Mitochondrial Aconitase Is a Source of Hydroxyl Radical AN ELECTRON SPIN RESONANCE INVESTIGATION*

Mitochondrial aconitase (m-aconitase) contains a [4Fe-4S] 2 1 cluster in its active site that catalyzes the stereospecific dehydration-rehydration of citrate to isocitrate in the Krebs cycle. It has been proposed that the [4Fe-4S] 2 1 aconitase is oxidized by superoxide, gen-erating the inactive [3Fe-4S] 1 1 aconitase. In this reaction, the likely products are iron(II) and hydrogen peroxide. Consequently, the inactivation of m-aconitase by superoxide may increase the formation of hydroxyl radical ( z OH) through the Fenton reaction in mitochondria. In this work, evidence for the generation of z OH from the reaction of m-aconitase with superoxide is provided using ESR spin trapping experiments with 5-diethoxypho-sphoryl-5-methyl-1-pyrroline N -oxide and a -phenyl- N tert -butylnitrone. Formation of free z OH was verified with the z OH scavenger Me 2 SO, which forms methyl radical upon reacting with z OH. The addition of Me 2 SO to incubation mixtures containing m-aconitase and xan-thine/xanthine oxidase yielded methyl radical, which was detected by ESR spin trapping. Methyl radical formation was further confirmed using [ 13 C]Me 2 SO. Paral- lel low temperature ESR experiments demonstrated that the generation of the [3Fe-4S] 1 1 cluster increased with increasing additions of superoxide to m-aconitase. This reaction was ESR spin trapping The use of a loop-gap the utilization of very small of purified ESR spin trapping experiments at room and instrumental set- tings as follows: milliwatts;

There is much debate in the literature on the relative importance of hydroxyl radical ( ⅐ OH) and peroxynitrite in free radical pathology (1). Clarification of the mechanism centered on this subject is of considerable importance, especially in mitochondria, cellular organelles that are constantly exposed to low levels of superoxide anion (2,3). Several neurodegenerative diseases (e.g. Alzheimer's disease, Parkinson's disease, and Lou Gehrig's disease or amyotrophic lateral sclerosis) and aging have been linked to mitochondrial oxidative damage that results in decreased mitochondrial function (3,4). However, in biological systems, it is nearly impossible to associate a specific damage to a single oxidant. For example, superoxide and nitric oxide ( ⅐ NO) co-generated at very low levels (Ϸ10 Ϫ8 M) in cells will form peroxynitrite (ONOO Ϫ ) via a nearly diffusion-con-trolled reaction (5,6). The toxicological significance of these species is clearly dependent on cell type, the biological targets, and their relationship to one another. One of the sensitive biological targets in oxidative damage to mitochondria is aconitase, an iron-sulfur protein that catalyzes the stereospecific dehydration-hydration of citrate to isocitrate in the Krebs cycle (7).
Aconitase activity in mitochondria has been reported to be a sensitive redox sensor of reactive oxygen and nitrogen species in cells (8 -11). Aconitase contains a cubane-type [4Fe-4S] 2ϩ cluster in its active site with three iron atoms bound to cysteinyl groups and inorganic sulfur atoms and a fourth labile iron atom (Fe-␣). This Fe-␣ is unique in that it is not bound to a protein cysteine, but rather to a hydroxyl group of substrate and water (7). The labile Fe-␣ is released upon oxidation of the [4Fe-4S] 2ϩ cluster with the concomitant formation of inactive [3Fe-4S] 1ϩ enzyme. Aconitase is inactivated rapidly by superoxide (k Ϸ 10 7 M Ϫ1 s Ϫ1 ) (12) in the presence and absence of substrate and relatively slowly by peroxynitrite (k Ϸ 10 5 M Ϫ1 s Ϫ1 ) and ⅐ NO (13,14). However, the reaction between aconitase and peroxynitrite is strongly inhibited by the addition of substrate that binds to the enzyme with high affinity (14).
