Optimization of a broad and potent decoy receptor for SARS-associated viruses

ABSTRACT The spike protein (S) of SARS coronavirus 2, the pathogen responsible for COVID-19, binds angiotensin- converting enzyme 2 (ACE2) as an entry receptor, triggering conformational changes in S that drive fusion of the viral envelope and host cell membrane. Infection is inhibited by neutralizing antibodies that block the ACE2-binding site on S, yet escape mutations in S rapidly emerge towards monoclonal antibodies in tissue culture. Furthermore, monoclonal antibodies are generally strain specific, and many do not recognize with high affinity both human SARS-CoV-1 and SARS-CoV-2, yet alone exotic bat betacoronaviruses that are a reservoir for future outbreaks. As an alternative for biologic drug development, we have used deep mutagenesis to guide the engineering of an exceptionally broad soluble decoy receptor that binds with tight picomolar/low- nanomolar affinity to S from all bat and human SARS-associated coronaviruses tested. The engineered decoy potently neutralizes authentic SARS-CoV-1 and SARS-CoV-2 with an efficacy that rivals monoclonals under commercial development, and has desirable properties for manufacture at scale. The engineered decoy also catalytically converts angiotensin II to vasodilatory peptide products that might directly address symptoms of COVID-19, providing us with a unique potential therapeutic that has dual mechanisms of action. Our proposal investigates whether the SARS-CoV-2 spike can mutate to escape neutralization by the engineered decoy receptor, and addresses final optimization of the engineered protein as an IgG1-Fc fusion before advancement to an IND-enabling program. For SARS-CoV-2 to become resistant to the engineered decoy, mutations in S must decrease affinity to the decoy while maintaining binding to human ACE2 receptors. To identify such S variants, we have used saturation mutagenesis of the receptor-binding domain coupled with a selection for tight binding to wild type ACE2 in the presence of competing soluble decoy. Following deep sequencing, a small number of mutations were found to be enriched, but it is unclear whether any of these mutations do indeed preferentially bind wild type ACE2 and if so, to what degree they have achieved specificity. Based on this preliminary data, (Aim 1) we will validate whether mutations in S can be found that discriminate between human ACE2 and the engineered decoy, and characterize the variants for their affinities and expression levels. Thus far, we can conclude that possible resistance mutations appear to be very rare and generally require more than one nucleotide change within a codon, but further quantitative characterization is needed. Simultaneously, (Aim 2) we will rapidly optimize fusions of the engineered decoy with the Fc region of IgG1 for enhanced serum stability. Our current IgG1-fusion construct (which was based on rational, structure-guided design) is highly expressed, stable and binds SARS-CoV-2 S with picomolar affinity. We will finalize optimization of the protein by scanning suitable fusion sites between the engineered decoy and IgG1, assessing protein quality by activity, stability and expression.