INHIBITORS OF GLYCOGEN PHOSPHORYLASE: IN SEARCH OF AN ANTIDIABETIC DRUG. K.A. Watson, M. Gregoriou and L.N. Johnson. Universtiy of Oxford, Laboratory of Molecular Biophysics, South Parks Road, Oxford, OX1 3QU U.K.

The aim of this work has been to design specific regulators of glycogen phosphorylase (GP) that shift the balance from glycogen degradation to glycogen synthesis. Such compounds have potential therapeutic interest in the treatment of non-insulin dependent (Type II) diabetes mellitus (NIDDM). Several compounds have been designed based on the three-dimensional rabbit muscle GP structure (Watson et al., Biochemistry 1994, 33, 5745-5758). Analysis of the crystallographically refined complexes has offered some understanding of the factors influencing sugar-protein interactions. To date, the best inhibitor, a glucopyranose analogue of hydantocidin, shows a 1000 fold improvement over the parent glucose moiety (Bichard et al.,Tetrahed. Lett. 1995, 36, 2145-2148). It has been shown, with an earlier analogue (showing two orders of magnitude improvement over glucose) that the compound is an effective regulator of liver glycogen metabolism and is considerably more effective than glucose (Board & Johnson, Biochem. J. 1995, 311, 845-852). The results indicate a positive hypoglyceaemic effect and suggest a potential use for these analogues in the treatment of Type II diabetes.

Both structure-based and quantitative structure-activity relationship (QSAR) drug design strategies have been used (Watson et al., Acta Crystallogr. 1995, D51, 458-472). The additional information from the QSAR approach provides a quantitative method for the design and prediction of new potential drug molecules. We have used the program combination GRID (Wade & Goodford, J. Med. Chem. 1993, 36, 140) and GOLPE (Baroni et al., Quant. Struct.-Act. Relat. 1993, 12, 9-20). The latest version of GOLPE (V3.0) has been adapted to use structural information of both the ligand and the protein molecules (traditional methods focus only on the ligand). To date, there is no QSAR methodology that simultaneously uses the inhibitor-macromolecular complex structures and biological data to predict the activity of new drug molecules. Our GP dataset, together with the latest program GOLPE, have provided a unique opportunity to explore such an approach.