Design, synthesis, and quantitative structural activity relationships of carboxylate ligands for the HisB10 binding site of hexameric insulin and the use of dynamic nuclear magnetic resonance (DNMR) spectroscopy to probe the phenolic ligand binding mechan

The insulin hexamer is an allosteric protein that displays positive and negative cooperativity and half-site reactivity that is controlled at two different binding sites by homotropic and heterotropic binding interactions. These binding sites consist of two anion sites situated along the three-fold axis of symmetry in R3 units of T3R3 and R6 and phenolic ligand sites located at the dimer-dimer interface of all allosteric forms of the protein. In this thesis we show that carboxylate ligands can be designed computationally using molecular mechanics and dynamics techniques that bind with affinities greater than carboxylates previously reported. We also report the Quantitative Structure Activity Relationship (QSAR) between both computationally and experimentally derived descriptors and the experimentally determined binding affinity of various carboxylates. These analyses show that in addition to the molecular volume of the ligand, the polarizability of the electrons of the molecule and the basic nature of the carboxylate coordinating the Lewis acid Zn(II) control the affinity of the carboxylate for the HisB10 site. Finally, the binding mechanism of phenolic ligands was investigated through the use of 19F Dynamic NMR studies with the phenolic ligand alpha,alpha,alpha-Trifluoro-m-cresol. Ligand association and dissociation is not obvious when viewing X-ray crystal structures of R6 hexameric insulin phenolic ligand complexes. Amino acid residues cover the ligand when bound to the phenolic ligand binding site making discovery of a mechanism for dissociation structurally difficult. We determined in the liquid phase that the rate of exchange between ligand free in solution and ligand bound to protein is fast to moderately fast on the NMR timescale. At -15°C, the exchange slowed enough to allow deconvolution of the free and bound peaks whereby equilibrium constants, populations, and T2 relaxations were determined. These parameters were then used along with the program DNMR5 to fit theoretical lineshapes onto experimental lineshapes at varying concentrations of alpha,alpha,alpha-Trifluoro-m-cresol to determine binding off rates. These data along with previously reported dynamic and structural work were used to develop possible mechanistic schemes by with phenolic ligands can bind to the phenolic ligand binding sites on R-state hexameric insulin.