The Lipoxygenase Activity of Myoglobin OXIDATION OF LINOLEIC ACID BY THE FERRYL OXYGEN RATHER THAN PROTEIN RADICAL*

Linoleic (9(Z),12(Z)-octadecadienoic) acid is oxidized by sperm whale myoglobin and HzOz to an (9S):(9R) enantiomer mixture of g-hydroperoxy-10(E),12(Z)-octadecadienoic acid. Neither the 9,lO- nor 12,13-epoxide of linoleic acid, nor 9-hydroxy-lO(E),12(2)-octadecadienoic nor 13-hydroperoxy-9(Z),ll(E)-octa- deca-dienoic acids, is detectably formed. Incubations with [(llR)-’H]- and [(11S)-2H]linoleic acids indicate that the pro-R hydrogen is abstracted 76% of the time. An H64V mutant in which access to the heme crevice is increased oxidizes linoleic acid exclusively by abstrac- tion of the pro-R hydrogen to give the (9s)-hydroperox-ide. Spectroscopic studies show that the Kd value for binding of linoleic acid to myoglobin is similar to the K , value for its oxidation and indicate that linoleic acid reduces the ferryl species to the ferric state. The stereochemical results, supported by 180-labeling studies, de- finitively Y146FIY151F (22, 23). [(llR)-2Hl- and [(11S)-2Hllinoleic acids were obtained as reported (28). Mass spectrometric analysis indicates that the 11(R) isomer was 16.1% monodeuterated and 83.9% undeuteriated. The 11(S) isomer was 19.7% deuteriated and 80.3% undeuteriated. and m-chloroperbenzoic and with glass-distilled, and were pretreated (Bio-Rad).

Linoleic (9(Z),12(Z)-octadecadienoic) acid is oxidized by sperm whale myoglobin and HzOz to an 84:16 (9S):(9R) enantiomer mixture of g-hydroperoxy-10(E),12(Z)-octadecadienoic acid. Neither the 9,lO-nor 12,13-epoxide of linoleic acid, nor 9-hydroxy-lO(E), 12(2)octadecadienoic nor 13-hydroperoxy-9(Z),ll(E)-octadeca-dienoic acids, is detectably formed. Incubations with [(llR)-'H]-and [(11S)-2H]linoleic acids indicate that the pro-R hydrogen is abstracted 76% of the time. An H64V mutant in which access to the heme crevice is increased oxidizes linoleic acid exclusively by abstraction of the pro-R hydrogen to give the (9s)-hydroperoxide. Spectroscopic studies show that the Kd value for binding of linoleic acid to myoglobin is similar to the K , value for its oxidation and indicate that linoleic acid reduces the ferryl species to the ferric state. The stereochemical results, supported by 180-labeling studies, definitively rule out a significant role for singlet oxygen in the myoglobin-catalyzed, HzOz-dependent oxidation of linoleic acid. The myoglobin protein radical formed with HzO2 also plays no part in the reaction because the K , and V, , values for the oxidation of linoleic acid are similar for native myoglobin and two mutants (K102Q/ Y103FIY146FN151F and H64V/K102Q/Y103F/Y146F/ Y151F) with no tyrosine residues. Furthermore, the rate of formation of the 9-hydroperoxide is not changed if the protein radical is allowed to decay before linoleic acid is added. The results establish that linoleic acid is oxidized within the heme crevice by reaction with the ferryl oxygen rather than a protein radical. They indicate, furthermore, that hydrogen abstraction and oxygen addition occur in an antarafacial manner and suggest a specific model for binding of linoleic acid within the myoglobin active site.
Myoglobin and hemoglobin function as reversible carriers for molecular oxygen but, in the presence of HzOz or alkyl peroxides, catalyze the oxidation of small substrates. Myoglobin and hemoglobin are generally limited to the catalysis of facile reactions such as arylamine oxidation (11, thioether sulfoxidation (2), and olefin epoxidation (3), although hydroxylation of cyclohexane by hemoglobin has been reported (4). Potentially the * This work was supported by National Institutes of Health Grant most important of these reactions, in physiological terms, is the oxidation of unsaturated fatty acids (5)(6)(7)(8). The initiation of lipid peroxidation by myoglobin has been postulated to contribute to myocardial reperfusion injury (9)(10)(11) and the corresponding hemoglobin-catalyzed reaction to erythrocyte lysis (12).