It was recently proposed that the reaction between mitochondrial aconitase (m-aconitase) 1 and superoxide plays a major role in mitochondrial oxidative damage (15)(16)(17). During this reaction, it has been proposed that iron is released from maconitase as iron(II) with the concomitant generation of hydrogen peroxide. This facilitates the formation of "free" hydroxyl radical in mitochondria. In the presence of intracellular reducing agents (e.g. glutathione, ascorbate, and NADPH), iron(II) is reincorporated into the inactive form of m-aconitase to regenerate the active form. According to this proposal, hydroxyl radical should be continuously generated in mitochondria as a result of the reaction between superoxide and aconitase. However, the experimental verification of this intriguing mechanism has so far been lacking.
The objective of this study is to provide evidence, using ESR, for the formation of hydroxyl radical and inactive [3Fe-4S] 1ϩ species from the reaction between superoxide and purified maconitase. Direct low temperature ESR was used to quantify [3Fe-4S] 1ϩ species. 2 Hydroxyl radical was detected by ESR spin trapping using a novel phosphorylated spin trap, 5-diethoxyphosphoryl-5-methylpyrroline N-oxide (DEPMPO), and a loop-gap resonator, which makes it possible to obtain ESR spectra using exceedingly small amounts of enzyme (19). Our results indicate that the reaction between m-aconitase and superoxide releases iron(II) from the [4Fe-4S] 2ϩ cluster, which can subsequently catalyze the formation of free hydroxyl radical. The biological implications of these reactions are discussed.

EXPERIMENTAL PROCEDURES
Materials and General Methods-Bovine heart mitochondrial aconitase was purified according to published procedures (20). Protein purity (Ն95%) was determined by dye staining of SDS-polyacrylamide gels. The cluster content was measured from iron, sulfide, and protein content analyses as described previously (21,22). Active enzyme was prepared immediately before use by anaerobically incubating the protein with iron and dithionite. To remove excess reagent, the preparation was desalted anaerobically on Sephadex G-50 columns equilibrated with buffer (50 mM phosphate, pH 7.4). The same protocol was followed for reactivation studies, except that dithiothreitol or glutathione replaced dithionite as the reductant. Enzyme activity was measured following the formation of cis-aconitate from isocitrate at 240 nm, and the concentration was calculated using an extinction coefficient of 3.6 mM Ϫ1 cm Ϫ1 . The assay was performed at 25°C in 90 mM Tris-HCl buffer, pH 8.0, containing 20 mM DL-isocitrate. Rates of superoxide generation by a xanthine (1-0.5 mM)/xanthine oxidase (0.1 unit/mg of solid) system (Roche Molecular Biochemicals catalog no. 1048180) were determined by following the reduction of ferricytochrome c (50 M) at 550 nm, and the concentration was calculated using an extinction coefficient of 21 mM Ϫ1 cm Ϫ1 .
Electron Spin Resonance Measurements-Direct ESR spectra were obtained at 10 -15 K using calibrated quartz ESR tubes. The spectra were analyzed on a Varian E109 Century Series spectrometer operating at 9.17-GHz and 100-kHz field modulation. A frequency counter (EIP Model 548) was used to determine frequency, and a gaussmeter (MH-110R Radiopan, NMR magnetometer) was used to determine field positions. Instrumental conditions for detecting the [3Fe-4S] 1ϩ cluster were as follows: microwave power, 0.1 milliwatts; modulation amplitude, 5 G; time constant, 0.128 s; and scan rate, 0.83 G/s. Quantification of spin concentration was carried out using a 1.0 mM copper perchlorate standard by comparing the double integral of the spectra and standard under similar conditions. ESR spin trapping experiments were recorded at room temperature on a Varian E109 spectrometer operating at 8.9-GHz and 100-kHz field modulation equipped with a loop-gap resonator (18). The use of a loop-gap resonator enabled the utilization of very small quantities of purified enzyme. ESR spin trapping experiments were analyzed at room temperature, and the instrumental settings were as follows: microwave power, 2 milliwatts; modulation am- . In A, the dashed line is a computer simulation fitted by considering two isomers of DEPMPO-OOH (in gauss): Isomer 1, a N ϭ 13.1, a H ϭ 12.1, and a P ϭ 49.7; and Isomer 2, a N ϭ 13.1, a H ϭ 10, and a P ϭ 49.5. In B, the dashed line is a computer simulation fitted by considering two spin adducts and obtained using hyperfine splitting constants (in gauss): DEPMPO-OH (38% contribution), a N ϭ 14, a H ϭ 13.4, and a P ϭ 46.9; and DEPMPO-OOH (62% contribution), Isomer 1, a N ϭ 13.2, a H ϭ 11.5, and a P ϭ 50; and Isomer 2, a N ϭ 13.1, a H ϭ 13.2, and a P ϭ 48. C is the same as B in the presence of 100 M m-aconitase. The dashed line is a computer simulation obtained using hyperfine splitting constants (in gauss): a N ϭ 14, a H ϭ 13.1, and a P ϭ 46.9. Inset, direct low temperature ESR of the [3Fe-4S] 1ϩ species. The spectrum is from a diluted sample (5 M) of C.
