Outsmarting influenza evolution: predicting antigenic escape for better vaccines

About the Project

Influenza A viruses are constantly evolving. Changes in their major surface protein, haemagglutinin (HA), can allow them to escape antibodies generated by previous infection or vaccination, which is one reason influenza vaccines need regular updating. However, HA cannot change freely because it must retain essential functions, including binding to host cells. This raises an important question: can influenza evolution be predicted?

Our recent work on H3 and H7 influenza viruses shows that antibody responses focus mainly on the head of HA, and that changes at a small number of positions near the receptor-binding site can determine whether antibodies recognise and neutralise a virus. H3 and H7 viruses circulate in birds and several mammalian species, including humans, allowing us to test whether the same evolutionary constraints apply across hosts. This project will identify antibody-targeted regions of HA that are difficult for the virus to change, predict which mutations permit immune escape while retaining HA function, and test whether these predictions can inform broader vaccine design.

Specifically, the student will combine machine learning, viral sequence data, protein structures and antibody measurements to predict important sites and viable variants. Selected variants will be synthesised and tested using archived sera from humans, poultry and wild birds, followed by virus neutralisation assays. These results will feed back into the predictive models. Finally, selected variants, alone and in rational combinations, will be tested in mice to determine whether they generate antibodies that recognise a wider range of H3 and H7 viruses.

The long-term goal is to move from reacting to influenza evolution to anticipating its likely routes, thus providing a new strategy for vaccine design.

[1] medRxiv. doi:10.1101/2025.09.08.253353