The mechanisms responsible for the HzOz-supported oxidation of organic substrates by myoglobin have been partially elucidated. Reaction of myoglobin with HzOz generates a moderately stable ferryl (Fe'" = 0 ) species and a transient protein radical (13,141. Strong support for the ferryl species is provided by NMR (151, resonance Raman (161, and x-ray edge absorption (17) studies. The protein radical is readily detected by freezequench or stopped flow methods, although the measured intensity of the radical signal corresponds to no more than half of that expected from the extent of ferryl complex formation (14,18). The EPR properties of the radical suggest that the unpaired electron is located on 1 or more tyrosine residues (18, 191, a n inference supported by the finding that 2 of the 3 tyrosine residues (Tyr-103 and Tyr-151) are involved in free radical protein-protein and heme-protein cross-linking reactions (20, 21). We have recently shown by site-specific mutagenesis, however, that a protein radical is still formed in mutants in which the 3 tyrosines have been singly, in pairs, or collectively replaced by phenylalanines (22). The protein radical is also observed when His-64, a residue that blocks access to the heme crevice, is replaced by a valine, but not when the His-64 mutation is combined with the 3 Tyr -Phe replacements (23). It therefore appears that the unpaired electron density resides primarily on the tyrosine residues but is readily transferred from 1 to another of a subset of oxidizable residues in the protein.
At least two mechanisms exist for the oxidation of aryl olefins. One mechanism produces the epoxide with retention of the olefin stereochemistry and incorporation of an oxygen atom from the peroxide. This epoxidation reaction is undoubtedly the result of a cytochrome P450-like ferryl oxygen transfer mechanism (3,23,24). The second epoxidation mechanism results in scrambling of the olefin stereochemistry and predominant incorporation of an atom of molecular oxygen. The nature of this second epoxidation mechanism is uncertain, but the properties of the reaction are consistent with a cooxidation mechanism in which epoxidation is mediated by the peroxy radical formed when molecular oxygen binds to the protein radical (3,23,24).
It is not known whether the ferryl species or the protein radical initiates the myoglobin-catalyzed oxidation of unsaturated lipids. Formation of a species with A, , , = 412 nm when linoleic acid is added at pH 7.3 to metmyoglobin (A,,, = 408 nm) or ferryl myoglobin (A,,, = 420 nm) suggests that binding of the fatty acid to either state of the hemoprotein produces a common low-spin, ferric complex (8). Galaris et al. ( 8 ) reported that linoleic acid is oxidized by equine myoglobin and Hz02 to 9-hydroperoxy-1O(E),12(Z)-octadecadienoic acid and a second major, unidentified product. A linoleic acid hydroperoxide also appears to be formed in incubations of linoleic acid with hemoglobin and H202 (5). Heme loss is observed during these H202dependent lipid peroxidation reactions (5,8).  reported that the ability to form malondialdehyde from unsaturated lipids decreases 50% within 10 min of adding the peroxide to myoglobin but then decreases very slowly over more than 1 h, suggesting that two distinguishable oxidizing species are formed. Davies (25) has spin-trapped a radical in incubations of myoglobin with HzOz and 5,5-dimethyl-l-pyrroline-Noxide and has proposed that the trapped radical is a tyrosine peroxy radical formed by addition of oxygen to the Tyr-103 phenoxy radical. This identification has been questioned because the EPR signal is the same under an atmosphere of I7O2 as 1 6 0 2 (261, but Davies and co-workers argue from the decrease observed in the intensity of the spin-trapped EPR signal upon the addition of erythrocyte membranes that lipid peroxidation is initiated by the tyrosine peroxy radical (27).
