H₂O₂-Driven Hydroxylation of Steroids Catalyzed by Cytochrome P450 CYP105D18: Exploration of the Substrate Access Channel

ABSTRACT

CYP105D18 supports H₂O₂ as an oxygen surrogate for catalysis well and shows high H₂O₂ resistance capacity. We report the hydroxylation of different steroids using H₂O₂ as a cosubstrate. Testosterone was regiospecifically hydroxylated to 2β-hydroxytestosterone. Based on the experimental data and molecular docking, we predicted that hydroxylation of methyl testosterone and nandrolone would occur at position 2 in the A-ring, while hydroxylation of androstenedione and adrenosterone was predicted to occur in the B-ring. Further, structure-guided rational design of the substrate access channel was performed with the mutagenesis of residues S63, R82, and F184. Among the mutants, S63A showed a marked decrease in product formation, while F184A showed a significant increase in product formation in testosterone, nandrolone, methyl testosterone, androstenedione, and adrenosterone. The catalytic efficiency (k_cat/K_m) toward testosterone was increased 1.36-fold in the F184A mutant over that in the wild-type enzyme. These findings might facilitate the potential use of CYP105D18 and further engineering to establish the basis of biotechnological applications.

IMPORTANCE The structural modification of steroids is a challenging chemical reaction. Modifying the core ring and the side chain improves the biological activity of steroids. In particular, bacterial cytochrome P450s are used as promiscuous enzymes for the activation of nonreactive carbons of steroids. In the present work, we reported the H₂O₂-mediated hydroxylation of steroids by CYP105D18, which also overcomes the use of expensive cofactors. Further, exploring the substrate access channel and modifying the bulky amino acid F184A increase substrate conversion while modifying the substrate recognizing amino acid S63 markedly decreases product formation. Exploring the substrate access channel and the rational design of CYP105D18 can improve the substrate conversion, which facilitates the engineering of P450s for industrial application.

INTRODUCTION

Steroid derivatives are the most prescribed drugs after antibiotics, and they are widely used to treat inflammation, cancers, and allergy and are used for contraception (1, 2). The biological activity of steroids is well differentiated based on the C-17 side chain groups and oxygenation of the core rings. The oxy-functionalization of the nonactivated carbon of the steroid core is a challenging chemical reaction. Cytochrome P450 (CYP or P450)-mediated hydroxylation of steroids is one of the crucial tools for their modification and functional diversification (2). Hydroxylated steroids improve their biological activity over that of their nonhydroxylated counterparts due to their increase in polarity (3).

CYPs are a superfamily of heme-containing monooxygenase proteins. They catalyze multiple reactions that include the reduction, dehydration, N-oxidation, epoxidation, dealkylation, desaturation, formation, and cleavage of the C–C bond, among others (4–6). Given the ability of chemical modification to alter the properties of chemical derivatives, CYPs have attracted the attention of many investigators, and it is expected that their potency will expand the range of applications in biomedicine and biotechnology (7). For example, the 11-hydroxylation of a steroid by CYPs from fungi in the 1950s was the first industrial biotransformation in the synthesis of steroid drugs. Further, mutation in CYP (encoded by pahA) from Penicillium chrysogenum was discovered to be responsible for the overproduction of penicillin by reducing the degradation of phenylacetate, which is a precursor for penicillin G (8). Many specific and nonspecific peroxygenase reactions catalyzed by CYPs utilize H₂O₂ as a cosubstrate. The investigation and industrial adaptation of H₂O₂-dependent reactions, particularly by prokaryotic P450s, are of clinical concern for drug modification (9). Bacterial CYPs from the CYP152 family are mostly investigated for H₂O₂-mediated reactions for fatty acid modifications (10); however, CYP154C8, CYP116B5, and CYP105D18 are also reported for peroxygenase activity (11–13). The use of H₂O₂ may be economically viable with some P450s, but it also leads to heme oxidation and enzyme inactivation, thus resulting in poor catalysis (10, 11). CYP105D18 showed high H₂O₂ tolerance and resulted in high turnover prior to heme oxidation (13).

