Crystal Structure of H₂O₂-dependent Cytochrome P450_SPα with Its Bound Fatty Acid Substrate

Abstract

Cytochrome P450_SPα (CYP152B1) isolated from Sphingomonas paucimobilis is the first P450 to be classified as a H₂O₂-dependent P450. P450_SPα hydroxylates fatty acids with high α-regioselectivity. Herein we report the crystal structure of P450_SPα with palmitic acid as a substrate at a resolution of 1.65 Å. The structure revealed that the C_α of the bound palmitic acid in one of the alternative conformations is 4.5 Å from the heme iron. This conformation explains the highly selective α-hydroxylation of fatty acid observed in P450_SPα. Mutations at the active site and the F–G loop of P450_SPα did not impair its regioselectivity. The crystal structures of mutants (L78F and F288G) revealed that the location of the bound palmitic acid was essentially the same as that in the WT, although amino acids at the active site were replaced with the corresponding amino acids of cytochrome P450_BSβ (CYP152A1), which shows β-regioselectivity. This implies that the high regioselectivity of P450_SPα is caused by the orientation of the hydrophobic channel, which is more perpendicular to the heme plane than that of P450_BSβ.

Introduction

Cytochrome P450s (P450s)³ are ubiquitous heme-containing monooxygenases that play crucial roles in the oxidative metabolism of many exogenous and endogenous compounds (1–4). X-ray crystal structure analysis is one of the most powerful methods for visualizing the structures of P450s and their interactions with substrates in the heme cavity at the atomic level. Since the first crystal structure of P450, P450_cam (CYP101A1) was reported by Poulos et al. (5), the crystal structures of P450s from mammals (6–14), archaea (15–17), and bacteria (18–27) have been reported, and interactions between their substrates and amino acid residues at substrate recognition sites have been clarified. Most P450s accomplish monooxygenation by reductive activation of molecular oxygen using NADPH or NADH to produce compound I (oxoferryl porphyrin π cation radical). P450s also use H₂O₂ to generate compound I, but the efficiency of this reaction is poor compared with that of reductive activation of molecular oxygen. In 1994, Matsunaga et al. (28) isolated P450_SPα (CYP152B1) from Sphingomonas paucimobilis and reported that it exclusively uses H₂O₂ as the oxidant and catalyzes α-selective (100%) hydroxylation of long alkyl chain fatty acids (29). Although P450_SPα is the first P450 to be classified as a family of H₂O₂-dependent P450, its crystal structure has not been determined despite its potential as a biocatalyst. The first crystal structure of H₂O₂-dependent P450, P450_BSβ (CYP152A1), which has 44% amino acid identity to P450_SPα, was reported in 2003 by Lee et al. (30). The crystal structure of a substrate-bound form of P450_BSβ (Protein Data Bank code 1IZO) revealed that P450_BSβ lacks general acid-base residues around the distal side of the heme, although this arrangement is highly conserved among peroxidases and peroxygenases (31–35). Instead of the general acid-base residues, the terminal carboxylate group of the bound fatty acid interacts with the guanidinium group of Arg²⁴² located near the heme group. The distance between an oxygen atom of the carboxylate group of palmitic acid and the heme iron is 5.3 Å, which is similar to that observed in chloroperoxidase (CPO) from Caldariomyces fumago; the distance between an oxygen atom of glutamic acid side chain and the heme iron is 5.1 Å (34, 35). The location of the carboxylate group of palmitic acid bound to P450_BSβ suggests that the general acid-base function for the facile formation of compound I is accomplished by the carboxylate group of the substrate (Scheme 1). Recently, we have shown that P450_BSβ is able to catalyze H₂O₂-dependent monooxygenation of foreign compounds such as styrene and ethylbenzene in the presence of a carboxylic acid with a short alkyl chain (C₄–C₁₀), a so-called “decoy molecule” (36). The crystal structure of a heptanoic acid (C₇)-bound form of P450_BSβ was analyzed, and an interaction between Arg²⁴² and the carboxylate group of heptanoic acid was detected. We also resolved the crystal structure of the substrate-free form of P450_BSβ and found that binding of fatty acid or substrate analogues did not induce any notable structural change, whereas the substrate-free form of P450_BSβ never reacts with H₂O₂ (37). These observations further confirm that substrate binding initiates the formation of compound I via a salt bridge between Arg²⁴² and the carboxylate group at the active site.

