An Electron Spin Resonance Spin-trapping Investigation of the Free Radicals Formed by the Reaction of Mitochondrial Cytochromec Oxidase with H2O2 *

The reaction of purified bovine mitochondrial cytochrome c oxidase (CcO) and hydrogen peroxide was studied using the ESR spin-trapping technique. A protein-centered radical adduct was trapped by 5,5-dimethyl-1-pyrrolineN-oxide and was assigned to a thiyl radical adduct based on its hyperfine coupling constants of a N = 14.7 G and a β H = 15.7 G. The ESR spectra obtained using the nitroso spin traps 3,5-dibromo-4-nitrosobenzenesulfonic acid (DBNBS) and 2-methyl-2-nitrosopropane (MNP) indicated that both DBNBS/⋅CcO and MNP/⋅CcO radical adducts are immobilized nitroxides formed by the trapping of protein-derived radicals. Alkylation of the free thiols on the enzyme with N-ethylmaleimide (NEM) prevented 5,5-dimethyl-1-pyrroline N-oxide adduct formation and changed the spectra of the MNP and DBNBS radical adducts. Nonspecific protease treatment of MNP-d 9/⋅NEM-CcO converted its spectrum from that of an immobilized nitroxide to an isotropic three-line spectrum characteristic of rapid molecular motion. Super-hyperfine couplings were detected in this spectrum and assigned to the MNP/⋅tyrosyl adduct(s). The inhibition of either CcO or NEM-CcO with potassium cyanide prevented detectable MNP adduct formation, indicating heme involvement in the reaction. The results indicate that one or more cysteine residues are the preferred reductant of the presumed ferryl porphyrin cation radical residue intermediate. When the cysteine residues are blocked with NEM, one or more tyrosine residues become the preferred reductant, forming the tyrosyl radical.

Oxidative stress in mitochondria has been widely recognized as an important cause of a variety of diseases (1) including aging (2,3), neurodegeneration (3,4), and cancer (5,6). Under certain conditions such as incomplete oxygen reduction during oxidative phosphorylation, ischemia-reperfusion, and inflammation, the production of reactive oxygen species by the mitochondrial electron transport chain can increase severalfold (7,8). It is generally thought that several events of oxidative stress, such as calcium imbalance, lipid peroxidation, glutathione depletion, and DNA and protein damage, arise from excess production of reactive oxygen species in mitochondria (8 -11).
Several reports have demonstrated that reactive oxygen spe-cies can cause oxidative inactivation of the components of the mitochondrial electron transport chain during exposure of submitochondrial particles to fluxes of O 2 . , ⅐ OH, or H 2 O 2 (12)(13)(14)(15).
However, the molecular mechanism regarding the effect of reactive oxygen species on the mitochondrial bioenergetic enzyme complex still remains unknown. Recently we employed the ESR spin-trapping technique to investigate the oxidative damage of horse heart cytochrome c (16,17). It was clearly shown that protein-centered radicals are formed on cytochrome c following its reaction with hydrogen peroxide, and the protein-derived radicals are expected to become sites of permanent oxidative damage. The cytochrome c-derived radicals were able to oxidize residues on peptides other than the cytochrome c (18). Cytochrome c oxidase (CcO), 1 the terminal enzyme of the mitochondrial respiratory chain, is a multicomponent membrane protein with a molecular weight of 200,000, comprising 13 different polypeptide subunits. The enzyme functions to reduce dioxygen to H 2 O at the active center of heme a 3 -Cu B and couples proton pumping across the mitochondrial inner membrane to provide the driving energy for ATP synthesis (19). It has been reported that a peroxidase-like reaction can occur at the heme a 3 -Cu B center of CcO with a very high K m for H 2 O 2 (20,21). The addition of excess H 2 O 2 can oxidize CcO to yield a peroxy intermediate (or "compound P"), and a one-electron reduction of compound P with either ferrocyanide or cytochrome c yields the ferryl-oxo species (or "compound F"). Finally, a one-electron reduction of compound F returns the CcO to the oxidized state and completes the catalytic cycle. The structure of compound P remains unclear and is under debate (22)(23)(24). Compound F was proposed to be structurally related to a horseradish peroxidase-compound II-like species (25)(26)(27). One would expect that both compounds P and F are highly reactive if the chemistry of the H 2 O 2 /CcO system is similar to that of peroxidases. However, no direct evidence indicates that the compound P or compound F species of CcO can oxidize the protein and initiate free radical-mediated lipid peroxidation as seen with metmyoglobin and cytochrome c. Herein we report that protein-centered radical formation on CcO was detected using the ESR spin-trapping technique. the UV absorption at 240 nm (⑀ ϭ 43.6 M Ϫ1 cm Ϫ1 ). Pronase (from Streptomyces griseus) was purchased from Boehringer Mannheim and used as received. The DMPO, MNP, NEM, potassium cyanide, and sodium dithionite were purchased from Aldrich. DMPO was vacuumdistilled twice and stored under nitrogen at Ϫ70°C until needed. DB-NBS was synthesized using the method of Kaur et al. (28). Prepacked Sephadex G-25 (PD-10) gel filtration cartridges were purchased from (Amersham Pharmacia Biotech).

