CYP106A2 from Bacillus megaterium ATCC 13368 is a bacterial cytochrome P450 that is capable of transforming steroid hormones. It can be easily expressed in Escherichia coli with a high yield. Its activity in vitro can be achieved by using the adrenal redox proteins adrenodoxin and adrenodoxin reductase. So far, it was not possible to crystallize CYP106A2 because of degradation during the crystallization process. Nevertheless, CYP106A2 is an interesting enzyme for biotechnological use. It hydroxylates pharmaceutically important steroids such as progesterone and 11-deoxycortisol. However, it will be necessary for efficient application of CYP106A2 in biotechnology to improve the hydroxylation activity and manipulate the regiospecificity. The present paper gives an overview of recent developments in protein engineering of CYP106A2.
Cytochromes P450 comprise a superfamily of related enzymes, which are responsible for the metabolism of a wide variety of hydrophobic compounds [1,2]. Cytochromes P450 are found nearly ubiquitously and are involved in numerous biological processes, which include the biosynthesis of lipids, steroids and antibiotics and the degradation of xenobiotics. Their ability to hydroxylate non-activated hydrocarbons is unique.
A characteristic property of prokaryotic cytochromes P450 is their solubility. Therefore prokaryotic P450s are easy to purify and to handle. To display functional activity they, however, as external mono-oxygenases need to interact with their soluble redox partners . In contrast, the eukaryotic P450s are, with some exceptions, integral membrane proteins. This makes the purification and handling of eukaryotic cytochromes P450 very difficult. Additionally, stability and activity are low compared with most of the prokaryotic P450s. The genus Bacillus contains some species with highly interesting cytochromes P450. Besides CYP106A2 that will be described in more detail in the present paper, P450BM-3 (CYP102A1) has been identified in Bacillus megaterium ATCC 14581. CYP102A1 was the first fusion protein found consisting of a cytochrome P450 reductase domain and a cytochrome P450 domain [4,5]. It functions as fatty acid hydroxylase [6,7]. Because of its self-sufficient catalytic properties, it is highly interesting for biotechnological processes . Meanwhile, two other catalytically self-sufficient P450s (CYP102A2 and CYP102A3) have been identified in Bacillus subtilis . Other members of the CYP102 family have been found, e.g. in Bacillus cereus and Bacillus anthracis .
The steroid 15β-hydroxylase (CYP106A2) from B. megaterium ATCC 13368 is another interesting prokaryotic cytochrome P450. The natural function of CYP106A2 in B. megaterium ATCC 13368 is still not known, but its ability to hydroxylate steroids makes this enzyme a candidate for industrial production of useful steroid intermediates. So far, most of the filamentous fungi are known to hydroxylate steroids . However, in most cases, the enzyme responsible for catalysis is not known yet. Only very few steroid hydroxylating bacteria have been described so far, among them Streptomyces roseochromogenes (CYP163A2) , Streptomyces griseus (CYP105D1)  and Arthrobacter simplex . On the other hand, multistep combinations of chemical and biochemical processes are used nowadays in the industrial production of steroids . Many steps can only be performed by the use of micro-organisms. Chemical synthesis of these steps would require for example additional chemical reactions to protect groups. This makes the chemical synthesis time-consuming and expensive. Furthermore, the used chemicals are often hazardous to health and environment. Micro-organisms or their isolated enzymes are thus a cheap and efficient alternative to chemical synthesis. Established processes in steroid-producing industry are 11α-, 11β-, 15α- and 16α-hydroxylations, which are realized by micro-organisms mainly for the production of adrenal cortex hormones and their analogues . Other important hydroxylation positions include the 9α, 7α and 7β positions. Those hydroxylated steroids are useful intermediates in the production of other pharmaceutically active compounds, or they are themselves biologically active. In this context, CYP106A2 is a very challenging tool for the development of new biotechnological processes for steroid production.
