Direct Magnetic Resonance Evidence for Peroxymonocarbonate Involvement in the Cu,Zn-Superoxide Dismutase Peroxidase Catalytic Cycle*

Cu,Zn-superoxide dismutase (SOD1) is a copper- and zinc-dependent enzyme. The main function of SOD1 is believed to be the scavenging and detoxification of superoxide radicals. Nevertheless, the last 30 years have seen a rapid accumulation of evidence indicating that SOD1 may also act as a peroxidase, an alternative function that was implicated in the onset and progression of familial amyotrophic lateral sclerosis. Although SOD1 peroxidase activity and its dependence on carbon dioxide have been well described, the molecular basis of the SOD1 peroxidase cycle remains obscure, because none of the proposed catalytic intermediates have so far been identified. In view of recent observations, we hypothesized that the SOD1 peroxidase cycle relies on two steps: 1) reduction of SOD-Cu(II) by hydrogen peroxide followed by 2) oxidation of SOD-Cu(I) by peroxymonocarbonate, the product of the spontaneous reaction of bicarbonate with hydrogen peroxide, to produce SOD-Cu(II) and carbonate radical anion. This hypothesis has been investigated through electron paramagnetic resonance and nuclear magnetic resonance to provide direct evidence for a peroxycarbonate-driven, SOD1-catalyzed carbonate radical production. The results gathered herein indicate that peroxymonocarbonate (\batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \(\mathrm{HOOCO_{2}^{-}}\) \end{document}) is a key intermediate in the SOD1 peroxidase cycle and identify this species as the precursor of carbonate radical anions.

Cytosolic Cu,Zn-superoxide dismutase (SOD1) 2 is a metal-dependent enzyme capable of accelerating the rate of spontaneous superoxide dismutation into O 2 and H 2 O 2 through the redox cycling of its copper ion (1,2). SOD1 is widely distributed in mammalian cells and tissues and has been demonstrated to be located in the cytosol and in the intermembrane space of the mitochondria (see Ref. 3 and references therein). Because of that, SOD1 is believed to be a major player in the first line defense against reactive oxygen species, in particular superoxide anion.
In addition to its dismutase activity, SOD1 possesses a well described but incompletely understood peroxidase activity which is dependent on hydrogen peroxide and markedly stimulated by small oxidizable anions such as nitrite and the ubiquitous carbon dioxide (3)(4)(5)(6)(7)(8)(9)(10)(11)(12). The peroxidase activity of SOD1 has been proposed to impact the onset and progression of familial amyotrophic lateral sclerosis, a severely debilitating fatal disease characterized by the selective death of motor neurons (13)(14)(15)(16)(17)(18). Although several reports exist in the literature indicating the formation of SOD1 aggregates and accumulation as a potential cause in the pathology progression, conflicting hypotheses are still under debate concerning the mechanisms that lead to the formation of SOD1-protein aggregates (19 -21). Although some support the suggestion that free radical-induced covalent cross-links among SOD1 amino acids play a fundamental role in aggregate formation (22,23), others support the view that metal loss from the enzyme structure leads to an unstable apo-form of SOD1 with increased capacity to form aggregates (24,25). A detailed understanding of the SOD1 peroxidase cycle is essential to unraveling the mechanisms through which SOD1 aggregates are produced.
The SOD1 peroxidase cycle is initiated when SOD1-Cu(II) is reduced by H 2 O 2 or its deprotonated form (12), the peroxide anion (HOO ؊ ), to SOD1-Cu(I). This latter species is subsequently oxidized to a hypervalent intermediate (proposed to be either SOD1-Cu(III), SOD1-Cu(II)-⅐ OH, or SOD1-CuϭO) (8,9) that remains to be characterized. The reduction of this hypervalent intermediate by small anions is supposed to close the cycle, leading to the native enzyme and diffusible highly reactive radicals derived from the anionic substrates (6,10).
