All-atom Metadynamics Simulations of the Hierarchical Folding Dynamics of Proteins on the Timescale of Seconds

Molecular dynamics simulations have proven instrumental to the understanding of molecular biophysics. However, the temporal constraints of molecular dynamics simulations have limited attempts to capture the protein folding process at the atomic scale. Herein, we circumvent this limitation by using our all-atom metadynamics algorithm to directly sample the potential-energy landscape of numerous proteins. Previous applications of the original metadynamics method to proteins penalized select collective variables or dihedral angles assumed to be principle to the folding process. Rather, our method penalizes the full coordinate space of the protein resulting in 3N-dimensional sampled energy-landscape, unbiased by a priori assumptions. With our all-atom sampling, a single simulation captures the folding and unfolding process multiple times, enabling the simulation of protein dynamics on the timescale of seconds, orders of magnitude longer than recorded molecular dynamics simulations. From the sampled energy landscape, we predict the folded state, the activation barrier and the timescale associated with the folding processes of the proteins. Herein, we will present these findings for several well-studied proteins, to validate our results, and several new proteins, as a novel extension.