Sequence plasticity and antigenic escape in the highly-conserved influenza haemagglutinin stalk region

Abstract

Universal flu vaccines adopt a structural vaccinology approach to induce antibodies that target the antigenically conserved but immuno-subdominant HA2 stalk of viral haemagglutinin (HA). Broadly neutralising antibodies (bnAbs) that target this region are being developed as therapeutics. However, whether influenza virus could evolve to escape these mitigations is not clear. Deep mutational scanning revealed HA2 to be mutationally intolerant but not entirely intransient as previously thought. The focus of this study is to determine the likelihood of bnAbs inducing viral escape mutations by evaluating the fitness landscape of helix A, a highly conserved region of the HA stalk epitope. We first attempted to generate antisera against the stalk region using two different approaches. A chimeric HA influenza recombinant DNA vaccine with a HA head of different subtypes fused to a H3 stalk was used to immunise mice. In an alternative approach, mice were immunised with a pre-fusion headless HA mini-stem protein vaccine. Human bnAbs T1-3B, T2-6C, T3-5D and MEDI8852 were obtained from another lab to test the virus mutants generated later in the thesis. Structural bioinformatics was used to rationally design cDNA libraries altered in amino acids identified in helix A within A/England/195/2009 HA2 as targets of bnAbs. Co-crystal structures of different HAs bound to different bnAb modalities published on RCSB Protein Data Bank were analysed for epitope-paratope interactions. We also performed in silico residue scanning to predict changes in protein stability and binding affinity. Using reverse genetics and site saturation mutagenesis, the cDNAs were used to rescue recombinant viruses with single or double mutations in this region. Sequencing of virus libraries revealed more than 22 different variants with changes at a total of 10 residues were viable, and a selection of 10 recombinant viruses were taken forwards for deeper phenotyping to probe for fitness costs and antibody escape. Viral replication kinetics, plaque phenotype and pH stability were characterised. Several of the virus mutants with changes in HA2 had altered pH stability. Some differences in viral replication that were not apparent in MDCK cell culture were observed in primary human airway epithelial tissue culture. One HA2 genotype, V52I, showed a significant increase in viral fitness. Interestingly this mutation has been reported on sequence databases. None of the viable viruses with HA2 mutations evaded bnAb recognition. One example of a double HA2 mutation in helix A that did evade antibody was poorly expressed, and viable virus was not rescued. BnAbs are thought to inhibit virus in endosomes, but two bnAbs out of four that were tested did not retain full HA binding at low pH. Overall, this work highlights that mutations that mediate escape such as those selected by universal influenza vaccine and studied bnAbs are hampered by restrictive conditions in the local fitness landscape of H1N1pdm09 HA stalk epitope. Immune complex sensitivity under low pH has underpinnings in protein acid stability and paratope immunogenetics which are not well-understood systematically. Although bnAbs carry great potential in controlling H1N1pdm09-lineage flu outbreaks, it would be prudent and timely to re-evaluate this confidence by testing their binding retention at low pH and include human-adapted HAs from most recent seasonal strains when testing such immune complexes.