There have been extensive studies with human and other eukaryotic P450s for the hydroxylation of steroids. The main findings are that steroidogenic P450s, such as CYP11B1, CYP11B2, CYP17A1, CYP19A1, and CYP27A1, are involved in steroid metabolism (14). The major human drug-metabolizing P450s, including CYP2C, CYP2D, and CYP3A subfamilies, are well defined for the hydroxylation of different steroids (15). For example, CYP3A4 can catalyze the monohydroxylation of testosterone at 10 different positions in human liver and intestine (16). Even though the numbers of human CYPs are characterized for steroid metabolism, their association with a separate NADPH-cytochrome P450 reductase (CPR) and general membrane-bound characteristics limit good expression and solubility (17). Bacterial CYPs are mainly cytosolic, and they are well expressed in the heterologous host, thus making them good candidates for industrial applications (18). The bacterial CYP105 family (CYP105A1, CYP105D5, and CYP105D7), CYP106 family (CYP106A1, CYP106A2, and CYP106A6), CYP107 family (CYP107D1 and CYP107E1), and CYP154 family (CYP154C2, CYP154C3, CYP154C4, CYP154C5, and CYP154C8) have been extensively explored in terms of steroid hydroxylation (2, 12, 16, 18–20). CYPs from different bacterial species have been well defined for the specific position hydroxylation of steroids. CYP106A1 and CYP106A2 from Sorangium cellulosum can hydroxylate steroids at positions 6β, 7β, 9α, and 15β, while CYP260B1 from the same species acts as 9-hydroxylase for 11-deoxycorticosterone (18, 21). CYP109B1 has been reported to have excellent activity and selectivity for the 15β-hydroxylation of testosterone derivatives (22), whereas CYP109E1 can hydroxylate testosterone at position 16β (23). Moreover, a few studies have suggested the diversification of fungal CYPs for steroid biotransformation in terms of species, family, and subfamily (24, 25). The CYP5311B family from Absidia sp. leads to the 11α- and 11β-position hydroxylation of steroids (25, 26). Regarding CYPs from the CYP5150 family, CYP5150AP3 enzyme possessed steroidal 7-hydroxylase and CYP5150AN1 catalyzed the 2-hydroxylation of 11-deoxycortisol whereas CYP5150AP2 possessed 19- and 11-hydroxylase activities (27). While various filamentous fungi, Aspergillus niger, Aspergillus ochraceus, Absidia coerulea, Absidia orchidis, Rhizopus nigricans, and Curvularia lunata, have been reported in terms of steroid hydroxylation, the molecular aspects of the reaction mechanism are poorly understood (28).

The efficient industrial applications of bacterial CYPs typically suffer from narrow substrate range, low region selectivity, and/or poor conversion rate. The recent laboratory evolution and rational design of CYPs are playing crucial roles to improve their activities, stability, and regioselectivity and even expanding the substrate scope (2, 29–31). Some approaches, like random mutagenesis, single target mutagenesis, alanine scanning, and knowledge-guided direct evolution based on saturation mutagenesis, have been developed for the engineering of some bacterial CYPs to improve steroid hydroxylation (2, 31, 32).

In this present study, we screened CYP105D18 for the hydroxylation of different steroids through in vitro reactions and molecular docking analysis. We reported high H₂O₂ resistance capacity and peroxygenase activity for steroid hydroxylation by CYP105D18. In addition, the structure-guided site-directed mutagenesis to the substrate access channel improved the product formation rate for 3-keto-4-ene steroids without a side chain at C-17.

RESULTS

Substrate screening.

The purified single band of CYP105D18 was obtained and characterized by CO reduction assay (see Fig. S1B and C in the supplemental material). The purified enzyme had an Rz value of 1.45, thus indicating high purity. The steroids were divided into two groups, as listed in Scheme S1 in the supplemental material. 3-Keto-4-ene steroids with no side chain at the 17′ position were listed as group A. 3-Keto-4-ene steroids with side chains at the 17′ position (progesterone, corticosterone, and cortisone), 3-keto-1,4-diene steroids with a side chain at the 17′ position (prednisone, prednisolone, and, dexamethasone), and the 3β-ol-5-ene steroid with a side chain at the 17′ position (pregnenolone) are placed in group B. Initially, all the steroids were screened for in vitro biotransformation by CYP105D18 using putidaredoxin reductase (Pdr) and putidaredoxin (Pdx) (purified from the P450cam system) (Fig. S1A) and ferredoxin (Fdx)/ferredoxin reductase (Fdr) (commercial) as redox partners. No product formation by CYP105D18 was observed in any of the steroids using the Pdr/Pdx and Fdr/Fdx systems. CYP105D18 was well supported by H₂O₂ for the N-oxidation of papaverine in our previous experiment (13). Therefore, we screened for the in vitro biotransformation of all the listed steroids by using H₂O₂ as an oxygen surrogate. Notably, all the steroid derivatives from group A were transformed to the hydroxylated product by using H₂O₂ as an oxygen surrogate (Scheme S1). CYP105D18 was unable to transform all the derivatives with an aliphatic side chain at the 17′ position (group B), even by using an oxygen surrogate.

H₂O₂ optimization and testosterone bioconversion.

The results of an experimental approach and the structural feature of CYP105D18 show high H₂O₂ resistance capacity (13). The incubation of protein in 200 mM H₂O₂ shows a very low heme dissociation constant k (<0.3 min⁻¹) compared to other peroxygenase CYPs (13). We optimized testosterone conversion by using different concentrations of H₂O₂. The heme oxidation rate was very low (0.014 min⁻¹) (Fig. 1A), while the highest product conversion was achieved with 40 mM H₂O₂ (Fig. 1B). Therefore, all the in vitro reactions were initiated with 40 mM H₂O₂. Further, a time-dependent assay and kinetic analysis for hydroxyl testosterone were performed using 40 mM H₂O₂. There was a single product peak at the retention time (t_R) of 13.4 min of the high-pressure liquid chromatography (HPLC) chromatogram (Fig. S2A), and liquid chromatography-mass spectrometry (LC-MS) spectral analysis showed the monohydroxylated mass of the product peak (Fig. S3). The nuclear magnetic resonance (NMR) analysis of the purified product results in 2β-OH testosterone (Fig. S4).