Reagents-Ascorbic
Preparations of Mitochondrial Cytochrome c Oxidase-Highly purified beef heart mitochondrial cytochrome c oxidase was prepared and assayed according to the methods reported by Yu et al. (29). Submitochondrial particles were used as the starting material and subjected to ammonium sulfate precipitation in the presence of 1.5% sodium cholate. The cytochrome c oxidase as prepared is essentially in the delipidated form and spectrophotometrically free of complex III (cytochrome bc 1 complex). The cytochrome c oxidase contains 10 -12 nmol of heme a/mg of protein and 7-8 g of phospholipid/mg of protein. Oxidase activity was confirmed by measuring the oxidation of ferrous cytochrome c and by oxygen consumption in the presence of ascorbate and ferric cytochrome c (29).
ESR Spin-trapping Measurements-All ESR spectra were recorded with a Bruker ESP 300 spectrometer, using a quartz flat cell and operating at 9.8 GHz with a modulation frequency of 100 kHz and a TM 110 cavity. The reactions were initiated by the addition of hydrogen peroxide to the mixture of cytochrome c oxidase and spin trap in 50 mM sodium/potassium phosphate buffer (pH 7.4) containing 200 M diethylenetriaminepentaacetic acid. For the preparation of thiol-blocked cytochrome c oxidase, the NEM (0.25 M in H 2 O) was mixed with the enzyme at a molar ratio of 50:1, and the reaction was allowed to proceed for 10 min at room temperature before the addition of spin traps and H 2 O 2 . The spectral simulations were done using the WinSim program of the NIEHS public ESR software package that is available on the World Wide Web (http://epr.niehs.nih.gov).
Proteolytic Digestion of Cytochrome c Oxidase-Enzymatic digestion of cytochrome c oxidase with Pronase was carried out at room temperature, and phospholipase D was included to a final concentration of 50 units/ml for 20 min before proteolysis. The Pronase (100 mg/ml), suspended in 50 mM phosphate buffer, pH 7.4, was then added to the reaction mixture at a final concentration of 10 mg/ml.
Oxygen Consumption Experiments-Oxygen consumption measurements were made using a Clark-type oxygen electrode fitted to a 1.8-ml Gilson sample cell and monitored by a Yellow Springs Instrument Company (Yellow Springs, OH) model 53 oxygen monitor. The reagents were added in the following order: 20 M cytochrome c in 50 mM sodium/potassium phosphate buffer (pH 7.4) containing 1 mM ascorbate. Then, after establishing the measurement of a 1-min base line, an appropriate amount of cytochrome c oxidase was added to initiate oxygen uptake.

