Effects of frustration on the kinetics of helix formation in alanine polypeptides

The folding kinetics of a seven-residue long alanine polypeptide are investigated using a fully atomic protein model and molecular dynamics simulations. The peptide adopts helical conformations in the native state when simulated in two different implicit solvents: a model with a distance dependent dielectric constant and the generalized Born (GB) model. Although the two solvation models correctly predict the native state of the peptide, they lead to distinct folding kinetics. The model in which the dielectric constant is distance-dependent produces much broader distributions of the first passage time, as measured by the intermittency ratio, than the generalized Born model. In addition, the conformational diffusion search folding mechanism, introduced earlier to study the folding kinetics of helical peptides in explicit solvent, is seen to describe more closely the folding kinetics of the GB model than of the distance-dependent dielectric constant model. Based on these observations, we argue that the implicit solvent model that treats the dielectric constant as dependent on interatomic distances introduces excessive amounts of frustration into the underlying potential energy landscape and is thus far less suitable for modelling the kinetics of protein folding than the more sophisticated GB implicit solvent.