FIG 1.

Analysis of the H₂O₂ tolerance capacity and optimum condition for CYP105D18. (A) Heme degradation pattern of CYP105D18 in the presence of 40 mM H₂O₂ for 30 min. k is the heme oxidation rate constant. (B) Percent conversion of 2β-hydroxytestosterone using different concentrations of H₂O₂. All experiments were conducted using 2 μM CYP105D18 and 200 μM testosterone at 30°C for 1 h. The reactions were conducted in triplicate under similar conditions. S, substrate; P, product.

Bioconversion of other steroids.

Methyl testosterone and nandrolone were catalyzed to a single monohydroxylated product, while adrenosterone was catalyzed to two monohydroxylated derivatives. Androstenedione was converted to a single monohydroxylated product (P3) and two double-hydroxylated products (P1 and P2) (Fig. S3). Methylated testosterone was transformed by CYP105D18, with the product at a 14.2-min retention time showing the exact monohydroxylated mass of 319.22 m/z (Fig. S2B and Fig. S3). Further, the HPLC chromatogram (t_R, 9.9 min) and LC-MS analysis confirmed the monohydroxylated product of mass 291.19 m/z of nandrolone by CYP105D18 (Fig. S2C and Fig. S3). The adrenosterone was transformed to two monohydroxylated products at a t_R of 9.5 min (P1) and t_R of 12.4 min (P2) (Fig. S2D and Fig. S3), while androstenedione showed double-hydroxylated product P1 (t_R, 13.2 min) and two single-hydroxylated products P2 (t_R, 14.3 min) and P3 (t_R, 14.5 min) (Fig. S2E and Fig. S3). Further, molecular docking analysis was performed to predict product formation for methyltestosterone, nandrolone, adrenosterone, and androstenedione (Fig. S5).

Molecular docking analysis for product prediction.

The results of HPLC and NMR analysis confirmed the 2β-hydroxylated product for testosterone. The possible products formed by CYP105D18 for all the other steroids were predicted based on the molecular docking analysis and compared to the results obtained using testosterone as a substrate. Molecular docking was performed to observe the possible orientation of steroid substrates (group A) to the active site that has been hydroxylated by CYP105D18. Testosterone, methyltestosterone, nandrolone, adrenosterone, and androstenedione were accommodated in the active site surrounded by L87, A235, T239, I281, A282, V286, I386, T385, and I386 residues. The A-ring of testosterone, methyltestosterone, and nandrolone was closely situated toward the iron of the heme group. The binding energy for testosterone was −9.23 kcal/mol, and the distance between iron and C-2 was 3.7 Å, indicating suitability as a site for hydroxylation, as suggested by the experimental (HPLC and NMR) data (Fig. 2C). Similarly, the distances between C-2 of the A-ring and iron for methyltestosterone and nandrolone were 4.2 and 4.0 Å, respectively, representing the possible hydroxylation at position 2 in the A-ring (Fig. S5). The binding energies for adrenosterone (−9.35 kcal/mol) and androstenedione (−9.27 kcal/mol) were both high. Adrenosterone and androstenedione were positioned in the active site pocket facing the B-ring toward the iron. For adrenosterone, the distance between C-6 and iron was 3.2 Å and, that between C-7 and iron was 3.6 Å, while for androstenedione, the distance between C-6 and iron was 3.2 Å and that between C-7 and iron was 3.7 Å. (Fig. S5). This indicates that the possible sites for hydroxylation by CYP105D18 for adrenosterone and androstenedione are C-6 and C-7, respectively. The HPLC chromatogram also showed two hydroxylated products, as was suggested by the molecular docking analysis.

FIG 2.

Rational design toward the substrate entry channel. (A) Superposition of CYP105D18 (7DLS) with CYP105D7 (4UBS). The aligned sticks represent the residues of CYP105D18 corresponding to the conserved arginine residue of CYP105D7 toward the substrate entry channel. Sky blue, 4UBS; red, 7DLS. (B) Closeup view showing the active site with the amino acid residues involved in the substrate access channel. Testosterone is colored cyan, heme is colored red, F184 is colored green, and S63 is colored pink. (C) Molecular docking of testosterone with CYP105D18 showing the accommodation of the substrate. Active site residues are represented by colored sticks in zoom form, and testosterone is shown in yellow. The distance between heme and C-2 was calculated.

Exploration and rational design of the substrate access channel.

