BPS2026 – Developing a simulation pipeline to predict phosphoinositide lipid binding sites on ion channels

Ion channels function as molecular gatekeepers, controlling passage of ions in and out of cells. Phosphoinositides-biphosphates (PIP 2 ) are minor phospholipid components of cellular membranes but play critical roles in modulating the functions of numerous ion channels. However, due to limited numbers of experimentally resolved ion channel structures bound to PIP 2 ; where PIP 2 binds, and if/how it modifies ion channel functions remain unelucidated for most ion channel families. Here, we present an automatic CGMD simulation PIPeline to predict PIP 2 binding site across ion channel families in a high-throughput manner. Using this PIPeline, we conducted the first systematic evaluation of CGMD’s performance in identifying lipid binding sites in comparing against both experimental data and AI-based predication tools. In our validation set of 12 experimentally resolved ion channel-PIP 2 complexes, PIP 2 successfully remained at their resolved binding sites after 60 μs CGMD simulation in 11 cases. Furthermore, these binding sites were reproduced when PIP 2 starts in the bulk membrane with high occupancy and extended binding duration (> 5 μs), demonstrating the sensitivity and accuracy of CGMD. We also showed these resolved binding sites can be identified using a combination of quantitative measurements such as binding duration and spatial clustering. Comparing to AI-based tools, we highlight the reliability of CGMD in predicting PIP 2 binding sites on proteins with few resolved structures and poorly understood mechanisms, which are underrepresented in the AI training data set. Finally, we applied the PIPeline and AI methods to ion channels with known PIP 2 regulation but unresolved binding sites, predicting putative sites and revealing potential regulatory effects. Our PIPeline enables identification and characterization of PIP 2 binding for various ion channels, offering valuable insights into molecular basis of lipid-protein interactions and potentially opening a door for lipid-like drug development.