Reaction of Myoglobin with Hydrogen Peroxide Forms a Peroxyl Radical Which Oxidizes Substrates*

Evidence is presented that the radical observed upon reaction of myoglobin with hydrogen peroxide is a peroxyl radical. Simulation of this spectrum gives principal values for the g tensor of g, = 2.0357, g, = 2.0082, and g, = 2.0016, which are consistent with those of a peroxyl radical. Use of molecular oxygen isotopically labeled with "0 confirmed that the radical observed was a peroxyl radical. Removal of oxygen from the incubation by use of glucose and glucose oxidase revealed two radicals, one at giso = 2.0028 and the other at giso = 2.0073. Addition of various amounts of the spin trap 5,5-dimeth- yl-l-pyrroline N-oxide revealed that the spin trap and oxygen compete for the same radical site. Four model substrates, glutathione, styrene, arachidonic acid and linoleic acid, were individually added to both the aero- bic and anoxic systems. Glutathione reacted with the peroxyl radical, reducing its intensity by 98%, and en- tirely eliminated the giso = 2.0028 line from the spectrum of the anoxic incubation. Styrene, arachidonic acid and linoleic acid reacted with the peroxyl radical, reducing its amplitude by 84,57, and 35%, respectively, but did not decrease the amplitude of either radical species in the anoxic

Evidence is presented that the radical observed upon reaction of myoglobin with hydrogen peroxide is a peroxyl radical. Simulation of this spectrum gives principal values for the g tensor of g, = 2.0357, g, = 2.0082, and g, = 2.0016, which are consistent with those of a peroxyl radical. Use of molecular oxygen isotopically labeled with "0 confirmed that the radical observed was a peroxyl radical. Removal of oxygen from the incubation by use of glucose and glucose oxidase revealed two radicals, one at giso = 2.0028 and the other at giso = 2.0073.
Addition of various amounts of the spin trap 5,5-dimethyl-l-pyrroline N-oxide revealed that the spin trap and oxygen compete for the same radical site. Four model substrates, glutathione, styrene, arachidonic acid and linoleic acid, were individually added to both the aerobic and anoxic systems. Glutathione reacted with the peroxyl radical, reducing its intensity by 98%, and entirely eliminated the giso = 2.0028 line from the spectrum of the anoxic incubation. Styrene, arachidonic acid and linoleic acid reacted with the peroxyl radical, reducing its amplitude by 84,57, and 35%, respectively, but did not decrease the amplitude of either radical species in the anoxic incubation. The giso = 2.0028 species detected in the anoxic incubation appears to be the original radical site to which molecular oxygen binds to form the peroxyl radical. This myoglobin-derived peroxyl radical species is responsible for the advent of lipid peroxidation as proposed in ischemidrepurfusion iqiury, as well as other reactions, as exemplified by the 02-dependent epoxidation of styrene.
There has been increasing interest in the identification of the radical species that forms on metmyoglobin upon exposure to hydrogen peroxide (1)(2)(3)(4)(5)(6)(7)(8)(9)(10). This interest is stimulated by evidence that this reaction might be of importance in elucidating the origin of the lipid peroxidation and cell damage resulting from ischemiaheperfusion of the heart (11)(12)(13)(14)(15). While myoglobin is not primarily a catalytic protein, it has been demonstrated to possess peroxidase-like activity (1,2,(16)(17)(18). Similar to classical peroxidases such as horseradish peroxidase, myoglobin reductively cleaves hydrogen peroxide, retaining one of the oxidizing equivalents on the iron of the heme in the form of a n oxo-ferry1 species (FeTV=O). However, the second oxidizing equivalent is located on the globin as a free radical (19-211, whose structure has been the subject of speculation since 1958 (18). This is different from the classical horseradish peroxidase, where the second oxidizing equivalent is the porphyrin M cat-* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
ll To whom correspondence should be addressed. ion radical (22). The nature of the metmyoglobin radical site has been proposed to be a tyrosine-based peroxyl radical using spin trapping (2, 8); however, this assignment has been contested (3,9). Utilizing direct EPRl spectroscopy, we demonstrated that a peroxyl radical is formed, and that this species is derived from the reaction of molecular oxygen with a globin free radical, which is not a tyrosyl radical. Both free radicals oxidize cellular constituents, although with different specificities.

