Protein sequence design by conformational landscape optimization
Christoffer Norn, Basile I. M. Wicky, David Juergens, Sirui Liu, David E. Kim, Doug Tischer, Brian Koepnick, Ivan Anishchenko, Foldit Players, David Baker, Sergey Ovchinnikov, Alan Coral, Alex J. Bubar, Alexander Boykov, Alexander Uriel Valle Pérez, Alison MacMillan, Allen Lubow, Andrea Mussini, Andrew Cai, Andrew John Ardill, Aniruddha Seal, Artak Kalantarian, Barbara Failer, Belinda Lackersteen, Benjamin Chagot, Beverly R. Haight, Bora Tastan, Boris Uitham, Brandon G. Roy, Breno Renan de Melo Cruz, Brian Echols, Brian Edward Lorenz, Bruce G. Blair, Bruno Kestemont, Charles Eastlake, Callen Joseph Bragdon, Carl Vardeman, Carlo Salerno, Casey Comisky, Catherine Louise Hayman, Catherine R. Landers, Cathy Zimov, Charles D. Coleman, Charles Robert Painter, Christopher Ince, Conor Lynagh, Dmitrii Malaniia, Douglas Craig Wheeler, Douglas Robertson, Vera Simon, Emanuele Chisari, E. Kai, Farah Rezae, Ferenc Lengyel, Flavian Tabotta, Franco Padelletti, Frisno Boström, G. Gross, George Victor McIlvaine, Gil Beecher, Gregory Hansen, Guido de Jong, Harald Feldmann, Jami Lynne Borman, Jamie Quinn, Jane Norrgard, Jason Truong, Jasper A. Diderich, Jeffrey M. Canfield, Jeffrey Photakis, Jesse Slone, Joanna Madzio, Joanne Mitchell, John Charles Stomieroski, John H. Mitch, Johnathan Robert Altenbeck, Jonas Schinkler, Jonathan Barak Weinberg, Joshua David Burbach, João C. Sequeira, Juan F. Bada Juarez, Jón Pétur Gunnarsson, Kathleen Diane Harper, Keehyoung Joo, Keith Clayton, Kenneth E. DeFord, Kevin F. Scully, Kevin M. Gildea, Kirk J. Abbey, K. L. Kohli, Kyle Stenner, Kálmán Takács, LaVerne Poussaint, Larry C. Manalo, Larry C. Withers, Lilium Carlson, Linda Wei, Luke Ryan Fisher, L. A. Carpenter, Ma Ji-hwan
Abstract
The protein design problem is to identify an amino acid sequence that folds to a desired structure. Given Anfinsen's thermodynamic hypothesis of folding, this can be recast as finding an amino acid sequence for which the desired structure is the lowest energy state. As this calculation involves not only all possible amino acid sequences but also, all possible structures, most current approaches focus instead on the more tractable problem of finding the lowest-energy amino acid sequence for the desired structure, often checking by protein structure prediction in a second step that the desired structure is indeed the lowest-energy conformation for the designed sequence, and typically discarding a large fraction of designed sequences for which this is not the case. Here, we show that by backpropagating gradients through the transform-restrained Rosetta (trRosetta) structure prediction network from the desired structure to the input amino acid sequence, we can directly optimize over all possible amino acid sequences and all possible structures in a single calculation. We find that trRosetta calculations, which consider the full conformational landscape, can be more effective than Rosetta single-point energy estimations in predicting folding and stability of de novo designed proteins. We compare sequence design by conformational landscape optimization with the standard energy-based sequence design methodology in Rosetta and show that the former can result in energy landscapes with fewer alternative energy minima. We show further that more funneled energy landscapes can be designed by combining the strengths of the two approaches: the low-resolution trRosetta model serves to disfavor alternative states, and the high-resolution Rosetta model serves to create a deep energy minimum at the design target structure.