PROJECT SUMMARY The rapid spread of the highly-pathogenic, novel SARS-coronavirus 2 (SARS-CoV-2) has caused a global health emergency. Thus, there is a desperate need for effective antiviral therapeutics to counteract this virus. The SARS-CoV-2 virus enters cells using the ACE2 receptor1 which binds the viral spike protein2. In its soluble form, ACE2 (sACE2) has the potential to be used as a stable and non-immunogenic competitive inhibitor to SARS-CoV-2 and is presently being explored in clinical trials3. Due to the potential negative side effects of anti-spike mAbs18, and the fact that ACE2 exhibits other biological roles4–6 including integrin signaling regulation7,8, spike-specific receptor mimics would yield novel therapeutics for SARS-CoV-2 and potentially other highly infectious diseases. This proposal seeks to use machine learning and directed evolution to develop high affinity, yet endogenously-inactive mimics of sACE2 in order to create rapidly implementable therapeutics to combat SARS-CoV-2 and potential corona-like viruses. This approach would allow for the generation of scalable and translatable biologics, and provide a platform to rapidly course-correct for potential mutations that may arise in the future. Utilizing deep-learning with UniRep49, will design and generate sACE2 variants that tightly bind the SARS-CoV2-2 spike protein but do not cross-interact with endogenous targets such as integrins [Aim 1]. Simultaneously, we will perform directed evolution to optimize spike-binding and select against variants that bind endogenous proteins [Aim 2]. Finally, we will identify lead candidates and evaluate the tolerance and immunogenicity of engineered sACE2 variants in mice [Aim 3]. Collectively, this proposal will develop highly-specific ACE2 receptor mimics in order to create novel antivirals with minimal immunogenicity in time to save lives and prevent future outbreaks. 10