The highly stereoselective oxidation of polyunsaturated fatty acids by cytochrome P450BM-3.

Cytochrome P450BM-3 catalyzes NADPH-dependent metabolism of arachidonic acid to nearly enantiomerically pure 18(R)-hydroxyeicosatetraenoic acid and 14(S), 15(R)-epoxyeicosatrienoic acid (80 and 20% of total products, respectively). P450BM-3 oxidizes arachidonic acid with a rate of 3.2 +/- 0.4 micromol/min/nmol at 30 degrees C, the fastest ever reported for an NADPH-dependent, P450-catalyzed reaction. Fatty acid, oxygen, and NADPH are utilized in an approximately 1:1:1 molar ratio, demonstrating efficient coupling of electron transport to monooxygenation. Eicosapentaenoic and eicosatrienoic acids, two arachidonic acid analogs that differ in the properties of the C-15-C-18 carbons, are also actively metabolized by P450BM-3 (1.4 +/- 0.2 and 2.9 +/- 0.1 micromol/min/nmol at 30 degrees C, respectively). While the 17,18-olefinic bond of eicosapentaenoic acid is epoxidized with nearly absolute regio- and stereochemical selectivity to 17(S),18(R)-epoxyeicosatetraenoic acid (>/=99% of total products, 97% optical purity), P450BM-3 is only moderately regioselective during hydroxylation of the eicosatrienoic acid omega-1, omega-2, and omega-3 sp3 carbons, with 17-, 18-, and 19-hydroxyeicosatrienoic acid formed in a ratio of 2.4:2.2:1, respectively. Based on the above and on a model of arachidonic acid-bound P450BM-3, we propose: 1) the formation by P450BM-3 of a single oxidant species capable of olefinic bond epoxidation and sp3 carbon hydroxylation and 2) that product chemistry and, thus, catalytic outcome are critically dependent on active site spatial coordinates responsible for substrate binding and productive orientation between heme-bound active oxygen and acceptor carbon bond(s).

P450BM-3 contains heme, FAD, and FMN in a stoichiometry of 1:1:1, respectively (3). Both purified and recombinant P450BM-3 catalyze NADPH-dependent oxidation of medium and long chain saturated fatty acids, with optimum chain lengths of 14 -16 carbons (3,5). The regiochemistry of fatty acid hydroxylation by P450BM-3 is more or less chain length-dependent, i.e. as chain length increases, regioselectivity shifts from the -1 to the -2 carbon of the fatty acid (5). The turnover number of purified P450BM-3, in the monooxygenation of palmitic acid, is approximately 1.6 mol/min/nmol at 25°C (3,5), which is similar to that of P450 cam for camphor hydroxylation under similar conditions (6) and 100 -1,000 times greater than that seen for most substrates with mammalian P450s. Monooxygenation of fatty acids by P450BM-3 is tightly coupled with a stoichiometry of NADPH, oxygen, and fatty acid consumed to hydroxylated fatty acid product formed of 1:1:1:1 (5). Under conditions of limiting substrate, P450BM-3 will further oxidize the initial metabolites to products that include diols and ketoalcohols (7).
In eukaryotes, AA serves both a structural role as a component of cellular membranes, and a critical functional role by participating in a variety of receptor/agonist-mediated signaling cascades (8). The latter role is a consequence of regio-and stereoselective oxygenations of AA that is utilized by mammalian cells to transduce the signal (8 -13). The transduced chemical information is then decoded either by receptor mediated processes or, alternatively, by the direct effects of the oxygenated products on metabolic pathways (8 -13). In addition to prostaglandin synthases and lipoxygenases, well recognized members of the AA cascade, the contribution of microsomal P450 to the metabolism of endogenous AA pools is now well established (9 -13).
