ABSTRACT
Designing covalent inhibitors is a task of increasing importance in drug discovery. Efficiently designing irreversible inhibitors, though, remains challenging. Here, we present covalentizer, a computational pipeline for creating irreversible inhibitors based on complex structures of targets with known reversible binders. For each ligand, we create a custom-made focused library of covalent analogs. We use covalent docking, to dock these tailored covalent libraries and to find those that can bind covalently to a nearby cysteine while keeping some of the main interactions of the original molecule. We found ~11,000 cysteines in close proximity to a ligand across 8,386 protein-ligand complexes in the PDB. Of these, the protocol identified 1,553 structures with covalent predictions. In prospective evaluation against a panel of kinases, five out of nine predicted covalent inhibitors showed IC50 between 155 nM - 4.2 M. Application of the protocol to an existing SARS-CoV-1 Mpro reversible inhibitor led to a new acrylamide inhibitor series with low micromolar IC50 against SARS-CoV-2 Mpro. The docking prediction was validated by 11 co-crystal structures. This is a promising lead series for COVID-19 antivirals. Together these examples hint at the vast number of covalent inhibitors accessible through our protocol.
ABSTRACT
The COVID-19 pandemic is a stark reminder that a barren global antiviral pipeline has grave humanitarian consequences. Future pandemics could be prevented by accessible, easily deployable broad-spectrum oral antivirals and open knowledge bases that derisk and accelerate novel antiviral discovery and development. Here, we report the results of the COVID Moonshot, a fully open-science structure-enabled drug discovery campaign targeting the SARS-CoV-2 main protease. We discovered a novel chemical scaffold that is differentiated from current clinical candidates in terms of toxicity, resistance, and pharmacokinetics liabilities, and developed it into noncovalent orally-bioavailable nanomolar inhibitors with clinical potential. Our approach leveraged crowdsourcing, high-throughput structural biology, machine learning, and exascale molecular simulations. In the process, we generated a detailed map of the structural plasticity of the main protease, extensive structure-activity relationships for multiple chemotypes, and a wealth of biochemical activity data. In a first for a structure-based drug discovery campaign, all compound designs (>18,000 designs), crystallographic data (>500 ligand-bound X-ray structures), assay data (>10,000 measurements), and synthesized molecules (>2,400 compounds) for this campaign were shared rapidly and openly, creating a rich open and IP-free knowledgebase for future anti-coronavirus drug discovery.