During its peroxidase cycle, a considerable fraction of SOD1 is inactivated due to the oxidation of the copper-binding histidines to oxohistidine, presumably by the hypervalent intermediate, in a process that can be prevented by the presence of reducing substrates and, in their absence, unavoidably leads to copper loss (26). Although this process is well described for the heme-dependent peroxidase cycle, current literature data (9,(27)(28)(29)(30) and the fact that the proposed SOD1-bound hypervalent copper states (Cu(III), Cu(II)ϭO, and Cu(II)/ ⅐ OH) have never been characterized suggested to us that an alternative mechanism may take place, leading to CO 3 . production from HCO 3 Ϫ and H 2 O 2 by the enzyme, a process that does not involve copper oxidation beyond the thermodynamically stable Cu(II) form. In the presence of HCO 3 Ϫ , a significant fraction of H 2 O 2 is promptly converted to HOOCO 2 Ϫ through the perhydration of CO 2 (27,28,31,32). The peroxo bond in peroxymonocarbonate can be cleaved by reduced metals to produce CO 3 . and H 2 O (30, 33) (Reactions 1-5), where Reactions 1 and 2 represent SOD1 reduction, Reactions 3 and 4 represent SOD1 oxidation, and Reaction 5 represents peroxycarbonate formation. Interestingly, studies employing molecular modeling of the SOD1 active site indicate that HCO 3 Ϫ and H 2 O 2 gain access to the SOD1 active site, where they react to produce HOOCO 2 Ϫ in close proximity to the copper ion (29). This interaction of HOOCO 2 Ϫ with Cu(I) may result in CO 3 . production.
On the basis of these new data, we hypothesized that SOD1-Cu(I), which is the predominant form of SOD1 exposed to excess hydrogen peroxide (8,9), is oxidized back to the native form of the enzyme by HOOCO 2 Ϫ more efficiently than by H 2 O 2 itself or HOO ؊ . The latter two oxidations would slowly produce ⅐ OH radicals (or the equivalent SOD1-Cu(III), SOD1-Cu(II)ϭO, or Cu(II)/ ⅐ OH) in the enzyme active site, leading to the observed inactivation of SOD1 (see Scheme 1). Here we present data that strongly support this hypothesis; they indicate that HOOCO 2 Ϫ is a key substrate for reduced SOD1, which mediates SOD1-Cu(I) reoxidation back to the resting SOD1-Cu(II) severalfold faster than H 2 O 2 itself and, in doing so, serves as the carbonate radical anion precursor.

MATERIALS AND METHODS
Chemicals-Superoxide dismutase from bovine erythrocytes, sodium bicarbonate, and sodium dithionite were purchased from Sigma. Acetonitrile (spectroscopy grade) was purchased from Caledon Laboratories (Georgetown, Canada). Sodium phosphate was purchased from Mallinkrodt Baker Inc. (Paris, KY). Chelex 100 resin was purchased from Bio-Rad. Phosphate buffers (0.1 M) were treated with Chelex 100 resin to remove traces of transition metal ions, and 200 M diethylenetriaminepentaacetic acid (DTPA) was added to minimize the possibility of metal interference, except for the NMR studies, where cyclohexanediaminetetraacetic acid (CDTA) was used.
Electron Paramagnetic Resonance Experiments-EPR spectra were recorded on a Bruker EMX EPR spectrometer (Billerica, MA) operating at 9.81 GHz with a modulation frequency of 100 kHz and equipped with an ER 4122 SHQ cavity. All reactions were performed at room temperature and transferred to 1-ml polyethylene syringes. After the indicated incubation times, samples were frozen in liquid nitrogen, the tips of the syringes were cut off, and the frozen contents were pushed into a Dewar flask containing liquid nitrogen. EPR spectra were recorded at 77 K. All samples contained 15% acetonitrile to minimize cracking of the frozen content. Linear base-line correction of the EPR spectra was computationally performed using Bruker's WINEPR software version 2.11. For kinetic studies, estimation of SOD-Cu(II) concentrations at each time point was performed by doubly integrating the acquired spectra after smoothing the noise using the same software. Solution pH was controlled by the use of phosphate buffer. Incubations were performed in sealed containers to avoid loss of CO 2 . All incubations were recovered at the end of each experiment and had their pH double-checked in order to guarantee that the pH remained unchanged over the course of the experiment.
