Cyclic peptides are promising modulators of protein-protein interactions and their value as potential drugs would increase dramatically if we could accurately predict their conformations de novo. Here we report a computational method that integrates bias-exchange metadynamics, a novel Boltzmann reweighting scheme, dihedral principal component analysis, and a modified density peak-based cluster analysis, that can be used to obtain and characterize the structural ensemble of highly constrained systems such as cyclic peptides. We further applied this method to evaluate the performance of a number of popular peptide force fields on modeling cyclic peptides. Our results suggest that current peptide force fields may over-stabilize the α-helix and β/PPII regions, commonly populated by linear peptides and proteins. Due to the structural constraint, one or several residues in a cyclic peptide likely populate a high free energy, rarely sampled region in a typical Ramachandran plot of linear peptides. Our findings thus call for a re-parameterization of peptide force fields to accurately describe the full Ramachandran plot, as new classes of peptides such as cyclic peptides emerge.

Toward structure prediction of cyclic peptides

249th National Meeting (2015)

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.