plitude, 1G; time constant, 0.128 s; and scan rate, 1.67 to 0.83 G/s. These results suggest that the catalytically active iron concentration was increased during the reaction of m-aconitase with superoxide. To determine that iron is indeed released from the [4Fe-4S] 2ϩ cluster as has been previously shown (12), the effect of substrate on DEPMPO-OH yields was investigated. The addition of saturating concentrations of D-isocitrate (1 mM) to incubation mixtures of m-aconitase with superoxide diminished the yield of DEPMPO-OH by ϳ58% (Fig. 2, trace C; cf. trace A). Control experiments demonstrated that isocitrate had no effect on the formation of DEPMPO-OH from incubation mixtures containing iron and xanthine/xanthine oxidase (data not shown). It is known that binding of substrate stabilizes the [4Fe-4S] 2ϩ cluster of aconitase and thereby protects the enzyme against inactivation (7,23). It is therefore likely that the inhibitory effect of isocitrate on the formation of DEPMPO-OH is due to a decreased rate of iron release from the [4Fe-4S] 2ϩ cluster.

Spin Trapping of Hydroxyl Radical Formed from the Reaction between Mitochondrial
To investigate the contribution of hydrogen peroxide formed from superoxide on iron release from m-aconitase, spin trapping experiments were performed with a bolus addition of hydrogen peroxide. The reaction between hydrogen peroxide and m-aconitase generated DEPMPO-OH (Fig. 2, trace D), which represented ϳ31% of the yield obtained during the reaction of m-aconitase with superoxide. This result is consistent with the relatively slower second-order rate constant for the reaction of hydrogen peroxide with aconitase (Ϸ10 3 M Ϫ1 s Ϫ1 ) (12) as compared with the rate constant for the reaction between aconitase and superoxide (Ϸ10 7 M Ϫ1 s Ϫ1 ). Taken together, these results indicate that the reaction between maconitase and superoxide is primarily responsible for the release of iron from the [4Fe-4S] 2ϩ cluster, which subsequently catalyzes the generation of hydroxyl radical.
Effect of Dimethyl Sulfoxide on Radical Adduct Formation-Because 5,5-dimethyl-1-pyrroline N-oxide (DMPO)-and DEPMPO-hydroxyl adducts can be formed through several mechanisms, additional experiments were performed using a hydroxyl radical scavenger, dimethyl sulfoxide (Me 2 SO). It is well known that hydroxyl radical reacts with Me 2 SO to form methyl radical ( ⅐ CH 3 ); and therefore, the detection of the DMPO-or DEPMPO-methyl adduct is a diagnostic indicator of free hydroxyl radical formation (Reaction 1).
As shown in Fig. 3B, the addition of Me 2 SO to incubation mixtures of m-aconitase and superoxide produced a DEPMPOcarbon-centered adduct whose ESR parameters are consistent with trapping of methyl radical. The identity of the trapped radical was confirmed using 13 C-labeled Me 2 SO. Substitution of [ 12 C]Me 2 SO for [ 13 C]Me 2 SO generated the 13 C-labeled DEPMPO-methyl adduct, DEPMPO-13 CH 3 (Fig. 3C). In additional experiments, the spin trap ␣-phenyl-N-tert-butylnitrone (PBN), which forms persistent spin adducts with carbon-centered radicals, was substituted for DEPMPO. As shown in Fig.  3D, no ESR signal was detected from the reaction between m-aconitase and superoxide in the presence of PBN. This is consistent with the instability of PBN-oxygen-centered adducts at physiological pH. However, the addition of Me 2 SO to the incubation mixtures generated a spectrum corresponding to the PBN-methyl adduct (Fig. 3E), and a 13 C-labeled PBNmethyl adduct was detected with [ 13 C]Me 2 SO (Fig. 3F). These results demonstrate that the reaction between m-aconitase and superoxide generates free hydroxyl radical.