NMR Study-NMR data were collected using a Varian Inova 600 spectrometer (Palo Alto, CA) with a 5-mm broadband probe tuned to carbon (150.83 MHz). Bovine kidney SOD1 was purchased from Calzyme Laboratories (San Luis Obispo, CA). 13 C-Labeled sodium hydrogen carbonate was purchased from Isotec/Sigma. Samples were placed in 5-mm NMR tubes, sealed with septum caps (Wilmad Glass, Buena, NJ), and then allowed to equilibrate for 30 min. Additions to the samples were made by injecting compounds through the septum cap using a Hamilton syringe. 10% D 2 O was added to the samples for the purpose of locking and shimming. All studies were run at 25°C.
Kinetics of SOD1-mediated H 2 O 2 Consumption-H 2 O 2 uptake experiments were performed using an H 2 O 2 electrode attached to an Apollo 4000 free radical analyzer unit (World Precision Instruments, Sarasota, FL). Amperometric detection of H 2 O 2 was carried out using a poise voltage of ϩ400mV. The reaction was constantly stirred throughout the course of the experiment. SCHEME 1. Schematic representation of the peroxidase catalytic cycle of Cu,Zn-SOD in the presence of HCO 3 ؊ /CO 2 . Native Cu,Zn-SOD is reduced by the peroxide anion, which gains access to the copper through the enzyme's anion channel. Reduced Cu,Zn-SOD is reoxidized by the peroxycarbonate anion (HOOCO 2 Ϫ ), which is in equilibrium with H 2 O 2 /HCO 3 Ϫ /CO 2 , leading to carbonate radical anion production. Superoxide anion (O 2 . ), the product of HOO Ϫ -induced SOD1 reduction, can oxidize SOD1-Cu(I) back to its resting SOD-Cu(II) state at diffusion-limited rates; however, it can alternatively reduce another molecule of SOD1-Cu(II) to SOD1-Cu(I), considerably accelerating the rate of SOD1-Cu(II) reduction by H 2 O 2 . Whether O 2 . will act as a reductant of SOD1-Cu(II) or an oxidant of SOD1-Cu(I) will depend on the ratio of SOD1-Cu(II)/SOD1-Cu(I) at a given time, because the rate constants for the reaction of O 2 . with both SOD1 states are close to the diffusion limits.
The data were fit to a monoexponential decay or growth model using the following equations, (for the exponential growth), where A 1 is the initial and final concentration of the enzyme. The rate is given by Equation 3, the reactants being either H 2 O 2 or HCO 4 Ϫ . Assuming that the observed rates were linearly dependent upon the reactant concentration, the slopes of these plots give the apparent secondorder rate constants (k) for the decay and recovery of SOD-Cu(II).
Computer Simulation-The computer simulation of kinetic curves was performed using Gepasi (version 3.30) software, available to the public through the Internet (34,35).

RESULTS
pH Dependence of SOD-Cu(II) Reduction by H 2 O 2 -The first step in the peroxidase catalytic cycle of SOD1 is the reduction of the active site Cu(II) cation to Cu(I). SOD1-Cu(II) has a very characteristic EPR signal at 77 K, which is lost upon the reduction of Cu(II) (23). Therefore, mixtures of SOD1 with H 2 O 2 were prepared at different pH values to evaluate the rate of H 2 O 2 -and/or HOO Ϫ -mediated reduction of the enzymebound Cu(II). As seen in Fig. 1, increasing the pH from 5.4 to 9.8 increased the reduction of SOD1 by H 2 O 2 . Indeed, the intensity of the SOD-Cu(II) signal remained virtually unchanged after a 1-min incubation of the enzyme with H 2 O 2 at the acidic pH 5.4. At pH 7.4, a significant fraction of the enzyme was reduced by H 2 O 2 , as indicated by the decrease of the EPR signal intensity. After only 1 min of incubation at pH 9.8 with H 2 O 2 , the SOD1-Cu(II) spectrum was decreased by 70% (peak-to-peak intensity), indicating that most of the enzyme had been reduced by the peroxide.