Eukaryotic P450s oxidize AA by one or more types of reactions: 1) allylic oxidation to generate six different regioisomeric hydroxyeicosatetraenoic acids containing a cis,trans-conjugated dienol, 2) hydroxylation at sp 3 carbons near or at the methyl terminus to generate 16-, 17-, 18-, 19-, or 20-OH-AA, and 3) olefinic bond epoxidation to generate four regioisomeric EETs. While the physiological significance of allylic oxidation remains obscure, products of hydroxylation and olefinic bond epoxidation reactions display a variety of potent biological activities and have been implicated in processes ranging from hormonal signaling to the pathophysiology of hypertension (9 -13). As with other metabolites of the AA cascade, e.g. prostanoids and leukotrienes, regio-and stereochemical features define biological activity and/or potency. Thus, understanding active site topology and the structural determinants of asymmetric catalysis by P450s are prerequisites for modifying their mechanism of action, and ultimately, for rational pharmacological intervention. Studies using microsomal fractions, and/or solubilized and purified mammalian P450 isoforms have demonstrated that the hemoprotein controls the regio-and stereoselectivities of oxidation of AA in an isoform-specific fashion at three different levels: 1) the type of reaction catalyzed, i.e. epoxidation to form EETs by CYP2B and CYP2C isoforms (9,11,13), or hydroxylation at C-16, C-17, and C-18 by CYP1A1 and CYP1A2 isoforms (9,11,13), and at C-19 and C-20 by the CYP4A isoform (11,(13)(14)(15); 2) positional selectivity, i.e. differentiation among four chemically equivalent olefinic bonds or between five sp 3 hybridized carbons (hydroxylations at C-16 -C-20) (11,13); and 3) absolute configuration. These processes are all the more remarkable since AA is an unbiased, acyclic molecule with high rotational freedom, and it is epoxidized by several mammalian P450 isoforms with unprecedented stereoselectivity (9,11,13).
As a guide to the protein structural determinants of polyunsaturated fatty acid metabolism by P450s, we describe herein the highly regio-and stereoselective metabolism of AA by recombinant P450BM-3, one of the four soluble P450s for which an atomic structure has been determined (16 -19). Based on these results and similar analyses done with the AA cogeners EPA and ETA, we propose a model of the active site that accounts for the asymmetric catalysis by P450BM-3.

EXPERIMENTAL PROCEDURES
Materials-The fatty acids were obtained from NuCheck Prep, Inc. All other chemicals were obtained in the purest form available from Sigma.
Methods-The original plasmid, containing the gene encoding P450BM-3, was a gift from Dr. A. Fulco (Department of Biochemistry, UCLA, Los Angeles, CA). In preparation for our studies of the mechanism of P450BM-3, we constructed a new expression plasmid that contained only the gene encoding P450BM-3 and no extraneous B. megaterium DNA. We had previously constructed a plasmid for the expression of the heme domain of P450BM-3 and a construct from this project, pIBI-BMP2 (20), was linearized with SalI. Likewise, a vector had been constructed for the expression of the reductase domain of P450BM-3, pIBI-BMR (21). A 1.78-kilobase pair fragment from SalIdigested pIBI-BMR was recovered and ligated into the linearized pIBI-BMP2, giving pIBI-BM3. The sequence of the gene encoding P450BM-3 in this plasmid was confirmed. The plasmid was used to transform Escherichia coli strain DH5␣. The protein was overexpressed and purified from these cells using published procedures (5). The concentration of the purified protein was estimated from the difference absorbance spectrum of the carbonyl complex of the ferrous form versus the ferrous form using the molar absorptivity of 91 mM Ϫ1 cm Ϫ1 for the wavelength pair of 450 versus 490 nm (22). The substrate binding spectra used for the determination of spectral binding constants (K s ), were recorded with a Hewlett Packard diode array spectrophotometer maintained at 25°C. The absorbance data for substrate binding were analyzed with the program Microsoft Excel.© Enzyme Activity Studies-The rates of P450BM-3-dependent oxygen and NADPH utilization were measured using 20 -50 nM solutions of the enzyme in 50 mM MOPS buffer, pH 7.4, containing 100 M fatty acid. After a 5-min preincubation, reactions were started by addition of NADPH. Oxygen concentration was measured with a Clark-type oxygen electrode instrument (Yellow Springs Instrument Co.). NADPH concentration was measured spectrophotometrically at 340 nm (⑀ ϭ 6.22 mM Ϫ1 cm Ϫ1 ).
For product quantification and structural characterization, incubations were performed at 30°C under atmospheric air and with vigorous mixing. Reaction mixtures in 50 mM Tris-Cl buffer, pH 7.4, containing 10 mM MgCl 2 , 150 mM KCl, 8 mM sodium isocitrate, isocitrate dehydrogenase (1.0 IU/ml), dilauroyl phosphatidyl choline (0.05 g/ml), and P450BM-3 (2-10 nM, final concentration) were incubated 2.5 min prior to the addition of the sodium salts of either AA, EPA, or ETA (25 mM each in 0.05 mM Tris-Cl buffer, pH 8.0) to final concentrations of 50 -100 M each. After 1 min, reactions were started by the addition of NADPH (1 mM, final concentration). At different time points, aliquots were withdrawn, and the organic soluble products were extracted three times with equal volumes of ethyl ether containing HOAc (0.05%, v/v). After solvent evaporation under a stream of nitrogen, the products were resolved by RP-HPLC on a 5-m Dynamax Microsorb C 18 column (4.6 ϫ 250 mm, Rainin Instruments Co., Woburn, MA) using a linear solvent gradient from 49.9% CH 3 CN, 49.9% H 2 O, 0.1% HOAc to 99.9% CH 3 CN, 0.1% HOAc over 40 min at 1 ml/min. Products were quantified by on-line liquid scintillation using a Radiomatic Flo-One ␤-Detector (Radiomatic Instruments, Tampa, FL).