We solved the substrate-free crystal structure of CYP105D18 (7DI3) and papaverine bound CYP105D18 (7DLS) in the space group C-2 diffracting to 1.7 Å. The active site of CYP105D18 is composed of three regions: the B/C loop, the turn region between αF and αG (residues 173 to 187), and the C-terminal loop (residues 336 to 396) (13). Although the CYP105D subfamily members share high structural homology, the amino acids in those three regions are barely conserved, which may account for the substrate recognition and selectivity. Comparing all the available CYP105D subfamily members revealed some residue differences in those substrate recognition regions, B/C loops, and αG helices (Fig. S6). The arginine residues (Arg70, Arg88, and Arg190) in the B/C loop and αG of the substrate-binding pocket of CYP105D7 are conserved in the CYP105 family (19, 33) and may be important residues for substrate recognition. Likewise, CYP105D7, CYP105D1, CYP105D2, CYP105D4, CYP105D5, CYP105D10, CYP105D11, CYP105D13, CYP105D14, and CYP105D19 conserve the arginine residues. Unlike CYP105D7, CYP105D18 contains Ser63 and Phe184, which correspond to Arg70 and Arg190 of CYP105D7, respectively (Fig. 2A). These variations in amino acids in the CYP105D series suggest that the residual recognition of the substrate at the active site is more complicated and varied.

The previous study suggested that the mutation of conserved arginine residues in the B/C loop and αG of the substrate-binding pocket of CYP105D7 significantly enhanced the catalytic activity (19). In this study, we chose the S63 and F184 residues, which are specially conserved in CYP105D18, to replace the arginine residue that is conserved in most of the CYP105D subfamily (Fig. 2A and Fig. S6). Interestingly, these two amino acids make the substrate access channel in CYP105D18 (Fig. 2B). Initially, when serine and phenylalanine were replaced with well-conserved arginine residues, they showed decreased product formations for all five substrates from group A (Fig. 3A and B). Then, an alanine mutation was prepared for both S63 and F184. S63A showed a marked decrease in product formation, while F184A showed a marked increase in product conversion for all five steroids relative to group A (Fig. 3A and B). Detailed comparisons between the wild type and all the mutants for steroid bioconversion are shown in Fig. 3 and Fig. S7, respectively. There were no such significant results from other mutants in steroid bioconversion (Fig. S7). The details of product distribution patterns by the wild type and mutants are presented in Table S1.

FIG 3.

Rate of product formation by CYP105D18 and its mutant. (A) HPLC chromatogram showing product formation for testosterone by the wild type and its mutant. P is the 2β-hydroxytestosterone peak for the testosterone (S). (B) Product conversion (percent) for different steroids (testosterone, methyltestosterone, nandrolone, adrenosterone, and androstenedione) by the wild type and its mutant. All reactions were performed using 2 μM CYP and a 200 μM concentration of the substrate and initiated by 40 mM H₂O₂ for 1 h. The mean and standard deviation were calculated from three independent reactions under similar conditions.

Kinetic analysis for 2β-hydroxyltestosterone.

The F184A mutant showed the highest conversion rate for testosterone. Therefore, kinetic parameters were analyzed to compare between wild type and F184A in terms of the production of 2β-hydroxyltestosterone. The time-dependent in vitro conversion rate was higher for F184A (Fig. 4A). The derived data from the experiment were best fit with the substrate inhibition model defined in Materials and Methods. Assuming that a high concentration of the substrate also inhibits the enzyme activity when two molecules of the substrate can bind to the enzyme, K_i (inhibition constant) was calculated. K_i is indicative of inhibitory potency. K_i values for the wild type and F184A mutant were 465 ± 188 μM and 240 ± 61 μM, respectively, indicating that testosterone has a higher potency for inhibition of the mutant F184A. The K_m and k_cat values for the wild type were 127 ± 40 μM and 0.6 ± 0.1 min⁻¹, respectively. Similarly, the K_m and k_cat values for F184A were 158 ± 37 μM and 1.0 ± 0.16 min⁻¹, respectively. For testosterone hydroxylation, the catalytic efficiency (k_cat/K_m) for the wild type was 0.28 μM⁻¹ S⁻¹ while that for the F184A mutant was 0.38-μM⁻¹ S⁻¹ (Fig. 4B and C and Table S2). The catalytic efficiency was increased 1.36-fold for the F184A mutant over that for the wild type.

FIG 4.

Kinetics analysis for testosterone. (A) Time-dependent conversion assay for wild type versus F184A mutant. (B) Substrate inhibition of CYP105D18-catalyzed 2β-hydroxylation of testosterone. (C) Substrate inhibition of mutant F184A-catalyzed 2β-hydroxylation of testosterone. All reactions were performed using 1 μM CYP and 40 mM H₂O₂. K_m, K_i, and k_cat values were calculated from three independent reactions under similar conditions.