We report here studies of the H202-supported oxidation of linoleic acid by sperm whale myoglobin which establish that the ferry1 oxygen rather than the protein radical initiates the reaction and suggest a specific binding mode for linoleic acid in the myoglobin heme crevice.
solution of NaBH, in ethanol (75 pg/ml, 2 pmol). The solution was allowed to stand at 0 "C for 20 min and a t -25 "C for a n additional 20 min before brine (1 ml) was added and the mixture was acidified to pH 3 with HCI. The alcohol product was extracted with two 3-ml portions of diethyl ether, the combined extracts were dried over MgSO,, and the solvent was removed under a stream of argon. The residue was taken up i n 1 ml of a diethyl ether solution of diazomethane, and the solution was allowed to stand a few minutes before the ether was removed. The residue was treated with BSTFA (55 "C for 120 min) to derivatize the alcohol. The product thus obtained was analyzed by gas-liquid chromatography on a DB-5 capillary column programed to rise from 190 to 240 "C at 1 "C min-'. The single product peak that eluted at 35.5 min was shown by GC-MS analysis to be methyl 13-trimethylsilyloxy-S(Z),Il(E)-octadecadienoate (31).
Analytical Methods-Absorption spectra were recorded on a n SLM-Aminco DW-2000 UV-Vis spectrophotometer. HPLC was done on a Hewlett-Packard model 1090 Series I1 instrument equipped with a diode array detector and interfaced to a Hewlett-Packard HPLC ChemStation. Gas chromatography was done on a Hewlett-Packard model 5890 gas chromatograph equipped with a flame ionization detedor and interfaced to a Hewlett-Packard 3365 ChemStation. GC-MS analyses were performed on a Hewlett-Packard model 5890 gas chromatograph coupled to a VG-70 mass spectrometer.
Spectroscopic Studies-Varying amounts of linoleic acid in ethanol were added to the sample cuvette, and a n equal amount of ethanol to a matched reference cuvette, each of which contained 1 ml of 0.2 M phosphate buffer (pH 7.4) and 10 p~ myoglobin. The solutions were mixed and the difference spectra recorded. The value of & was computed from is the concentration of myoglobin, & is the binding constant of linoleic acid to myoglobin, AA is the peak-to-trough absorbance in the difference spectrum, and A€ is the difference in molar absorptivity of the free and bound myoglobin.
Oxidation of Linoleic Acid by Myoglobin-Reaction mixtures ( 1 ml) contained myoglobin (60 p~) , the indicated amount of linoleic acid, and 900 HzOz in 0.2 M phosphate buffer (pH 7.4). The linoleic acid was added as a n ethanol solution to give a final ethanol concentration of 10%. Ethanol up to a concentration of 10% had no effect on the oxidation of linoleic acid by sperm whale myoglobin. The myoglobidinoleic acid solution was preincubated for 10 min a t 4 "C before the reaction was initiated by adding the precooled peroxide solution. Incubations were carried out at 4 "C for 30 min. The reaction mixtures were then extracted twice with 4 ml of ether, the combined ethereal extracts concentrated to dryness, and the residue dissolved in 0.25 ml of ethanol. A 50-111 aliquot was analyzed by HPLC. Control incubations containing DETAPAC (1,5, or 25 m~) or mannitol (25 or 50 m~) , or without either myoglobin or H,Oz, were camed out in a similar manner. Mixtures containing known 13-hydroperoxyoctadecadienoic acid were similarly incubated and worked up to generate standard curves for product quantitation. Product analysis was done using a 5 pm, 250 x 4.6 mm Partisil ODS-3 column (Alltech) eluted with acetonitrile/water/phosphoric acid (65:35:0.1) at a flow rate of 1 ml min" (8). The eluent was monitored a t 217 and 234 nm with a diode array detector. Linoleic acid is detected a t 217 nm 9-, and 13-hydroperoxyoctadecadienoic acid is detected at both 217 and 234 nm (with a higher extinction coefficient at 234 nm). Under these conditions, 9-and 13-hydroperoxyoctadecadienoic acids elute at 8.7 min and linoleic acid a t 26 min. 13-Hydroperoxyoctadecadienoic acid and 13-hydroxyoctadecadienoic acid were resolved on the same Partisil ODS-3 column with a mobile phase consisting of acetonitrile/water/ phosphoric acid (55:45:0.1) a t a flow rate of 1 d m i n : 13-hydroxyoctadecadienoic acid, 16 min; 13-hydroperoxyoctadecadienoic acid, 17.6 min.
