An integrated in silico and in vitro genotype to phenotype pipeline to predict and characterise new RSV F site zero escape mutants

The new respiratory syncytial virus (RSV) interventions target the Fusion (F) protein and may therefore impose a selective pressure upon the F gene. Identifying monoclonal antibody-resistant mutants (MARMs) of concern is a priority to ensure continued antibody effectiveness. Here we evaluated genomes of RSV isolates sampled in the UK prior to vaccine or nirsevimab implementation. We observed a low frequency of the K68E mutation in site Ø which we confirmed had increased resistance to a nirsevimab-like monoclonal antibody. To predict other MARMs of concern we used bioinformatic tools to model the interface between nirsevimab and RSV-F. There was a very strong correlation between antibody-antigen Kd and in vitro neutralisation data. Performing in silico deep mutational scanning of each viral contact residue identified new mutations of concern in RSV-A (N63T, I64V, D200E, K201N) which are already circulating; increased resistance to a site Ø targeting antibody was confirmed using reverse genetics derived RSV. To explore the universality of the pipeline, we also tested correlations between predicted and actual binding for licensed antibodies targeting SARS-CoV-2. We therefore demonstrate the ability to determine previously unidentified escape mutations in silico and flag them for further surveillance.