GPCRdb in 2023: state-specific structure models using AlphaFold2 and new ligand resources

Abstract

G protein-coupled receptors (GPCRs) are physiologically abundant signaling hubs routing hundreds of extracellular signal substances and drugs into intracellular pathways. The GPCR database, GPCRdb supports >5000 interdisciplinary researchers every month with reference data, analysis, visualization, experiment design and dissemination. Here, we present our fifth major GPCRdb release setting out with an overview of the many resources for receptor sequences, structures, and ligands. This includes recently published additions of class D generic residue numbers, a comparative structure analysis tool to identify functional determinants, trees clustering GPCR structures by 3D conformation, and mutations stabilizing inactive/active states. We provide new state-specific structure models of all human non-olfactory GPCRs built using AlphaFold2-MultiState. We also provide a new resource of endogenous ligands along with a larger number of surrogate ligands with bioactivity, vendor, and physiochemical descriptor data. The one-stop-shop ligand resources integrate ligands/data from the ChEMBL, Guide to Pharmacology, PDSP Ki and PubChem database. The GPCRdb is available at https://gpcrdb.org.

Machine Learning-based Modeling of Olfactory Receptors in their Inactive State: Human OR51E2 as a Case Study

“…e ., without ligands), we therefore chose the inactive conformation. In practice, such model was obtained from GPCRdb 24 (https://gpcrdb.org), which maintains an updated database of GPCR models generated with AlphaFold-MultiState.…”

“…23 As a last candidate, we considered a model of the receptor in its inactive state (AF in ), generated with AlphaFold-MultiState. 15,24 For all the predictors considered, we tried as much as possible to use the models already available to the public ( i . e ., without directly using the ML algorithm or modifying the default parameters – see details in the Methods section).…”

“…Extensive details on the initial structures, simulation protocol steps, description of the A 100 index for hOR51E2 and its implementation in PLUMED, figures with the time evolution of A 100 and RMSD, a table with the water passage results, a plot that shows the amino acid conservation for positions 2.50 and 3.39, as well as a table with the Ballesteros-Weinstein numbering for hOR51E2, as listed in the GPCRdb. 24…”

Machine Learning-Based Modeling of Olfactory Receptors in Their Inactive State: Human OR51E2 as a Case Study

“…For homology modeling, we relied on the SwissModel (SM) Web server, while for ML-based prediction, we considered AlphaFold (AF), , RoseTTAFold (RF), OmegaFold (OF), and ESMFold (EF) . As a last candidate, we considered a model of the receptor in its inactive state (AF in ), generated with AlphaFold-MultiState. , For all the predictors considered, we tried to use the models already available to the public (i.e., without directly using the ML algorithm or modifying the default parameterssee details in the Supporting Information). In Figure , we show the initial predicted structures and a similarity representation among all six models, based on the calculation of the mutual backbone RMSD, followed by a 2D projection using Multidimensional Scaling (MDS) .…”

The 2023 Nucleic Acids Research Database Issue and the online molecular biology database collection

“…The RCSB Protein Data Bank ( 26 ) reports that structure predictions, including AFDB, are now available via its website, alongside its core experimentally determined structures which now number almost 200 000. The update from GPCRdb ( 27 ) reports state-specific AF2 models alongside other new features such as lists of ligands for each receptor, both endogenous and surrogate. Even MobiDB ( 28 ), focusing on intrinsically disordered proteins, benefits from two AF2-related predictions unforeseen by the original methods developers: predictions of disordered regions and potential interaction motifs contained within them.…”