Incubations with ['801H,0z (98% l80), carried out as described above, were extracted with 4 ml of ether, and the combined extracts were concentrated to dryness. The residue was taken up in 1 ml of a cold solution of NaBH, in ethanol. The solution was allowed to stand at 0 "C for 20 min and at -25 "C for a n additional 20 min before brine (1 ml) was added and the mixture acidified to pH 3 with HCl. The acidic solution was extracted with ether (2 x 3 ml), the combined extracts were dried over anhydrous MgSO,, and the solvent was removed under a stream of argon. The residue was taken up in 1 ml of a solution of diazomethane in ether and the mixture allowed to stand a few minutes before the ether was removed. The residue was treated with BSTFA (55 "C for 120 mid to derivatize the methylated alcohol. The product was analyzed by GC-MS as described for the analysis of 13-hydroperoxyoctadecadienoic acid. Incubations with [(11R)-2H]-and [(US )-2Hllinoleic acids were carried out as above using the native protein and the H64V/K102~103F/Y146F/Yl5lF mutant. To determine the kinetic constants for linoleic acid, a minimum of seven concentrations of linoleic acid, including concentrations higher and lower than the K , value, were assayed. Concentrations of linoleic acid above 1 n m were turbid, however, so the kinetic curves reflect nominal rather than actual concentrations at the higher concentrations.

RESULTS
Do Linoleic a n d Related Fatty Acids Bind in the Heme Crevice of Myoglobin?-Spectroscopic binding studies confirm the report of Galaris et al. ( 8 ) that binding of linoleic acid to sperm whale myoglobin perturbs the heme chromophore (Fig. 1). The spectroscopic dissociation constant (Kd -1 mM) determined from the concentration dependence of the difference spectrum is comparable to the value of K, (2 mM) determined by kinetic analysis of product formation ( Table I). The Kd values (0.2-0.3 mM) for binding of linoleic acid to two mutants of sperm whale myoglobin, KlO2Q/YlO3F/Y146F/Y151F, in which the 3 tyrosine residues have been replaced by phenylalanines, and H64V/ K102Q/Y103F/Y146F/Y151F, in which His-64 has additionally been mutated to a n alanine, are 3-and &fold lower, respectively, than that for the native protein ( Table I) The crystal structure shows that these two mutations cause little alteration of the protein structure.
Nature of the Products Formed in the Oxidation of Linoleic Acid by Sperm Whale Myoglobin-The oxidation of linoleic acid by sperm whale myoglobin, as reported by Galaris et al. (€9, produces 9-hydroperoxyoctadecadienoic acid.3 Control experiments in which the authentic hydroperoxides were incubated with myoglobin indicate that both 9-hydroperoxy-and 13-hydroperoxyoctadecadienoic acids are stable under the incubation conditions. The identity of the 9-hydroperoxide product was confirmed by reduction of the hydroperoxide to the alcohol, trimethylsilylation, methyl esterification, and comparison of the mass spectrum of the resulting product with that of an authentic sample (34). The mass spectrum is distinguished by major differences in several ion intensities from that of the corresponding derivative of 13-hydroperoxyoctadecadienoic acid prepared by oxidation of linoleic acid with soybean lipoxygenase. Although the 9-and 13-hydroperoxide products are not resolved by the chromatographic method used for routine monitoring of the reaction, they are well resolved by the chiral HPLC method used to determine product stereochemistry (see below). The retention time of the methylated alcohol derivative of the metabolite in the chiral HPLC system clearly differs from that of the corresponding derivative of 13-hydroperoxyoctadecadienoic acid (Fig. 2). Neither the 9,lO-nor 12,13-epoxide of linoleic acid, nor 9-hydroxyoctadecadienoic acid, was detected as a metabolite despite a specific search for them with the help of authentic standards.
