Leveraging Integrative Modeling and AI-Based Tools to Develop Safer Opioids

PROJECT SUMMARY The misuse of opioid painkillers targeting the -opioid receptor (MOR) continues to result in record numbers of overdose deaths, underscoring the pressing need to identify safer analgesics, as well as effective therapies for opioid use disorder (OUD). Although efficacy with (e.g., preferable drugs that can sufficiently engage pain pathways to provide therapeutic relief while leading to lower side effects through reduced engagement of their mediating circuits. Understanding not free from liabilities, several MOR small-molecules that exhibit low- agonism for relevant G protein subtypes (e.g., buprenorphine and oliceridine) are powerful analgesics i mproved safety profiles compared to other opioids used clinically (e.g., morphine) or commonly abused fentanyl). This supports the emerging paradigm in the field that low-efficacy MOR agonists may be candidates for developing howlow-efficacy agonists activate MOR-mediated G protein pathways differently from high-efficacy opioid drugs is essential for informing the design of safer opioid therapeutics. Achieving this understanding requires an unprecedented level of spatial resolution of both the conformational dynamics and kinetics of ligand-specific MOR-Gi1 coupling and activation, which cannot currently be obtained by a single technique. Instead, it necessitates the integration of data from both experimental and computational sources. Our team is uniquely positioned to tackle this challenge using an innovative Bayesian inference framework to integrate comprehensive conformational ensembles from physics-based enhanced molecular dynamics simulations with state-of-the-art single-molecule imaging data collected at unprecedented time resolution both inpurified protein preparations and in living cellsby long-standing experimental collaborators. This integration will yield more accurate atomistic information regarding ligand-specific MOR-G protein conformations and their kinetic relationships, from which we can derive accurate predictions about the unique attributes of low-efficacy opioids. Experimental validation of these predictions will not only help us achieve the overarching goal of elucidating the mechanistic basis of opioid efficacy but also bolster the architectures and inform optimal training of guidelines for developing novel chemotypes in the pursuit of safer opioid artificial intelligence medications.