Systematic Analysis of Large Enzyme Families: Identification of Specificity- and Selectivity-Determining Hotspots

A general strategy is described that integrates data mining of enzyme families and molecular modeling of enzyme-substrate interactions to identify selectivity hotspots and to design focused variant libraries for changed regio- and stereoselectivity. This strategy is demonstrated for two case studies; the design of cytochrome P450 monooxygenases with improved regioselectivity and of thiamine diphosphate dependent enzymes with improved stereoselectivity. In both families, two selectivity hotspots are found in almost all sequences, and simple, generic rules are established to predict the effect of mutations at these positions on selectivity. The crucial role of the hotspot positions is validated for an increasing number of enzymes by designing variants with improved or switched selectivity.