Accurate Structure Prediction and Conformational Analysis of Cyclic Peptides with Residue-Specific Force Fields

Cyclic peptides (CPs)
are promising candidates for drugs, chemical
biology tools, and self-assembling nanomaterials. However, the development
of reliable and accurate computational methods for their structure
prediction has been challenging. Here, 20 all-trans CPs of 5–12
residues selected from Cambridge Structure Database have been simulated
using replica-exchange molecular dynamics with four different force
fields. Our recently developed residue-specific force fields RSFF1
and RSFF2 can correctly identify the crystal-like conformations of
more than half CPs as the most populated conformation. The RSFF2 performs
the best, which consistently predicts the crystal structures of 17
out of 20 CPs with rmsd < 1.1 Å. We also compared the backbone
(ϕ, ψ) sampling of residues in CPs with those in short
linear peptides and in globular proteins. In general, unlike
linear peptides, CPs have local conformational free energies and entropies
quite similar to globular proteins.