The strong dependence of SOD1-Cu(II) reduction on the pH and the much faster reduction rate observed at alkaline pH are consistent with HOO Ϫ rather than molecular H 2 O 2 being the species that gains access to the SOD1 anion channel to reduce the active site's Cu(II), which is in agreement with activity experiments showing that high pH accelerates SOD1 reduction by H 2 O 2 (36,37).
In addition to studying the pH effect on the H 2 O 2 -mediated reduction of SOD1, experiments were performed to roughly estimate the rate of disappearance of the EPR g ϳ 2 signal at physiological pH. For that purpose, the remaining SOD1-Cu(II) (100 M) concentrations were measured by the double integration of the EPR spectra as a function of time.
From the pseudo-first-order rates of disappearance of SOD-Cu(II) (0.037 s Ϫ1 ) and assuming that the second-order rate constants are linearly dependent upon hydrogen peroxide concentration, we estimated k ϳ 37 M Ϫ1 s Ϫ1 for the reduction of SOD1 by H 2 O 2 (Fig. 1, inset). Although this value should be taken as a rough estimate, it is in good agreement with previous reports that determined k ϭ 50 M Ϫ1 s Ϫ1 by evaluating H 2 O 2 -induced SOD1 bleaching (38).

Effect of HCO 3
Ϫ /CO 2 on the Yield of SOD-Cu(I) from the H 2 O 2 -mediated Reduction of SOD1-Cu(II)-To assess the effect of HCO 3 Ϫ /CO 2 upon the H 2 O 2 -mediated reduction of SOD1-Cu(II) to SOD1-Cu(I) and its reoxidation back to SOD1-Cu(II), the experiments shown in Fig. 1 were repeated in the presence of HCO 3 Ϫ . For most of the experiments, HCO 3 Ϫ was allowed to equilibrate in solution for 15 min before the addition of H 2 O 2 . The only exception was the pH 9.8 spectrum (Fig. 2C), where a pH jump experiment was performed in order to specifically evaluate the effects of CO 2 rather than HCO 3 Ϫ (predominant at pH 9.8) upon SOD-Cu(I) reoxidation, as discussed below. For that, bicarbonate dissolved in water was added immediately before hydrogen peroxide so as to attain considerably higher initial concentrations of CO 2 in solution than would be present at equilibrium at pH 9.8. Different incubation times were used because, as demonstrated in Fig. 1, the SOD1-Cu(II) reduction rate is strongly dependent on pH. Therefore, to attain appreciable reduction yields, incubations at lower pH values required longer times.
As shown in Fig. 2, HCO 3 Ϫ /CO 2 had only a minor influence on the reduction yield of SOD1-Cu(II) at pH 5.4. At pH 5.4, both hydrogen peroxide and bicarbonate are predominantly in their protonated acidic forms (see Table 1), thus limiting the formation of HOOCO 2 Ϫ . This is likely to have limited the effect of HCO 3 Ϫ /CO 2 upon SOD1-Cu(I) reoxidation. At pH 7.4, bicarbonate had a marked effect on the yield of SOD1-Cu(II) reduction by H 2 O 2 after a 3-min incubation. At this pH, bicarbonate is mostly in its monoprotonated form, HCO 3 Ϫ (CO 2 ϳ 10%, HCO 3 Ϫ ϳ 90%, pK a ϭ 6.4). The fraction of the H 2 O 2 that is in the anionic form is ϳ0.01% (pK a 11.8). Therefore, since CO 2 and HOO Ϫ coexist at pH 7.4, significant yields of HOOCO 2 Ϫ are produced from the nucleophilic attack of HOO ؊ on CO 2 , which will then oxidize SOD1-Cu(I) back to its resting state.