SCHEME 1.

A proposed catalytic reaction mechanism for hydroxylation of long alkyl chain fatty acid. For the generation of compound I (oxoferryl porphyrin π-cation radical), the carboxylate group of the fatty acid (blue) that serves as a general acid-base catalyst first accepts a proton from H₂O₂, producing the ferric hydroperoxy complex (Fe⁺³-OOH). Subsequently, a proton is donated to the distal oxygen atom of the ferric hydroperoxy complex followed by the O–O bond cleavage to produce compound I.

P450_BSβ oxidizes the α- and β-positions of fatty acids in a 40:60 ratio, whereas P450_SPα exclusively oxidizes the α position. To elucidate the cause of this selectivity, we need to study the crystal structures of both enzymes. Although the crystal structure of the substrate-bound form of P450_BSβ is known, that of P450_SPα has not been resolved. Therefore, we crystallized P450_SPα and succeeded in preparing high quality crystals of P450_SPα. Herein we describe the x-ray crystal structure of P450_SPα containing a fatty acid at a resolution of 1.65 Å and examine enzymatic properties of its mutants to study its highly selective α-hydroxylation of fatty acids. We also compare the structure of P450_SPα with that of P450_BSβ in the context of similarities and differences among H₂O₂-dependent P450s.

EXPERIMENTAL PROCEDURES

Crystallization of P450_SPα

P450_SPα WT was concentrated to 13.4 mg/ml in 50 mm MES (pH 7.0) containing 20% (v/v) glycerol by centrifugation using Amicon Ultra filter units (Millipore, Co.). A 2-μl aliquot of the concentrated P450_SPα solution was mixed with 2 μl of a reservoir solution composed of 0.1 m HEPES (pH 7.0) and 35% (v/v) MPD. Crystals of P450_SPα were grown by a sitting-drop vapor diffusion method at 20 °C for 6 days. The P450_SPα L78F and F288G mutants were crystallized under the same conditions used for the WT.

Data Collection, Phasing, and Refinement of P450_SPα

Crystals were flash-cooled in liquid nitrogen. X-ray diffraction data sets were collected on a beam line BL41XU instrument equipped with an ADSC Quantum 315 CCD detector at the RIKEN SPring-8 (Hyogo, Japan) with a 1.0 Å wavelength at 100 K. The HKL2000 (38) program was used for integration of diffraction intensities and scaling. Initial phases were calculated and refined using the SHELXE program (39) and the hkl2map graphical interface (40). In the calculated electron density, the main chain was clearly traceable, and the initial polypeptide chain was built using ARP/wARP (41). Model building and refinement were performed using COOT (42), CNS (43), and REFMAC5 (44). TLS refinement (45) was performed in the final stages of the refinement, defining each chain in the asymmetric unit as a separate TLS group. The resulting model had a final R_fact of 15.1% and an R_free of 17.3% (Table 1). The final model consisted of one polypeptide chain with residues 9–415 of P450_SPα, one heme, one palmitic acid, one MPD, and 371 water molecules. Structure validation was performed using WHAT-IF (46) and PROCHECK (47). X-ray diffraction data sets of the L78F and F288G mutants were collected using a beam line BL26B1 instrument equipped with an ADSC Quantum 210 CCD detector at SPring-8. The structures of L78F and F288G were solved using a molecular replacement method using MOLREP (48), followed by refinement using COOT and REMAC5. The final refinement statistics are summarized in Table 1.

TABLE 1.