MATERIALS AND METHODS
Horse heart myoglobin, D-glucose, DMPO, glutathione, and hydrogen peroxide were obtained from Sigma. Glucose oxidase (250 unitdmg, 1 unit converts 1 mmol of glucose to product in 1 min at 25 "C and pH 7) was purchased from Boehringer Mannheim. O2 enriched 37% with I' O was obtained from ICON Services Inc. (Summit, NJ). Styrene was obtained from Aldrich. Arachidonic and linoleic acids were obtained from Nu Chek (Elysian, MN). The DMPO was further purified before use by vacuum distillation at room temperature.
All EPR spectra were recorded at liquid nitrogen temperature using a fingertip liquid nitrogen Dewar flask (Wilmad, Buena, NJ) on a Bruker ESP300 spectrometer equipped with a TMllo cavity. All incubations were 532 PM metmyoglobin in 0.1 M sodium phosphate buffer, pH 7.4, and initiated with 400 PM hydrogen peroxide. EPR spectral simulations were performed using the powder pattern simulation program POW, part of the Brown University Powder Pattern package (23,24).
Aerobic and Anoxic Incubations-The aerobic system was initiated with the addition of hydrogen peroxide and frozen in liquid nitrogen within 10 s after the initiation of the reaction. The anoxic system was achieved as follows. Metmyoglobin solution was placed in a 20-ml vial, and a stream of nitrogen gas was passed on to the solution without bubbling by use of a Pasteur pipette. The oxygen-consuming system of glucose (15 m~) and glucose oxidase (50 unitdml) was added, hydrogen peroxide was added immediately afterward, and the incubation was frozen in liquid nitrogen within 10 s. In order to accurately determine the values of the g tensor for each species, a g standard of C13' in MgO was employed (gis0 = 1.9800 f 0.0006) (25).
Time Dependence-The stability of the radicals formed in both the aerobic and anoxic conditions was explored. Incubations were prepared as above, and times between initiation of the reaction with hydrogen peroxide and freezing in liquid nitrogen were varied.
DMPO Competition-DMPO was added at various concentrations to both aerobic and anoxic incubations prior to the addition of the hydrogen peroxide. These incubations were then frozen in liquid nitrogen within 10 s.
Isotopic Substitution with z70-Myoglobin in buffer was blown with nitrogen gas as described above. Oxygen gas enriched 37% with " 0 was bubbled into the incubation for 3 min. Hydrogen peroxide was added to initiate the reaction, and the incubation was frozen immediately in liquid nitrogen.
Reaction with GZutathione-Hydrogen peroxide was added to myoglobin in buffer, glutathione (2 mM) was added immediately afterward, and the incubation was frozen immediately in liquid nitrogen. The anoxic incubations were treated as above.

7458
This is an Open Access article under the CC BY license. Reaction with Styrene-Styrene (10 mM) was added to myoglobin in buffer, and the incubation was initiated with the addition of hydrogen peroxide, then frozen immediately in liquid nitrogen. Similarly, anoxic incubations made with glucose and glucose oxidase, as described above, were treated with styrene.
Reaction with Arachidonic and Linoleic Acids-Arachidonic acid (82 mM) or linoleic acid (80 mu) in absolute ethanol was pipetted into a test tube to coat the bottom of the tube. The ethanol was then evaporated with a stream of nitrogen gas. ~yoglohin in buffer was added and the solution vortexed for 30 s. The final concentration of the fatty acids was 400 p~. Hydrogen peroxide was added to initiate the reaction, and the incubation frozen immediately in liquid nitrogen. Anoxic incubations were treated similarly.

