Comparison of random mutagenesis and semi-rational designed libraries for improved cytochrome P450 BM3-catalyzed hydroxylation of small alkanes

Three semi-rational approaches, combinatorial site-saturation mutagenesis (CSSM) using a reduced amino acid set and two libraries based on C(orbit) and CRAM computational design algorithms targeting up to 10 active site residues, were used to engineer cytochrome P450 BM3 to demethylate dimethyl ether and hydroxylate propane and ethane. These small libraries (343-1028 variants) were all enriched with respect to the fraction functional and maximal activities compared with a random mutagenesis library and individual site-saturation libraries targeting the same residues. Despite high average amino acid substitution levels of 2.6, 5 and 7.5, the CSSM, C(orbit) and CRAM libraries had at least 75% of library members properly folded. Propane- and ethane-hydroxylating P450 BM3 variants were identified using all three mutagenesis approaches, with as few as two amino acid substitutions. The library designed using the CRAM algorithm, which sought to reduce the size of the binding pocket, produced both a higher number of active variants and variants supporting the greatest number of catalytic turnovers. The most active variant E32 supports 16 800 propane turnovers at 36% coupling, which rivals the activity of variants obtained after 10-12 rounds of directed evolution using random and site-saturation mutagenesis. None of the variants in this study achieved the complete re-specialization for propane hydroxylation (including 93% coupling) previously obtained via multiple rounds of mutagenesis and screening. However, these semi-rational approaches allowed for large jumps in sequence space to variants with the desired functions.