For the characterization of EPA metabolites, the organic soluble material extracted from solutions containing [1-14 C]EPA (100 M final concentration, 0.1 Ci/mol), 5 nM P450BM-3, and 1 mM NADPH was resolved by RP-HPLC as described. The radioactive fraction eluting from the HPLC column with the retention time of authentic 17,18epoxy-EPA (18.7 min) was collected batchwise and further characterized. To confirm the epoxide nature of the metabolite, an aliquot of the purified material (2-5 g) was incubated, under an argon atmosphere and with constant mixing, with 0.25 ml of a mixture containing 20% EtOH, 40% H 2 O, and 40% glacial HOAc. After 12 h at room temperature, the reaction mixture was diluted with 1 ml of 0.1 M KCl and extracted twice with equal volumes of ethyl ether. The resulting product co-eluted in RP-HPLC with synthetic vic-17,18-dihydroxy-5, 8,11,14eicosatetraenoic acid (R t Ϸ 10 min) and, after derivatization to the corresponding TMS ether, PFB ester, showed a NICI/GC/MS fragmentation pattern identical to that of an authentic standard (Fig. 1A). For regiochemical analysis, an aliquot of the hydrated epoxide (5 g) was hydrogenated over PtO 2 , derivatized to the corresponding PFB ester (24), and purified by SiO 2 chromatography. The dry residue was dissolved in 200 l of NaIO 4 (10 mg/ml in 70% CH 3 OH) and, after 2 h at 50°C, the product was extracted into hexane and purified by RP-HPLC using a linear solvent gradient from 49.9% CH 3 CN, 49.9% H 2 O, 0.1% HOAc, to 99.9% CH 3 CN, 0.1% HOAc over 40 min at 1 ml/min (R t ϳ 43 min). The purified aldehyde, resulting from oxidative cleavage of the vic-diol precursor, was dried under a stream of N 2 , mixed with 200 l of 0.5% solution of methoxylamine hydrochloride in pyridine (Pierce), incubated 3 h at 30°C, extracted into hexane, and then characterized by NICI/GC/MS (Fig. 1B).
For structural analysis, the organic soluble products extracted from solutions containing [1-14 C]ETA (100 M, final concentration, 0.2 Ci/ mol), 2 nM P450BM-3, and 1 mM NADPH were purified by RP-HPLC as above. The radioactive material eluting from the RP-HPLC column between 18 and 20 min was collected batchwise and, after solvent evaporation, resolved into fractions a, b, and c (R t ϳ 16.9, 23.1, and 37.1 min, respectively for a, b, and c) by NP-HPLC on a 5-m Dynamax Microsorb Silica column (4.6 ϫ 250 mm) using an isocratic solvent mixture composed of 0.5% 2-propanol, 0.1% HOAc, 99.4% hexane at a flow rate of 2 ml/min. After catalytic hydrogenation and derivatization to the corresponding PFB esters, fractions a, b, and c were characterized by NP-HPLC on the above silica column using an isocratic solvent mixture composed of 0.2% 2-propanol, 99.8% hexane at a flow rate of 2 ml/min. In this chromatographic system, the PFB esters of hydrogenated a, b, and c (R t ϳ 20, 23, and 35 min, respectively) co-eluted with the PFB esters of synthetic 17-, 18-, and 19-hydroxyeicosanoic acid, respectively. The PFB esters of hydrogenated a, b, and c were further characterized by NICI/GC/MS.
Stereochemical Analysis of 17,18-Epoxyeicosatetraenoic Acid-A stream of O 3 in oxygen was passed for 2 h through a solution of enzymatically generated 17,18-epoxyeicosatetraenoic acid (1.87 mg) in 90% CH 3 OH, 10% CH 2 Cl 2 cooled to 0°C. NaBH 4 (2 mg) was added, and the mixture was stirred at room temperature. After 30 min, the reaction mixture was diluted with Et 2 O (2 ml) and H 2 O (2 ml). The organic phase was separated, and the aqueous phase was extracted once more with Et 2 O. The combined organic phases were evaporated in vacuo, and the residue was dried azeotropically with anhydrous benzene, then dissolved in dry pyridine (100 l) to which was added benzoyl chloride (15 l). After 12 h, the reaction mixture was diluted with CH 2 Cl 2 (2 ml), and a saturated aqueous solution of CuSO 4 (300 l) was added to effect phase separation. The organic layer was collected and concentrated in vacuo, and the residue was purified by PTLC (15% EtOAc, 85% hexane, R F ϳ 0.26) using standards of synthetic epoxy-benzoate in adjacent lanes as guides to the location of the enzymatically derived 3,4-epoxyhexan-1-yl benzoate.