RESULTS
Aerobic and Anoxic Systems- Fig. 1 shows both the aerobic and anoxic incubations with their controls. A three-line anisotropic spectrum is observed in the complete aerobic system (Fig. a). This spectrum is not observed if either the hydrogen peroxide or the metmyoglobin is omitted. Forcing the system to be anoxic by use of a stream of nitrogen gas and addition of glucose and glucose oxidase results in the disappearance of the anisotropic spectrum, which was replaced by two isotropic spectra due to species with different g values (Fig. l D ). The broader line has a g value of 2.0073, whereas the upfield line has a g value of 2.0028 and a peak-to-trough linewidth of 5 gauss. This spectrum is not observed if metmyoglobin is eliminated from the incubation (Fig. LE). This demonstrates that f o~a t i o n of the species detected in the complete incubation requires the presence of all the components. In addition, it is shown that the species observed in the aerobic system is indeed dependent on the presence of oxygen.
In order to determine the kinetic properties of the radicals detected, incubations of both the aerobic and anoxic systems were initiated with hydrogen peroxide and the time before freezing the sample was varied. Fig. 2 shows the time dependence of the radical species in the aerobic and anoxic systems. As the time between initiation and cessation of the reaction increases, the intensity of the aerobic species decreases with an apparent half-life of 7 s, as determined by kinetic analysis of the low field extrema. In the anoxic system, the sharper, higher   is little or no evidence of the species observed in the absence of DMPO. As the concentration of DMPO decreases, the peroxyl radical becomes more evident. At 500 VM DMPO, the radical adduct is virtually absent. This indicates that DMPO is competing with oxygen for a radical site on the globin. Fig. 5 shows the results of adding varying concentrations of DMPO to the anoxic system. Similar to the aerobic system, only the radical adduct, which has the same spectrum, is detected a t the higher concentrations of the spin trap. However, unlike the aerobic system, presence of the radical adduct can be observed at concentrations as low as 50 p . Without oxygen present, DMPO does not have to compete for the radical site, and therefore the radical adduct can be detected at much lower concentrations of DMPO. Fig. 6 demonstrates that the radical adduct detected at low temperature is the same as that previously detected at room temperature (9). An incubation of metmyoglobin, hydrogen peroxide, and DMPO was frozen, and the spectrum was recorded (Fig. 6A). The parallel nitrogen hyperfine coupling constant, u r , was determined to be 32.5 gauss. The sample was allowed to thaw into a flat cell, and the room temperature spectrum was then recorded (Fig. 6B). The nitrogen hyperfine coupling constant (a: = 14.5 gauss) and the @-hydrogen hyperfine coupling constant (a; = 8.2 gauss) are similar to the reported values (9).
Isotopic Substitution with I7O-A powerful method in determining the structure of a radical is the use of isotopic substitution. Fig. 7 shows the experimental results of the substitution of air by molecular oxygen enriched with 37% I7O after a nitrogen gas purge. Additional hyperfine structure due to the 512 nuclear spin of the 170 can be seen in Fig. 7A. Two sets of hyperfine structure arise from the fact that the majority of the 170 is in the form of x70160, and the hyperfine coupling constant depends on which oxygen the unpaired electron resides, which in turn depends upon which oxygen reacted with the original radical site (24). A third set of hyperfine coupling from peroxyl radicals formed by molecular oxygen comprised of i70170, which makes up 12% of the peroxyl radical observed, can be detected at x40 gain fFig. 7Bf. While one would expect that 40% of the peroxyl radical would be from l6O1"0 due to the isotopic ratio of oxygen in the gas used, there is actually a higher percentage due to the catalase-like activity of myoglobin. Myoglobin generates molecular oxygen from hydrogen peroxide as determined by oxygraph measurements (9); thus, a portion of the added hydrogen peroxide will be consumed in this manner. This leads to a percentage of the peroxyl radical being formed from 160160 higher than 40%. Corresponding lower percentages of 1 7 0 containing peroxyl radicals are found in the simulation. The percentages of each type of peroxyl species, as The DMPO radical adduct can be detected at a much lower concentration of DMPO than in the aerobic system. Spectrometer conditions were similar to Fig. 4, except that gain for 100 m M was 2.0 x lo4 and all other gains were 5.0 x IO4. globin radical adduct. A, the spectrum from a frozen incubation of Fic. 6.77 I(: and room temperature spectra of the D1MpO-myo-DMPO, myoglobin, and hydrogen peroxide yielded aif" = 32.5 gauss. B, the spectrum of the same sample after it had been thawed to room temperature yielded azo = 14.5 gauss, u; = 8.2 gauss. Spectrometer parameters for the frozen sample were the same as in Fig. 4. Spectrometer parameters for the room temperature spectrum were as follows:  Fig. 8 shows the results of the addition of 2 m M glutathione to both the aerobic and anoxic systems. In the aerobic system, the signal intensity is decreased by 98% ( n = 4,p c. 0.00005) (Fig. 8B). In the absence of oxygen, the sharp line at giso = 2.0028 disappears entirely, leaving the broad line at giso = 2.0073 (Fig. 80).

