Free energy calculations on the leucine zipper domain of yeast transcription factor GCN4 using molecular dynamics

Molecular dynamics (MD) simulations under physiological conditions [explicit water, a 0.15 M sodium chloride concentration, no constraints, and a 14-A distance cutoff on non-bonded interactions at constant pressure and temperature (CPT)] using CHARMM were applied to both the native parallel and non-native antiparallel alignments of GCN4-p1, a 33 amino acid leucine zipper peptide derived from a DNA-binding protein, using continuum models. GCN4-p1, forms a parallel left-handed supercoil composed of two a-helices under physiological conditions. The non-native antiparallel orientation was provided through a mapping technique using Seryl tRNA Synthetase (STS) helix 3 and 4. The purpose of such a calculation was to determine individual energetic contributions, hydrophobic and hydrophilic, to the stability of parallel vs antiparallel alignments as a measure of structural preference. Structural analysis revealed the presence of similar polar and non-polar interactions in the hydrophobic core of the dimer (leu, val and asn) for both orientations of the leucine zipper. However, a critical hydrogen bonding interaction between Asn16 and Asn16' was absent in the antiparallel core. Free Energy of Solvation calculations (Gsolute and G solvent) revealed that the simulated native alignment of GCN4-p1 was more stable (∼200 kcal/mol) than the simulated non-native antiparallel structure. This value was influenced by the intrahelical hydrophilic energy of the solute possibly arriving from the "salt bridges", which confer specificity of alignment, present in the parallel rather than the antiparallel orientation. All other solute-solute interaction energies were comparable for both alignments. The solute-solvent hydrophilic energy significantly favored the non-native antiparallel superhelix (∼100 kcal/mol), but this did not outweigh the solute-solute hydrophilic contributing energy. Electrostatic interactions were the overriding factors, which determined the alignment preference of the DNA-binding protein GCN4-p1, in accord with work published by Oakley and Kim (1998). A parallel investigation on Acanthamoeba myosin II wildtype aimed to provide a theoretical assessment of the reasons behind the loss of coiled-coil conformation of the dimer segment of wildtype myosin II after phosphorylation of the 3 serine residues at the tip of the tail of each heavy chain. Molecular dynamics simulations at 300 K for 200 ps using the Langevin algorithm were carried out for five parallel dimers of myosin II with a variety of adjusted realignment structures of helix B with respect to helix A of the coiled-coil (3.5, 7, 10.5, and 14 amino acids) with subsequent energetic and structural analysis. The energetic approach demonstrated that the 3.5 residue shift was an energetically favored realignment over all other shifted alignments, as well as the native conformation, by a solvation free energy (SFE) of approximately 1kcal/mol. The structural approach revealed torsion angle values for heptad repeats within the range of expected values, therefore no irregularities in the helix-to-helix packing of all alignments.