Electrostatic models for polypeptides: can we assume transferability?
Abstract
In most simulations of flexible molecules, the electrostatic model, usually a set of atomic point charges, is assumed to be independent of conformation. The electrostatic models for large molecules are usually constructed by assuming that the atomic charge densities can be transferred from smaller model molecules. Both assumptions neglect polarization of the atoms by the rest of the molecule. We test the accuracy of these assumptions for peptide systems. This is done using ab initio charge densities for N-acetyl, N′-methylamide blocked derivatives of alanine (CH3CO · NHCHCH3CO · NHCH3) and diglycine [CH3CO ·(NHCH2CO)2· NHCH3] in different conformations. The electrostatic fields are calculated from sets of multipoles (charge, dipole, quadrupole etc.) at each atomic site, obtained from a distributed multipole analysis (DMA) of the wavefunction. Model DMAs can be constructed by transforming the atomic multipoles from the wavefunction of one conformer to represent the structure of another, or by transferring multipoles from a blocked single peptide to a polypeptide. The accuracy of the transferability assumptions are then tested by comparing the electrostatic potential around the molecule, as predicted by the models which assume transferability, with that calculated from the DMA of the entire molecule in the chosen conformation.
Our results show that large errors in the electrostatic potential outside the molecule can result from the assumption that the charge density is independent of the (ϕ, ψ) torsion angles. However, the different models agree well on the positions (although not the relative energies) of the probable water binding sites. This implies that there is some limited utility in the crude approximation that the charge density is independent of conformation. However, model charge distributions for blocked diglycine transferred from the DMAs of blocked single peptides with the same torsion angles are much more successful, and provide a promising route forward to accurate electrostatic models for polypeptides.