RESULTS AND DISCUSSION
Rates of AA Oxidation by P450BM-3-The rate of fatty acid oxidation by P450BM-3 can be easily measured either polarographically using an oxygen electrode or spectrophotometrically monitoring absorbance changes at 340 nm. Seen in Fig. 2 is an oxygen electrode trace of O 2 consumption during the NADPH-dependent metabolism of AA by P450BM-3. The order of addition of reactants is crucial, i.e. NADPH must be added last to avoid inactivating the reductase domain of P450BM-3 (3,29). During the first few seconds after addition of NADP, the turnover number of P450BM-3 is 3.5 mol of O 2 consumed/min/ nmol, but decreases as oxidized products accumulate (Fig. 2). In experiments not shown here, the rate of NADPH oxidation, measured at 340 nm, was approximately the same as the rate of oxygen consumption per mol of P450 per min indicating a tight coupling of NADPH oxidation, O 2 consumption and substrate oxidation.
Within the first 30 s of incubation, the rates of AA utilization and product formation began to decrease showing the lack of a clear linear relationship between product formation and incubation time. As the rate of AA oxidation decreased, the recovery of polyoxygenated products, derived from secondary oxygenations, increased concomitantly, and became predominant 2-3 min after initiation. When limiting amounts of AA were added, as shown in Fig. 2, the ratio of O 2 consumed per mol of AA added was approximately 2, indicating that fatty acid polyoxy- genation had occurred as shown for palmitic acid (5,7). The limited solubility of AA made impractical attempts to increase its concentration, and thus prolong enzyme-substrate saturation. The rates of product formation shown in Table I were obtained at 30°C and after a 30 s incubation, and are approximations of the initial velocities. They are useful only for comparative purposes.
Among the fatty acids metabolized by P450BM-3, AA showed the highest oxidation rate, 3.2 mol of product formed/min/ nmol of P450BM-3 (Table I). To the best of our knowledge, the rates at which P450BM-3 catalyzes the redox coupled activation of molecular oxygen, the cleavage of the oxygen-oxygen bond, and the insertion of a reactive oxygen atom into the AA molecule are the highest ever reported for a mixed function oxidase, and in particular for a P450, NADPH-dependent, catalyzed reaction. It is of interest that these rates of metabolism (k cat Ϸ 50 s Ϫ1 ) are similar to those of the electron transfer from NADPH to the FMN cofactor of the reductase domain of P450BM-3 (k Ϸ 80 s Ϫ1 ) (29). More importantly, as shown below, under the reaction conditions employed, P450BM-3 displayed high regiochemical selectivity for the fatty acid -2 carbon and 14,15-olefinic bond, and generated 18-OH-AA and 14,15-epoxyeicosatrienoic acid (14,15-EET) as major reaction products (80 and 20% of the total products, respectively) (Table II). While small and variable amounts of 17-OH-AA were also formed, they accounted for less than 1% of the total products.
At difference with AA, the rates of epoxygenation of EPA at 17,18-olefinic bond remained constant within the first 2 min of incubation. Thus, initial velocities were measured 1 min after NADPH addition. However, as with AA, within the first 30 s of incubation at 30°C, the relationship between the incubation time and the extent of ETA oxidation was nonlinear. Consequently, the reaction rates shown in Table I, were obtained after 30 s and are given as estimates of the initial rates and are useful only for comparative purposes. As with AA and palmitic (5, 7) acid, extended incubation times (2-5 min) resulted in polyoxygenation of ETA.
