Necessity of High Physical Resolution in the Development of Flexible Coarse-Grained Protein Models

Abstract

The Martini force field is a one of the most widely used coarse-grained (CG) model. An important strength of the Martini force field is that it includes explicit, microscopic representation of solvent, which allows proper description of the context-dependence of protein-protein interactions. However, in the Martini force field the secondary structures of the protein are fixed, which makes it is unsuitable for simulating dynamic processes such as protein folding and membrane insertion. In this work, we examine the possibility of developing a flexible protein model within the Martini CG framework, such as by supplementing the force field with Go-like potentials. It was found the current Martini CG representation of proteins was insufficient to support a flexible model. The model does not provide a sufficient resolution to properly describe the volume and packing of protein backbone and sidechains. The interior voids resulting from improper packing would lead to excessive structural collapse in absence of structural restraints. Along this line, we further examined the PACE model, where an atomistic protein model is deployed in the Martini solvent. The results suggest that, while atomistic protein representation does dramatically improve the ability to describe specific protein-protein interactions, the low physical resolution of the Martini water molecules does not allow sufficient discrimination of various open and compact conformational states. As a result, PACE is not capable of describing context dependent protein structure transition, such as helical transition induced by membrane absorption and/or insertion. In conclusion, high physical resolution is likely necessary for developing flexible protein models and the coarse-graining needs to focus on exploiting possible simplification of interaction potentials.