Direct ESR Analysis of [3Fe-4S] 1ϩ Cluster Formation during Reversible Inactivation of Mitochondrial Aconitase by Superoxide-
The characterization and quantitative analysis of [3Fe-4S] 1ϩ species generated from the reaction of m-aconitase with superoxide were based on the comparative analysis with standards obtained from the stoichiometric oxidation of m-aconitase with ferricyanide (24) ( Table I,  different time points and immediately frozen in liquid nitrogen to quench the reaction. As shown in Table I 2ϩ form, the reversibility of the oxidation of m-aconitase by superoxide was investigated by examining the recovery of enzyme activity. The samples used in low temperature experiments were thawed and reactivated by anaerobic reduction with either glutathione (5 mM) or dithiothreitol (10 mM) in the presence of iron(II). As shown in Table I (part B), reactivation of samples from a 5-min incubation of m-aconitase with superoxide led to the recovery of 90% of the initial enzyme activity. This result demonstrates that inactivation of m-aconitase generates [3Fe-4S] 1ϩ and is therefore a reversible reaction. The addition of isocitrate (1 mM) to incubation mixtures of m-aconitase did not prevent superoxide-mediated oxidation of the cluster and loss of activity. This result indicates that although isocitrate causes a decrease in the formation of hydroxyl radical (Fig. 2), the extent of the inhibition is not high enough to prevent the inactivation of the enzyme. Under these conditions, therefore, the recovery of enzyme activity was maximal (Table I, part B). As shown in Fig. 1, oxidation of m-aconitase by superoxide generated hydroxyl radical, a strong oxidant that can further oxidize amino acids critical for m-aconitase function. To investigate whether or not hydroxyl radical generated from m-aconitase contributes to the inactivation of m-aconitase by superoxide, the effect of hydroxyl radical scavengers on inactivation and reactivation of the enzyme was investigated. The addition of Me 2 SO (10%, v/v) or DEPMPO (0.1 M) did not diminish superoxide-mediated formation of inactive [3Fe-4S] 1ϩ m-aconitase, and neither DEPMPO or Me 2 SO prevented enzyme inactivation. A slight increase in m-aconitase activity was de- tected in the presence of Me 2 SO, an effect that seems related to the effect of Me 2 SO on enzyme stability. Taken together, these results indicate that hydroxyl radical generated during the reaction of m-aconitase with superoxide does not irreversibly inactivate the enzyme (22,25). Thus, in the presence of cellular reductants, it is likely that m-aconitase will recycle iron(II) and enhance the generation of hydroxyl radical.
The effect of superoxide on [3Fe-4S] 1ϩ formation and the recovery of enzyme activity was further investigated in incubation mixtures containing m-aconitase at 10-fold lower concentrations than described above (Table I, part C). Under this condition, m-aconitase was oxidized to the [3Fe-4S] 1ϩ species, but the recovery of enzyme activity was lower than that previously detected (Table I, part C; cf. part B). Hydrogen peroxide (1 mM) also oxidized [4Fe-4S] 2ϩ aconitase to form the [3Fe-4S] 1ϩ species. However, the recovery of enzyme activity was much lower than that measured for superoxide under similar incubation conditions (Table I, part D). These results indicate that inactivation of m-aconitase by hydrogen peroxide generated from superoxide dismutation or added as a bolus is due to oxidation of the [4Fe-4S] 2ϩ cluster, leading to protein destabilization. DISCUSSION Superoxide-mediated Redox Cycling of Mitochondrial Aconitase-The intriguing hypothesis on the reaction between su-peroxide and [4Fe-4S] 2ϩ clusters of dehydratases leading to hydroxyl radical formation has been previously proposed (12,26). However, to date, the experimental verification of this hypothesis has not been provided. In this study, we present, for the first time, ESR spin trapping evidence for the generation of hydroxyl radical during superoxide-induced oxidation of maconitase ( Figs. 1 and 3). This work was made possible because of recent breakthroughs in ESR technology and spin trap synthesis. First, the use of a loop-gap resonator in spin trapping has made it feasible to obtain ESR spectra from exceedingly small sample volumes (Ϸ10 l) (27). As a result, we were able to perform a large number of experiments using highly purified m-aconitase. Clearly, this would not be possible with conventional ESR using cavity resonators since they use large sample volumes. Second, the newly synthesized DEPMPO spin trap reacts with superoxide to form a persistent DEPMPO-OOH adduct (28). Unlike DMPO-OOH, the DEPMPO-OOH adduct does not spontaneously decay to form the DEPMPO-OH adduct (18,27). In this study, in which a mixture of superoxide and hydroxyl radical was generated, the use of DEPMPO has made spin trapping interpretations less confounding.