The absolute stereochemistry of the hydroperoxide product was determined by chiral HPLC. For this purpose, the linoleic acid-derived hydroperoxide was reduced to the alcohol with NaBH4, the carboxyl group was methylated, and the esterified 9-hydroxyoctadecadienoic acid isomers were chromatographed on a system that resolves the (13S), (13R), (9S), and (9R) isomers (32). Control experiments confirmed that soybean lipoxygenase produces the (13s) isomer with no more than traces of the other three isomers. In contrast, the myoglobincatalyzed oxidation of linoleic acid yields the (9s) and (9R) enantiomers in an 84:16 ratio without significant formation of the 13-hydroperoxide enantiomers ( Fig. 2A). The enantioselectivity of the reaction is even greater with the H64V and H64V/ K102Q/Y103F/Y146F/Yl5lF myoglobin mutants, both of which only detectably yield the (9s) enantiomer (Fig. 2B 1. The absolute stereospecificity of the hydrogen abstraction step was examined by analyzing the deuterium content of the 9-hydroperoxide obtained in incubations of wild type myoglobin or the H64V/K102Q/Y103F/Y146F/Yl5lF mutant with HzOz and either [(llR)-'H]or [(11S)-2Hllinoleic acid. Oxidation of [(llR)-'H]linoleic acid, which is 16.1% monodeuterated, by the wild type protein yielded hydroperoxide that was only 3.7% deuterated. The reaction thus proceeds with 77% abstraction of the pro-R deuterium atom. Analogous oxidation of [(11S)-2H]linoleic acid, which is 19.7% monodeuterated, yielded hydroperoxide that retained 15% of the label. Again, the oxidation proceeds with 76% abstraction of the pro-R hydrogen. Product analysis is simplified for the mutant protein because it gives exclusively the (9S)-hydroperoxide product. Mass spectrometric analyses of the products formed by the H64V/K102Q/Y103F/ Y146F/Y151F mutant indicate that the 9-hydroperoxide derived from [( llR)-'H]linoleic acid retains none of the deuterium label, whereas the 9-hydroperoxide from [(11S )-2Hllinoleic acid is 19.5% monodeuterated. This corresponds to essentially 100% retention of the deuterium label. The results unambiguously establish for the mutant that the hydrogen abstraction involves  highly stereospecific removal of the pro-R hydrogen and formation of the S hydroperoxide. Hydrogen abstraction and oxygen addition thus occur from opposite sides of the plane defined by the pentadienyl radical intermediate. The oxidation catalyzed by the wild type protein also involves antarafacial hydrogen abstraction and oxygen addition, but the absolute stereospecificity of the reaction is not as high as that for the mutant protein.
Nature of the Oxidizing Species Involved in Myoglobin-catalyzed Linoleic Acid Oxidation-Singlet oxygen (71, the hydroxyl radical, the myoglobin protein radical, and the myoglobin ferryl complex are the species that could be involved in the oxidation of linoleic acid by myoglobin and HzOz. The stereochemical results discussed above, however, argue strongly against a diffusible oxidizing species. Additional evidence against the involvement of singlet oxygen is provided by the finding that the oxygen incorporated into the hydroperoxide product in incubations with myoglobin and [1801H202 derives primarily from molecular oxygen and not from the peroxide. Mass spectrometric analysis of the hydroperoxide thus obtained, after reduction, methylation, and trimethylsilylation, showed that less than 9% of the hydroperoxide group was lsO-labeled, and thus that no more than 9% of the product could have arisen by reaction of linoleic acid with H202-derived singlet oxygen. It is likely, however, that the labeled hydroperoxide is obtained by reaction of the lipid radical with triplet molecular oxygen produced from the peroxide by the catalase-like activity of myoglobin (14). Reaction with oxygen formed by this mechanism is the most likely explanation for the reported finding that the hydroperoxide of linoleic acid is formed with myoglobin and Hz02 under anaerobic conditions (8). Earlier evidence against the involvement of hydroxyl radicals in myoglobin-catalyzed oxidation of unsaturated fatty acids (6, 10, 35) has been confirmed in our studies by the demonstration that the rate of formation of 9-hydroperoxyoctadecadienoic acid is not affected by up to 1 m~ concentrations of DETAPAC (not shown) or by high concentrations of mannitol. The rates of formation of g-hydroperoxyoctadecadienoic acid in the presence of 0,20, and 50 m~ mannitol were thus, respectively, 176 2 25, 190 * 33, and 182 2 21 pmol min" p"l. The protein radical or the ferryl species must therefore be directly responsible for initiating linoleic acid oxidation.
