(Ala)4-X-(Ala)4as a model system for the optimization of theχ1andχ2amino acid side-chain dihedral empirical force field parameters

Abstract

Amino acid side-chain fluctuations play an essential role in the structure and function of proteins. Accordingly, in theoretical studies of proteins, it is important to have an accurate description of their conformational properties. Recently, new side-chain torsion parameters were introduced into the CHARMM and Amber additive force fields and evaluated based on the conformational properties of the individual side-chains using protein simulations in explicit solvent. While effective for validation, molecular dynamics simulations of proteins must be extended into the microsecond regime to obtain full convergence of the side-chain conformations, limiting their use for force field optimization. To address this, we systematically test the utility of explicit solvent simulations of (Ala)(4)-X-(Ala)(4) peptides, where X represents the amino acids, as model systems for the optimization of χ(1) and χ(2) side-chain parameters. The effect of (Ala)(4)-X-(Ala)(4) backbone conformation was tested by constraining the backbone in the α-helical, C5, C7(eq), and PPII conformations and performing exhaustive sampling using Hamiltonian replica exchange simulations. Rotamer distributions from protein and the (Ala)(4)-X-(Ala)(4) simulations showed the highest correlation for the C7(eq) and PPII conformations, although agreement was the best for the α-helical conformation for Asn. Hydrogen bond analysis indicates the utility of the C7(eq) and PPII conformations to be due to specific side-chain-backbone hydrogen bonds not being oversampled, thereby allowing sampling of a range of side-chain conformations consistent with the distributions occurring in full proteins. It is anticipated that the (Ala)(4)-X-(Ala)(4) model system will allow for iterative force field optimization targeting condensed-phase conformational distributions of side-chains.