Reaction with Glutathione-
Reaction with Styrene-The effect of adding styrene to the incubations is seen in Fig. 9. When styrene is added to the aerobic system, the amplitude of the peroxyl spectrum is reduced by 84%, as determined by the signal intensity of the low field extrema ( n = 3, p c 0.00005), and a radical with less anisotropy is formed (Fig. 9B). No signal was observed with styrene and hydrogen peroxide alone. In the absence of oxygen, there is no significant decrease in the signal intensity. This indicates that styrene is reacting with the peroxyl radical rather than with the original radical site. Reaction with Arachidonic and Linoleic Acids- Fig.  10 shows that the addition of arachidonic acid to the incubation results in a decrease in the peroxyl radical. Arachidonic acid reduced the spectrum amplitude by 57% (n = 4, p < 0.005).
Similarly, linoleic acid reduced the spectrum amplitude by 35% ( n = 4, p < 0.05) (data not shown). There was no significant difference in the signal intensities from the anoxic incubations.

DISCUSSION
The g tensor for the species observed in the aerobic system closely matches those observed for peroxyl free radical species (26). It is unclear as to why the g, value is lower than the expected value of 2.0023, though this may be due to the uncertainty in the g standard or an effect of the paramagnetic heme. The addition of glucose and glucose oxidase causes a dramatic change in the spectrum of myoglobin oxidized with hydrogen peroxide. Glucose oxidase oxidizes D-glucose to D-gluconic acid, consuming a molecule of oxygen concomitantly. Studies have shown that this system can decrease the oxygen content of an aqueous solution to 10 n~ (27).
Binding of DMPO to the radical site produces a spin trap radical adduct, while the binding of oxygen produces the per-oxy1 radical. Kelman and Mason (9) have shown that the radical that is trapped by DMPO is not a peroxyl radical. The radical adduct detected at low temperature is the same species previously detected at room temperature, as shown by the thawing experiment. Immobilization of the radical adduct at 77 K broadens the spectrum (Fig. €A), hence obscuring the hydrogen hyperfine coupling apparent in the room temperature spectrum (Fig. 6B). Fig. 7 clearly demonstrates that the radical detected directly with EPR is a peroxyl species. The parallel 170 hyperfine coupling constants, 97 and 55 gauss, are consistent with those found by Sevilla and co-workers (26) for a carbon-centered radical formed by x-irradiation, with the 55-gauss hyperfine coupling being due to the oxygen bound to a carbon of the myoglobin. The two different sets of coupling constants arise from the two different configurations that can occur with a molecule of oxygen containing one each of the two isotopes. Thus there will While not all these lines can be identified in the spectrum observed ( Fig. 7B 1, enough of these lines are visible to confirm the presence of both oxygen atoms in the same radical. Fig. 8 shows that glutathione is a very efficient scavenger of both of the radicals that form on myoglobin (or a precursor of these radicals). Within seconds, glutathione decreases the peroxyl radical concentration nearly 2 orders of magnitude. Ro-mer0 et al. (28) have trapped the glutathione thiyl radicals with DMPO after the addition of glutathione to myoglobin and hydrogen peroxide. The authors added hydrogen peroxide to myoglobin, then added catalase 1 min later in order to eliminate excess hydrogen peroxide. Glutathione was added subsequent to the catalase. Given the lifetime of the myoglobin peroxyl radical, it is likely that only a small portion of the glutathione radicals were derived from the reaction of glutathione with this peroxyl radical, and that the majority of the radicals were generated through oxidation by the ferryl heme (Compound 11).