DISCUSSION

CYP105s are predominantly present in the phylum Actinobacteria, and they are associated with a wide variety of metabolic pathways ranging from antibiotic biosynthesis and xenobiotic metabolism to steroid biotransformation (34). Here, we reported the use of CYP105D18 from Streptomyces laurentii for the efficient biotransformation of steroids. CYP105D18 was able to catalyze the A-ring of testosterone to a 2β-hydroxytestosterone derivative by using the H₂O₂ as a cosubstrate (Fig. 5). The regiospecific hydroxylation of testosterone to the 2β position by bacterial CYPs has rarely been reported. Modification of the A-ring to form a 2β-hydroxylated derivative of testosterone by CYP105D, CYP105D4, CYP105D5, CYP127A3, and CYP219A1 at a low conversion level was reported (35). The BM3 mutant F87A catalyzed testosterone at an equal mixture of 2β- and 15β-hydroxylated products, and the 2β selectivity was enhanced to 97% by further saturation mutagenesis (31). Recently, CYP105D7 from Streptomyces avermitilis has been reported for the A- and D-ring modification of steroids with less than 10% conversion. Overall, the hydroxylation of testosterone to the 2β position by bacterial CYPs is summarized in Table 1. Further, the structure-guided site-directed mutagenesis to the conserved arginine residue enhanced the product formation (19). Interestingly, CYP105D18 does not support the redox partners (Pdx/Pdr and Fdx/Fdr) for the catalytic cycle, even though it shares 73% homology with CYP105D7. CYP105D18 was the first CYP from the CYP105D subfamily that was shown to support an oxygen surrogate, H₂O₂, for the N-oxidation of papaverine and hydroxylation of steroids, and it was also able to perform modification of the A-ring of testosterone to a 2β-hydroxylated product. In a previous work, we reported that CYP154C8 was active in the presence of high H₂O₂ concentration and that it was capable of hydroxylating steroids using an H₂O₂ shunt (12). Fewer bacterial CYPs have the ability to support the peroxide shunt, and the challenge remains with the inherent heme stability (9). CYP105D18 showed high stability (k of <0.3 min⁻¹) even with the 200 mM H₂O₂ concentration for the papaverine and steroid biotransformation. This may offer potential toward biological applications, like the production of biofuels or molecules that are difficult or expensive to synthesize.

FIG 5.

Testosterone bioconversion by CYP105D18 and its mutants.

TABLE 1.

Bacterial CYPs responsible for hydroxylation of testosterone to 2β-hydroxyltestosterone

Origin	CYP or mutant	Accession no.	2β-Hydroxyltestosterone		Reference
			% conversion	% selectivity
Streptomyces lividans	CYP105D4	AAC25766	<5	NDa	35
Streptomyces coelicolor	CYP105D5	NP 625076	<10	ND	35
Mesorhizobium loti	CYP127A3	BAB52249	<5	ND	35
Novosphingobium aromaticivorans	CYP219A1	ZP 00093310	<5	ND	35
Streptomyces avermitilis	105D7/R70AR190A	Q825I8	6/41	100/78	19
Bacillus megaterium	BM3 A330WF87A	P14779	76.6	79	31
Streptomyces laurentii	105D18/F184A	BAU86794	46/79	100/100	This study

^aND, not defined.

The substrate recognition orientation of steroids is influenced by the presence of the side chain. CYP105D18 does not display any reaction with steroid with the bulky side chain at C-17 (group B). The steroid with the hydroxyl group at C-17 (testosterone, methyltestosterone, and nandrolone) demonstrated the exposure of the A-ring toward the heme, and it also showed the regiospecificity and same pattern of product formation on the docking analysis. Meanwhile, steroids with a keto group at C-17 displayed the exposure of the B-ring, and they also showed a similar pattern of product formation. Testosterone was placed in the cavity of CYP105D7, surrounded by all hydrophobic amino acids, involved in the A- and D-ring modifications (19). Previously, in the case of CYP105D18, we reported that papaverine was buried in the active site and surrounded by all the nonpolar amino acids. In contrast, the molecular docking analysis here demonstrated that testosterone was surrounded by hydrophilic residue T239, along with all the other hydrophobic residues. The exposure of the A-ring facing the C-2 toward the heme, and the role of this hydrophilic amino acid T239 in binding the polar groups (methyl) of testosterone, nandrolone, and methyltestosterone, may be crucial for regiospecific biotransformation (2).

