Enzymes catalyzing asymmetric carboligation reactions typically show very high substrate specificity for their nucleophilic donor substrate components. Structure-guided engineering of the thermostable transketolase from Geobacillus stearothermophilus by directed in vitro evolution yielded new enzyme variants that are able to utilize pyruvate and higher aliphatic homologues as nucleophilic components for acyl transfer instead of the natural polyhydroxylated ketose phosphates or hydroxypyruvate. The single mutant H102T proved the best hit toward 3-methyl-2-oxobutyrate as donor, while the double variant H102L/H474S showed highest catalytic efficiency toward pyruvate as donor. The latter variant was able to complement the auxotrophic deficiency of Escherichia coli cells arising from a deletion of the dxs gene, which encodes for activity of the first committed step into the terpenoid biosynthesis, offering the chance to employ a growth selection test for further enzyme optimization.

Transketolase variants were engineered to utilize arylalkanals and benzaldehyde as substrates with up to 28-fold rate acceleration for C–C bond formation with good yields (50–73%) and virtually complete (3S)-stereoselectivity (>99% ee).