computational models to understand and predict the antigenic evolution of seasonal A/H3N2 influenza viruses

PROJECT SUMMARY Seasonal influenza virus vaccines have to be reformulated most years primarily due to immune escape caused by mutations in the surface hemagglutinin (HA) protein. The genetic variation in HA only occasionally causes change in antigenic phenotype and consequent immune escape. For extended periods of time strains with genetic differences remain in a single antigenic cluster. In 2013 we (Koel et al.)1 found that the amino acid substitutions responsible for antigenic cluster transitions in human A/H3N2 viruses occurred at only seven key positions on the periphery of the HA receptor binding site (RBS), and that seven out of ten A/H3N2 cluster transitions were caused by just single amino acid substitutions. Furthermore, major antigenic change in other (sub)types of human influenza, as well as influenza viruses in other species, is also primarily due to single amino acid substitutions at the same seven key HA sites, and nearby, on the periphery of the HA RBS. This discovery raises an immediate, and not previously obvious question: If just one amino acid change is typically required to escape immunity, why is the antigenic evolution of influenza viruses so slow? Human seasonal influenza A/H3N2 viruses remain in an antigenic cluster for an average of 3.1 years, and occasionally as long as eight years. This is especially perplexing given that, as an RNA virus, influenza viruses have a fast rate of molecular evolution. A possible explanation for the delay in fixation of cluster transition substitutions is that antigenic change incurs a fitness cost. The proximity of escape mutations to the RBS offers a mechanism for this cost: the virus needs to change close to the RBS as antibodies targeting the RBS need to be escaped, but change in this area also affects receptor-binding function. Substitutions which advance a strain antigenically may only be competitive when sufficient population immunity has built to contemporary circulating variants, such that the gain in fitness from escaping immunity (the "extrinsic" fitness gain) outweighs the potential fitness loss associated with the concomitant distortion of the receptor binding site (the "intrinsic" fitness loss). We refer to this as the "fitness exchange" hypothesis. To understand the evolutionary dynamics of influenza requires understanding what paces antigenic change. In this proposal we set out to test the fitness exchange hypothesis, to gain understanding of the variation and impact of viral intrinsic fitness, and to determine the relative importance of intrinsic fitness and stochastic effects in novel mutations becoming fixed in viral populations, and to integrate empirical measurements of these effects into a probabilistic framework for predicting the antigenic evolution of seasonal influenza viruses.