The L-lactate dehydrogenase of Bacillus stearothermophilus (BSLDH) is a highly stereospecific enzyme that catalyzes synthetically useful reductions of 2-keto acids to the corresponding L-2-hydroxy acids. The strategy of probing the factors controlling enzyme stereospecificity by evaluating BSLDHs resistance to being induced to reverse its stereospecificity preference from L to D has been extended by the introduction of new site-directed mutations designed to induce the natural substrate pyruvate to bind in an alternative, 180-rotated, orientation that would lead to D-lactate production. With this "flipping" of the natural substrate-binding orientation as the goal, residue Ile240 was changed to Arg and Lys to introduce a new carboxyl-binding site, while an Argl71Tyr replacement removed the natural Arg171-COO--binding interaction, thereby creating the double mutants I240R/R171Y and I240K/R171Y. The validity of the strategy was demonstrated by significant, up to 2.3%, proportions of D-lactate produced by preparative-scale I240K/R171Y-catalyzed reductions of pyruvate. This represents an estimated similar to 10(5)-fold relaxation of the natural L-stereoselectivity of BSLDH, thereby demonstrating the feasibility of eventually applying rational protein engineering to optimize and control the stereoselectivity of enzymes used in asymmetric synthesis. Interestingly, modeling studies based on comparisons of the active site region of the D-stereospecific enzyme, glycerate dehydrogenase (GDH), with that of BSLDH reveal that in GDH, the position of the catalytically important substrate COO--binding Arg-residue approximates closely that of the Arg/Lys240 residues introduced into BSLDH. It thus appears that the strategy of changing stereospecificity by reversing 2-keto acid binding orientation adopted in the current study parallels the one employed by Nature to introduce L- and D-configurations in 2-keto acid reductions.