Spectral Binding Constants for P450BM-3-To assess the binding of AA and EPA to P450BM-3, the enzyme was titrated with substrate, and absorbance changes in the Soret region were measured. As shown in Fig. 3, the AA-dependent decrease in absorbance at 418 nm and the increase in absorbance at 390 nm is indicative of substrate binding to P450s and the conversion of the heme iron from the low to the high spin state (30). The K s for AA binding to P450BM-3 is 1.2 Ϯ 0.1 M and that for EPA 1.6 Ϯ 0.5 M (Fig. 3, Table I). Both have approximately the same binding constants as those of saturated fatty acids (3,5); however, the latter substrates are oxidized at a substantially lower rate than the eicosanoids. (The K s of ETA was not determined because of its limited availability.) AA Oxidation by P450BM-3-Incubations of AA with P450BM-3 resulted in the rapid, NADPH-dependent formation of two predominant metabolites with RP-HPLC retention times of 15.5 and 22.6 min (Fig. 4, fractions A and B). Additional amounts of a radiolabeled product (Ͻ1% of the total products), with a retention time identical to authentic 17-OH-AA (approx-imately 14.5 min) were recovered. To minimize secondary metabolism due to the unusually high reaction rates, the hemoprotein and the fatty acid concentrations were maintained at Յ5 nM and Ն50 M, respectively. For the first 2 min of incubation under these conditions, primary oxygenation products were recovered almost exclusively, i.e. derived from the insertion of a single atom of atmospheric oxygen. Higher enzyme concentrations, longer incubation times, and/or lower fatty acid concentrations led to complex product profiles resulting from reiterative oxygenation of primary metabolites. Among these products, 14,15-epoxy-18-hydroxyeicosatrienoic acid was identified by GC/MS.
For structural analysis, the reaction products shown in Fig.  4 were collected and, after solvent evaporation, characterized utilizing a combination of functionality-specific chemical derivatization reactions and chromatographic and NICI/GC/MS techniques. The reaction product, with a RP-HPLC elution time of 15.5 min (Fig. 4, fraction A), was conclusively identified as 18-OH-AA based on the following: 1) co-elution on reversed and normal phase HPLC with an authentic standard, 2) coelution of its PFB-TMS derivative on capillary GC with a similarly derivatized standard, 3) NICI/GC/MS analysis of its PFB-TMS derivative showed the presence of an intense ion fragment at m/z 407 (loss of PFB, base peak) indicative of a monohydroxylated metabolite, 4) catalytic hydrogenation followed by conversion to the corresponding PFB-TMS derivative yielded a product that co-eluted on capillary GC with a similarly derivatized authentic standard, and 5) NICI/GC/MS analysis of its hydrogenated PFB-TMS derivative yielded a fragment ion at m/z 415 (loss of PFB, base peak), consistent with the presence of four double bonds in the parent molecule that had been reduced upon hydrogenation.
Fraction B (Fig. 4, retention time 22.6 min) was shown to contain 14,15-EET based on the following chromatographic and mass spectral evidence: 1) co-elution in reversed and normal phase HPLC with authentic 14,15-EET (23, 24); 2) co-elution in capillary GC of its PFB ester derivative with authentic 14,15-EET-PFB (23, 24); 3) the NICI mass spectrum of the PFB ester of fraction B was identical to that of synthetic 14,15-EET-PFB with major ion fragments at m/z: 319 (base peak, loss of PFB), 303 (20% of base peak, loss of PFB and oxygen) and 301 (12% of base peak, loss of PFB and water) (24); and 4) the NICI mass spectrum of the PFB ester of hydrogenated B was nearly identical to that of the PFB ester of authentic 14,15-epoxyeicosanoic acid with major ion fragments at m/z: 325 (base peak, loss of PFB), 309 (10% of base peak, loss of PFB and oxygen) and 307 (8% of base peak, loss of PFB and water) (24,26).
Hydroxylation of saturated fatty acids (e.g. palmitic acid) by P450BM-3 occurs preferentially at the -2 carbon atom, with -1 and -3 hydroxylated products accounting for a substantial portion of the total metabolism (3,5). In contrast, P450BM-3 hydroxylates the AA C18 carbon in a nearly exclusive fashion, i.e. 18-OH-AA accounts for better than 99% of the overall sp 3 carbon hydroxylation. Hydroxylation at the fatty acid -3 carbon accounted for less than 1% of the total products and no metabolism was observed at the -1 or carbons (Fig. 5). Even though amino acid sequence analysis indicates a 25-30% sequence similarity between P450BM-3 and CYP4A gene subfamily isoforms (a group of mammalian microsomal fatty acid -hydroxylases) (4), none of the CYP4A isoforms catalyzes fatty acid or AA -2 oxidation or epoxidation (11,13). However, the formation of 18-OH-AA and of 18(R)-OH-AA by rat hepatic and monkey seminal vesicle microsomal fractions, respectively, has been reported (31). Recent studies have also shown that hydroxylation at the C-18 position of AA was catalyzed by CYP1A1, CYP1A2, and CYP2E1 (11,13,25). Hydroxylations at the sp 3 carbons near or at the fatty acid methyl end require the delivery by the protein catalyst of a reactive, heme-bound oxygen species to a ground state sp 3 carbon atom. It is therefore likely that for all these reactions the oxygen chemistries and the mechanism(s) are similar yet independent of the fatty acid chain length and/or degree of saturation. Nevertheless, compared to saturated fatty acids, the AA molecule imposes additional steric requirements on the active site of the enzyme. During AA C18 hydroxylation, the P450 binding/active site must position the acceptor carbon atom not only in optimal proximity to the heme-bound active oxygen, but also with complete segregation of the fatty acid reactive bis-allylic methylene carbons at C-7, C-9, and C-13 and the 5,6-, 8,9-, 11,12-, and 14,15-olefinic bonds.