The possible access channels can be studied through the in silico analysis of the crystal structure of P450 by analyzing the network of channels connecting the buried catalytic activity with the protein distal surface (36). In our previous report, we observed the substrate access channel accommodated by F184 and S63 in the crystal structure of CYP105D18 (13). The properties of these amino acid networks in the channel may account for and control the substrate accessibility and selectivity (37). We performed site-directed mutagenesis toward the substrate access channel while targeting the F184, S63, and R82 residues. A previous study tested the single alanine amino acid mutant for the conserved arginine residue and found that R70A/R190A enhanced the steroid conversion rate (19). Interestingly, R70A and R190A from CYP105D7 are not conserved in CYP105D18, and they correspond to S63 and F184 residues, respectively. We replaced the S63 and F184 residues with arginine, which is almost conserved in the CYP105D subfamily, to observe the role of these amino acids in CYP105D18 for the substrate accessibility. We found a similar pattern of product formation, but the conversion rate was markedly decreased compared to the wild type for all five kinds of steroids (Fig. 3B). The BC-loop was involved in substrate recognition and stabilization in CYP107D1 (OleP), and the introduction of polar and more flexible amino acids like Q and T in F84 enhanced the flexibility of the BC-loop (38). Similarly, the BC-loop forming the substrate access channel in CYP105D18 already contains polar amino acid S63, which may play a role in attracting the polar group of steroids for substrate recognition (13). To assess the role of S63, we mutated S63A and S63R, and they both decreased the conversion rate for all steroids. Therefore, the S63 in CYP105D18 plays a significant role in substrate recognition in the substrate entry channel. F184A mutation markedly increased the product formation for all five kinds of steroids from group A (Fig. 3A and B). Replacing a bulky amino acid with the simplest amino acid in the substrate entry channel may increase the accessibility of the substrate to the active site (37). Substrate inhibition occurs in approximately 25% of known enzymes. The impact of substrate inhibition depends on the K_i/K_m ratios (39). Here, mutant F184A suffers from strong substrate inhibition (K_i/K_m = 1.5) compared to the wild type (K_i/K_m = 3.6) (Table S2). Despite that, we still achieved an approximately 1.36-fold increase in 2β-hydroxyltestosterone formation using the F184A mutant, in comparison to the wild type. These results suggest the future rational design of the substrate access channel to enhance the product formation.

In conclusion, we explored high-H₂O₂-resistant CYP105D18 for the modification of different steroids. We observed the H₂O₂-mediated regiospecific conversion of testosterone to 2β-hydroxyltestosterone. Based on the experimental data, the structure of CYP105D18, and molecular docking analysis, we predicted the product formation pattern for methyltestosterone, nandrolone, adrenosterone, and androstenedione. Moreover, for a rational design of the substrate access channel, we found that the F184A mutant enhanced the product formation rate for the steroids. We believe that the functionalization of C-2 in testosterone by CYP105D18 and the rational design for product enhancement may be applicable over chemical methods for drug modification. Hydroxylated steroids have been reported to be diverse in function. C-14-hydroxylated steroids have been confirmed to have carcinolytic and antigonadotropic activities (40), while 7α-hydroxyl is essential for regulating the immune system (41). The specific activities of 2β-hydroxylated steroids have yet to be reported and require further investigation. Therefore, the production of a wide variety of hydroxylated steroid derivatives is of significant interest. CYP105D18 provides further scope for targeting the other nonactivated positions of steroids through further protein engineering.

MATERIALS AND METHODS

Chemicals and reagents.

All the steroid substrates (listed in Scheme S1 in the supplemental material) of >99% HPLC grade were purchased from Tokyo Chemical Industry Co., Ltd. (Korea). T4 DNA ligase, DNA polymerase, and deoxynucleoside triphosphates (dNTPs) were obtained from TaKaRa Bio (Japan). The α-aminolevulinic acid (ALA), hydrogen peroxide (H₂O₂), (diacetoxyiodo)benzene (DIB), ampicillin (Amp), chloramphenicol (Cm), NADH, NADPH, catalase, formate dehydrogenase, sodium formate, spinach ferredoxin (Fdx), and spinach ferredoxin reductase (Fdr) were obtained from Sigma-Aldrich (Korea). Isopropyl-1-thio-β-d-galactopyranoside (IPTG) and kanamycin were bought from Duchefa Bohemie (Korea). Restriction enzymes were purchased from TaKaRa Clontech (Korea).

Cloning and overexpression of CYPs and redox partner.

CYP105D18 encoding 396 amino acids (GenBank accession no. BAU86794.1), from the Streptomyces laurentii strain, was cloned in pET28a(+) vector. The construct pET28a-CYP105D18 was transformed in C41 cells, and overexpression was performed as described in our previous work (13). The cell pellets were harvested by centrifugation (3,500 rpm) for 30 min at 4°C, washed twice with 50 mM potassium phosphate buffer (pH 7.4), and stored at −50°C. Redox partners putidaredoxin reductase (PdR) (CamA) and putidaredoxin (Pdx) (CamB) (P450cam system) from Pseudomonas putida were expressed as His383-tagged proteins in Escherichia coli BL21(DE3) using the plasmid constructs pET28a(+) and pET32a(+) according to a previously described method (42). Cells were harvested by centrifugation (3,500 rpm) for 30 min at 4°C, washed twice with 50 mM potassium phosphate buffer (pH 7.4), and stored at −50°C.

Site-directed mutagenesis.

