Further efforts toward a molecular dynamics force field for simulations of peptides in 40% trifluoroethanol-water

A set of force field parameters for trifluoroethanol (TFE) was shown in earlier work from this lab to give a good accounting of system density, translational diffusion coefficients, and the solvent fluorine-solute hydrogen NMR cross-relaxation parameter (∑HF) for acetate dissolved in 40% TFE-water. It has since been found that this parameter set performs poorly when used to predict ∑HF for interactions of TFE with the hydrogens of an octapeptide, [val(5)]angiotensin. In the present work, adjustment of nonbonded force field parameters for interactions of fluorine with hydrogen was explored with the goal of improving these predictions. Six sets of TFE parameters were examined. When used in conjunction with the TIP5PE water model, all gave values for system density, translational diffusion coefficients, and ∑HF for acetate-TFE interactions that were similar to experimental results. Increasing the Lennard-Jones σHF for fluorine-hydrogen interactions by 10% led to calculated solvent-solute cross-relaxation parameters that are in better agreement with experiment for many of the peptide hydrogens. Changing the Lennard-Jones εHF parameters for the same interactions had little effect on calculated ∑HF values. There was no discernible influence of the TFE model used on the radius of gyration of the peptide or on the backbone conformational angles of the peptide, implying that the conformational properties of the peptide are not strongly influenced by changes in the force field description of TFE in 40% TFE-water. A recent parameter set for TFE proposed by Vymetal and Vondrasek (2014), which reproduces well various physical properties of neat TFE and TFE-water mixtures, was shown to predict cross-relaxation terms for this system poorly.