Simple Method for Simulating the Mixture of Atomistic and Coarse-Grained Molecular Systems.

Abstract

Combining fine-grained (FG) all-atom and coarse-grained (CG) systems in a single simulation in a hybrid manner is of immense interest in recent times, owing to the possibility of overcoming the limitations of both FG simulations as well as CG simulations. The existing methods for combining these two resolutions tend to require heavy parametrizations or sometimes lack in transferability to other systems of interest, and further developments toward such directions are highly required. We report here a simple protocol to combine CG and FG systems in a single simulation, using the standard FG and CG force field models by adopting a series of small proteins as test cases. Our method makes use of virtual sites as reported earlier for relatively simple butane and dialaine systems (Rzepiela et al. Phys. Chem. Chem. Phys. 2011, 13, 10437-10448), to bridge the interaction between FG protein atoms and CG water. We find that the conventional CG model (MARTINI potentials) couples too strongly with the FG model and that it leads to complete unfolding of a test protein within very short time. We find that reducing the Lennard-Jones potential between CG atoms and virtual site atoms stabilizes the secondary and tertiary structures, sometimes almost to a comparable level with the fully atomistic simulations. However, detailed inspection reveals that this reduction is not enough for satisfactory consistency of the hybrid scheme against the FG simulation. As a remedy, we observe that the addition of as small as 4 Å thick position-restrained FG water layer in the hybrid simulation can further improve the structural behaviors in many respects, with its results closely mimicking those of the FG-only simulations. However, free energy landscapes reveal that this agreement with a restrained solvent layer is still accompanied by the overstabilization of the protein native structure, which will likely pose limitations for studying protein dynamics with the scheme. We show various test results that we have tried in optimizing the FG-CG mixing scheme over the course and discuss future prospects as concluding remarks of the present work.