To further delineate the structural determinants responsible for AA active site binding and productive spatial orientation, the chiral properties of the metabolites formed by P450BM-3 were characterized. In previous studies it was shown that AA epoxidation and -2 carbon hydroxylation by mammalian P450 isoforms proceeded with a degree of enantiofacial selectivity unprecedented for P450 catalyzed oxidations of acyclic, unbiased molecules such as AA (9,11,13). As shown in Table II, P450BM-3 hydroxylates the -2 carbon atom of AA in a highly asymmetric mode generating 18(R)-OH-AA with 96% optical purity. Similarly, P450BM-3 catalyzed AA epoxidation yields 14(S),15(R)-EET with 99% optical purity. It thus appears that P450BM-3 evolved a highly structured and spatially rigid substrate binding site, capable of accommodating a polyunsaturated fatty acid such as AA in optimal orientation with regards to the heme-bound active oxygen. The unprecedented high enantiofacial selectivity of this protein, in conjunction with its moderate regioselectivity, indicates that during catalytically productive binding: 1) the AA carbon-carbon rotational freedom is substantially restricted, and 2) the active site spatial coordinates allow for a moderate degree of substrate lateral displacement. Finally, while 18(R)-OH-AA is the predominant enantiomer formed by mammalian microsomal enzymes (31), 14(S),15(R)-EET is the predominant enantiomer found in vivo in rat liver, plasma, and kidney (26,32,33).
EPA and ETA Oxidation by P450BM-3-The metabolism of EPA and ETA, two AA analogs (Fig. 5), was studied to probe the role that substrate structural features, in particular C-H bond acceptor chemistry, play as determinants of P450BM-3 catalytic outcome. We selected these AA analogs because: 1) radiolabeled EPA and limited amounts of radiolabeled ETA were commercially available, 2) all of the molecules are of the same carbon length, 3) between carbons C-1 and C-13, all three fatty acids (AA, EPA, and ETA) are structurally identical, 4) ETA, with its sp 3 C-14 -C-15 carbons, was used to probe the role of the AA 14,15-olefinic bond in inducing oxygenation of  C-H bonds distal to the fatty acid -2 carbon, the substrate's most metabolically active carbon atom (Table II), and 5) EPA, with the extra olefinic bond at C-17-C-18 allowed for the analysis of C-H bond acceptor reactivity in the enzyme's regioselectivity of oxygen insertion.
Incubation of EPA with P450BM-3 and NADPH resulted in the time dependent formation of a radioactive product with a RP-HPLC retention time of 18.7 min (Fig. 6). Importantly, and at difference with AA, longer incubation times (Ն4 min) and/or higher enzyme concentrations (Ն10 nM) did not result in substantial changes in the profile shown in Fig. 6, indicating that the EPA oxygenated metabolite is a poor substrate for P450BM-3. Structural analysis demonstrated that the sole product of EPA metabolism by P450BM-3 was 17,18-epoxyeicosatetraenoic acid (17,18-epoxy-EPA), i.e. no other product was generated during the first 3 min of incubation at 30°C; however, during prolonged incubations, the 17,18-epoxy-EPA underwent partial chemical hydration to 17,18-dihydroxyeicosatetraenoic acid. Additionally, as shown in Table I, EPA was metabolized at approximately half the rate estimated for AA. On the other hand, the EPA 17,18-olefinic bond is epoxidized at more than double the rate of the AA 14,15-olefinic bond (Tables  II and III). Significantly, under the experimental conditions used, epoxidation of the EPA 14,15-olefinic bond was negligible (less that 1% of the total reaction products). The identification of the P450BM-3 metabolite as 17,18-epoxy-EPA was based on the following: 1) co-elution in reversed and normal phase HPLC with an authentic standard; 2) co-elution in capillary GC of its PFB ester with a similarly derivatized standard; 3) the mass spectrum of its PFB ester, under NICI conditions, was nearly identical to that of authentic 17,18-epoxy-EPA with major ion fragments at m/z: 317 (base peak, loss of PFB), 301 (16% of base peak, loss of PFB and O) and 299 (22% of base peak, loss of PFB and H 2 O); 4) acid catalyzed hydrolysis (in 50% HOAc) yielded a radioactive product with RP-HPLC characteristics identical to that of synthetic 17,18-dihydroxyeicosatetraenoic acid; 5) the NICI/GC/MS spectra of the PFB ester-TMS ether derivative of the metabolite hydration product matched that of a similarly derivatized standard with major ion fragments at m/z: 479 (base peak, loss of PFB), 407 (25% of base peak, loss of PFB and TMSO), 317 (15% of base peak, loss of PFB, TMS, and TMSOH) (Fig. 1A); 6) acid catalyzed hydrolysis followed by catalytic hydrogenation and NaIO 4 oxidative cleavage of the diol carbon-carbon bond afforded a radioactive aldehyde with a RP-HPLC retention time identical to that of the synthetic PFB ester of 17-oxo-heptadecanoic acid; and 7) incubation of the above aldehyde with methoxylamine hydrochloride yielded the corresponding PFB ester-methoxime derivative with a diagnostic molecular weight of 313 (Fig. 1B).
