A new active site model for predicting and interpreting the stiuctural-and stereospecificity of horse liver alcohol dehydrogenase-catalyzed oxidations and reductions has been formulated from the specificity, X-ray, and kinetic data now available. The new model is based on a cubic-space system and is convenient and easy to use. The model is completely reliable and permits HLADH specificity towards all of its currently known substrates to be predicted with confidence. Its application is illustrated by analyses of representative examples of major substrate types.J. BRYAN JONES et IGNAC J. JAKOVAC. Can. J. Chem. 60, 19 (1982) A partir des donnees de specicite, de diffraction de rayons X et de cinetique disponibles. on a elabore un nouveau modele de site actif pour predire et interpreter la structure et la stereospecifite des oxydations et des reductions catalysee par la deshydrogenase de l'alcool du foie de cheval (DHAFC). Le nouveau modele est fonde sur un systeme d'espace cubique et est d'emploi facile. Le modele est completement fiable et permet de predire en toute confiance la specifite de la DHAFC envers tous ses substrats connus. On illustre son utilisation par I'analyse d'exemples representatifs des principaux types de substrats.[Traduit par le journal]The asymmetric synthetic opportunities provided by the chiral catalytic properties of enzymes are becoming increasingly recognized (2a,b). Alcohol dehydrogenases, which catalyze oxidoreductions of the type depicted in reaction are among the most useful enzymes for organic synthetic purposes, with the commercially available enzyme from horse liver the best documented.HLADH is an extremely versatile alcohol dehydrogenase, capable of effecting highly stereospecific oxidoreductions on a broad structural range of aldehyde, ketone, and alcohol substrates (2-6).In order for any enzyme to become widely adopted as a practical synthetic catalyst, it is essential that a reliable active site model be available for predicting the structural-and stereospecificity of the enzyme towards both known and potential substrate structures. Predictive systems of varying degrees of sophistication have been developed for a number of enzymes, including proteases (2a), mono-oxygenases (7, 8), and some alcohol dehydrogenases (2a, 3 , 4 , 9 ) . For HLADH, the diamond lattice section approach devised by Prelog ( 3 , and subsequently refined ( 2 a , 4 ) , has been very successful for interpreting the steric course of catalyses of cyclohexane and decalin substrates. However, for many of the other types of cyclic structures now known to be substrates ( 2 ) , stereochemical predictions based on diamond lattice analyses have become increasingly subjective and ~n c e r t a i n .~ It is clear that, after serving as a valuable active site model for nearly twenty years, the diamond lattice section has become totally inadequate for interpreting the stereochemical course of HLADH-catalyzed oxidoreductions of many of the recently identified, and still expanding, range of substrate structures (2, 6b)....