Computational design of peptide allosteric modulators

G protein-coupled receptors (GPCRs) represent the largest family of membrane proteins in the human genome. Located in the plasma membrane GPCRs sense a plethora of ligands ranging from ions, photons and small molecules to lipids, proteins and peptides. Their activation by ligand binding leads to coupling to intracellular proteins including heterotrimeric G proteins, ß-arrestins and kinases and initiation of downstream signaling pathways. Since GPCRs regulate fundamental physiological processes that vary from heart rate and blood pressure control to learning, memory, and cognition, it is not surprising that they constitute a prime target class in modern pharmacotherapy of human illnesses. In fact, approximately 34% of all approved therapeutics act by modulating these receptors. Canonically, GPCR ligands have been classified into orthosteric agonists or antagonists that either activate or block GPCR signaling pathways. Recent studies have identified biased allosteric modulators of GPCRs which not only preferentially activate a subset of signaling pathways but also target allosteric sites. Integrating signaling bias with allosteric modulation holds promise to design safer and more effective therapeutics. In the last decade, groundbreaking advances in structural elucidation of GPCRs provided high-resolution structures and created an opportunity to apply structure-based ligand design. While to date a series of ligands have been developed by using structural information of GPCRs, rational design of ligands with biased and allosteric properties remains a long-standing challenge. Over the past years, computational methodology has advanced to the point that peptide structures may be engineered to tackle current challenges in GPCR ligand development. Hence, the proposed project aims to rationally develop GPCR targeting peptide ligands with biased and allosteric profiles using state-of-the-art computational design technologies. The computationally designed peptides will then experimentally be validated and first-in-class structure of a GPCR in complex with a biased allosteric modulator will be elucidated. This study will provide proof-of-concept that a generalizable computational pipeline can be developed to rationally design peptide-based biased allosteric modulators of GPCRs which will illuminate molecular mechanisms and open up new opportunities for drug development of the largest family of cell membrane receptors.

At the core of how our bodies function is a group of cell surface receptors known as G protein-coupled receptors (GPCRs). These receptors help us sense pain, taste, hormones, mood, and much more. In fact, more than one-third of all approved medicines work by targeting GPCRs. Yet computationally designing protein-based molecules that can switch these receptors "on" or "off" has remained an outstanding challenge. In this project, we developed powerful new computational methods to build tiny proteins-called miniproteins-entirely from scratch. Using advanced deep learning tools, we designed miniproteins that either activate specific GPCRs (acting as agonists) or block them (acting as antagonists). Our process began with small residue pieces that fit deep into the receptor's binding pocket, and then we "grew" the rest of the protein around that core. We also explored a wide variety of natural protein shapes to discover new design strategies. To test whether these designs worked, we employed pharmacological assays but also created a new screening method called Receptor Diversion, which uses fluorescence in human cells to see whether a miniprotein binds to its target receptor. Using these tools, we successfully designed agonists for the pain-related receptor MRGPRX1, achieving atomic-level accuracy between the computer models and real protein structures. We also designed antagonists for a range of important GPCRs: CXCR4 (involved in cancer and immune response), GLP1R GCGR and GIPR (key in metabolism), and CGRPR (linked to migraine pain). One of our newly designed CGRPR blockers shows high potency and selectivity-comparable to approved antibody drugs, but in a much smaller format. These custom-built miniproteins could soon lead to precise, targeted therapies for conditions like diabetes, migraine, pain, and cancer, potentially with fewer side effects than current treatments. They also provide scientists with powerful new tools to study how GPCRs function-working like tiny molecular switches. By designing these proteins entirely on computers and validating them rapidly in cells, we have created a faster, more flexible, and less trial-and-error-dependent path to drug discovery. This work represents a major step forward in medicine and biotechnology-opening the door to a new generation of fully designed protein-based drugs that can control how our cells respond to the world around us.