For chiral analysis of the 17,18-epoxy-EPA, we initially utilized degradative ozonolysis followed by derivatization to the corresponding 3,4-epoxyhexan-1-yl benzoates. The Chiralcel OC HPLC properties of the synthetic standards were then compared to those of the biologically derived sample (see "Experimental Procedures" for further details). For routine nondestructive analysis, the optical antipodes of methyl 17,18-epoxyeicosatetraenoate were resolved with baseline separation by chiral phase HPLC on a Chiracel OB column as shown in Fig. 7. Absolute configurations were assigned based on the results obtained by the above degradative ozonolysis procedure. Chiral analysis of the EPA epoxygenase metabolite demonstrated that its biosynthesis was highly asymmetric and generated 17(S),18(R)-epoxy-EPA with 97% optical purity (Table III). The high degree of regio-and stereochemical selectivity shown by P450BM-3 during the metabolism of EPA, illustrates the key role played by the C-H acceptor in directing catalytic outcome. Thus, assuming similar active site binding coordinates for AA and EPA, the electron cloud associated with the EPA 17,18-olefinic bond is an efficient trap for the heme-bound oxygenating intermediate, more so than that associated with the AA -2 sp 3 carbon orbital. Therefore, C-H acceptor chemistry and active site binding coordinates, as opposed to reactive oxygen chemistry and/or heme-oxygen redox properties, provide a coherent explanation for product chemistry. It is of interest that for both the AA 14,15-and EPA 17,18-epoxygenases, oxygen was delivered to the olefinic bond's si,re-face.
As mentioned, ETA lacks olefinic bonds in the vicinity of its -3 sp 3 carbon (i.e. C-17-C-18), as well as at C-14 -C-15 (Fig. 5), and as a result the carbons at or near its methyl terminus are chemically similar to those of most saturated fatty acid substrates. Incubations of [1-14 C]ETA (50 -100 M) with P450BM-3 resulted in the NADPH-dependent formation of radioactive metabolites with an average HPLC retention time of approximately 19.3 min (Fig. 8, fraction A). The broad, asymmetric nature of fraction A (Fig. 8) suggested the presence of more than one product. Fraction A was collected from the HPLC eluate and resolved by normal phase HPLC into three radioactive metabolites (Fig. 9). After HPLC purification, fractions a, b, and c ( Fig. 9)    For ETA, as was the case for AA and palmitic acid, extended incubation times (Ն2 min) resulted in further metabolism of 17-, 18-, and 19-hydroxy-ETA. Among the secondary oxidation products, 18-oxo-ETA was identified using GC/MS. As shown in Table I, P450BM-3 metabolized ETA at rates comparable to those obtained with AA. Under conditions favoring primary metabolism, the enzyme generated 17-, 18-, and 19-hydroxy-ETA in a 2.4:2.2:1 molar ratio, respectively (Table IV). Importantly, the moderate degree of regioselectivity of the enzyme for the ETA -1, -2 or -3 sp 3 carbon atoms is similar to that previously obtained with several saturated fatty acid substrates (5,34).