At pH 9.8, CO 2 (spectrum C), but not HCO 3 Ϫ (spectrum D, equilibrated), strongly influenced the steady-state concentration of SOD1-Cu(I). At pH 9.8, ϳ1% of the H 2 O 2 is in the anionic form, HOO Ϫ . Therefore, in the presence of CO 2 (in nonequilibrated mixtures, such as in spectrum C), both HOO ؊ and CO 2 are present in high enough concentrations to produce significant yields of HOOCO 2 Ϫ . In the equilibrated solution, the CO 2 concentration is very limited, and, therefore, no effect of HCO 3 Ϫ on the reduction yield of SOD1-Cu(II) was noted.
Taken together, the studies shown in Fig. 2 indicate that the formation of HOOCO 2 Ϫ is dependent on both CO 2 and the peroxide anion HOO ؊ and that it increases the steadystate SOD1-Cu(II) concentration, presumably by reoxidizing the enzyme back to its SOD1-Cu(II) state and concomitantly producing CO 3 . .

Effect of H 2 O 2 and H 2 O 2 /HCO 3
Ϫ /CO 2 on the Reoxidation of SOD1-Cu(I) to SOD1-Cu(II)-To determine whether HOOCO 2 Ϫ oxidizes SOD1-Cu(I) more efficiently than H 2 O 2 or HOO ؊ alone, we incubated preformed SOD1-Cu(I) with H 2 O 2 in the presence and in the absence of HCO 3 Ϫ /CO 2 . SOD1-Cu(I) was prepared by briefly incubating SOD1-Cu(II) with sodium dithionite at a molar ratio of 1:1.2 under anaerobic conditions (for details, see "Materials and Methods").
As demonstrated in Fig. 3, dithionite at this concentration led to an almost complete reduction of SOD1-Cu(II); no appreciable traces of remaining SO 2 . , which is in equilibrium with dithionite, were detected by EPR. The addition of H 2 O 2 to prereduced SOD1 solutions had a minor effect upon the SOD1 EPR signal intensity (Fig. 3), suggesting that H 2 O 2 by itself is not an efficient mediator of SOD1-Cu(I) oxidation to SOD1-Cu(II). Indeed, the rate constant for the oxidation of SOD1-Cu(I) by H 2 O 2 has been estimated to be 13 M Ϫ1 s Ϫ1 (38). The addition of HCO 3 Ϫ /CO 2 to the samples markedly increased the intensity of the SOD1-Cu(II) EPR signal in a concentration-dependent manner (Fig. 3, D and E), indicating a key role for HOOCO 2 Ϫ as the mediator of SOD1-Cu(I) reoxidation back to the native enzyme with the consequent formation of carbonate radicals. In order to estimate the efficiency of was 1 mM. Incubations were performed at room temperature and contained 15% acetonitrile. Samples were frozen at liquid nitrogen temperature at the indicated times prior to acquisition of spectra. Instrumental conditions were as follows: power, 2 milliwatts; time constant, 163.84 ms; scan rate, 18 G/s; receiver gain, 5 ϫ 10 4 . Each spectrum is the result of four accumulations. For the inset, each point represents the average of three independent experiments whose EPR spectrum was accumulated eight times. Please note that our measurements refer to the disappearance of the g ϳ 2 signal attributable to SOD1-Cu(II). Although we believe that this disappearance is due to the reduction of SOD1-Cu(II) to SOD1-Cu(I), it is not possible to exclude the formation of other EPRsilent copper-derived species. Ϫ and the rapid freezing of solutions prepared at pH 9.8 (after 1 min), spectrum D was prepared with solutions in which HCO 3 Ϫ was allowed 10 min for equilibration. All mixtures contained 15% acetonitrile.

