CONFORMATIONAL-ANALYSIS OF A HIGHLY POTENT, CONSTRAINED GONADOTROPIN-RELEASING-HORMONE ANTAGONIST .2. MOLECULAR-DYNAMICS SIMULATIONS

Molecular dynamics simulations in vacuo and in solvent have been used, in combination with the NMR data presented in the preceding paper in this issue, to analyze the conformational behavior of a highly potent antagonist of gonadotropin-releasing hormone (GnRH). An initial conformational search in vacuo yielded two fundamentally different classes of structures that differ in the location of the tail formed by residues 1-3, above or below the cyclic part of the molecule. NMR restraints were applied progressively on both families of structures, leading to a consistent conformational model that confirms and refines the interpretation of the experimental data. The restraints force the orientation of the tail above the ring and induce a beta-hairpin structure in residues 5-8, as expected from the NMR analysis. The simulations support the presence of a gamma-turn around D-Trp3 and indicate the presence of high mobility in residues 1 and 2. Two different conformational equilibria have been characterized in the Asp4-Dpr10 bridge. Frequent contacts between the tail, Tyr5, and Arg8 indicate new bridging constraints to obtain more rigid GnRH antagonists. The biological activity of some bicyclic GnRH analogues that include these constraints supports our refined model for the bioactive conformation of GnRH. Our molecular dynamics results show that only a careful choice of NMR restraints, and their continuous evaluation, can lead to reliable structures when considerable flexibility exists in the molecule. In vacuo, large energy differences are observed between structures with different aromatic side chain rotameric states. We find that conformations that should be visited frequently in our experimental conditions can have energies more than 30 kcal/mol above the lowest energy conformation found in vacuo. These energy differences are mainly due to nonbonding interactions that highly favor compact structures. Simulations carried out in a solvent bath can overcome this problem and yield improved structures, with a lower tendency to form intramolecular hydrogen bonds.