Data collection and refinement statistics

	WT	L78F	F288G
Data collection
Wavelength (Å)	1.000	1.000	1.000
Space group	P3121	P3121	P3121
Cell dimensions
a, b, c (Å)	94.440, 94.440, 113.553	94.137, 94.137, 113.402	94.58, 94.58, 113.449
α, β, γ (°)	90.000, 90.000, 120.000	90.000, 90.000, 120.000	90.000, 90.000,120.000
Resolution (Å)	50.00-1.68 (1.68-1.65)	20.0-1.90 (1.97-1.90)	20.0-1.80 (1.86-1.80)
No. of total observed reflections	1,516,965	493,946	582,013
No. of unique reflections	70,272	46,225	54,812
Rmerge (%)a,b	6.4 (48.9)	7.5 (38.0)	6.2 (34.9)
Completeness (%)a	99.1 (82.8)	100.0 (100.0)	99.9 (100.0)
I/σ(I)a	84.5 (6.2)	31.8 (6.2)	36.4 (6.1)
Redundancya	21.6 (17.7)	10.7 (11.0)	10.6 (10.6)
Refinement statistics
Resolution range (Å)	19.85–1.65	19.98–1.90	19.67–1.80
No. of monomer/asymmetric unit	1	1	1
Rfact/Rfree (%)c,d	15.1/17.3	15.9/18.8	16.0/19.0
RMSD bond length (Å)e	0.012	0.014	0.012
RMSD bond angles (°)e	1.279	1.295	1.199
No. of atoms	3749	3576	3697
Average B-factor (Å2)	16.8	16.6	20.1

^a The values in parentheses are for the highest resolution shell.

^b R_merge = ΣΣ|I − <I>|/ΣI.

^c R_fact = Σ||F_o| − k|F_c||/Σ|F_o|, where F_o and F_c are the observed and calculated structure factor amplitudes, respectively.

^d R_free was calculated as the R_fact for 5% of the reflections that were not included in the refinement.

^e RMSD, root mean square deviation.

Structural Analysis of P450_SPα and P450_BSβ

Electrostatic potentials were calculated using GRASP2 (49). Probe-occupied voids were calculated using VOIDOO (50), and a probe of 1.1 Å and a grid mesh of 0.3 Å were used unless otherwise specified. The accessible channels were calculated using CAVER (51). All of the protein figures were depicted using PyMOL (52).

Hydroxylation of Myristic Acid

The standard reaction mixture contained 0.1 m potassium phosphate (pH 7.0), 0–120 μm myristic acid (C₁₄) (0–60 μm for F288G and A172F/F288G), 50 nm P450_SPα or P450_BSβ, and 200 μm H₂O₂ in a total volume of 1 ml. The reaction mixture was incubated at 37 °C for 1 min, and then the reaction was quenched by adding 500 μl of dichloromethane followed by vigorous mixing. After the addition of 12-hydroxydodecanoic acid as an internal standard, the products were extracted with dichloromethane. For derivatization of the extract, 50 μl of N,O-bis(trimethylsilyl)trifluoroacetamide (BSTFA) containing 1% (v/v) trimethylchlorosilane (TMCS) was added, and the mixture was incubated in the dark at room temperature for 2 h. The derivatized products were analyzed using a Shimadzu GC-17A (Shimadzu Corp., Kyoto, Japan) equipped with a Shimadzu GC/MS-QP5000 and Rxi^TM-5ms capillary column (30 m × 0.25 mm; Restek Corp., Bellefonte, PA) to identify the products. The GC/MS analytical conditions were as follows: column temperature, 50 °C (1 min) to 40 °C/min (5 min) to 250 °C (8 min); injection temperature, 250 °C; interface temperature, 280 °C; carrier gas, helium; flow rate, 0.9 ml/min, mode, split mode; and split ratio, 1/50. To quantify the products, derivatization of the extract was performed by adding 9-anthryldiazomethane and incubating the solution in the dark at room temperature for 1 h. For quantification of the products, reverse phase HPLC analysis was performed using an Inertsil® ODS-3 column (4.6 mm × 250 mm; GL Sciences, Inc., Tokyo, Japan) installed on a Shimadzu SCL-10A_VP system controller equipped with Shimadzu LC-10AD_VP pump systems, a Shimadzu RF-10A_XL fluorescence spectrometer, a Shimadzu CTO-10A_VP column oven, and a Shimadzu DGU-12A degasser. The HPLC analytical conditions were as follows: flow rate, 1.0 ml/min; acetonitrile/water = 99/1; column temperature, 30 °C; excitation wavelength, 365 nm, emission wavelength, 412 nm; and retention times, 12-hydroxydodecanoic acid (6.41 min), β-OH C₁₄ (10.7 min), α-OH C₁₄ (11.9 min), and C₁₄ (21.5 min). Chiral separation of the products was performed on a CHIRALPAK AD-RH column (Daicel Chemical Industries, Ltd., Osaka, Japan) installed on the same reverse phase HPLC system as in the case of the quantification. The absolute configuration was assigned by comparison of the product ratios in the hydroxylation of C₁₄ by P450_SPα WT (53) and P450_BSβ WT (54). The HPLC conditions for the chiral separation were as follows: flow rate, 0.9 ml/min; linear gradient, MeOH/water = 85/15 (0–10 min) to 100/0 (100–120 min); column temperature, 40 °C; excitation, 365 nm; emission, 412 nm; retention times, (R)-α-OH C₁₄ (34.2 min), (S)-α-OH C₁₄ (37.6 min), (S)-β-OH C₁₄ (49.8 min), and (R)-β-OH C₁ (53.2 min).