SOD1-Cu(I) oxidation by preformed HOOCO 2
Ϫ , samples of SOD1-Cu(I) were prepared as previously described. A preincubated mixture of H 2 O 2 /HCO 3 Ϫ was added to reduced SOD1 solutions and allowed to react for different times as indicated in Fig. 3, inset. Reaction mixtures were then frozen, and SOD1-Cu(II) concentrations were measured by EPR spectroscopy. From the observed rate of SOD1-Cu(II) formation extracted from Fig. 3 (inset) (k obs ϭ 0.12 s Ϫ1 ), we roughly estimated the second-order rate of SOD1-Cu(I) oxidation by HOOCO 2 Ϫ to be ϳ150 M Ϫ1 s Ϫ1 . Importantly, the rate constant for the reaction of SOD1-Cu(I) with HOOCO 2 Ϫ was more than 10 times higher than that determined for the reaction of SOD1-Cu(I) and H 2 O 2 (13 M Ϫ1 s Ϫ1 ) (38), which is consistent with the observation that HCO 3 Ϫ /CO 2 stimulates SOD1 peroxidase activity, apparently by accelerating SOD1-Cu(I) oxidation back to the resting state with concomitant formation of CO 3 . .
Recently, we became aware of studies performed by Medinas et al. (47) with human SOD1 in which the rate constant for the reaction between SOD1-Cu(I) and HOOCO 2 Ϫ was obtained. In this study, the authors measured a rate constant of 2 ϫ 10 3 M Ϫ1 s Ϫ1 for the reaction between reduced SOD1 and HOOCO 2 Ϫ . Although the reasons for this discrepancy of more than 1 order of magnitude are not clear to us at this point, this study suggests a considerably higher rate constant for the reaction between the human isoform of SOD1-Cu(I) and HOOCO 2 Ϫ .

Effect of Carbonic Anhydrase upon H 2 O 2 /HCO 3
Ϫ /CO 2 -mediated Oxidation of SOD1-Cu(I) to SOD1-Cu(II)-So far, our results indicate an important role for CO 2 , but not HCO 3 Ϫ , in the formation of HOOCO 2 Ϫ from the reaction of HOO ؊ with CO 2 to oxidize SOD1-Cu(I). To further test the potential effect of HCO 3 Ϫ , we evaluated the effect of carbonic anhydrase upon preformed SOD1-Cu(I) oxidation by H 2 O 2 /HCO 3 Ϫ /CO 2 . As shown in Fig. 3, spectrum F, the addition of carbonic anhydrase to the samples increased the SOD1-Cu(II) EPR signal intensity. Carbonic anhydrase accelerated the relatively slow equilibration of HCO 3 Ϫ /CO 2 by several orders of magnitude, replenishing CO 2 as it is consumed by HOO ؊ . The more efficient supply of CO 2 by carbonic anhydrase is likely to increase HOOCO 2 Ϫ production and accelerate SOD1-Cu(I) oxidation.

Effect of HCO 3
Ϫ /CO 2 on the Overall SOD1 Activity-To gather further information about HCO 3 Ϫ /CO 2 effects upon SOD1 peroxidase activity, the rate of H 2 O 2 decay was directly measured using an H 2 O 2 electrode. As shown in Fig. 4, HCO 3 Ϫ addition led to a concentration-dependent acceleration of H 2 O 2 uptake by SOD1. The maximum initial rates of SOD1mediated H 2 O 2 degradation measured at 0 and 100 mM HCO 3 Ϫ were 0.4 and 1.7 M x min Ϫ1 , respectively, per active site.

Direct Nuclear Magnetic Resonance Demonstration of HOOCO 2
Ϫ Interaction with Active Site Copper Cation-Taken together, the EPR experiments indicated that SOD1-Cu(II) is promptly reduced by excess H 2 O 2 to SOD1-Cu(I), especially at nearly neutral to alkaline pH, suggesting a peroxide anion (HOO ؊ ) requirement for copper ion reduction. Moreover, the oxidation of SOD1-Cu(I) to SOD1-Cu(II) was shown to be strongly dependent on CO 2 as a precursor of HOOCO 2 Ϫ . To confirm the interaction of HOOCO 2 Ϫ with SOD1-copper, we designed NMR experiments using 13 C-labeled bicarbonate (shown in Figs. 5 and 6). For these experiments, H 13 CO 3 Ϫ was dissolved in phosphate buffer and left undisturbed for at least 15 min for equilibration. After this time, H 2 O 2 was added and incubated for at least an additional 10 min to allow equilibration. The yields of HOOCO 2 Ϫ are dependent on pH and the concentrations of H 2 O 2 and HCO 3 Ϫ , so experimental conditions were varied so as to attain an appreciable HOO 13 CO 2 Ϫ NMR peak intensity. The conditions shown in Fig. 5 were the ones that gave good HOO 13 CO 2 Ϫ yield with significant retention of SOD1 activity over the time required to complete the analysis (Fig. 8) and were used for the NMR studies.
