Abstract
Allosteric pockets that typically only emerge in the presence of a binder, known as cryptic pockets, can provide an avenue for drug discovery in challenging pharmaceutical targets. However, protein conformations exposing cryptic pockets are generally short-lived and can require significant structural rearrangements that complicate their discovery in experiment and simulation. Here, we investigate the structural basis of cryptic pocket formation in drug targets characterized by extensive dynamics using simulation-based methods. We find that functional protein segments can be anchored by local intramolecular contacts and that disrupting these interactions drives undirected large conformational changes to form cryptic pockets in PRMT5, PRMT6, SMARCA2, Abl1, and PI3Kα. Perturbing the contact networks with benzene probes, elevated temperature, or scaled protein-water interactions could not facilitate these structural dynamics here, indicating that complex mechanisms involving high-energy barriers are necessary to form ligandable cryptic pockets. Based on these limitations, a new computational approach was developed to guide conformational sampling by local interactions surrounding functional protein segments, termed "SLICE" (sampling by local interaction-guided conformational exploration). Across multiple pharmaceutically relevant proteins, our simulations aid in understanding and rapidly exploring the large-scale structural plasticity governed by the local protein environment around functional segments that can be leveraged for drug discovery.