UV-visible and EPR Measurements

UV-visible spectra were recorded using a Shimadzu UV-2400 PC spectrophotometer at room temperature. X-band EPR spectra were recorded using an E500 X-band CW-EPR instrument (Bruker, Ettlingen, Germany) at 10 K. A cryostat (ITC503; Oxford Instruments Co., Abingdon, UK) was used for measurements at low temperatures.

RESULTS AND DISCUSSION

Overall Structure and Substrate Binding

The structure of P450_SPα was resolved at a resolution of 1.65 Å (Fig. 1). One molecule was observed in the asymmetric unit. The overall structure exhibited typical P450 folding with 17 α helices and three β sheets. It has a trigonal prism-shaped structure with the heme buried deep inside the protein. The I helix lays across the interior of the P450 molecule on the distal side of the heme group. Two channels connecting the active site cavity with the protein surface were identified (Fig. 1). Channel I is composed of hydrophobic residues (Ile⁷³, Leu⁷⁷, Leu⁷⁸, Phe¹⁶⁹, Ala¹⁷², Ala²⁴⁵, Phe²⁸⁷, Phe²⁸⁸, Pro²⁸⁹, Leu³⁹⁸, and Pro³⁹⁹) (Fig. 2A). Channel II includes hydrophilic residues (Gln⁸⁴ and Asn²³⁸) as its constituent residues. A cluster of water molecules with a hydrogen-bonding network was observed in Channel II (Fig. 2D). We expect that Channel II would be used for the ingress of H₂O₂ and the egress of water during the reaction. Phe²⁸⁸ is located at the border of the two channels, but the two channels are not clearly separated because the entrances of the channels are wide (Figs. 1, B and C, and 3A).

FIGURE 1.

Overall structure of P450_SPα with palmitic acid. The B′ helix and F helix-loop-G helix are colored in blue and green, respectively. Heme (red) and palmitic acid (yellow and cyan) are represented as stick models. The Channel I (magenta) and Channel II (orange) locations were calculated using CAVER (51). Water molecules in Channel II are represented as red spheres. A, the distal face of P450_SPα. B, surface model of P450_SPα. C, two channels of P450_SPα. Arg²⁴¹ (orange) and Phe²⁸⁸ (purple) are labeled.

FIGURE 2.

The heme cavity of P450_SPα. The 2F_o − F_c electron density maps of heme, palmitic acid, and a water molecule coordinated to the iron are contoured at the 1 σ level (blue mesh). The hydrophobic amino acid residues, heme, and Cys³⁶¹ are represented as stick models. A water molecule coordinated to the iron is represented by a red sphere. The distance between the iron and water is shown in red. A, two alternative conformations of palmitic acid shown as yellow and cyan stick models. B, palmitic acid with 0.7 occupancy (Conformation A). C, palmitic acid with 0.3 occupancy (Conformation B). The interaction of Arg²⁴¹ with the oxygen atoms of the carboxylate group of palmitic acid are shown as black dashed lines. The distance of the C_α carbon and the oxygen atoms of palmitic acid from the iron and are shown as red dashed lines and blue dashed lines, respectively. D, hydrogen bond network in Channel II of P450_SPα. Heme (pink), palmitic acid (yellow and cyan), and hydrophilic residues and Val²⁹² (orange) are represented as stick models. Water molecules are represented as red spheres. Hydrogen bonds are shown as dashed lines.

FIGURE 3.

The substrate access channel of P450_SPα. A, the substrate access channel (Channel I) of P450_SPα. Heme, palmitic acid, Arg²⁴¹, and Phe²⁸⁸ are represented as stick models. B, B-factor of P450_SPα.

