Unraveling G Protein-Coupled Receptor Allosteric Modulation through Computational Approaches

The development of new drugs is a multidisciplinary process that is time-consuming and expensive. Computer-Aided Drug Design (CADD) is essential for accelerating the drug discovery process. GPCRs represent a large and pharmacologically important family of proteins involved in essential cellular signaling pathways. As of mid-2025, approximately 36% of all FDA-approved drugs target GPCRs, 99% of which bind to the primary endogenous ligand site, known as the orthosteric site. However, the growing availability of experimental GPCR structures has revealed a rich diversity of allosteric binding sites—regions distinct from the orthosteric site that modulate receptor function in subtle and often highly selective ways. Allosteric modulation offers promising opportunities to enhance drug efficacy, selectivity, and chemical diversity; however, a systematic understanding of its mechanisms is still developing. This work is structured around three main chapters. The first part introduces an annotation scheme for GPCRs that structurally classifies binding sites based on receptor class, contacts with transmembrane helices, and topology relative to the membrane. Applied to 100 GPCR-ligand complexes, this framework identifies 24 distinct allosteric sites and evaluates site-detection algorithms (BioGPS, SiteMap, and FTMap), with BioGPS performing best. Property analysis shows that extrahelical allosteric pockets tend to be shallow and low in volume, and that their ligands are enriched in halogens, indicating a distinct chemical space. Combining receptor and ligand similarity enables effective prediction of ligandability, despite challenges posed by ligand-induced conformational changes at the binding sites. In the second part, the Atlas algorithm, originally developed for globular proteins, was refined by integrating known extra-helical ligand fragments for GPCRs and recalibrating cavity contributions to account for shallow allosteric pockets at the protein-membrane interface. This optimized version successfully identified 96% of validated allosteric sites in GPCRs and was retrospectively validated through structure-activity relationship (SAR) analysis. Prospective validation included the synthesis and binding confirmation of hybrid probe ligands targeting hot spots in the complement C5a receptor (C5AR1) and the cannabinoid receptor 1 CNR1), highlighting Atlas’s utility in allosteric drug design without requiring prior knowledge. The third and final study explores how allosteric modulators influence protein dynamics. A total of 45 μs of molecular dynamics simulations were analyzed across four Class A GPCRs C5AR1, P2YR1), CNR1/CNR2 to capture dynamic features of shallow extra-helical sites. Focusing on Allosteric Communication Networks (ACNs), the shortest paths were calculated between orthosteric site and G-protein interface. This analysis revealed communication patterns specific to conformational state and ligand. The ACN analysis guided a mutagenesis study on CNR2, which revealed the crucial role of this residue G2105.59 in the differential pharmacological profile of Ec21a in CNR1 versus CNR2. This thesis demonstrates how the combination of GPCR crystallographic structures, computational methods, and experimental techniques can yield new insights into allosterism, enabling the rational design of novel allosteric ligands in the future.