The anoxic incubation with glutathione shows that treatment of myoglobin with hydrogen peroxide actually results in the formation of two distinct radicals. One radical, with giso = 2.0028, is readily reducible by glutathione, as this signal disappears entirely when glutathione is added to the incubation. The other species, at giso = 2.0073, does not react with glutathione. This species was detected in incubations with glutathione as long as 5 min before freezing, with no reduction in signal intensity (data not shown). This may be due to either the radical being inaccessible to glutathione, or the radical being very stable and thus unreactive.
Styrene appears to react with the peroxyl radical, but not with the original radical site, as there is little decrease in the spectrum amplitude in the absence of oxygen. It has been shown by Ortiz de Montellano and Catalan0 (16) that the ad- dition of styrene to myoglobin and hydrogen peroxide will result in the formation of styrene oxide, with 78% of the oxygen added to the styrene oxide derived from molecular oxygen. Rao et al. (4) demonstrated that when oxygen was added to cis+ methylstyrene, it could do so in such a manner that the stereochemistry was not preserved, resulting in trans-p-methylstyrene oxide. Although the ferry1 iron of the myoglobin did oxidize &-&methylstyrene under these conditions, it preserved the stereochemistry of the molecule, resulting in cis-&methylstyrene oxide. These results showed that oxygen was being added to cis-P-methylstyrene by myoglobin from both the ferryl iron and another site. Given that the reaction of a methylenyl carbon with a peroxyl radical will produce an epoxide (29), it is likely that the peroxyl radical reported here is the site at which the epoxidation reaction is occurring.
The addition of either arachidonic or linoleic acids, like styrene, decreases the amount of peroxyl radical present, but they do not react with the original radical site. Lipid peroxidation due to the exposure to myoglobin and hydrogen peroxide has been reported (11, 14); however, evidence that the hydrogen extraction was due to interaction with a peroxyl radical located on myoglobin had not been previously demonstrated. The peroxyl radical formed on myoglobin upon reaction with hydrogen peroxide is the likely site that initiates lipid peroxidation proposed to result from ischemidreperfusion of the heart.
The identity of the original radical site, characterized by giso = 2.0028, is not yet resolved. Wilks and Ortiz de Montellano used site-directed mutagenesis to replace one, two, or all three tyrosine residues from sperm whale myoglobin without detecting a change in the EPR spectrum (3). The authors postulated that the radical was therefore not based upon a tyrosine residue. However, since the radical species detected by these authors was a peroxyl radical, there would be no noticeable change in the spectrum if the peroxyl radical formed on different residues in different mutants. Peroxyl radicals can form on different residues, yet have similar g values (26,30). However, the initial radical site is indeed not a tyrosyl radical, based on spectroscopic grounds. DeGray et al. (31) have shown that the linewidth of the immobilized tyrosyl radical is at least on the order of 20 gauss, much broader than the 5-gauss linewidth of the anoxic species, confirming that the original radical site is not a tyrosyl radical. Given an apparent lack of hyperfine coupling, this radical may be a tertiary carbon-centered radical, as would be consistent with its g value and the absence of nearby nuclei with nuclear spin. The relative stability of the myoglobin peroxyl is also consistent with the known stability of tertiary peroxyl ra.dicals, which are known to be much more persistent than either primary or secondary peroxyl radicals (29): HOWever, an assignment to a particular residue cannot be made from these data.

Peroxyl Radical from
Myoglobin and Hydrogen Peroxide 7463 CONCLUSION A summary for these reactions is shown in Scheme 1. In the absence of DMPO, the radical formed reacts with molecular oxygen to form a peroxyl radical and then decays to an EPR silent species. In the presence of DMPO, the spin trap competes with molecular oxygen for the radical site. In the absence of molecular oxygen, DMPO binds to the radical site without competition, allowing the radical adduct to be observed with much lower concentrations of the spin trap. The addition of glutathione demonstrated that the peroxyl species is readily accessible to molecules in solution and that the narrow, single-line species detected in the anoxic incubation is likely the original radical site. Use of styrene, arachidonic acid, and linoleic acid showed that the myoglobin peroxyl radical reacts in a manner typical of smaller peroxyl radicals. This indicates that myoglobin-derived radicals are potential sources of damage to cellular constituents such as membranes and protein sulfhydryl groups.