Although no substrates were added to the purified P450_SPα, the initial 2F_o − F_c electron density map showed a long continuous electron density in Channel I (Fig. 2A). One of the ends of this electron density, located near Arg²⁴¹ in the active site, has a Y-shape. Because the shape of this electron density is very similar to that of a long alkyl chain fatty acid observed in P450BM3 (21) and P450_BSβ (30), we assumed that this electron density corresponds to a long alkyl chain fatty acid originating from Escherichia coli cells (55). Indeed, GC/MS analysis of the extract of the purified P450_SPα with dichloromethane showed that palmitic acid and stearic acid were coexistent, even after purification (supplemental Fig. S1). It was difficult to deduce the length of the alkyl chain of the fatty acid based on the electron density of the substrate(s) because the electron density of the substrate was shorter than the alkyl chain of palmitic acid, possibly because of disordering. Therefore, we tentatively assigned this electron density to palmitic acid. In addition, because the Y-shaped electron density adjacent to the Arg²⁴¹ was accompanied by an additional electron density, two alternative conformations with occupancies of 0.7 (Conformation A) and 0.3 (Conformation B) were placed and refined (Fig. 2, B and C). The terminal alkyl chain of palmitic acid in the final structure is highly disordered, indicating that the terminal alkyl chain is loosely fixed. It is noteworthy that the A′-helix, B′-helix, and F–G loop have relatively high B-factors (Fig. 3B) and that the entrance of Channel I is open wide (Fig. 3A), suggesting that this region is flexible even though the substrate was accommodated. The substrate access channel and B-factor of P450_BSβ are shown in supplemental Fig. S2 for comparison.

The charge distribution on the surfaces of the proximal and distal sides of P450_SPα is shown in Fig. 4. In contrast to the charge distribution typical of P450s such as P450BM3 (56, 57), negative surface charges were observed on the proximal side of P450_SPα. A positive surface potential on the proximal side of regular P450 is important for the recognition of reductases in the electron transfer step of oxygen activation in the P450 catalytic cycle (58). The negative surface potential on the proximal side of P450_SPα may preclude binding of a reductase. This unique charge potential distribution of P450_SPα is an indication that P450_SPα does not need binding of a reductase and prefers the H₂O₂ shunt pathway. P450_BSβ also has no positively charged region on the surface of the proximal side. In addition, a long loop structure was observed on the proximal side (³⁴⁹QGGGDHYLGHRC³⁶¹) (Fig. 5), whereas P450s that require electron transfer from reductases have short loop structures. P450_cam, for example, has a short loop structure, ³⁵⁰FGHGSHLC³⁵⁷. P450_BSβ (³⁵²QGGGHAEKGHRC³⁶³) (30) and allene oxide synthase (⁴⁵⁵WSNGPETETPTVGNKQC⁴⁷²) (59) also have long loop structures at the proximal side of the heme, implying that the long loop structure is a common feature among P450s that do not require electron transfer systems for the activation of molecular oxygen.

FIGURE 4.

Electrostatic potential surface of P450_SPα. Electrostatic surfaces of P450_SPα were calculated using GRASP2 (49). The negatively charged surface is represented in red, and the positively charged surface is represented in blue. Palmitic acid is shown as a yellow stick model.

FIGURE 5.

The proximal side of the heme of P450_SPα. A, the main chain composed of the loop at the proximal side of the heme is represented as a green stick model. Heme is shown as a pink stick model. The water molecule on the heme is represented as a red sphere. The distances between the heme iron and nitrogen atoms of the main chain from the sulfur atom of Cys³⁶¹ are shown in blue. B, comparison of the amino acid sequence of P450_SPα with other P450s. Conserved amino acid residues are depicted in bold type.