As shown in Fig. 5, dissolving H 13 CO 3 Ϫ into phosphate buffer led to the detection of two pronounced peaks attributable to H 13 CO 3 Ϫ (160.5 ppm) and 13 CO 2 (124.8 ppm) (27,32). The addition of H 2 O 2 to bicarbonate solutions led to a pronounced diminution of the 13 CO 2 peak and the appearance of a new peak at 158.9 ppm (as in previous observations) attributed to peroxymonocarbonate (HOO 13 CO 2 Ϫ ) (32). Interestingly, the addition of SOD1 to the H 13 CO 3 Ϫ / 13 CO 2 /H 2 O 2 led to a nearly complete disappearance of the HOO 13 CO 2 Ϫ peak over time (5-20 min), indicating peroxy monocarbonate and, indirectly, peroxide consumption by SOD1 (Fig. 5A).
Because the NMR experiments with SOD1 were performed in the presence of CDTA, and copper is released by the action of H 2 O 2 upon SOD1, control experiments using an aqueous copper-CDTA complex (Fig. 5B) were performed to confirm SOD1 activity as a fundamental requirement for HOO 13 CO 2 Ϫ con-  ; reaction with reduced cuprous ion. c This rate constant was estimated from reaction rate constants determined for CO 3 . reaction with reduced Co 2ϩ complexes based on cited references. d This rate was assumed to be 0.65 M Ϫ1 s Ϫ1 based on the cited reference.
sumption. As shown in Fig. 5B, Cu(II)-CDTA failed to diminish the HOO 13 CO 2 Ϫ peak, demonstrating that this chelated copper does not mediate HOO 13 CO 2 Ϫ decomposition. In contrast to SOD1, copper addition to the H 13 CO 3 Ϫ /H 2 O 2 led to a strong broadening of the H 13 CO 3 Ϫ peak, which over-lapped the HOO 13 CO 2 Ϫ NMR signal (data not shown). Nevertheless, the addition of CDTA promptly narrowed the H 13 CO 3 Ϫ peak and restored the HOO 13 CO 2 Ϫ NMR signal to its initial levels (data not shown). Peroxycarbonate formation depends on relatively slow equilibria; therefore, these data indicate that the copper CDTA complex is not able to consume HOOCO 2 Ϫ at appreciable rates, as opposed to SOD1. It has been shown that at H 2 O 2 /SOD1 concentration ratios of ϳ10:1, DTPA (25-50 M) significantly inhibit SOD-bound radical formation assayed through immunospin trapping (39). This inhibition indicates that copper release is involved in protein radical formation when SOD is exposed for prolonged times (Ͼ10 min) to high H 2 O 2 concentrations (Ͼ100 M). Therefore, the relevance of Cu(II) loss by SOD induced by H 2 O 2 in vivo requires further investigation, because the rapid detoxification of H 2 O 2 in the cellular milieu is likely to limit both the attainable concentrations and lifetime of the peroxide.
Together with the data reported here, these findings support SOD1-produced CO 3 . as a major oxidant mediating SOD1 peroxidase activity, as previously proposed (5,10,11). Nevertheless, in light of the relatively slow equilibria and rates measured previously (see Table 1) and in this study, it seems unlikely that the SOD1 peroxidase activity will be a relevant source of CO 3 . in

vivo.
To further explore the effects of CO 2 and HOO Ϫ upon HOOCO 2 Ϫ formation and decay, NMR experiments were repeated at different pH values. As seen in Fig. 6, raising the pH from 6.4 to 8.4 induced a marked acceleration in the rate of HOOCO 2 Ϫ decay. This result further characterizes the reduction of SOD1-Cu(II) by HOO Ϫ as the rate-limiting step in the peroxidase catalytic cycle of SOD1.
