ACS Comp Spring 2024, Oral talk

This is a the work presented on ACS Spring 2024 (Comp Divsion). The abstract of the talk is: PIKfyve is the third member (type III) of the PIPK lipid kinase family which catalyzes the phosphorylation reaction of PI3P to generate PI(3,5)P2 molecules. It has been reported that pharmacological inhibition of PIKfyve reduces the degenerative phenotype of the induced motor neurons derived from the patients with Amyotrophic lateral sclerosis (ALS). Hence, PIKfyve is a drug target for the treatment of ALS and potentially other neurodegenerative diseases with similar biomolecular origin. PIKfyve is activated in presence of membrane where the activation loop of the kinase binds to the membrane in a particular conformation. Some of the critical features such as the secondary structure of the activation loop or the orientation of the membrane bound activation loop are not clearly understood yet due to the underlying technical challenges associated with the complex biological system. Moreover, the molecular mechanism of the PIKfyve kinase domain-membrane binding/unbinding are also central to the development of novel PIKfyve kinase inhibitors. The AlphaFold derived structure of the hPIKfyve kinase domain is consistent with NMR and mutagenesis-based studies but lacks the environmental effects due to membrane binding. To establish the stability of this predicted structure, a set of three molecular dynamics simulations of 50 ns were carried out with the activation loop of the kinase bound to a lipid bilayer (membrane). We observe no substantial conformational changes in the structural features of the PIKfyve activation loop. MD simulation of the same protein in aqueous solution, in absence of the membrane, drastically rearranges the activation loop leading to the loss of the required helicity. This suggests that the membrane-bound state of hPIKfyve is essential to preserve the functionality of the protein. Since the timescale of this membrane-protein binding/unbinding processes is beyond that reachable by standard MD simulations, we have employed weighted ensemble based enhanced sampling methods to generate the pathways of these events. Specifically, we use REVO (Resampling of Ensembles by Variation Optimization), which has been efficient and successful in deciphering binding/unbinding pathways in other complex biological systems as well. We utilize the REVO binding/unbinding trajectories for building Markov state models in order to fully characterize the hPIKfyve-membrane bilayer interactions during the binding process and its globally stable membrane-associated state.