AlphaFold2 versus experimental structures: evaluation on G protein-coupled receptors.

Area of Science:

• Structural Biology
• Computational Biology
• Pharmacology

Background:

• G protein-coupled receptors (GPCRs) are crucial drug targets involved in numerous physiological processes.
• Experimental determination of GPCR structures is challenging due to inherent instability.
• AlphaFold2 is a recent AI tool for predicting protein structures, including GPCRs.

Purpose of the Study:

• To evaluate the accuracy of G protein-coupled receptor (GPCR) structure models generated by AlphaFold2.
• To assess the suitability of AlphaFold2-predicted GPCR models for functional studies and drug design.

Main Methods:

• Comparison of AlphaFold2-predicted GPCR structures with experimentally determined structures.
• Analysis of differences in extracellular and transmembrane domains, ligand-binding pockets, and transducer interfaces.

Main Results:

• AlphaFold2 successfully captures the overall backbone topology of G protein-coupled receptors (GPCRs).
• Significant discrepancies exist between predicted and experimental structures concerning domain assembly and interface conformations.
• Ligand-binding pocket shapes in predicted models differ from those in experimental structures.

Conclusions:

• While AlphaFold2 provides a useful starting point for GPCR structural analysis, its predictions have limitations.
• The observed differences between predicted and experimental GPCR structures hinder direct application in structure-based drug design.
• High-resolution experimental structural data remains essential for accurate functional studies and drug development targeting GPCRs.