INTRODUCTION
COVID-19 became a global pandemic partially as a result of the lack of easily deployable, broad-spectrum oral antivirals, which complicated its containment. Even endemically, and with effective vaccinations, it will continue to cause acute disease, death, and long-term sequelae globally unless there are accessible treatments. COVID-19 is not an isolated event but instead is the latest example of a viral pandemic threat to human health. Therefore, antiviral discovery and development should be a key pillar of pandemic preparedness efforts.
RATIONALE
One route to accelerate antiviral drug discovery is the establishment of open knowledge bases, the development of effective technology infrastructures, and the discovery of multiple potent antivirals suitable as starting points for the development of therapeutics. In this work, we report the results of the COVID Moonshot—a fully open science, crowdsourced, and structure-enabled drug discovery campaign—against the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) main protease (Mpro). This collaboration may serve as a roadmap for the potential development of future antivirals.
RESULTS
On the basis of the results of a crystallographic fragment screen, we crowdsourced design ideas to progress from fragment to lead compounds. The crowdsourcing strategy yielded several key compounds along the optimization trajectory, including the starting compound of what became the primary lead series. Three additional chemically distinct lead series were also explored, spanning a diversity of chemotypes.
The collaborative and highly automated nature of the COVID Moonshot Consortium resulted in >18,000 compound designs, >2400 synthesized compounds, >490 ligand-bound x-ray structures, >22,000 alchemical free-energy calculations, and >10,000 biochemical measurements—all of which were made publicly available in real time. The recently approved antiviral ensitrelvir was identified in part based on crystallographic data from the COVID Moonshot Consortium.
This campaign led to the discovery of a potent [median inhibitory concentration (IC50) = 37 ± 2 nM] and differentiated (noncovalent and nonpeptidic) lead compound that also exhibited potent cellular activity, with a median effective concentration (EC50) of 64 nM in A549-ACE2-TMPRSS2 cells and 126 nM in HeLa-ACE2 cells without measurable cytotoxicity. Although the pharmacokinetics of the reported compound is not yet optimal for therapeutic development, it is a promising starting point for further antiviral discovery and development.
CONCLUSION
The success of the COVID Moonshot project in producing potent antivirals, building open knowledge bases, accelerating external discovery efforts, and functioning as a useful information-exchange hub is an example of the potential effectiveness of open science antiviral discovery programs. The open science, patent-free nature of the project enabled a large number of collaborators to provide in-kind support, including synthesis, assays, and in vitro and in vivo experiments. By making all data immediately available and ensuring that all compounds are purchasable from Enamine without the need for materials transfer agreements, we aim to accelerate research globally along parallel tracks. In the process, we generated a detailed map of the structural plasticity of Mpro, extensive structure-activity relationships for multiple chemotypes, and a wealth of biochemical activity data to spur further research into antivirals and discovery methodologies. We hope that this can serve as an alternative model for antiviral discovery and future pandemic preparedness.
Further, the project also showcases the role of machine learning, computational chemistry, and high-throughput structural biology as force multipliers in drug design. Artificial intelligence and machine learning algorithms help accelerate chemical synthesis while balancing multiple competing molecular properties. The design-make-test-analyze cycle was accelerated by these algorithms combined with planetary-scale biomolecular simulations of protein-ligand interactions and rapid structure determination.