Active Site Structure

At the active site, Arg²⁴¹ is located above the heme, and its guanidinium group interacts with the carboxylate group of palmitic acid (Fig. 2A); the distances between two oxygen atoms of the carboxylate group and the guanidinium group of Arg²⁴¹ (N_η2 and N_ϵ) are 2.9 and 3.0 Å in Conformation A (Fig. 2B) and 3.3 and 3.2 Å in Conformation B (Fig. 2C), respectively. In contrast to other heme enzymes that utilize H₂O₂, P450_SPα lacks general acid-base residues around the distal side of the heme. As an alternative, the terminal carboxylate group of palmitic acid is located above the heme. The distance between the oxygen atoms close to the heme iron is 5.2 Å for Conformation A and 5.5 Å for Conformation B, indicating that the location of the oxygen atom is similar to that of the glutamic acid moiety of CPO (34, 35) and that of the terminal carboxylate group of palmitic acid observed in P450_BSβ. These observations indicate that participation of the terminal carboxylate group of the fatty acid in the generation of active species using H₂O₂ (Scheme 1) is common among H₂O₂-dependent P450s. The distal side of the heme is hydrophilic because of Gln⁸⁴, Asp²³⁸, Arg²⁴¹, and the carboxylate group of palmitic acid (Fig. 2D). The polar environment is expected to facilitate the heterolytic cleavage of the O–O bond of H₂O₂ to generate compound I, as is proposed for heme peroxidases such as cytochrome c peroxidase (31, 32), HRP (33), CPO (34, 35), and myoglobin mutants (60, 61). A water molecule is located 2.1 Å away from the heme iron and could function as a sixth ligand on the heme even though the palmitic acid occupies the distal side of the heme cavity (Fig. 2A). The UV-visible spectra of P450_SPα in the absence and presence of 120 μm of myristic acid showed a Soret absorption peak at 417 nm, which is consistent with a typical six-coordinate low-spin ferric heme (Fig. 6). The EPR spectrum of the substrate-free form of P450_SPα showed a ferric low spin state having g value 2.59 (g_z), 2.25 (g_y), and 1.85 (g_x) (Fig. 7), suggesting that the electronic environment of the heme iron of P450_SPα resembles that of CPO (2.61 (g_z), 2.26 (g_y), 1.83 (g_x), pH 5.2) (62) rather than those of bacterial P450s such as P450_cam (2.45 (g_z), 2.26 (g_y), 1.91 (g_x)) (63) and P450BM3 (2.42 (g_z), 2.26 (g_y), 1.92 (g_x)) (64). The EPR spectral change of P450_SPα upon addition of myristic acid (120 μm) was very small, indicating that the low spin state is essentially retained irrespective of substrate binding. Minor signals at g = 2.67 in the substrate free-form and at g = 2.53 in the myristic acid-bound form might reflect P420 species of P450_SPα (supplemental Fig. S5), whereas signals are different from that of P420 species of P450_cam (2.46 (g_z)) (64).

FIGURE 6.

UV-visible absorption spectra of P450_SPα. The spectra of P450_SPα in 0.1 m potassium phosphate buffer (pH 7.0) containing 0.3 m KCl and 20% (v/v) glycerol in the absence (solid line) and presence of 120 μm of myristic acid (dotted line). The concentration of P450_SPα was 3.1 μm.

FIGURE 7.

EPR spectra of P450_SPα recorded at 10 K. The spectra of P450_SPα in the absence (A) and presence of 120 μm of myristic acid (B) are shown. The concentrations of P450_SPα were 44 and 24 μm for the substrate-free form and for the myristic acid-bound form, respectively.

Regioselectivity for Hydroxylation of Fatty Acid

P450_SPα exclusively catalyzes the hydroxylation of fatty acid at the C_α position and produces the corresponding α-hydroxy fatty acid, whereas P450_BSβ produces α and β hydroxy products in the ratio of 43:57. In the crystal structure of P450_SPα, the distances of the C_α carbon and the C_β carbon in Conformation B from the heme iron are 4.5 and 5.5 Å, respectively. Because the C_α carbon in Conformation B is clearly close to the iron atom, and the distance of 4.5 Å agrees well with the distance between the C5 position of d-camphor and the heme iron in P450_cam (65), Conformation B is expected to produce the α-hydroxy fatty acid selectively. Because the C_α and C_β carbons in Conformation A are both far away from the heme iron in respect of the hydroxylation reaction, we assume that Conformation A is a nonproductive conformation. In the crystal structure of P450_BSβ, the C_α and C_β carbons of palmitic acid are located at distances of 5.0 and 6.2 Å from the heme iron, respectively. The substrate observed in P450_BSβ needs to be closer to the heme iron to be hydroxylated, as was observed for P450BM3 (66). The C_α and C_β carbons of the possible productive conformation of palmitic acid in P450_BSβ may be equally close to the heme iron.