If linoleic acid oxidation is catalyzed by the ferryl species, the reaction should reduce the ferryl species back to the ferric form. The Soret band of the ferric enzyme, as shown by the difference spectrum, is shifkd from 408 nm to 412 nm when linoleic acid is bound. Addition of 1.5 equivalents of H2O2 to myoglobin shifts the Soret from 408 to 420 nm, the position characteristic of the ferryl species. Addition of linoleic acid to the ferryl complex returns the Soret band to 414 nm. Addition of more H202 to the reaction mixture does not shift the Soret maximum back to 420 nm, presumably because any ferryl species that is formed is reduced by reaction with the unsaturated fatty acid. Addition of a second aliquot of HzOz decreases the intensity of the Soret band, however. Addition of ascorbic acid, which reduces the ferryl species to the ferric state, does not cause a shift in the Soret maximum. The results are consistent with reduction of the ferryl species to the ferric state by linoleic acid.
Earlier evidence suggested that the protein radical generated in the reaction of HzOz with sperm whale myoglobin resides primarily on the tyrosine residues (Tyr-103, Tyr-146, Tyr-151) and on the histidine (His-64) that serves as a gate to the heme crevice (22,231. The oxidation of linoleic acid by proteins without these residues was examined to explore their role in Linoleic Acid Oxidatic myoglobin and 60 H202 in 1 ml of total volume of 0.2 M phosphate buffer (pH 7.4) were preincubated at 4 "C for varying amounts of time. Linoleic acid was then added to initiate the reaction, and the reaction was allowed to proceed at 4 "C for 30 min. The reactions were worked up and analyzed as described under "Experimental Procedures." linoleic acid oxidation. Three mutants were examined: Y151F, in which 1 of the tyrosines involved in protein-protein crosslinking is replaced by a phenylalanine, K102Q/Y103F/Y146F/ Y151F, in which all the tyrosines are replaced by phenylalanines, and H64V/K102Q/Y103F/Y146F/Yl5lF in which all the tyrosines and His-64 are replaced by less redox-active residues. The K102Q mutation is present in the latter two mutants because it is required to stabilize the Y103F mutation (22). A protein radical is readily observed by freeze-quench EPR in the reactions of the first two mutants with HzOz but is not detected with the last mutant. Despite the differences in the native and mutant proteins, the K, and V , , , values for the oxidation of linoleic acid by all four recombinant proteins are very similar (Table I). This suggests that the protein radical is not critical for myoglobin-catalyzed oxidation of linoleic acid.