Mutagenesis was performed using the QuikChange II site-directed mutagenesis kit method. The PCR mixture contained 5 μL 10× reaction buffer, 0.25 μM forward and reverse primer (each 1.25 μL), 1 μL dNTP mixture, and 50 ng (4 μL) of the template (pET28a-CYP105D18), and double-distilled water (ddH₂O) was added to a final volume of 50 μL. One microliter of PfuUltra HF DNA polymerase (20 U/μL) was added to each reaction mixture. After initial denaturation for 30 s at 95°C, 12 cycles of denaturation at 95°C for 30 s, annealing at 55°C for 1 min (all primers were designed to anneal at this temperature, and these are listed in Table 2), and extension at 68°C for 7 min, there was a final elongation step at 68°C for 10 min. After digestion with the DpnI restriction enzyme (10 U/μL) at 37°C for 1 h, the PCR mixtures (1 μL) were used for heat shock transformation into 50-μL aliquots of NEB5α competent E. coli cells. SOC broth (50 μL) was added to the transformed mixture and incubated at 37°C for 1 h with shaking at 225 to 250 rpm. NEB5α cells were plated on Luria-Bertani broth (LB; Miller) agar containing 25 μg/mL kanamycin and then incubated at 37°C overnight. Three colonies for each mutagenesis reaction were picked, cultured in LB supplemented with 25 μg/mL kanamycin at 37°C, and shaken at 220 rpm for 10 h, followed by plasmid isolation, digestion, and confirmation by automated sequencing (Macrogen, South Korea). The confirmed mutants were transformed into E. coli C41(DE3) and overexpressed as described in our previous study (13).

TABLE 2.

Primers used for site-directed mutagenesis

Sequence	Namea
CATCGAGAAGGCGGCCCGGTCGCTGGAAG	F184A_FP
CTTCCAGCGACCGGGCCGCCTTCTCGATG	F184A_RP
CATCGAGAAGGCGCAGCGGTCGCTGGAAGG	F184Q_FP
CCTTCCAGCGACCGCTGCGCCTTCTCGATG	F184Q_RP
CATCGAGAAGGCGCGCCGGTCGCTGGAAGG	F184R_FP
CCTTCCAGCGACCGGCGCGCCTTCTCGATG	F184R_RP
GCTGTCCACCGACGCCACGCGGGAGGA	S63A_FP
TCCTCCCGCGTGGCGTCGGTGGACAGC	S63A_RP
GCTGTCCACCGACCGCACGCGGGAGGA	S63R_FP
TCCTCCCGCGTGCGGTCGGTGGACAGC	S63R_RP
GCTGCGCCGACAGGCCCGCGGGGCACTCC	R82A_FP
GGAGTGCCCCGCGGGCCTGTCGGCGCAGC	R82A_RP

^aFP, forward primer; RP, reverse primer.

Purification of CYPs and redox partners.

Harvested cell pellets of CYP105D18 as well as all the mutants were suspended in potassium phosphate buffer (pH 7.4) solution and lysed by ultrasonication. The soluble protein-containing fraction was separated by centrifugation at 24,650 × g for 20 min at 4°C and then purified through Ni²⁺ affinity chromatography with the use of Talon His tag. Resin-bound proteins were eluted using elution potassium buffer (pH 7.4) containing 10% glycerol, 100 mM NaCl, and different concentration gradients of imidazole of 10 and 100 mM. The purity of protein was checked by SDS-PAGE in all the fractions, and the purified fraction was concentrated by ultrafiltration with Amicon centrifugal filters with a 30-kDa-molecular-weight cutoff for CYP105D18, all the mutants, and Pdr, along with a 10-kDa-molecular-weight cutoff for Pdx.

Determination of enzyme concentration.

The concentrations of the CYPs (CYP105D18 and all mutants) were determined by CO difference spectra, as described previously. The amount of CYP was calculated from an ε_449–489 of 91 mM⁻¹ cm⁻¹ (43). To determine the purity (Rz value) of all the mutants, the sample containing the CYP enzyme was scanned in the range of 200 to 500 nm, whereas the reference contained only buffer. The purity (Rz value) of all mutants was calculated by the ratio of absorbance at λ_max of the Soret band to the absorbance value at 280 nm. Pdr concentration was determined as the average of the concentrations calculated from wavelengths of 378, 454, and 480 nm, using the respective extinction coefficients (ε) of 9.7, 10.0, and 8.5 mM⁻¹ cm⁻¹. Pdx concentration was also determined as the average of concentrations calculated from wavelengths of 415 and 454 nm by using extinction coefficients of 11.1 and 10.4 mM⁻¹ cm⁻¹, respectively (44). All samples were scanned using Biochrom Libra S35PC UV-visible spectrophotometry (Cambridge, UK).

In vitro biotransformation by CYPs.

The in vitro reactions by CYP105D18 and its mutant for all the steroids listed (see Scheme S1 in the supplemental material) were screened using redox partners (Pdr/Pdx and Fdr/Fdx system) and the oxygen surrogate H₂O₂. All the steroids were prepared in 100 mM stock in dimethyl sulfoxide (DMSO) solvent. All in vitro biotransformations were carried out in a 250-μL reaction mixture for all the steroid substrates (200 μM) with 3 μM CYP in 50 mM phosphate buffer (pH 7.4).

The reaction mixture contained CYP, Pdr, and Pdx (a 1:2:16 ratio), catalase (100 mg/mL), MgCl₂ (1 mM), substrate (400 μM), and an NADH regeneration system (1 U formate dehydrogenase and 150 mM sodium formate). Similarly, CYP, Fdr, and Fdx (a 1:2:10 ratio) were used with the NADPH regeneration system (glucose dehydrogenase, glucose). The reaction was initiated by NAD(P)H, and the reaction mixture was incubated for 2 h at 30°C with vigorous shaking at 1,000 rpm. The reaction for testosterone was optimized by using different concentrations of H₂O₂ for stochastic use to achieve maximum catalytic efficiency. Reactions were initiated by 1 to 200 mM H₂O₂.