DMPO Spin Trapping of CcO Thiyl Radial
Formed by H 2 O 2 -When the purified CcO was reacted with 5 eq (based on the amount of heme a and heme a 3 in CcO) of hydrogen peroxide in the presence of DMPO, a broad four-line spectrum was obtained (Fig. 1A). The spectrum was totally dependent on protein (Fig. 1E). However, the removal by dialysis of the excess DMPO and any small radical adducts did not affect the spectrum (data not shown), indicating that a protein-derived radical(s) had been detected. To identify the DMPO adduct formed in H 2 O 2 -treated CcO, the spectrum of Fig. 1A was analyzed by computerized simulation. Fig. 1B shows the simulated spectrum of a partially immobilized nitroxide. This DMPO adduct can be assigned to a thiyl radical adduct based on its hyperfine coupling constants (a N ϭ 14.7 and a ␤ H ϭ15.7 G) (30). In addition, the ESR spectrum from the reaction of H 2 O 2 with NEM-pretreated CcO was clearly different from and weaker than that of native protein (Fig. 1C), indicating that sulfhydryl group(s) are necessary for the radical adduct formation. A spectrum similar to that shown in Fig. 1E was obtained when either NEM, H 2 O 2 , and DMPO or NEM-CcO and DMPO were present in the reaction mixture (data not shown). The addition of NEM to the DMPO/CcO thiyl radical adduct did not affect the spectral characteristics of Fig. 1A, thus ensuring that NEM does not react with the CcO thiyl radical adduct (data not shown). To further confirm that the Fig. 1A spectrum is of a radical adduct of a sulfur-centered radical, Ce 4ϩ was used to selectively oxidize the thiol residues of CcO, which were subsequently spin-trapped with DMPO (31). The resulting partially immobilized ESR spectrum with hyperfine coupling constants a N ϭ 14.7 and a ␤ H ϭ 15.7 G ( Fig. 2A) is nearly identical to that of native CcO treated with H 2 O 2 . The spectrum observed in Fig. 2A was completely abolished when the NEMpretreated CcO was treated with Ce 4ϩ (Fig. 2E).  (1.5 mM) to the mixture at room temperature. 0.3 ml of the sample was withdrawn and submitted to ESR measurement. B, computer simulation of the spectrum in A with hyperfine constants a N ϭ 14.7 and a ␤ H ϭ 15.7 G. C, same as A, except the CcO was pretreated with 50 eq of NEM before H 2 O 2 addition. D, same as A, except the H 2 O 2 was omitted from the system. E, same as A, except the enzyme was substituted with buffer. F, same as A, except the DMPO was omitted from the reaction mixture. Instrumental parameters were as follows: modulation amplitude, 1 G; time constant, 1.3 s for A and C, 327 ms for D-F; gain, 1 ϫ 10 5 ; modulation frequency, 100 kHz; microwave frequency, 9.8 GHz; microwave power, 20 milliwatts. broad ESR spectrum typical of an immobilized nitroxide. The signal persisted after dialysis to remove excess H 2 O 2 , MNP, and small radical adducts (data not shown), demonstrating that a protein-bound radical adduct was formed. When the MNP/ ⅐ CcO-derived adduct was subjected to nonspecific proteolysis, the intensity of immobilized nitroxide signal decreased, and an isotropic three-line spectrum was observed with a nitrogen hyperfine constant of 17.1 G, which is assigned to di-tbutylnitroxide, a decomposition product of MNP (data not shown).
Cyanide can form a low-spin ferric complex with CcO and inhibit the enzymatic activity detected with cytochrome c oxidation and oxygen consumption. Accordingly, adding 10 eq of potassium cyanide/heme before adding H 2 O 2 inhibits the heme-catalyzed reaction and prevents the production of MNP and DBNBS adducts (Figs. 3D and 4D). A four-line ESR spectrum with hyperfine coupling constants a N and a H of 14.4 G from MNP/ ⅐ H was obtained (Fig. 3D). The addition of KCN to the protein-centered radical adducts of either MNP or DBNBS did not effect the spectra of these radical adducts in any way (data not shown).