The marked difference in the lifetimes of the ferryl complex (>1 h) and the protein radical ( < 2 3 min) formed in the reaction of sperm whale myoglobin with HzOz have been exploited to obtain independent evidence of the relative roles of the protein radical and the ferryl species in the oxidation of linoleic acid. In these experiments, myoglobin was incubated with 1.5 equivalents of H202, and the protein radical was allowed to decay for increasing periods of time before linoleic acid was added to the incubation mixture. As shown in Fig. 3, the yield of 9-hydroperoxyoctadecadienoic acid does not depend on the length of the preincubation of myoglobin with HzOz before linoleic acid is added. DISCUSSION Earlier studies demonstrated that myoglobin catalyzes the HzOz-dependent oxidation of linoleic acid ( 5 4 , identified 9-hydroperoxyoctadecadienoic acid as a metabolite (8), and provided evidence that the hemoprotein chromophore is perturbed by the binding of fatty acids (8). Nevertheless, the details of this oxidative process have remained unclear despite its proposed involvement in myocardial reperfusion injury (9)(10)(11). The purpose of the present study was to clarify the roles of the possible oxidizing species, particularly the ferryl oxygen and the protein radical formed in the reaction of myoglobin with Hz02 (13,141, in the peroxidation of linoleic acid. Two independent lines of evidence unambiguously establish that linoleic acid is oxidized within the heme crevice of myoglobin rather than free in solution. First, spectroscopic studies confirm the earlier observation (8) that linoleic acid causes a change in the environment of the heme chromophore and dem-3n by Myoglobin-H20P onstrate that the spectroscopically determined dissociation constant (0.9 m) is similar to the K,,, value (2 m) for oxidation of linoleic acid (Table I). Second, oxidation of linoleic acid yields the 9-but not 13-hydroperoxide and, most importantly, does so in a strongly stereospecific manner. The (9S):(9R) enantiomers are obtained in an 84:16 ratio with the native protein, and the (9s) enantiomer is exclusively formed with the H64V and H64V/K102Q/Y103F/-Y146F/Yl5lF mutants in which His-64 has been replaced by a valine (Fig. 2). Furthermore, the pro-R hydrogen at position 11 of the substrate is stereospecifically removed by the H64V/K102Q/Y103F/Y146F/Yl51F mutant, and the extent to which the pro-R hydrogen is removed by the native protein (76%) is comparable to the proportion of (9s)hydroperoxide product. The strong regio-and stereospecificity of the reaction requires oxidation within the chiral confines of the protein and unambiguously rules out chain reactions in which the propagation step occurs in solution. Furthermore, the increase in stereospecificity caused by mutation of His-64, a residue that serves as a gate to the heme crevice, is consistent with oxidation of linoleic acid within the heme crevice.
The regio-and stereochemistry of the oxidation of linoleic acid by myoglobin suggests that the fatty acid interacts with the hemoprotein as shown in Fig. 4. Displacement of His-64 (or Val-64 in the mutants) toward the region above pyrrole ring C (36) allows penetration of the hydrocarbon terminus of the fatty acid in the geometry required to form a trans-l0,ll-double bond into the heme crevice above pyrrole ring D. Abstraction of the pro-R hydrogen yields a planar pentadienyl radical intermediate, with the face from which the pro-R hydrogen was abstracted directed toward the heme ring and protected from reaction with molecular oxygen. Oxygen addition therefore occurs from the opposite face of the pentadienyl radical to give the S-hydroperoxide. The minor extent of pro-S hydrogen abstraction by the native enzyme results in similar antarafacial oxygen addition to give the (9R)-hydroperoxide. The specificity for formation of the 9-rather than 13-hydroperoxide is readily rationalized by the model in Fig. 4 because the C-9 terminus of the pentadienyl radical is exposed to molecular oxygen in the medium, whereas the C-13 terminus is inside the heme crevice and is shielded from oxygen by both the protein and the terminal atoms of the fatty acid chain bound within the heme crevice. Formation of a small amount of the R-hydroperoxide with the wild type protein but not with the two His-64 mutants (Fig.  2) indicates that the residue at position 64 influences the extent to which pro-R and pro-S hydrogens are presented to the ferryl oxygen.
