STRUCTURE OF HUMAN GLYOXALASE I, A ZINC ENZYME, SOLVED BY MIR METHODS. A.D. Cameron, B. Olin, M. Ridderström, B. Mannervik & T.A. Jones. Departments of Molecular Biology and Biochemistry, Uppsala University, BMC, Box 590, S-751 24, Uppsala, Sweden

The glyoxalase system catalyses the conversion of methyglyoxal into D-lactic acid using glutathione as coenzyme. It is found at all levels of evolution in the cytosol of cells and has been targetted for the development of anti-cancer drugs and anti-malarial agents. There are two enzymes involved in the pathway, glyoxalase I and glyoxalase II. Glyoxalase I catalyses the conversion of the hemithioacetal formed from the non-enzymatic reaction between methylglyoxal and reduced glutathione, into S-D-lactoylglutathione. Glyoxalase II, in turn catalyses the hydrolysis of the product of the first reaction to form D-lactic acid and regenerate the reduced glutathione.

We have solved the structure of human glyoxalase I by MIR coupled with four-fold non-crystallographic averaging and are currently refining the structure against data from an inhibitor complex extending to a resolution of 2.2Å. The enzyme is a dimer of molecular weight 50,000 and contains one zinc ion per monomer which is essential for activity. The inhibitor, benzyl glutathione, is clearly defined in the electron density. It is situated in the middle of an eight stranded beta-sandwich reminiscent, of the retinol binding proteins, and juxtaposed to the zinc ion. There is no structural homology with other glutathione binding sites. As expected from EPR and EXAFS studies there are four protein ligands to the zinc (His, Glu, Glu and Gln), and one water molecule which has been implicated in catalysis. What was completely unsuspected, however, is that two of the ligands to the zinc come from one monomer and two from the other. Indeed, the whole of the active site is formed only on dimerisation. Each monomer contains two structurally very similar domains. Of the eight strands making up the beta-sandwich, four strands are from the N-terminal domain of one monomer and four from the C-terminal domain of the other. What is most remarkable, however, is that if the N- and C- terminal domains are superimposed, the two zinc ligands of one domain overlap almost exactly with the those of the other domain. This has important implications for the evolution of the enzyme.