The three-dimensional structure prediction of human bitter taste receptor using the method of AlphaFold3

Bitter taste receptors (T2Rs), a subfamily of G protein-coupled receptors, are expressed not only in oral tissues but also in extraoral sites, playing key roles in physiological processes such as the gut-brain axis. However, structural information on T2Rs is limited, with only two human T2Rs, T2R14 and T2R46, experimentally determined to date. This study explores the potential of AlphaFold3 (AF3), an advanced AI-based protein structure prediction tool, to predict the structures of 25 human T2Rs and compares them with those of the earlier AlphaFold2 (AF2). The accuracy of AF3 was evaluated by comparing the predicted structures of T2R14 and T2R46 with known experimental structures. Our results show that AF3 provides more accurate structural predictions than AF2 for these receptors, though the predicted local distance difference test scores for AF3 were unexpectedly lower across all T2R subtypes. Subsequent analysis indicated that significant structural variations were observed in the receptor's extracellular region, in contrast to a higher degree of structural consistency in the intracellular region. Clustering based on sequence identity and root mean square deviation highlighted distinct groupings among the receptors. The structural properties of these T2Rs may be related to their ability to recognize thousands of diverse bitter substances through interaction with the taste receptor-specific G protein, α-gustducin. The present study provides evidence that AF3 can advance our understanding of T2R structure and research into the biological activity of T2R-ligand interactions in health-related processes, including risk reduction of obesity and diabetes.