DMD4B-HYDRA: Toward a Novel Discrete Molecular Dynamics Protein Model

Abstract

The task of protein structure prediction is computationally demanding, particularly for intrinsically disordered proteins (IDPs), which by definition lack a well-defined folded structure. IDPs include amyloidogenic proteins associated with human disease, such as Alzheimer's and Parkinson's diseases, which are known to be critically involved in the corresponding pathologies. For most protein and peptide sequences with significantly more than 100 amino acids, it is not yet possible to examine the details of the folding landscape by fully atomistic molecular dynamics (MD) in explicit solvent. This problem can be solved through a two-stage MD, first exploring the protein dynamics using coarse-grain (CG) protein models to sample the phase space at a diminished computational effort and second by using a wide range of CG structures converted into fully atomistic representation as initial conformations for all-atom MD in explicit solvent. We here use the four-bead protein model in combination with discrete molecular dynamics (DMD), known as DMD4B-HYDRA force field, which has been previously successfully applied to studies of amyloid b-protein folding and assembly, to examine protein folding dynamics of five proteins with known native structures. We elucidate both the ability of DMD4B-HYDRA force field to capture the folded structure of these proteins and aspects that are less well reproduced by the current force field. We then demonstrate that the follow-up all-atom MD dynamics compensates for structure prediction difficulties of the DMD4B-HYDRA force field itself and provides feedback to be used in the future DMD4B-HYDRA force field development. Our preliminary results thus confirm that DMD4B-HYDRA combined with fully atomistic MD in explicit solvent presents a powerful computational approach to studying IDPs and elucidate structural dynamics of a broad range of amyloidogenic proteins.