Deep learning reveals endogenous sterols as allosteric modulators of the GPCR-Gα interface

Area of Science:

• Biochemistry
• Computational Biology
• Molecular Pharmacology

Background:

• Endogenous intracellular allosteric modulators of G protein-coupled receptors (GPCRs) are poorly understood due to limited data and computational tools.
• Identifying and designing ligands for these modulators is challenging across diverse GPCR structures.

Purpose of the Study:

• To develop a computational toolkit (Gcoupler) for predicting high-affinity ligands targeting the intracellular GPCR-Gα interface.
• To identify endogenous metabolites that modulate GPCRs and their downstream signaling pathways.

Main Methods:

• Developed Gcoupler, an AI-driven toolkit integrating de novo ligand design, Graph Neural Networks, and statistical methods.
• Utilized computational analysis and experimental validation (genetic screening, multi-omics, mutagenesis, biochemical assays) in yeast models.
• Employed molecular simulations and studied GPCR-mediated cardiac hypertrophy models in mammals.

Main Results:

• Identified endogenous hydrophobic metabolites, particularly sterols, as direct intracellular allosteric modulators of the Ste2p-Gpa1p complex in yeast.
• Confirmed that metabolite binding to the GPCR-Gα interface obstructs downstream signaling.
• Demonstrated that elevated metabolite levels attenuate GPCR-mediated cardiac hypertrophy in mammalian models.

Conclusions:

• Endogenous metabolites, such as sterols, function as intracellular allosteric modulators of GPCRs, regulating signaling pathways.
• The Gcoupler toolkit provides a robust platform for discovering novel GPCR-targeting ligands.
• This mechanism highlights an evolutionarily conserved pathway influencing cellular responses and disease states.