MNP Spin Trapping of a Protein-derived Tyrosyl Radical from the Reaction of NEM-CcO with H 2 O 2 -To evaluate whether the MNP/ ⅐ CcO and DBNBS/ ⅐ CcO adducts have properties similar to the DMPO/ ⅐ CcO adduct, we pretreated the CcO with 50 eq of NEM at room temperature to modify the cysteine residues on the surface of CcO. As indicated in Figs. 5A and 6A, the ESR spectral characteristics of DBNBS/ ⅐ CcO and MNP/ ⅐ CcO were significantly changed when the NEM-CcO was incubated with H 2 O 2 . The addition of NEM to preformed MNP/ ⅐ CcO or DBSBS/ ⅐ CcO did not effect the spectra of these radical adducts in any way (data not shown). These results are consistent with the spin-trapping results obtained with DMPO, where a sulfur-centered radical is formed in the oxidation of CcO by H 2 O 2 . The H 2 O 2 -dependent, persistent, immobilized nitroxides were detected in the reaction systems of NEM-CcO/ H 2 O 2 /DBNBS and NEM-CcO/H 2 O 2 /MNP (Fig. 5, A and C, and  Fig. 6, A and B). When the sample from either Fig. 5A or 5C was submitted to Pronase digestion, a species with a three-line ESR spectrum was observed (Fig. 5D). The hyperfine coupling constant of this proteolytic product is 13.6 G (as indicated by arrows in Fig. 5D), which is similar to that reported for proteolyzed DBNBS/ ⅐ metmyoglobin and DBNBS/ ⅐ cytochrome c (17,33,34). It should be noted that the spin adduct from the proteolytic product of DBNBS/ ⅐ NEM-CcO is relatively unstable. The intensity of the ESR signal (Fig. 5D) decayed during the process of scanning and Pronase digestion. Attempts to obtain a clear, three-line spectrum by proteolysis were unsuccessful. A protein-derived radical adduct was also detected in the NEM-CcO/H 2 O 2 /MNP system (Fig. 6A). In the ESR spectrum of Fig. 6A, two MNP adducts were generated when the NEMpretreated CcO reacted with H 2 O 2 . One species (as indicated by the solid lines in Fig. 6A) with a sharper ESR line shape is less immobilized than the other species (as indicated by the dashed lines) with a broad low-field extrema. However, nonspecific proteolytic treatment (up to 6 h) of the sample from Fig. 6A or 6B can gradually remove the species with the broad line width and yield an isotropic three-line spectrum with a hyperfine coupling constant of a N ϭ 15.6 G (Fig. 6C), which is basically similar to those observed in MNP/ ⅐ metmyoglobin and MNP/ ⅐ cytochrome c (17,33,34). The inclusion of potassium cyanide (up to 10 eq of NEM-CcO) in the system drastically inhibited the signal of the heme-catalyzed immobilized nitroxide, presumably because of its ability to form a ligand at the sixth position (Fig. 6D). However, KCN did not significantly change the spectral characteristics when the H 2 O 2 -dependent, proteincentered radical adduct of NEM-CcO had been previously formed (data not shown).
The isotropic three-line spectrum observed upon proteolysis of MNP/ ⅐ NEM-CcO indicates that the atoms ␣ and ␤ of the nitroxide had no nuclear spin (i.e. I ϭ 0) and, therefore, that the protein-centered radical was located on a tertiary carbon. In the previous investigation, Gunther et al. (34) were able to use a low modulation amplitude to resolve the super-hyperfine structure from the DBNBS/ ⅐ metmyoglobin adduct. Indeed, they concluded that the globin-derived radical trapped by DBNBS was centered on the C-3 of the tryptophan residue in metmyoglobin by comparison with the pure DBNBS/ ⅐ Trp adduct. In another report, Barr et al. (17) extended this technique and used perdeuterated MNP (i.e. MNP-d 9 ) and carbon-13 labeling of the aromatic ring positions of tyrosine to identify MNP/ ⅐ cytochrome c-derived adducts (17). We employed the same technique here, using MNP-d 9 to address the structure of the MNP/ ⅐ NEM-CcO adduct generated with H 2 O 2 . Because the hyperfine structure from the nine methyl hydrogens broadens the ESR spectrum of the MNP adduct, it can prevent the resolution of super-hyperfine structure. The fully deuterated MNP decreases this effect and increases the resolution of the spectrum. A three-line spectrum with a similar hyperfine coupling constant (a N ϭ 15.5 G) was detected in the MNP-d 9 / ⅐ NEM-CcO/H 2 O 2 system (data not shown). To resolve the hyperfine structure, an 8-G scan of the low-field line from this preliminary spectrum coupled with modulation amplitude as low as 0.25 G was chosen to record a super-hyperfine spectrum, which basically consisted of five resolved lines. Fragmentation of the protein by proteolysis to form