Bridging structure and function: artificial intelligence-based modelling of kidney proteins

The ability to predict the three-dimensional structure of a protein from its amino acid sequence has potential to provide insights into its function, and in the context of disease, its pathogenic mechanisms and potential drug targets. Artificial intelligence (AI)-driven algorithms, particularly AlphaFold and RoseTTAFold, have revolutionized the field of protein modelling, enabling rapid, high-confidence predictions of protein structures. In nephrology, these advances have clarified the molecular architecture of key renal systems such as podocyte slit diaphragm complexes, the conformational states of membrane transporters and the structural basis of channelopathies that affect polycystin channels. These developments have also enabled low-resolution modelling of complex macromolecular structures, providing insights into structural changes that might underlie the pathogenesis of disease mutants, and enabled virtual screening of drugs and toxins. Although these AI models have yielded important new insights, their integration with experimental methods, particularly cellular cryogenic electron tomography, remains crucial for capturing the domain flexibility, conformational dynamics and binding specificity of proteins in their native environment. Combining AI-based structure prediction with experimental validation will uncover novel pathophysiological mechanisms, guide drug discovery and reveal potential new avenues to target mechanisms of kidney disease.