Potential-Energy and Free-Energy Surfaces of Glycyl-Phenylalanyl-Alanine (GFA) Tripeptide: Experiment and Theory (original) (raw)

2008, Chemistry - A European Journal

The free energy surface (FES) of glycyl-phenyalanyl-alanine (GFA) tripeptide was explored with molecular dynamic simulations in combination with high level correlated ab initio quantum chemical calculations. The molecular dynamics employed the tight-binding DFT method, accounting for the dispersion energy, instead of the AMBER empirical force field, which yielded inaccurate results. We classified the minima, localized in the FESs, according to two different criteria namely: (a) the 2 backbone conformational arrangement and its resemblance to the secondary structure of proteins (families β L , 3 11 , γ and γ -3 11 ); and (b) the existence of a COOH···O=C intramolecular H-bond between the hydrogen of the terminus carboxyl group and the backbone CO of residue i+1 (families CO 2 H free and CO 2 H bonded ). Comparison with experiment shows that the theoretically predicted most stable minima in the FES correspond to the observed experimentally structures and the theoretically scaled frequencies match reasonably well those measured spectroscopically. Remarkably, however, we do not experimentally observe the CO 2 H bonded family, although its stability is comparable to that of the CO 2 H free structures. Motivated by this result we reinvestigated the FES of GFA with a completely different method, metadynamics, which includes the anharmonic effects. This is the first combination of the metadynamics approach with the tight-binding DFT-D procedure. Metadynamics confirms the existence and comparable stability of the two families of structures. The fact that we do not observe structures of the CO 2 H bonded experimentally was explained by their short excited state lifetime. Additionally, we also carried out ab initio calculations using DFT theory either in its augmented version by dispersion interaction (DFT-D) or by using the M06-2X functional. The importance of the dispersion energy in stabilizing peptide conformers is well reflected by our pioneer analysis using the DF-DFT-SAPT method on the nature of the backbone/side chain interactions. compensation of errors; the effect of improving the basis set at the MP2 level is compensated by the neglected higher-order correlation effects. We performed RI-MP2/cc-pVQZ 40 //RI-MP2/cc-pVTZ single-point calculations on these geometries and used the extrapolation scheme of Helgaker and co-workers 41 in order to obtain complete basis set (CBS) limit energies (MP2 CBS ). Additionally we added higher-order contributions to the correlation energy beyond the second perturbation order, the MP2 CBS energies, by calculating the difference between CCSD(T) and MP2 relative energies (CCSD(T)-MP2) determined with the 6-31G*(0.25) basis set. This correction term is known to be essentially independent of the basis set size, contrary to the MP2 and CCSD(T) energies themselves. 42 We computed theoretical infrared (IR) spectra only for the conformers calculated at the highest level of theory, i.e. CCSD(T)/CBS. We employed scaled harmonic frequencies for the calculation of zero-point vibrational energies (ZPVE), enthalpies, entropies and Gibbs energies (T = 300K) in the context of RR-HO-IG approximation. The scaling factors 43 employed were 0.958, 0.951 and 0.956