(636b) Computational Screening for the Discovery of Small Molecule Regulators of Insulin Degrading Enzyme

Insulin degrading enzyme (IDE) is involved in the clearance of insulin of amyloid-β (Aβ) and insulin, which are peptides associated with Alzheimer's disease (AD) and diabetes respectively. In this study, our objective is to design novel regulators of IDE using structure-based drug design approach, which might facilitate the development of activators or inhibitors of IDE. This approach is a powerful method for designing inhibitors and/or activators with high specificity.

Three dimensional structure of IDE was obtained using X-Ray crystallography by Tang group at the University of Chicago. IDE is organized in two domains (IDE-N and IDE-C) that connect with each other by a flexible hinge. IDE can adopt at least two conformations; nominated as ?open? and ?closed?. The open state allows substrates to enter freely the catalytic cavity; however the closed (stable) state entraps the substrates inside to perform hydrolysis, and it is significant for the catalytic function. Additionally, N-terminal of the IDE substrates make contacts with a highly conserved exosite that is located ~30 Å away from the catalytic center prior to cleavage. In the study of Tang group, short peptide substrate bradykinin is found to be an activator of IDE by only binding to the exosite and not to the catalytic site, which suggested low affinity of bradykinin for IDE. This finding has been taken as the basis for the design of regulatory molecules for IDE activity. This might suggest that small chemical compounds that bind to the exosite would have the possible regulatory role in substrate binding and subsequent cleavage by IDE.

In this study, we applied both the computational docking methods and the experimental biological activity assays for the discovery of IDE regulators. The first step of the computational docking method involves minimization and equilibration of the crystal structure of IDE (PDB ID: 2G47), and molecular dynamics simulation was performed until the protein was stabilized. Second, virtual ligand screening with AutoDock molecular docking simulation program was performed by screening of a chemical library consisting of over 9 million compounds. Next, drug candidates were selected with respect to their binding energies and spatial conformations in exosite regions.

As a result, more than 100 small molecule drug candidates with lowest predicted binding free energies between -11 kcal/mol and -13 kcal/mol have been identified as potential hits which will be tested by using fluorogenic assay for the biological activity. Finally, the activity of the successful drug candidates will be characterized in a cellular assay to establish that they do not have cytotoxic effects.