The CYP106A2 hydroxylase system
B. megaterium ATCC 13368 has a special position among all so far tested B. megaterium strains. When several Bacillus species (B. megaterium, B. licheniformis, B. pumilis, B. alvei, B. brevis, B. cereus, B. macerans and B. subtilis) were tested for the induction of cytochromes P450 by pentobarbital , neither constitutive nor pentobarbitural-inducible P450s occurred except in B. megaterium. In 11 of 12 B. megaterium strains, an induction of cytochromes P450 by pentobarbital has been observed. The strain without cytochrome P450 induction was B. megaterium ATCC 13368. However, the ability of 15β-hydroxylation of various steroids by this strain has been known since 1958 . Berg et al. [18–23] investigated intensively the hydroxylase system from B. megaterium ATCC 13368. They were able to show that it consists of the cytochrome P450, a strictly NADPH-dependent FMN-containing protein (megaredoxin reductase), and an iron–sulfur protein (megaredoxin). The cytochrome P450 from B. megaterium ATCC 13368 was later classified as CYP106A2. The sequence of CYP106A2 shows high identity (63%) to CYP106, the P450BM-1 from B. megaterium ATCC 14581 . CYP106A2 has a molecular mass of 47.5 kDa. The redox partners of CYP106A2 have not been cloned yet, but enzymatic activity can be obtained using the adrenal redox system consisting of Adx (adrenodoxin) and AdR (Adx reductase) [20,25]. CYP106A2 is also able to interact with the electron transfer system from B. subtilis  as well as with putidaredoxin reductase and putidaredoxin from Pseudomonas putida .
Steroid hydroxylation by CYP106A2
CYP106A2 was found to hydroxylate 3-oxo-Δ4-steroids in 15β-position, for example DOC (11-deoxycorticosterone), androstenedione and testosterone. Therefore CYP106A2 is also called 15β-hydroxylase. In contrast, 3β-hydroxy-Δ5-steroids will not be converted by CYP106A2 [19,21]. The side chain of the D-ring does not have an effect on the substrate–CYP106A2 interaction . Besides the 15β-hydroxylation, a hydroxylation in 6β-position of progesterone could be shown, too. Very recently, also hydroxylation of progesterone in 11α- and 9α-position was found [27,28]. Berg et al.  obtained a progesterone hydroxylation activity of 0.8 nmol of 15β-hydroxyprogesterone·(nmol of CYP106A2)−1·min−1 in an assay where they used the partially purified natural redox partners, megaredoxin and megaredoxin reductase. Using an in vitro system consisting of Adx, AdR and a NADPH-regenerating system, we could observe a 15β-hydroxyprogesterone production of 337.3±47.7 nmol·(nmol of CYP106A2)−1·min−1 . Interestingly, the observed specificity of hydroxylation in β-position can be altered by the choice of the electron transfer system. Replacing the natural electron transfer partners by peroxides, a ratio of 15α-/15β-hydroxylation of progesterone of up to 1.3 could be found . A change of the regiospecificity is also possible by replacement of active site residues of this cytochrome P450. This will be discussed in more detail below.
Development of a high-level expression system
To characterize the function of CYP106A2 in more detail and to be able to engineer CYP106A2 for biotechnological purpose, an efficient high-level expression system is needed. In the supernatant of B. megaterium ATCC 13368, only 8 pmol of CYP106A2/mg of protein could be observed . Rauschenbach et al.  therefore cloned the cDNA of CYP106A2. Heterologous expression of this cDNA in Escherichia coli gave a yield of 260 pmol of CYP106A2/mg of total protein. The expression yield could be increased to 4400 pmol of CYP106A2/mg of protein by the work of our group . Further improvements by optimization of the expression conditions lead to 8000 pmol of P450/mg of total protein . This confers an optimization of a factor of 1000 compared with the level in the original bacterial strain.
Substrate binding to CYP106A2
It has been demonstrated that substrate binding to cytochromes P450 causes a high-spin shift of the haem iron, resulting in a peak at approx. 390 nm and a trough at approx. 420 nm [30,31]. Spectroscopic studies of CYP106A2 lead to an unexpected result. The spectrum of the substrate-free, oxidized form shows absorption maxima at 417, 534 and 566 nm. After addition of an excess of DOC, the expected shift of the Soret band to 392 nm, which is characteristic for the high-spin state of the haem in a substrate-bound P450 , was not observed. This result is in contradiction to the observed conversion of DOC by CYP106A2 . The absorption spectra seem to indicate that DOC might not bind in the haem pocket of the P450. Therefore FTIR (Fourier-transform IR) measurements of the stretch vibration of the CO iron ligand have been performed. The spectrum of the reduced substrate-free CYP106A2 shows an apparent maximum at approx. 1943 cm−1 and a shoulder at approx. 1975 cm−1. The DOC-bound form shows shoulders at 1960 and 1975 cm−1. The apparent maximum shifts to approx. 1935 cm−1. Thus, using FTIR spectroscopic measurements, a clear proof of DOC binding to the active site of cytochrome P450 could be obtained.
