Structure-based directed evolution of a monomeric triosephosphate isomerase: toward a pentose sugar isomerase.

Through structure-based and directed evolution approaches, a new catalytic activity has been established on the (β/α)8 barrel enzyme triosephosphate isomerase (TIM). This work started from ml8bTIM, a monomeric variant of TIM, in which the phosphate-binding loop (loop-8) had been shortened. Structure analysis suggested an additional point mutation (V233A), converting ml8bTIM into A-TIM. A-TIM has no detectable TIM activity, but it binds the TIM transition state analog, 2-phosphoglycollate. In an in vivo selection approach, we aimed at transferring the activity of three sugar isomerases (L-arabinose isomerase (L-AI), D-xylose isomerase A (D-XI) and D-ribose-5-phosphate isomerase (D-RPI)) onto A-TIM. Escherichia coli knockout variants were constructed, lacking E. coli L-AI, D-XI and D-RPI activities, respectively. Through a systematic approach, new A-TIM variants were obtained only from selection experiments with the L-AI knockout strain. Selection for D-RPI activity was impossible because of an impaired strain due to the gene knockouts. The selection for D-XI activity was unsuccessful, showing the importance of the starting protein for obtaining new biocatalytic properties. The L-AI-directed evolution experiments show that A-TIM already has residual in vivo L-AI activity. Most of the mutations providing A-TIM with enhanced L-AI activity are located in the loops between β-strands and the subsequent α-helices.