Turning high-throughput structural biology into predictive inhibitor design
Kadi L. Saar, William McCorkindale, D. Fearon, Melissa L. Boby, Haim Barr, Amir Ben‐Shmuel, Nir London, F. von Delft, John D. Chodera, Alpha A. Lee, Matthew C. Robinson, Nir London, Efrat Resnick, Daniel Zaidmann, Paul Gehrtz, Rambabu Reddi, Ronen Gabizon, Haim Barr, Shirly Duberstein, Hadeer Zidane, Khriesto A. Shurrush, Galit Cohen, Leonardo J. Solmesky, Alpha A. Lee, Andrew J. Jajack, Milan Cvitkovic, Jin Pan, Ruby Pai, Emily Grace Ripka, Luong Viet Nguyen, Mikhail Shafeev, Tatiana Matviiuk, Oleg Michurin, Eugene Chernyshenko, Vitaliy A. Bilenko, Serhii O. Kinakh, Ivan G. Logvinenko, Kostiantyn P. Melnykov, Victor D. Huliak, Igor S. Tsurupa, Marian V. Gorichko, Aarif L. Shaikh, Jakir Pinjari, Vishwanath Swamy, Maneesh Pingle, Sarma BVNBS, A. Aimon, F. von Delft, D. Fearon, Louise Dunnett, A. Douangamath, Alex Dias, A.J. Powell, José Brandão Neto, R. Skyner, Warren Thompson, T.J. Gorrie-Stone, Martin Walsh, David Owen, Petra Lukacik, Claire Strain‐Damerell, Halina Mikolajek, Sam Horrell, L. Koekemoer, T. Krojer, Mike Fairhead, Elizabeth MacLean, Andrew Thompson, Conor Wild, Mihaela Smilova, Nathan D. Wright, Annette von Delft, C. Gileadi, V.L. Rangel, Chris Schofield, E. Salah, Tika R. Malla, Anthony Tumber, Tobias John, Ioannis Vakonakis, A.L. Kantsadi, Nicole Zitzmann, Juliane Brun, J. L. Kiappes, Michelle L. Hill, Karolina D. Witt, Dominic S. Alonzi, Laetitia L. Makower, Finny S. Varghese, Gijs J. Overheul, Pascal Miesen, Ronald P. van Rij, Jitske Jansen, Bart Smeets, Susana Tomésio, Charlie Weatherall, M. Vaschetto, Hannah Bruce Macdonald, John D. Chodera, Dominic A. Rufa
Abstract
A common challenge in drug design pertains to finding chemical modifications to a ligand that increases its affinity to the target protein. An underutilized advance is the increase in structural biology throughput, which has progressed from an artisanal endeavor to a monthly throughput of hundreds of different ligands against a protein in modern synchrotrons. However, the missing piece is a framework that turns high-throughput crystallography data into predictive models for ligand design. Here, we designed a simple machine learning approach that predicts protein–ligand affinity from experimental structures of diverse ligands against a single protein paired with biochemical measurements. Our key insight is using physics-based energy descriptors to represent protein–ligand complexes and a learning-to-rank approach that infers the relevant differences between binding modes. We ran a high-throughput crystallography campaign against the SARS-CoV-2 main protease (M Pro ), obtaining parallel measurements of over 200 protein–ligand complexes and their binding activities. This allows us to design one-step library syntheses which improved the potency of two distinct micromolar hits by over 10-fold, arriving at a noncovalent and nonpeptidomimetic inhibitor with 120 nM antiviral efficacy. Crucially, our approach successfully extends ligands to unexplored regions of the binding pocket, executing large and fruitful moves in chemical space with simple chemistry.