In addition to hydroxyl radical trapping, we demonstrated that the [4Fe-4S] 2ϩ aconitase is quantitatively oxidized to the [3Fe-4S] 1ϩ cluster by superoxide ( Fig. 1 and Table I). From monitoring the ESR-active [3Fe-4S] 1ϩ aconitase, we showed that superoxide releases only one iron atom from the [4Fe-4S] 2ϩ cluster ( Table I), suggesting that the reaction between superoxide and m-aconitase is stoichiometric. As shown in Table I, this was also confirmed by measuring the activity recovery. It is known that in the presence of cellular reductants such as thiols, m-aconitase recycles iron by incorporating iron(II) (Scheme I and Table I) (7,16). Therefore, under in vivo conditions, it is likely that aconitase undergoes redox cycling between the [4Fe-4S] 2ϩ and [3Fe-4S] 1ϩ forms. Another aspect of this reaction is that inactivation of m-aconitase by superoxide is a reversible process and that hydroxyl radical generated during the reaction does not destroy the iron-sulfur cluster at the active site of the enzyme. This result lends further support for using m-aconitase activity as a measure of intracellular superoxide steady-state levels (8,16). Oxidation of m-aconitase by hydrogen peroxide also generates hydroxyl radical, albeit at lower yields compared with superoxide-mediated oxidation of aconitase. The reactivation of the enzyme was partial, however, indicating that hydrogen peroxide may irreversibly inactivate the enzyme by destabilizing the protein structure. Recently, it was reported that cytosolic aconitase inactivation by hydrogen peroxide is reversible, although no data were provided (24).
Pathological Consequences-Emerging data strongly suggest that an increase in superoxide in mitochondria causes severe dysfunction (3,4). Superoxide-mediated mitochondrial dysfunction may be explained by enhanced formation of hydroxyl radical and aconitase inactivation. It was reported that the anthracycline antibiotic doxorubicin causes a rapid inactivation of m-aconitase in cardiomyocytes (29). In this system, the superoxide dismutase mimetic manganese(III) tetrakis-(4-benzoic acid) porphyrin inhibited aconitase inactivation, whereas the nitric-oxide synthase inhibitor N w -nitro-L-arginine methyl ester exacerbated aconitase inactivation (28). These data reveal that conditions that lower intracellular nitric oxide enhance m-aconitase activation, suggesting that superoxide is more potent than peroxynitrite in the inactivation of m-aconitase (28).
The source of free iron in oxidative cellular pathology has not been determined. It is likely that the reaction of m-aconitase with superoxide releases iron, which, in turn, may increase the concentration of low molecular mass iron complexes with cit-SCHEME 1. Mechanism of superoxide-mediated generation of hydroxyl radical from m-aconitase (modified from Ref. 16). rate (29) or ATP or ADP (30). These iron complexes and iron bound to DNA (31) represent a redox-active pool of iron capable of catalyzing free radical reactions including the site-specific generation of hydroxyl radical. In this study, we have shown that m-aconitase is a plausible physiological source of free iron during oxidative stress resulting from increased superoxide formation. Summary-We have demonstrated that the reaction between m-aconitase and superoxide generates free hydroxyl radical. Superoxide releases iron(II) from the [4Fe-4S] 2ϩ aconitase, forming the inactive [3Fe-4S] 1ϩ aconitase and hydrogen peroxide, thus facilitating hydroxyl radical formation by the Fenton reaction. It is likely that the reaction between superoxide and m-aconitase may enhance mitochondrial oxidative damage associated with the pathophysiology of several chronic neurodegenerative and cardiovascular diseases.