Generic Coarse-Grained Model for Protein Folding and Aggregation

Abstract

The complexity involved in protein structure is not only due to the rich variety of amino acids, consistent with the random heteropolymer picture, but also the weak interactions involved, comparable to thermal energy, and important cooperative phenomena. This presents a challenge in computer simulations, as it is associated with high-dimensionality and ruggedness of the free energy landscape as well as long equilibration times, frequently exceeding what can be handled in atomistic studies. We have recently developed a coarse-grained (CG) implicit solvent peptide model which has been designed to reproduce key consequences of the abovementioned weak interactions. Its intermediate level of resolution, four beads per amino acid, allows for accurate sampling of local conformations, in particular secondary structure, by designing a force field that relies on simple interactions (e.g. hydrogen bonds, hydrophobicity). A realistic ratio of alpha-helix to beta-sheet content is achieved by mimicking a nearest-neighbor dipolar interaction. In the present study, we tune the model in order to fold helical proteins while systematically comparing the structure with NMR data. Very good agreement is achieved for proteins that have simple tertiary structures, which implies that the force field is able to reproduce important cooperativity features between amino acids. We further probe these effects by looking at peptide aggregation scenarios. Hydrophobic peptide fragments cooperatively form largescale beta-sheet structures. The model is able to reproduce features from atomistic simulations on a qualitative basis. The large-scale and long-term regime that this CG model offers, coupled with our design criteria (folding and realistic alpha/beta content), make it very suitable for many biological processes, such as misfolding and oligomerization involved in neurodegenerative diseases.