The energetics and dynamics of ligand binding such as found in myoglobin (CO, NO to Mb) or to insulin (glucose to insulin) constitute an active area of research in our group.
Despite its central role in glucose regulation in the human body, there is relatively little theoretical knowledge about ligand bindin and dynamics of insulins. From an experimental viewpoint much more is known, in particular from mutation studies. A recent NMR NOESY and fluorescence quenching study of monomeric insulin in the presence of low concentration of D-glucose suggested that there exists a binding site with specific interactions between the bound glucose and certain insulin side chains. Although the detailed structure was not determined, the authors suggested that the proton resonances involved in this interaction correspond to methyl protons of insulin and can be attributed to surface residue(s), probably Val, Ile or Leu. However, no specific assignments of the resonances were made. By molecular dynamics simulation of the free insulin dimer (although the experiments were done with the monomer), followed by manual docking of D-glucose on the insulin surface, the authors proposed that Val(B2) and Leu(B17) residues are good candidates for the observed NOEs with the D-glucose protons. Fluorescence studies of the system led to an estimate of the binding free energy of about -2.0 +/- 0.5 kcal/mol. However, the authors also suggested that there might exist multiple binding sites which operate in a synergistic way.
We theoretically investigated  putative binding sites of D-glucose to insulin have using two docking programs for small molecules: Multiple Copy Simultaneous Search (MCSS) and Solvation Energy for Exhaustive Docking (SEED). The MCSS resulting configurations were investigated with a scoring function initially developed to estimate the binding free energy of HIV-1 protease inhibitors. SEED calculations were performed using various values for the dielectric constant of the solute. It is found that scores emphasizing non polar interactions give preferential binding sites at locations also inferred from experiment. Molecular dynamics simulations were used to study the stability of the proposed binding sites. During the length of the simulations the D-glucose is stable in the binding pockets found by the docking programs. The theoretical results are compared with NMR studies and show that the same binding sites are favored. Glucose motions during molecular dynamics simulations are correlated to the motions of the insulin side chains that are in contact and also to larger scale motions of insulin. These correlated motions are also found by normal mode analysis.