Determination of kinetic parameters.

The optimized condition for in vitro biotransformation of testosterone was used for the time-dependent assay. The product formation of testosterone was determined every 5 min for 1 h using 1 μM CYP105D18/mutant, 40 mM H₂O₂, and a 200 μM concentration of the substrate. The product formation rate for testosterone was determined using 1 μM CYPs and 40 mM H₂O₂ at different substrate concentrations ranging from 5 to 500 μM for 30 min. All the reaction mixtures were extracted as described previously, and the conversion (percent) of each product at different time intervals was calculated from the area of the product peaks based on the HPLC chromatogram. Assuming that the products and substrates had the same absorbance properties, the products were quantified by correlating the peak area of the respective product(s) with the combined peak area of product(s) and the substrate. K_m, K_i, and k_cat values were calculated by plotting the product formation rate against the substrate concentration. This is under the assumption that a high concentration of the substrate also inhibits the enzyme activity when two molecules of the substrate can bind to the enzyme. The Michaelis-Menten substrate inhibition equation {v = V_max[S]/(K_m + [S] × (1 + [S]/K_i))} (empirical model) was applied to characterize the substrate inhibition kinetics. Here, V_max and K_m are defined as the maximum velocity and the substrate concentration at which the velocity is equal to half of the maximum velocity, respectively. Further, K_i is the substrate inhibition constant (45).

Docking simulation.

The CYP105D18 (PDB accession no. 7di3) and CYP105D7 (PDB accession no. 4UBS) structures were obtained from the Protein Data Bank (https://www.rcsb.org/). The structures of the substrates, testosterone (compound identification number [CID]: 6013), methyltestosterone (CID: 6010), nandrolone (CID: 9904), adrenosterone (CID: 223997), and androstenedione (CID: 6128), were downloaded from PubChem. The structural optimization of the substrate and rigid formation molecular docking were performed using Gnina. The grid box was set up (ADT ver. 1.5.6, http://mgltools.scripps.edu/) to include heme, and the box size was set to 35 Å × 35 Å × 30 Å, spacing of 0.375 Å, and locations of x = 14.721, y = −3.573, and z = 14.768. The whole-docking simulation parameters were set to num_modes 500, exhaustiveness = 8, and min_rmsd_filter = 0.5. The protein-ligand visualization and structural figures were prepared using PyMOL (46).

Product extraction, purification, and analysis.

The reaction mixtures were extracted using an equal volume of ethyl acetate twice, dried, and dissolved in HPLC-grade methanol for further analysis. The mixture was filtered with an 0.2-μm Whatman filter, injected into an ultrahigh-pressure liquid chromatograph (UHPLC), and separated with the use of a Mightysil reverse-phase C₁₈ column (4.6 mm by 250 mm, 5 μm; Kanto Chemical, Tokyo, Japan). The gradient system included water (A) and acetonitrile (B) as the mobile phase for separation. All the steroid samples were analyzed using a gradient of B for 0 to 10 min, 50% for 10 to 20 min, 70% for 20 to 25 min, and 15% for 25 to 40 min, at a flow rate of 1 mL/min. The substrates and their hydroxylated products were detected by UV-A at 242 and 245 nm, respectively. Following the HPLC analysis, the reaction mixtures were analyzed using Synapt G2-S/Acuity UPLC quadrupole time of flight-electrospray ionization mass spectrometry (water) in positive ion mode.

Large-scale (300-mL reaction mixture) in vitro reactions were carried out to determine the structure of a product. Reactions were carried out separately in a 15-mL volume using 5 μM CYP, 200 μM testosterone, and 40 mM H₂O₂. Reactions were performed in the presence of 50 mM potassium phosphate buffer (pH 7.4) at 30°C and 800 rpm for 1 h, and reaction mixtures were extracted with double-volume ethyl acetate. Further, the sample was dried and concentrated under reduced pressure, and the residue was dissolved in HPLC-grade methanol and subjected to preparative HPLC (Shimadzu) with a C₁₈ column (Mightysil RP-18 GP, 150 by 4.6 mm, 5 μm; Kanto Chemical, Japan) for purification of the product peak. The purified product was dissolved in [D6] DMSO and subjected to NMR analysis (800 MHz) by Varian Unity INOVA spectrometry (Varian, Palo Alto, CA, USA). One-dimensional NMR (¹H NMR and ¹³C NMR) was performed, followed by two-dimensional (2D) NMR, heteronuclear multiple-bond correlation (HMBC), rotating-frame nuclear Overhauser effect spectroscopy (ROESY), and heteronuclear single quantum coherence (HSQC) to elucidate the exact structures as appropriate when needed.

Data availability.

We confirm that the data supporting the findings of this study are available within the article and its supplemental material.