UCSF

It is now widely accepted that drugs achieve efficacy via interactions with multiple targets simultaneously, a serendipitous biological symphony that is typically only deconvoluted retrospectively. While pan-profiling of these molecules has discovered varying activity against different proteins, efforts to recapitulate therapeutic effects through focused, specific interactions with singular targets have largely failed. This is especially true for G protein-coupled receptors (GPCRs), whose conserved topologies have introduced widespread degeneracy among their ligands. Yet, 38% of GPCRs are orphans, with no elucidated physiological function or no illuminating modulatory chemical matter.

First, we use a robust yeast-screening platform to show that even in a small library of drugs, the NIH Clinical Collection, orphan GPCRs are potent targets for a variety of drugs. We focus on the benzodiazepine lorazepam and show that is also a strong positive allosteric modulator (PAM) of the pH-sensing orphan, GPR68. We generated over 3,000 receptor homology models for virtual screening campaigns that, among other hits, identified ogerin, a PAM that selectively and potently potentiated GPR68-Gs signaling (αβ= 27.5). Ogerin was prioritized to interrogate the in vivo function of GPR68, revealing that receptor activation suppressed recall in fear conditioning in wild-type, but not in GPR68 knockout, mice. Application of the same approach to a second orphan receptor, GPR65, led to the discovery of specific allosteric agonists and negative allosteric modulators.

Second, we pick subsets of GPCRs to demonstrate that structure-based virtual screening can be used to directly identify molecules with a desired receptor binding profile. Although only a small number of hits were selective over the anti-target, a result observed in two parallel studies between three aminergic receptors and 2 opioid receptor subtypes, this performance is still better than that of ligand-based methods. Inexhaustive and incorrect treatment of receptor flexibility was subsequently deemed to be the culprit, a potentially receptor-specific nuisance that can be leveraged. At its core, this thesis describes test cases and methods that would ultimately encompass the machinery for a pipeline to generate molecules with a complete, prescribed GPCR polypharmacology.