The cleavage of the donor substrate D-xylulose 5-phosphate by wild-type and H263A mutant yeast transketolase was studied using enzyme kinetics and circular dichroism spectroscopy. The enzymes are able to catalyze the cleavage of donor substrates, the first half-reaction, even in the absence of any acceptor substrate yielding D-glyceraldehyde 3-phosphate as measured in the coupled optical test according to Kochetov (Kochetov, G. A. (1982) Methods Enzymol. 90, 209-223) and compared with the H263A variant. Overall, the H263A mutant enzyme is less active than the wild-type. However, an increase in the rate constant of the release of the enzyme-bound glycolyl moiety was observed and related to a stabilization of the "active glycolaldehyde" (alpha -carbanion) by histidine 263. Chemically synthesized DL-(alpha,beta -dihydroxyethyl)thiamin diphosphate is bound to wild-type transketolase with an apparent K-D of 4.3 +/- 0.8 muM (racemate) calculated from titration experiments using circular dichroism spectroscopy. Both enantiomers are cleaved by the enzyme at different rates. In contrast to the enzyme-generated alpha -carbanion of (alpha,beta -dihydroxyethyl)thiamin diphosphate formed by decarboxylation of hydroxylactylthiamin diphosphate after incubation of transketolase with beta -hydroxypyruvate, the synthesized DL-(alpha,beta -dihydroxyethyl)thiamin diphosphate did not work as donor substrate when erythrose 4-phosphate is used as acceptor substrate in the coupled enzymatic test according to Sprenger.