Thus, for all the polyunsaturated fatty acids tested here, the preferred sites for the P450BM-3 catalyzed oxygen insertion were the -2 and -3 carbon atoms. P450BM-3 metabolized the -2 or the -3 carbons of ETA at rates comparable to those of EPA epoxidation (in micromoles of product/min/nmol of P450: 1.2 and 1.1 for 17-and 18-OH-ETA, and 1.4 for 17, 18-epoxy-EPA, respectively) (Tables III and IV). On the other hand, 19-OH-ETA and 14,15-EET are generated at lower rates ( Ta-TABLE IV P450BM-3 oxidation of ETA P450BM-3 was incubated with [1-14 C]ETA (1-5 Ci/mol) NADPH (5 nM, 100 M, and 1 mM final concentrations, respectively) and an enzymatic NADPH regenerating system. After 30 s at 30°C, organic soluble products were extracted into acidified ethyl ether, resolved by RP-HPLC and quantified by on-line liquid scintillation. Product distribution and rates of individual metabolite formation were obtained after normal phase resolution (Fig. 9) of the radioactive material eluting from the RP-HPLC column with a retention time of 18.6 min (Fig. 8)  Ci/mol) of a mixture of the synthetic and biological product were resolved using a Chiracel OB column (4.6 ϫ 250 mm) and a solvent mixture of 0.03% 2-propanol, 99.7% hexane at 2 ml/min, with UV detection at 210 nm. For quantification, the enantiomers were collected individually and, after solvent evaporation, their radioactivity determined by liquid scintillation. Absolute configurations were assigned as described in the text. bles II and IV). It is therefore likely that all three molecules occupy more or less similar spatial coordinates in the active site of P450BM-3.
Modeling AA Binding to the Active site of P450BM-P-Only the atomic coordinates of the substrate-free form of P450BM-P are known (19). Attempts to soak either AA or palmitic acid into preformed crystals of P450BM-P to obtain crystals of substrate bound enzyme has resulted in either low occupancy of the active site by the substrate fatty acid or disorder in the crystals. 2 Thus, to enable us to visualize substrate binding in the active site of P450BM-3, we have utilized a molecular modeling approach in which AA was "docked" into the substrate access channel and active site. The fatty acid carboxylate was positioned within charge coupling distance of the guanidinium group of Arg 47 at the mouth of the access channel (19). The remainder of the AA molecule was built into the volume of the substrate access channel defined by Arg 47 at the surface and Phe 87 above the heme, maintaining the appropriate bond angles and with reduced van der Waals contacts with amino acid side chains and backbone atoms. During energy minimization of AA in the substrate access channel and active site, Phe 87 moved slightly allowing the -end of the AA molecule to occupy more readily an existing active site cavity (Fig. 10). In this model, the -end is bent upward and toward the fatty acid 14,15-olefinic bond. The pro-R hydrogen of C-18 of AA was closest to the heme iron. An additional displacement of F87 would position the AA 14,15-olefinic bond in optimal proximity to the P450BM-3 heme iron. We propose that AA bound P450BM-3 oscillates between these two alternate conformers in response to the positioning of Phe 87 with respect to the heme iron. Clearly AA must be as close to the heme iron of P450BM-3 as depicted in Fig. 10 to exhibit the stereochemical oxidation of AA described here.
In conclusion, while epoxidation and a combination of epoxidation and sp 3 hydroxylation have been demonstrated in mammalian P450s of the CYP1 and CYP2 families (9,11,12,25,35,36), with highly enantioselective epoxidations catalyzed by CYP2C11 and CYP2C23 (36,37), none of the characterized eukaryotic AA epoxygenases appears capable of epoxidizing a single AA olefinic bond with complete exclusion of the other three. The present study has demonstrated that the common reactive oxygen intermediate that has been proposed for P450 monooxygenations (38) can catalyze epoxidation and sp 3 carbon hydroxylation of AA to 18-OH-AA and 14,15-EET (80 and 20% of total products, respectively) in the reaction catalyzed by P450BM-3. In addition, by comparing the stereoselective epoxidation of EPA at the 17,18-olefinic bond, and by contrast, the less selective 17-, 18-, and 19-hydroxylation of ETA with that of AA, we find that in contradistinction to the nature and chemical properties of the reactive oxygen intermediate(s), the chemistry of the reaction products are critically dependent on: 1) the chemical properties of the acceptor C-H bonds and 2) the optimal orientation of the C-H acceptor with respect to the heme-bound reactive oxygen intermediate. We also can conclude from these stereoselective oxidations of AA and EPA that the active site geometry responsible for substrate binding and orientation must restrict the freedom of substrate C-C bond rotation while, at the same time, allowing some degree of substrate lateral mobility that, for P450BM-3 permits 14,15-EET formation. This substrate lateral mobility may be controlled by the positional relationship of the active site residue Phe 87 to the heme iron. Thus, substrate chemistry and protein structural features, as opposed to oxidant chemistry, are the key determinants of catalytic outcome. Finally, in as much as the biological function(s) of P450BM-3 are yet to be determined, the high catalytic rates and unprecedented degree of regio-and stereochemical selectivity displayed by the enzyme during the metabolism of AA and EPA are indicative of evolutionary specialization and suggest a role for P450BM-3 in the metabolism of bacterial unsaturated fatty acids.