ИСТИНА

Small GTPases plays regulatory role in cellular signalling where they act as binary molecular switches. Their activity solely depends on which molecule (GTP or GDP) is currently bound with them. Proteins from Ras subfamily are prominent for their involvement in the processes of cell growth and division. The reaction GTP hydrolysis switches off Ras protein and is regulated by the GTPase activating protein (GAP). Malfunction of GTP hydrolysis in Ras leads to the uncontrollable cell division and is one of the most common oncogenic factors in human cancer. In current work we provide a computational study of the first reaction step of GTP hydrolysis in H-Ras protein and its oncogenic variant G13V. We have used QM/MM-based molecular dynamics along with umbrella sampling method to sample configurational space along the reaction coordinate. Energies and forces on QM atoms were calculated by DFT (BLYP/DZVP) method using CP2K software. Atoms in MM-subsystem were described by CHARMM force field. One-dimensional free energy reaction profile was obtained from statistical data by weighted histogram analysis method (WHAM) as well as umbrella integration. While these approaches rely on different methodology of the free energy calculation, they give very similar results on the data from our simulations. The first step of GTP hydrolysis comprises the nucleophilic attack of the oxygen atom of the reactive water molecule on the phosphorous atom on γ-phosphate group of GTP with simultaneous dissociation of the phosphate β-γ P-O bond. The calculated free energy barrier of this step in wild-type protein is ~10 kcal/mol. The corresponding barrier in G13V mutant is ~3 kcal/mol higher that agrees with an experimental two-order decrease of the reaction rate constant. We have also shown the importance of sampling comparing by calculating several minimum energy reaction profiles which were initiated from points in different regions of configurational space. A significant spread in energies is observed between corresponding stationary points from different reaction profiles. It implies that proper sampling is essential to obtain accurate results. This work is supported by the RFBR grant 14-03-31898. We also thank the Supercomputing Center of Lomonosov Moscow State University for providing computational resources.