Structural Comparison of P450_SPα with P450_BSβ

To gain further insight into H₂O₂-dependent P450s, the structure of P450_SPα was compared with that of P450_BSβ. The structure of P450_SPα is superimposed on that of P450_BSβ in Fig. 8. Except for the B′ helix and the F–G loop, the overall structures are well superposed with a root mean square deviation value of 1.4 Å. The locations of Channel I are notably different, resulting in different locations for the bound palmitic acid. The palmitic acid is more perpendicular to the heme plane in P450_SPα than in P450_BSβ. The active site cavity of P450_SPα is smaller than that of P450_BSβ (supplemental Fig. S3). The smaller active site cavity is mainly due to the presence of Phe²⁸⁸ in the heme cavity of P450_SPα. The corresponding amino acid residue in P450_BSβ is Gly²⁹⁰. The effect of Phe²⁸⁸ and the location of Channel I are discussed in the next section.

FIGURE 8.

Overall structure of P450_SPα superposed on that of P450_BSβ. A, the superposition of the structures. The structures of P450_SPα and P450_BSβ are represented in green and blue, respectively. Heme is depicted as a red stick model. Palmitic acids in P450_SPα and P450_BSβ are represented as a yellow and cyan stick models and a pink stick model, respectively. B and C, substrate access channel location of P450_SPα (B) and P450_BSβ (C) calculated using CAVER (51) are shown as a light blue surface for P450_SPα and a cyan surface for P450_BSβ.

Hydroxylation of Myristic Acid by Mutants

We carried out mutagenesis study to elucidate the structural requirement for the α-selective reaction of P450_SPα. Based on the comparison between amino acid residues in the active sites of P450_SPα and P450_BSβ, two amino acid residues in the active site of P450_SPα, Phe²⁸⁸ and Leu⁷⁸, were mutated to the corresponding P450_BSβ residues and vice versa. The active site cavity of P450_SPα is smaller than that of P450_BSβ, mainly because of the side chain of Phe²⁸⁸ (supplemental Fig. S3), which interacts directly with the fatty acid. Although Leu⁷⁸ does not interact with the fatty acid, it is located at the border of the channels and may affect the catalytic reaction. Three mutants of P450_SPα, L78F, F288G, and L78F/F288G, and three P450_BSβ mutants, F79L, G290F, and F79L/G290F, were prepared and their catalytic activities, regioselectivities, and stereoselectivities were examined (Table 2). P450_BSβ mutants F79L and G290F had 75 and 95% α selectivity, respectively, indicating that the regioselectivity of P450_BSβ was greatly altered. The substitution of Gly²⁹⁰ with Phe must induce a conformational change in myristic acid, and the C_α carbon is expected to move closer than the C_β carbon to the heme iron. In sharp contrast with the P450_BSβ mutants, no clear differences in regioselectivity were observed for P450_SPα mutants, whereas the amino acids at the active site were replaced with the corresponding amino acids of P450_BSβ, which has β-regioselectivity. Three mutants of P450_SPα showed >99% α selectivity. To further investigate the regioselectivity of P450_SPα, two mutants of P450_SPα, A172F and A172F/F288G, were prepared. Ala¹⁷² located in the F–G loop seems to be important for controlling fatty acid conformation. However, the regioselectivity was not affected by these mutagenesis experiments, indicating that the mutagenesis at the active site and at the F–G loop does not alter the position of bound palmitic acid. The x-ray crystal structures of the L78F and F288G mutants revealed that there is little difference in the location of the bound palmitic acid in Conformation B (the productive conformation) and the location of the hydrophilic channel (Fig. 9), whereas the position of the amino acids located near the active site, Met⁶⁹, Pro⁷⁰, and Pro²⁸⁹, are shifted (Fig. 10 and supplemental Fig. S4). The C_α positions of palmitic acid in both mutants are essentially the same as in the WT. Whereas the stereoselectivity of α hydroxylation of myristic acid was slightly reduced by mutagenesis (Table 2), the regioselectivity of P450_SPα was not affected, suggesting that the small perturbation induced by the mutagenesis is insufficient to alter the high regioselectivity of P450_SPα. We presume that the regioselectivity of P450_SPα is highly controlled by its hydrophobic channel. The orientation of the channel (almost perpendicular to the heme plane) and the wide open entrance appear to be crucial for the high α-selectivity of hydroxylation. The hydrophobic channel may control the direction of the fatty acid access. Further mutagenesis studies, especially at the F and G helices and at the F–G loop region, are necessary for elucidating the mechanistic details of the highly selective α-hydroxylation of fatty acid catalyzed by P450_SPα.

