Influence of Side Chain Conformations on Local Conformational Features of Amino Acids and Implication for Force Field Development

Statistical analysis of coil regions in protein structures has been used to obtain the local backbone ϕ, ψ preferences of amino acids, which agree well with the NMR experiments of unfolded peptides and proteins. We analyzed the conformational features of amino acid residues in a restricted coil library of 4220 high-resolution protein crystal structures. In addition to Gly, Ala, and Pro, the ϕ, ψ distribution (Ramachandran plot) of each amino acid is analyzed with respect to three side chain conformers: g+ (χ₁ ∼ −60°), g− (χ₁ ∼ 60°), and t (χ₁ ∼ 180°). The statistical study indicates that the effect of side chain conformations on ϕ, ψ distributions is even greater than the effect of amino acid types. On the basis of the χ₁, ϕ, ψ conformational preferences, the amino acids in addition to Gly, Pro, and Ala can be divided into five types: (1) ordinary amino acids, (2) Ser, (3) Asp and Asn, (4) Val and Ile, and (5) Thr, each with distinguished χ₁ rotamers. The α-helix, β-sheet, and type-I β-turn preferences of the different rotamers of various amino acid types can be captured by their intrinsic ϕ, ψ preferences from our coil library. Molecular dynamics simulations of dipeptide Ac−X−NHMe and tetrapeptide Ac−A−X−A−NHMe models give nearly the same side chain rotamer distributions. However, for many amino acids, both OPLS−AA/L and AMBER−FF03 force fields give very different χ₁ rotamer distributions from the coil library. This may partially explain why dipeptide models sometimes cannot reproduce those of protein structures well. The current coil library analysis may be valuable in improving the force field for protein simulations.