Quantifying the genetic basis of antigenic variation among human influenza A viruses

Influenza viruses are a major cause of morbidity and mortality worldwide, with seasonal epidemics of influenza resulting in around three to five million cases of severe illness globally each year. The evolution of influenza A viruses is characterised by rapid antigenic drift, which allows mutant viruses to evade host immunity acquired to previously circulating viruses. Antigenic variation is observed across a wide range of infectious organisms and can circumvent long-lasting immunity in hosts leading to repeated infection or non-clearance. Influenza A viruses can often be effectively combatted by the immune system and vaccines also exist to protect at-risk individuals, limiting the burden of disease. However, the effectiveness of the vaccine depends on constituents being antigenically similar to circulating viruses. Antigenic drift of influenza viruses therefore requires a global surveillance system responsible for the antigenic characterisation of circulating viruses. The identification of emerging antigenic variants is critical to the vaccine virus selection process and in addition experts must anticipate which viruses are likely to predominate in forthcoming epidemic seasons. Mutations to B-cell epitopes on the surface of haemagglutinin (HA) that facilitate escape from neutralising antibodies play a key role in influenza antigenic drift. Consequently the haemagglutination inhibition (HI) assay, which measures HA cross-reactivity, is commonly used to approximate antigenic phenotype. In this thesis, I investigate the genetic basis of antigenic variation among human influenza A viruses through analysis of HI data collected in recent decades and associated HA gene sequence data. In Chapter 2, I use phylogenetic methods and antigenic cartography to characterise the genetic and antigenic variation among the viruses studied and evaluate the usefulness of these methods for epitope identification. In Chapter 3, I extend a model developed to investigate antigenic differences among foot-and-mouth disease (FMD) viruses to former seasonal A(H1N1) viruses. By attributing variation in HI titre to amino acid differences between viruses, while accounting for phylogenetic relationships, I identify substitutions that have driven the antigenic evolution of the virus. Reverse genetics was then used to validate model predictions experimentally. In Chapter 4, I further extend the model and investigate the genetic drivers of antigenic drift among A(H3N2) viruses, comparing model results with published HI data generated using mutant recombinant viruses. In Chapter 5, I explore the power of the identified genetic determinants for predicting antigenic relationships among A(H1N1) and A(H3N2) viruses. Specifically I show that sequence-based models can be used to estimate the antigenicity of emerging viruses directly from their sequence and that by including substitutions of smaller antigenic impact, in addition to the high-impact substitutions that are often focused on, predictions were improved. I also demonstrate the versatility of these methods by extending this sequence-based approach to predict antigenic relationships among viruses of three serotypes of FMD virus. Determining phenotype from genotype is a fundamental challenge for virus research. It is of particular interest in the case of the antigenic evolution of influenza viruses, given the need to continually track changes in the virus population, anticipate which viruses will predominate in future seasons, and select vaccine viruses. Collectively, the results I present demonstrate an enhanced quantitative understanding of the molecular genetic basis of the adaptive phenotype of influenza viruses. The ability to quantify the phenotypic impact of specific amino acid substitutions should help to refine methods that predict, from genotype, the fitness and evolutionary success of influenza viruses from one season to the next, strengthening the theoretical foundations for vaccine virus selection. The techniques presented also have great potential to be extended to other antigenically variable pathogens and to elucidate the genetic basis of their antigenic variation.