TABLE 2.

Hydroxylation of myristic acid by P450_SPα and P450_BSβ

Enzyme	kcat	Km	Regioselectivity	Stereoselectivity
				At α-position (R):(S)	At β-position (R):(S)
	nmol/min/nmol P450	μm	α-OH %
P450SPαWT	1300 ± 20	43 ± 2	>99	3:97
P450SPαL78F	5500 ± 930	300 ± 70	>99	4:96
P450SPαF288G	770 ± 70	100 ± 20	>99	19:81
P450SPαL78F/F288G	1300 ± 40	85 ± 16	>99	21:79
P450SPαA172F	1200 ± 20	68 ± 3	>99	3:97
P450SPαA172F/F288G	100 ± 2	23 ± 2	>99	17:83
P450BSβWT	1400 ± 180	66 ± 20	43	76:24	98:2
P450BSβF79L	1400 ± 30	54 ± 3	75	62:38	99:1
P450BSβG290F	1900 ± 210	190 ± 30	95	29:71	85:15
P450BSβF79L/G290F	890 ± 20	95 ± 3	77	28:72	86:14

FIGURE 9.

The active site structures of P450_SPα L78F (A) and F288G (B). Palmitic acid, heme, Arg²⁴¹, Phe⁷⁸ or Leu⁷⁸, Phe²⁸⁸ or Gly²⁸⁸, and MPD are represented as stick models. A water molecule coordinated to the iron is represented as a sphere. The 2F_o − F_c electron density maps of palmitic acid, Phe⁷⁸ or Leu⁷⁸, Phe²⁸⁸ or Gly²⁸⁸, and MPD are contoured at the 1 σ level (blue mesh). The position of Phe²⁸⁸ in the WT is occupied by MPD in F288G. In both structures, palmitic acid with 0.7 occupancy and 0.3 occupancy are colored yellow and cyan, respectively.

FIGURE 10.

Comparison of WT (green), L78F (cyan), and F288G (purple). Superimposition of two mutants and WT P450_SPα. Side view from the opposite side of the propionate of the heme. Heme is represented as a line. MPD in F288G is represented as a blue stick model. Root mean square deviation values over 9–415 amino acid residues of C_α atoms are 0.225 for L78F and 0.197 for F288G, respectively.

Conclusion

We have determined the x-ray crystal structure of H₂O₂-dependent P450_SPα as a palmitic acid-bound form at a resolution of 1.65 Å. The crystal structure revealed that the carboxylate group of the fatty acid interacts with Arg²⁴¹, which is located above the heme. Previous studies on the reaction mechanism of P450_BSβ suggested that the carboxylate group of the fatty acid serves as a general acid-base catalyst for the generation of compound I using H₂O₂. Our crystal structure study confirms that this substrate-assisted activation mechanism is also conserved in P450_SPα, indicating that the substrate-assisted activation mechanism is common in the H₂O₂-dependent P450-catalyzed hydroxylation reaction with fatty acids. Notably, a water molecule was observed as the sixth ligand of the heme iron, even in the presence of palmitic acid. Consistent with the crystal structure, the ferric low spin state of P450_SPα was retained irrespective of the substrate binding. These results also indicate that the shift in redox potential of the heme that is induced by substrate binding, which is generally indispensable for the reductive activation of molecular oxygen, is not essential for the H₂O₂-dependent P450s. Crystallographic studies on substrate binding revealed that the C_α carbon of the bound palmitic acid in Conformation B is situated close to the heme iron (4.5 Å). This conformation explains the highly selective α-hydroxylation of fatty acid. Surprisingly, mutations at the active site and at the F–G loop of P450_SPα did not impair the high regioselectivity. The crystal structures of the L78F and F288G mutants revealed that the location of the bound palmitic acid was not affected by these mutations. These results imply that the orientation of the hydrophobic channel of P450_SPα, which is more perpendicular to the heme plane than that of P450_BSβ, is crucial for the highly selective α-hydroxylation. Although further mutagenesis studies are required to fully understand the high regioselectivity of P450_SPα, the structural studies reported here contribute to a better understanding of the relationship between the structure and function of H₂O₂-dependent P450s at the atomic level.