freely rotating peptides gave a well resolved super-hyperfine structure (Fig. 7A). Fig.  7B is the simulated spectrum using three nonequivalent hydrogens (a H ϭ 1.40, a H ϭ 0.90, and a H ϭ 0.69 G). The super- M DBNBS before being subjected to ESR measurement. B, the reaction mixture was the same as in A, except the hydrogen peroxide was omitted. C, the sample shown in A was recovered from the flat cell and dialyzed against 100 ml of 50 mM sodium/potassium phosphate buffer containing 1.0% sodium cholate, pH 7.4, for 2 h before the ESR spectrum was re-recorded. The dialysate was concentrated with a Centricon 30 microconcentrator and brought to the same volume as in A, prior to ESR analysis. D, same as C, except 10 mg/ml Pronase was added to the sample. The hyperfine coupling constant, a N ϭ 13.6 G, is indicated by arrows. The instrumental parameters were the same as described in the legend to Fig. 3, except the modulation amplitude is 2 G. hyperfine structure obtained from proteolytic MNP-d 9 / ⅐ NEM-CcO was similar to that of authentic MNP-d 9 / ⅐ Tyr formed by oxidation of the free amino acid by horseradish peroxidase as reported by Gunther et al. (33). A trace amount of another MNP-d 9 adduct was detected on the low-field side of the major spectrum shown in Fig. 7A (as indicated by asterisks), which was believed to be caused by incompletely hydrolyzed peptides that remain partially immobilized. The intensity of this more immobilized adduct gradually decreased when the incubation time for proteolysis was prolonged. Proteases have been shown to be incapable of complete digestion of membrane proteins because of the high hydrophobicity of the transmembranespanning helices (35). Inclusion of SDS (up to 0.2%) and 2 M urea did not improve proteolysis efficiency. DISCUSSION In the current investigation, our ESR spin-trapping results demonstrate that the heme-catalyzed reaction of native CcO with H 2 O 2 gave a sulfur-centered radical that was identified by spin trapping with DMPO. When the protein was pretreated with thiol-blocking reagent, a carbon-centered radical was de-tected in the reaction of NEM-CcO with H 2 O 2 . Presumably, the cysteine residue of CcO was not the initial site for oxidative damage, but a ferryl porphyrin cation radical (compound I of the peroxidases) was probably first generated in the reaction of CcO with H 2 O 2 . This reactive intermediate then oxidized the thiol group(s) of CcO to form sulfur-centered radical adduct(s). As suggested by Prü tz (36,37), cysteine can act as a "sink," being an important terminus for the cascading transfer of radical centers in proteins. Similar results were also found in the hemoglobin and bovine serum albumin when the oxidative damage of protein was initiated with an hydroxyl radical attack (38). The protective role of thiol group(s) in CcO can be further supported by two events observed in this study. First, direct oxidation of CcO with Ce 4ϩ led to the trapping of a thiyl radical but not a carbon-centered radical, as illustrated in Fig.  2A. Second, both Ce 4ϩ oxidation and NEM modification had no effect on the enzymatic activity of CcO (data not shown). As revealed by the recently determined high resolution, threedimensional structure of CcO, cysteine residues were found primarily on the surface of protein and were logically involved in this shielding effect (39). However, it may not be ruled out that oxidative damage can be thermodynamically transferred from thiyl radical intermediates to carbon-centered radicals as demonstrated by Prü tz et al. (40). Thiyl radicals are reactive species capable of addition even across carbon-carbon double bonds and possibly capable of initiating lipid peroxidation (41,42).
The elimination of the DMPO/thiyl radical adduct with NEM gave the spectrum shown in Fig. 1C, which is weak and apparently results from another protein-derived radical. In a like manner, NEM also inhibited the thiyl radical formation from the direct oxidation of CcO with Ce 4ϩ . In addition, the spectral characteristic was also greatly changed when either DBNBS or MNP was used as the spin trap in the reaction system of NEM-CcO/H 2 O 2 . It is clear that one or more thiols are involved in the reaction between CcO and H 2 O 2 . It is noteworthy that the MNP adduct of NEM-CcO is more easily fragmented with Pronase treatment than is the adduct of native CcO, and this observation is in line with the sheltering effect of sulfhydryl groups in CcO, as suggested previously.
