UCSF

Deep learning has greatly advanced design of highly stable static protein structures, but the controlled conformational dynamics that are hallmarks of natural switch-like signaling proteins have remained inaccessible to de novo design. In this dissertation, I review the fundamental principles and current advances in designing said conformational motions (Chapter 1) and then describe a general deep learning-guided approach for the de novo design of dynamic changes between intra-domain geometries of proteins, similar to switch mechanisms prevalent in nature, with atom-level precision (Chapter 2). In our study, we solved 4 structures validating the designed conformations, showed microsecond transitions between them, and demonstrated that the conformational landscape can be modulated by orthosteric ligands and allosteric mutations. Physics-based simulations were in remarkable agreement with deep learning predictions and experimental data, revealed distinct state-dependent residue interaction networks, and predicted mutations that tuned the designed conformational landscape. Our approach demonstrates that new modes of motion can now be realized through de novo design and provides a framework for constructing biology-inspired, tunable and controllable protein signaling behavior de novo. Finally, in Chapter 3, I discuss key areas where further multi-state tool development is needed and promising applications for de novo dynamics design in the near future.