Most bacterial cytochrome P450 monooxygenases (P450s or CYPs) require two redox partner proteins for activity. To reduce complexity of the redox chain, the Bacillus subtilis flavodoxin YkuN (Y) was fused to the Escherichia coli flavodoxin reductase Fpr (R), and activity was tuned by placing flexible (GGGGS)n or rigid ([E/L]PPPP)n linkers (n = 1–5) in between. P-linker constructs typically outperformed their G-linker counterparts, with superior performance of YR-P5, which carries linker ([E/L]PPPP)5. Molecular dynamics simulations demonstrated that ([E/L]PPPP)n linkers are intrinsically rigid, whereas (GGGGS)n linkers are highly flexible and biochemical experiments suggest a higher degree of separation between the fusion partners in case of long rigid P-linkers. The catalytic properties of the individual redox partners were best preserved in the YR-P5 construct. In comparison to the separate redox partners, YR-P5 exhibited attenuated rates of NADPH oxidation and heme iron (III) reduction, while coupling efficiency was improved (28% vs. 49% coupling with B. subtilis CYP109B1, and 44% vs. 50% with Thermobifida fusca CYP154E1). In addition, YR-P5 supported monooxygenase activity of the CYP106A2 from Bacillus megaterium and bovine CYP21A2. The versatile YR-P5 may serve as a non-physiological electron transfer system for exploitation of the catalytic potential of other P450s.

Natural redox partners of bacterial cytochrome P450s (P450s) are mostly unknown. Therefore, substrate conversions are performed with heterologous redox partners; in the case of CYP106A2 from Bacillus megaterium ATCC 13368, bovine adrenodoxin (Adx) and adrenodoxin reductase (AdR). Our aim was to optimize the redox system for CYP106A2 for improved product formation by testing 11 different combinations of redox partners. We found that electron transfer protein 1(516–618) showed the highest yield of the main product, 15β-hydroxyprogesterone, and, furthermore, produced a reduced amount of unwanted polyhydroxylated side products. Molecular protein–protein docking indicated that this is caused by subtle structural changes leading to alternative binding modes of both redox enzymes. Stopped-flow measurements analyzing the CYP106A2 reduction and showing substantial differences in the apparent rate constants supported this conclusion. The study provides for the first time to our knowledge rational explanations for differences in product patterns of a cytochrome P450 caused by difference in the binding mode of the redox partners.

Bacterial P450s have considerable potential for biotechnological applications. The P450 CYP106A2 from Bacillus megaterium ATCC 13368 converts progesterone to several hydroxylated products that are important precursors for pharmaceutical substances. As high yields of monohydroxylated products are required for biotechnological processes, improving this conversion is of considerable interest. It has previously been shown that the binding mode of the redox partner can affect the selectivity of the progesterone hydroxylation, being more stringent in case of the Etp1 compared with Adx(4–108). Therefore, in this study we aimed to improve hydroxylation selectivity by optimizing the binding of Adx(4–108) with CYP106A2 allowing for a shorter distance between both redox centers. To change the putative binding interface of Adx(4–108) with CYP106A2, molecular docking was used to choose mutation sites for alteration. Mutants at positions Y82 and P108 of Adx were produced and investigated, and confirmed our hypothesis. Protein–protein docking, as well as conversion studies, using the mutants demonstrated that the iron–sulfur(FeS) cluster/heme distance diminished significantly, which subsequently led to an approximately 2.5‐fold increase in 15β‐hydroxyprogesterone, the main product of progesterone conversion by CYP106A2.

No abstract