Using the available rate constants (for the reactions and rate constants, see Table 1), the NMR results were computationally simulated using Gepasi software at physiological pH values (Fig. 7). Interestingly, our simulations suggest that HOOCO 2 Ϫ is rapidly consumed by SOD1 and does not accumulate, because its rate of formation from H 2 O 2 and HCO 3 Ϫ is too slow at this pH to efficiently replenish consumed HOOCO 2 Ϫ . In addition, SOD1 inactivation is much slower in the presence of HOOCO 2 Ϫ   than in its absence, which is consistent with the protection exerted by HCO 3 Ϫ at high molar excess upon SOD1 inactivation by H 2 O 2 (10,26). Indeed, from the data shown in Fig. 8, it is clear that SOD1 inactivation is substantially accelerated after the 10 min time point, when HOOCO 2 Ϫ is considerably lower. Also relevant, the simulations demonstrate that SOD1 does not accumulate in its reduced form in the presence of HOOCO 2 Ϫ under the conditions of Fig. 6, a direct consequence of its rapid oxidation by HOOCO 2 Ϫ as opposed to a fairly slow reduction of SOD-Cu(II) by H 2 O 2 .

DISCUSSION
Although SOD1 peroxidase activity has been well characterized in recent years, important details about the enzyme's catalytic cycle remain obscure. This is in part because of the assumption that SOD1 peroxidase activity relies on the formation of hypervalent states in clear analogy to those identified in the heme-peroxidase catalytic mechanism. The hypothesis that the SOD1 peroxidase cycle closely paralleled that of the conventional peroxidases led to the proposal of hypothetical SOD1 hypervalent states, such as SOD1-Cu(III), SOD1-Cu(II)ϭO, or SOD1-Cu(II)/ ⅐ OH, which have never been characterized. Recent literature has reported several developments that have been important in unraveling the SOD1 peroxidase cycle: first, the demonstration that SOD1 peroxidase activity is greatly stimulated by HCO 3 Ϫ (5, 6) and dependent on CO 2 (8); second, the characterization by Richardson's group (27) of HOOCO 2 Ϫ as a product of the spontaneous reaction of HCO 3 Ϫ /CO 2 with H 2 O 2 ; and third, Valentine and co-workers' (29) molecular models of the SOD1 active site's interaction with HOOCO 2 Ϫ . These contributions led to the proposal in 2002 that HOOCO 2 Ϫ might be an important biological oxidant precursor of the carbonate radical anion CO 3 . (30,33). This idea has inspired continued investigation (30, 31, 39 -41). In the case of SOD1, it is hypothesized that the SOD1 peroxidase cycle is unique and depends on the reduction of SOD1-Cu(II) to SOD-Cu(I) followed by HOOCO 2 Ϫ -mediated oxidation of SOD1-Cu(I) to SOD1-Cu(II) with the formation of the carbonate radical in the second step of the catalytic cycle. This hypothesis is based on previous reports that metal ion-catalyzed heterolysis of HOOCO 2 Ϫ might lead to CO 3 . radical production (30,33). Through EPR investigations, we have unequivocally demonstrated that HOO ؊ mediates SOD1-Cu(II) reduction (Fig. 1)  the thermodynamic but also the kinetic and structural points of view.
Besides being an intriguing biochemical problem, the fact that SOD1 peroxidase activity was invoked as a possible factor leading to the development of amyotrophic lateral sclerosis, a severely debilitating disease (42,43), demands that the SOD1 peroxidase cycle be characterized in detail so as to shed light onto which biochemical processes are physiologically relevant and thus serve as a molecular basis for the development of new therapeutic approaches. Our results indicate that the reduction of SOD1-Cu(II) by HOO Ϫ followed by the oxidation of SOD1-Cu(I) by HOOCO 2 Ϫ is too slow of a process to account for significant carbonate radical production under physiological conditions, supporting the current view that SOD1 peroxidase activity is not a main contributor to the development of amyotrophic lateral sclerosis (24, 44 -46).