Crystallization and stability of CYP106A2
Since CYP106A2 could be expressed to a very high level and purification of the protein from E. coli is easy to achieve, attempts to crystallize this steroid hydroxylase have been made. Unfortunately, so far, all crystallization attempts did not lead to CYP106A2 crystals. Analysis of the protein by SDS/PAGE revealed degradation of the protein during storage at 4°C for 4 weeks (Figure 1). An additional protein lane at approx. 43 kDa was visible. Partial N-terminal sequencing of this fragment showed a deletion of the first 72 amino acids. The sequence of the cleavage site could not be assigned to a known protease. To avoid cleavage of the protein at this site, we constructed a S72A/V73I and a Δ72 truncated mutant. The Δ72 mutant, which lacks the first 72 amino acids, was not expressed in E. coli at a detectable amount, suggesting that the truncated mutant cannot fold properly within the bacterial cell. It was, however, possible to express the S72A/V73I mutant, but also this mutant did not show a better stability in the crystallization process than the wild-type protein. Moreover, we tried to stabilize the enzyme by addition of the substrate DOC and the typical cytochrome P450 inhibitors imidazole and metyrapone. However, also by adding these ligands, it was not possible to prevent CYP106A2 from degradation in the crystallization process. Unfortunately, neither the natural function in B. megaterium ATCC 13368 nor the natural substrate of CYP106A2 is known so far, so that a better stabilization of this enzyme by the natural substrate, which was demonstrated for other cytochromes P450 , is not possible yet.
Fragmentation of CYP106A2
Modelling of CYP106A2 and site-directed mutagenesis of the protein
Since crystallization of CYP106A2 is still problematic, we constructed a model of CYP106A2 (Figure 2) (M. Lisurek, C. Virus, I. Antes, B. Simgen and R. Bernhardt, unpublished work). As template proteins for the modelling of the structurally conserved regions, five cytochromes P450 with known structure were used. Comparison of the postulated structure elements of CYP106A2 with the results of a secondary-structure prediction program showed a good conformance. The quality of the model was checked using PROCHECK , PROSA II  and WHATIF . All programs revealed a high quality of the CYP106A2 model. The substrate progesterone was docked into the model by using the program FLEXX. After prediction of the substrate recognition sites  in CYP106A2, we were able to choose promising residues for mutagenesis. Site-directed mutagenesis was, on the one hand, used to check the accuracy of the computer-derived model of CYP106A2 and, on the other hand, used to change the regiospecificity of steroid hydroxylation. Amino acid positions 395 and 397 have been chosen for site directed mutagenesis and changed to the corresponding amino acids of CYP11B1, the 11β-hydroxylase. Using progesterone as a substrate for CYP106A2, in fact, a shift of the regiospecificity of progesterone hydroxylation from the 15th to the 11th position was shown. Mutations of these positions lead to a dramatic increase (up to 11-fold) of the 11α-OH progesterone production compared with the 15β-OH progesterone production. The effects on the other two known products of the CYP106A2 wild-type, 9α- and 6β-OH progesterones, were less dramatic (M. Lisurek, C. Virus, I. Antes, B. Simgen and R. Bernhardt, unpublished work).
Computer model of CYP106A2
Development of whole-cell screening system and directed evolution of CYP106A2
Since effective engineering of biocatalysts, in general, requires application of several techniques, it was tempting to develop methods for directed evolution of the CYP106A2 system. Directed evolution needs a fast screening system with a high accuracy. The main drawback in the work with steroids is, however, the lack of fast analytics. We developed an easy and fast system for the screening of improved steroid hydroxylation by CYP106A2 . This system is based on a method described in  using the observation that steroids exhibit fluorescence in an acidic environment. The intensity of the fluorescence depends on the number of hydroxy groups. Hydroxylated steroids show higher fluorescence than non-hydroxylated steroids. To simplify the screening procedure, a whole-cell biocatalyst has been developed . As mentioned above, CYP106A2 requires for its activity an electron transfer system. By constructing two plasmids, one carrying the sequence of CYP106A2 or its mutants and the other one carrying the sequences of AdR and Adx, we could enable E. coli to act as whole-cell biocatalyst. After co-transformation of E. coli with both plasmids, E. coli is able to express the three proteins and to hydroxylate DOC, RSS (11-deoxycortisol) as well as progesterone. Applying only one round of mutagenesis by error-prone PCR and screening of the obtained mutants using the fluorescence assay, a CYP106A2 mutant with an approx. 2-fold increased activity towards RSS could be obtained (Figure 3). Screening towards the substrate progesterone also revealed mutants with higher activity and, furthermore, changed the product pattern towards di- and/or poly-hydroxylated progesterones.
Kinetic characterization of a CYP106A2 mutant
8th International Symposium on Cytochrome P450 Biodiversity and Biotechnology: Independent Meeting held at Swansea Medical School, Swansea, Wales, U.K., 23–27 July 2006. Organized and Edited by D. Kelly, D. Lamb and S. Kelly (Swansea, U.K.).