Discovery of novel spike/ACE2 inhibitory macrocycles using in silico reinforcement learning

Abstract

Introduction: The COVID-19 pandemic has cast a heavy toll in human lives and global economics. COVID-19 is caused by the SARS-CoV-2 virus, which infects cells via its spike protein binding human ACE2. Methods: To discover potential inhibitory peptidomimetic macrocycles for the spike/ACE2 complex we deployed Artificial Intelligence guided virtual screening with three distinct strategies: 1) Allosteric spike inhibitors 2) Competitive ACE2 inhibitors and 3) Competitive spike inhibitors. Screening was performed by docking macrocycles to the relevant sites, clustering and synthesizing cluster representatives. Synthesized molecules were screened for inhibition using AlphaLISA and RSV particles. Results: All three strategies yielded inhibitory peptides, but only the competitive spike inhibitors showed “hit” level activity. Discussion: These results suggest that direct inhibition of the spike RBD domain is the most attractive strategy for peptidomimetic, “head-to-tail” macrocycle drug development against the ongoing pandemic.