Seasonal H3N2 influenza evolves rapidly, leading to extremely poor vaccine efficacy. Substitutions occurred in embryonated eggs during vaccine production (i.e., egg passage adaptation) can contribute to the poor vaccine efficacy (VE), but the evolutionary mechanism remains elusive. Using a probabilistic approach known as the mutational mapping and an unprecedented number of hemagglutinin sequences (n > 100,000), we found that the egg passage adaptation is driven by temporally-fluctuating convergent changes across different codons. Strikingly, the fitness landscape of passage adaptation is dominated by pervasive epistasis between two leading residues (186 and 194) and multiple other positions. Convergent evolutionary paths driven by strong epistasis explain most of the variation in VE, which has resulted in extremely poor vaccines for the past decade. Leveraging the unique fitness landscape, we developed a novel machine learning model that can predict egg passage substitutions for any candidate vaccine strain before the passage experiment, providing a unique opportunity for selecting optimal vaccine viruses. Our study demonstrated that that evolutionary trajectories can be harnessed for predicting Influenza adaptation and subsequently improving influenza vaccines.
Harnessing epistatic interactions in the fitness landscape for predicting adaptive evolution in the Influenza H3N2 virus