Langevin Dynamics Simulations of Protein Fencing of PIP2

Abstract

Anionic phosphotidylinositol 4,5-bisphosphate (PIP2) occupies only 2∼3% of the phospholipids in the inner leaflet of the plasma membrane, but participates in numerous cellular processes. Concentrated pools of PIP2 on the membrane surface appear to be the source for such PIP2-involved cellular processes, and recent experimental observations indicate that a “protein fence” prevents PIP2 from diffusing out of the pool. Here the necessary properties of the protein fence are examined using Langevin dynamics simulation with effective potential maps. PIP2 molecules are modeled as explicitly diffusing spheres, and the rest of the system represented by steric and electrostatic potentials. Simple models of charged or porous rods, as well as rigid proteins are considered. Retardation by electrostatic forces from a rod is only significant when the charges are highly concentrated and close to the diffusion plane; charge densities and geometries of proteins are insufficient to effectively block diffusion. Hence, fencing appears to be dominated by steric effects, though even a small (> 1%) opening in a porous rod in contact with the surface results in substantial leakage. The three protein fence candidates, actin, human septin, and yeast septin, are known to self-assemble into filaments on the cell surface and to interact with PIP2. Simulations reveal that actin is a poor fence for all depths and orientations simulated because its arch-type shape leaves passageways for diffusion. In contrast, the septins effectively block PIP2 diffusion when buried in the membrane to 10∼15 A from the lipid surface. Free energy calculations using an implicit membrane model support the possibility of burial of septins at these depths. It is also possible that the septins are not buried as deeply, but associate with other proteins or peptides to make a fence.