The myoglobin-catalyzed oxidation of linoleic acid is not mediated by hydroxyl-free radicals. This conclusion is supported by the finding that high concentrations of mannitol and low concentrations of DETAPAC do not inhibit linoleic acid oxidation and, most importantly, by the fact that the reaction proceeds with high regio-and stereospecificity. These results strengthen earlier reports that mannitol, benzoic acid, and other hydroxyl radical scavengers do not inhibit the metmyoglobin/HzOz-catalyzed oxidation of membranes or arachidonic acid to malondialdehyde (6,10,35). The results unambiguously establish, furthermore, that singlet oxygen is not involved in the myoglobin-catalyzed oxidation of linoleic acid. The participation of singlet oxygen produced by ferryl myoglobin-catalyzed oxidation of HzOz has been proposed from chemiluminescence and cholesterol oxidation studies (71, but significant participation of singlet oxygen in linoleic acid oxidation is definitively ruled out by the demonstration that (a) only a trace of 180-label from [1SO]H20z is incorporated into the hydroperoxide product and ( b ) the myoglobin reaction proceeds via an antarafacial reaction in which hydrogen abstraction and oxygen addition occur from opposite sides. The reaction of singlet Linoleic acid oxidation is thus mediated either by the ferryl species or the protein radical engendered by reaction of myoglobin with H202 (13,14). To differentiate between these two oxidizing species, we have examined the oxidation of linoleic acid by two sperm whale myoglobin mutants. In one mutant, the 3 tyrosines are replaced by phenylalanines. In the second mutant, the 3 Tyr + Phe mutations are combined with a His-64 + Val mutation. The first of the two mutants gives an EPRdetectable protein radical with HzOz but the second does not (22, 23). The two mutants bind linoleic acid with smaller Kd values than the native enzyme (Table I), but the K, and V, , values for oxidation of linoleic acid to the hydroperoxide are essentially the same as for the recombinant wild-type enzyme (Table I). Not only does replacement of the tyrosines on which the radical is normally centered have little effect on linoleic acid oxidation, but additional replacement of His-64 to give a protein that does not give a detectable protein radical also has no effect. These results strongly argue against a role for the protein radical in the oxidation of linoleic acid. Independent evidence that the protein radical does not catalyze the oxidation of linoleic acid is provided by the finding that the extent of the reaction is the same whether linoleic acid is present when 1.5 equivalents of the HzOz are added to initiate the reaction, or when linoleic acid is added 30 min after the HzOz (Fig. 3). The lifetime of the protein radical is in the order of 1-2 min (14, 371, so the radical is essentially absent when linoleic acid is added 30 min after the peroxide. The report that the EPR signal from spin trapping of the protein radical is only marginally decreased in the presence of linoleic acid is consistent with the present results (38).
The results indicate that linoleic acid is oxidized by the ferryl species rather than by the protein radical, a protein radicalderived oxidant, or a transient species (e.g. porphyrin radical cation) that precedes the protein radical. Direct abstraction of the hydrogen by the ferryl oxygen does not conflict with the inactivity of myoglobin as a general hydrocarbon hydroxylase because the bis-allylic C-H bond is much weaker (-75 kcal/mol) (39) than a normal secondary C-H bond (-95 kcal/mol) (39). The corresponding cytochrome P450-catalyzed oxidation produces a hydroxy-rather than hydroperoxyoctadecadienoic acid via recombination of the pentadienyl radical with the ironcoordinated hydroxyl radical (28). Myoglobin, however, does not directly form hydroxyoctadecadienoic acid. This is consistent with the fact that the myoglobin ferryl complex retains only one of the two oxidation equivalents provided by the peroxide (the other is dissipated in the protein), and thus is converted by hydrogen atom abstraction to a ferric-hydroxyl union rather than ferric-hydroxyl radical complex. Hydroxyl anion recombination with the dienyl radical is not favored (PPM = protoporphyrin E): P450, PPE(+.)Few = 0 + RH + PPMFew-OH + R. + PPIXFe"' + ROH; Mb, PPIXFew = 0 + RH + PPIXFe"'-OH + R. + no reaction.
Prostaglandin synthase initiates the oxidation of arachidonic acid to PGHz by abstracting one of the bis-allylic hydrogens of the polyunsaturated fatty acid (40). Abstraction of the hydrogen by a ferryl species (41) or a protein (tyrosine) radical (42,43) are currently under consideration for the first step in the catalytic mechanism of this enzyme. The present results establish that a ferryl species similar to that of prostaglandin synthase can, in the absence of additional catalytic assistance, directly abstract a bis-allylic hydrogen with subsequent stereospecific addition of oxygen to the pentadienyl radical thus formed.