Cryptic allosteric sites could provide new opportunities for activating or inhibiting proteins, even proteins classically considered undruggable. However, the promise of cryptic allosteric sites has not been fully realized because the population of the open form is often too low to detect experimentally unless ligands that can bind to and stabilize those forms are present. Computer simulations are a promising means to identify high-energy structures with pockets that may be cryptic allosteric sites. However, the difficulty of simulating slow conformational changes, such as the opening of many cryptic pockets, has hampered the discovery of cryptic sites. There are also limitations in how simulation data is analyzed. To overcome these limitations, we have developed a new sampling algorithm, called FAST, for accelerating the discovery of conformations with desirable geometric features, such as the presence of cryptic pockets. We have also developed new algorithms for robustly detecting cryptic pockets in simulation data. Our methods identify new cryptic pockets that could be exploited to restore the efficacy of existing antibiotics. We have experimentally verified the presence of these pockets, and designed and experimentally verified allosteric drugs that bind them.
This event is co-sponsored by The Center for Interdisciplinary Scientific Computation (CISC).