Prolyl-leucyl-glycinamide (PLG) is a unique endogenous peptide that modulates dopamine receptor subtypes of the D2 receptor family within the CNS. We seek to elucidate the structural basis and molecular mechanism by which PLG and its analogs modulate dopamine receptors, toward the development of new therapeutics to treat Parkinson's disease, tardive dyskinesia and schizophrenia. As a first step toward establishing a validated protocol for accurate computational modeling of PLG and associated peptidomimetic analogs, we evaluated the accuracy of density functional theory (DFT), wave function theory (WFT), and molecular mechanics (MM) calculations for PLG and for a library of structurally related small molecules. We first tested 12 local and nonlocal density functionals, Hartree-Fock (HF) theory, four "semiempirical" methods of the neglect of diatomic differential overlap (NDDO) type, and one self-consistent-charge nonorthogonal tight-binding (SCC-DFTB) method as implemented in two software suites, against coupled-cluster benchmark geometries for 4-methylthiazolidine, a small molecule that comprises key structural features present in our PLG-analog library. DFT and HF calculations were done with the MG3S augmented polarized triple-zeta basis set. We find that for 4-methylthiazolidine bond distances, DFT significantly outperforms NDDO, and both SCC-DFTB versions we evaluated perform worse than HF theory and are less accurate than 83% of the density functionals tested. The top five functionals for 4-methylthiazolidine were M05-2X, mPW1PW, B97-2, M06-2X, and PBEh, with mean unsigned errors (MUEs) in bond length of 0.0017, 0.0020, 0.0023, 0.0025 and 0.0027 Å, respectively. The widely used B3LYP functional ranked 11th out of 12 functionals evaluated, slightly below SCC-DFTB, and is significantly less accurate for 4-methylthiazolidine bond distances (MUE = 0.0095 Å) than the best local functional (M06-L, MUE = 0.0030 Å), which is far less computationally costly. Based on that initial analysis, we obtained new M05-2X benchmark geometric parameters for PLG and a library of 11 peptidomimetic derivatives, which we in turn used to examine the accuracy of thirty-four popular molecular mechanics (MM) force fields, four NDDO approaches, and SCC-DFTB for the full compound structures. Here, we found that ∼70% of the MM force fields tested superior to the best semiempirical and SCC-DFTB codings. Moreover, AMBER-type force fields proved most accurate among MM methods for this class of small-molecule peptidomimetics; the AMBER-type methods comprised eight out of the top 10 molecular mechanics options we tested.
Bibliographical noteFunding Information:
The authors express their appreciation to Donald G. Truhlar of the University of Minnesota, Department of Chemistry for helpful discussions. This work was supported in part by the University of Minnesota, Department of Medicinal Chemistry, by the Minnesota Supercomputing Institute for Advanced Computational Research, and by the National Institutes of Health under R01 NS020036 to R.L.J.
- Molecular mechanics