Directed evolution based on saturation mutagenesis at sites lining the binding pocket is ac ommonly practiced strategy for enhancing or inverting the stereoselectivity of enzymes for use in organic chemistry or biotechnology. However,a st he number of residues in ar andomization site increases to five or more,the screening effort for 95 %library coverage increases astronomically until it is no longer feasible. We propose the use of as ingle amino acid for saturation mutagenesis at superlarge randomization sites comprising 10 or more residues.When used to reshape the binding pocket of limonene epoxide hydrolase,t his strategy,w hichd rastically reduces the search space and thus the screening effort, resulted in R,R-and S,S-selective mutants for the hydrolytic desymmetrization of cyclohexene oxide and other epoxides.X-raycrystal structures and docking studies of the mutants unveiled the source of stereoselectivity and shed light on the mechanistic intricacies of this enzyme.Directed evolution is ap rotein-engineering technique for manipulating essentially any catalytic property of enzymes, [1] including their enantio-, diastereo-, and regioselectivity. [1b] Multiple cycles of gene mutagenesis,e xpression, and screening are generally required. Since the screening effort is the bottleneck of directed evolution, the design of higher-quality mutant libraries is crucial. In this endeavor, we developed the structure-based combinatorial active-site saturation test (CAST) in combination with iterative saturation mutagenesis (ISM) at sites lining the binding pocket of enzymes as am ethod for manipulating stereo-and regioselectivity. [1b,2] Nevertheless,t heoretical and practical issues remain unresolved. Since the degree of oversampling for 95 %l ibrary coverage [3] increases drastically as the number of residues of ag iven randomization site increases (see Table S1 in the Supporting Information), one can either settle for significantly less coverage [1b] and/or employ codon degeneracies encoding reduced amino acid alphabets,s uch as NDT (12 amino acids) [1b] or even smaller alphabets (5-7 amino acids), [1b,4] which require less screening. Thec hoice is currently unclear, especially when 10 or more residues surround the binding pocket:afrequently encountered situation.In the present curiosity-driven study,a10-residue site was simultaneously randomized by using the smallest possible amino acid alphabet composed of as ingle amino acid. In contrast to,f or example,N DT codon degeneracy, which would require approximately 210 11 transformants for 95 % library coverage,i nt his case only about 3000 transformants would be required, although diversity is greatly reduced. This approach is reminiscent of shotgun alanine scanning [5] and combinatorial alanine scanning [6] used to manipulate binding properties.O nly one example of the use of combinatorial alanine scanning in this way has been described:Itwas used in combination with point mutations evolved earlier and errorprone PCR to enlarge the substrate scope of P450-BM3. [...