Earlier reports indicated that an immobilized nitroxide was detected in the reaction of other heme proteins such as myoglobin and cytochrome c with H 2 O 2 when either DBNBS or sharper line shapes, are indicated by solid lines. B, sample A was recovered from the flat cell and dialyzed against 100 ml of 50 mM sodium/potassium phosphate buffer containing 1.0% sodium cholate, pH 7.4, for 2 h before the ESR spectrum was re-recorded. C, same as A, except the sample was treated with 10 mg/ml Pronase for 6 h at room temperature. D, same as A, except the NEM-CcO was pretreated with 10 eq of KCN for 5 min at room temperature. The instrumental parameters were the same as described in the legend to Fig. 3, except the modulation amplitude was 2 G for spectrum B. MNP was included in the reaction mixture (17,34). These protein-derived, immobilized nitroxides have been unambiguously identified as, respectively, the tryptophanyl radical of metmyoglobin and the tyrosyl radical of cytochrome c by resolving the low-field super-hyperfine structures of their radical adducts. Likewise, an analogous, immobilized nitroxide was also observed when the NEM-pretreated CcO was reacted with H 2 O 2 in the presence of either of the nitroso spin traps. The trapped radical site was located on a tertiary carbon because a three-line ESR spectrum was detected after nonspecific proteolysis of either the DBNBS-or MNP/ ⅐ NEM-CcO-derived adducts. Tryptophan and tyrosine are two likely amino acid residues that could form radical adducts and produce the triplet spectra observed in this study. In fact, the a N values calculated here for the DBNBS/ ⅐ NEM-CcO-derived and MNP/ ⅐ NEM-CcOderived adducts were very close to those reported for the corresponding adducts of metmyoglobin and cytochrome c (17,34).
The ESR spectrum of the DBNBS/ ⅐ NEM-CcO disclosed a stable adduct. The proteolysis of the DBNBS/ ⅐ NEM-CcO-derived sample yielded an unstable spin adduct with a three-line spectrum, raising the possibility that DBNBS was trapping a tyrosyl radical. However, it is unlikely for this short-lived species to be identified by resolving its super-hyperfine structure, because of its instability.
Because both the tryptophan and tyrosine radical adducts with MNP have similar primary a N values (17,33,34), the super-hyperfine structure was necessary to identify which MNP adduct was formed in the reaction of NEM-CcO with H 2 O 2 . We were able to address this issue by comparing the high resolution ESR spectrum from the MNP-d 9 / ⅐ NEM-CcO-derived adduct with that from the authentic MNP-d 9 / ⅐ Tyr adduct. The fact that both resolved spectra at low field are very similar, if not identical, served as indisputable proof that the tyrosyl radical was trapped by MNP in the peroxidation of NEM-CcO with H 2 O 2 . The formation of a tyrosyl radical intermediate in the thiol-blocked CcO can be explained in a parallel way to those in metmyoglobin and cytochrome c following their reaction with H 2 O 2 (17,33). That is, hydrogen peroxide was reduced to water, forming a compound I-like ferryl porphyrin cation radical, which subsequently accepted an electron from the tyrosine residue, forming a ferryl tyrosyl radical species from NEM-CcO. It is also possible that ferryl-oxo compound II could oxidize the amino acids in the vicinity of heme and subsequently produce a tyrosyl radical. However, in the case of native CcO, the cysteine residues were available and acted as a sink in an electron-transfer process that occurred in the protein following the reaction of heme a (or a 3 ) with H 2 O 2 . In fact, a ferryl-oxo species from the reaction of dioxygen with heme a 3 also had been proposed based on resonance Raman spectroscopic studies (25,43). It has been suggested by Proshlyakov et al. (44 -46) that the detected ferryl-oxo intermediates correspond to the so-called compound P and compound F, which are involved in the catalytic mechanism of CcO. Compound P and compound F are equivalent to compound I and compound II, respectively, of the peroxidases if the oxygen chemistry (such as homolytic cleavage of the O-O bond of peroxide species) is similar to that of the peroxidase (22)(23)(24). Furthermore, recent works have demonstrated that the addition of hydrogen peroxide to the fully oxidized CcO can result in formation of the same compound P and compound F species (44 -46). The tyrosyl radical is known to have a role in the electron transfer events in Photosystem II (47,48), prostaglandin synthase (49,50), and ribonucleotide reductase (51). Circumstantial evidence suggests that the participation of the tyrosyl radical in the enzyme turn-over of CcO is possible (24). As revealed by the implication of the recently reported x-ray structure of CcO (52, 53) and by the results of mutagenesis studies (54), a highly conserved tyrosine (Tyr-244 in bovine oxidase, Tyr-280 in Paracoccus oxidase) is located at the active site of CcO (24). Elucidation of the possibly catalytic role of Tyr-244 using the ESR spin-trapping technique is currently under investigation in our laboratory.