Article

AlphaFold3 versus experimental structures: assessment of the accuracy in ligand-bound G protein-coupled receptors

Xin-heng He1,2, Jun-rui Li1, Shi-yi Shen1,2, H. Eric Xu1,2
1 State Key Laboratory of Drug Research and CAS Key Laboratory of Receptor Research, Shanghai Institute of Materia Medica, Chinese Academy of Sciences, Shanghai 201203, China
2 University of Chinese Academy of Sciences, Beijing 100049, China
Correspondence to: H. Eric Xu: eric.xu@simm.ac.cn,
DOI: 10.1038/s41401-024-01429-y
Received: 13 September 2024
Accepted: 11 November 2024
Advance online: 6 December 2024

Abstract

G protein-coupled receptors (GPCRs) are critical drug targets involved in numerous physiological processes, yet many of their structures remain unresolved due to inherent flexibility and diverse ligand interactions. This study systematically evaluates the accuracy of AlphaFold3-predicted GPCR structures compared to experimentally determined structures, with a primary focus on ligand-bound states. Our analysis reveals that while AlphaFold3 shows improved performance over AlphaFold2 in predicting overall GPCR backbone architecture, significant discrepancies persist in ligand-binding poses, particularly for ions, peptides, and proteins. Despite advancements, these limitations constrain the utility of AlphaFold3 models in functional studies and structure-based drug design, where high-resolution details of ligand interactions are crucial. We assess the accuracy of predicted structures across various ligand types, quantifying deviations in binding pocket geometries and ligand orientations. Our findings highlight specific challenges in the computational prediction of ligand-bound GPCR structures, emphasizing areas where further refinement is needed. This study provides valuable insights for researchers using AlphaFold3 in GPCR studies, underscores the ongoing necessity for experimental structure determination, and offers direction for improving protein–ligand interaction predictions in future computational models.
Keywords: AlphaFold; structure-based drug design; artificial intelligence; GPCR; structural biology

Article Options

Download Citation

Cited times in Scopus