A Langevin Dynamics Algorithm for Coarse-Grain Modeling of Protein Clusters

Abstract

A variety of recent experimental results have highlighted the importance of spatial localization and clustering in cellular signaling pathways. We have called such non-stoichiometric dynamic clusters “pleiomorphic ensembles” (PE). The participation of PEs often significantly changes the dynamic and equilibrium properties of signaling networks, but the mechanisms behind such changes can be unclear, and computational modeling of these effects is hampered by the lack of efficient and appropriate methods to address such questions. Non-spatial models can reproduce some features of clustering, such as phase transitions, but cannot capture features such as steric hindrance and PE diffusion, which may have important biological consequences. Publicly available spatial stochastic solvers, such as Smoldyn or MCELL, treat all particles as points and thus cannot model spatial organization and excluded volume. In principle, clustering could be modeled with molecular dynamics simulations, but typical PE sizes and time scales preclude such models with current computational hardware.To overcome these limitations we have designed a general purpose Langevin dynamics simulator that models proteins as a set of sites connected by stiff links. The sites are modeled as impenetrable spheres, which captures the effects of excluded volume and steric hindrance. The program allows for zero order creation reactions, a variety of first order reactions such as dissociation reactions and transitions between states of a given site (e.g., active or inactive), and second order binding reactions between sites.We apply our program to study cluster formation in the nephrin-Nck-N-wasp system, which is important for kidney podocyte function. We compare our results to the previously published analysis of this system based on a non-spatial simulator (Falkenberg et al. Biophys. J. 2013), to explore the consequences of steric crowding and diffusion on cluster dynamics. (Supported by NIH grants TRO1DK087650 and P41GM103313)