Direct probing of copper active site and free radical formed during bicarbonate-dependent peroxidase activity of bovine and human copper, zinc-superoxide dismutases. Low-temperature electron paramagnetic resonance and electron nuclear double resonance studies.

Using X-band electron paramagnetic resonance (EPR) and electron nuclear double resonance (ENDOR) spectroscopy at liquid helium temperatures, the Cu(II) coordination geometry at the active site of bovine and human copper,zinc-superoxide dismutases (bSOD1 and hSOD1) treated with H(2)O(2) and bicarbonate (HCO(3)(-)) was examined. The time course EPR of wild type human SOD1 (WT hSOD1), W32F hSOD1 mutant (tryptophan 32 substituted with phenylalanine), and bSOD1 treated with H(2)O(2) and HCO(3)(-) shows an initial reduction of active site Cu(II) to Cu(I) followed by its oxidation back to Cu(II) in the presence of H(2)O(2). However, HCO(3)(-) induced a Trp-32-derived radical from WT hSOD1 but not from bSOD1. The mutation of Trp-32 by phenylalanine totally eliminated the Trp-32 radical signal generated from W32F hSOD1 treated with HCO(3)(-) and H(2)O(2). Further characterization of the free radical was performed by UV irradiation of WT hSOD1 and bSOD1 that generated tryptophanyl and tyrosyl radicals. Both proton ((1)H) and nitrogen ((14)N) ENDOR studies of bSOD1 and hSOD1 in the presence of H(2)O(2) revealed a change in the geometry of His-46 (or His-44) and His-48 (or His-46) coordinated to Cu(II) at the active site of WT hSOD1 and bSOD1, respectively. However, in the presence of HCO(3)(-) and H(2)O(2), both (1)H and (14)N ENDOR spectra were almost identical to those derived from native bSOD1. We conclude that HCO(3)(-)-derived oxidant does not alter significantly the Cu(II) active site geometry and histidine coordination to Cu(II) in SOD1 as does H(2)O(2) alone; however, the oxidant derived from HCO(3)(-) (i.e. carbonate anion radical) reacts with surface-associated Trp-32 in hSOD1 to form the corresponding radical.

Pioneering research from Fridovich and co-workers (1)(2)(3)(4) has demonstrated a unique peroxidase activity from bovine SOD1 (bSOD1) 1 that was dependent on bicarbonate levels in phosphate buffers. The peroxidase activity was proposed to result as shown in Equations 1 and 2.
The oxidant, E-Cu(II)-⅐ OH, derived from bSOD1 is able to oxidize several small anionic ligands, viz. azide, formate, and others (3)(4)(5)(6)(7)(8), that gain access to the active site through a narrow channel as shown in Fig. 1. It has been proposed that the bicarbonate (HCO 3 Ϫ ), a ubiquitous anion present in high concentrations (ϳ25 mM) in biological systems, may enter the active site and be oxidized to form the carbonate anion radical, CO 3 . , which can diffuse out and oxidize other substrates in the bulk solution (3,(11)(12)(13)  Previously, we reported that HCO 3 Ϫ enhances the peroxidase activity of human Cu,Zn-SOD (hSOD1) in a concentration-dependent manner, causing protein aggregation through Trp-32derived oxidation products (14).
Recently, Elam et al. (15) proposed an alternative mechanism for HCO 3 Ϫ -mediated peroxidase activity and inactivation of SOD1 via an enzyme-associated peroxycarbonate (HCO 4 Ϫ ) intermediate, which apparently oxidized the histidine coordinated to Cu(II) at the active site causing its inactivation (15). It was proposed that the active species (HCO 4 Ϫ or CO 3 Ϫ ) formed during the peroxidase activity is enzyme-associated and nondiffusible (15). The changes in geometry and coordination that occur at the Cu(II) site during the HCO 3 Ϫ -dependent SOD1 peroxidase activity have not yet been investigated directly. To obtain a better understanding of the interaction between the histidine ligands and the oxidant generated at the active Cu(II) site during SOD1 peroxidase activity, we used the low temperature EPR and ENDOR techniques that can directly probe the paramagnetic metal ion at the active site, the coordinating ligands containing the NMR active nuclei, and the enzymederived free radical.
Although EPR is used to detect and characterize unambiguously paramagnetic metal ions at the active site in metalloenzymes, the EPR spectra are, in general, inhomogeneously broadened due to unresolved g-anisotropy and hyperfine cou-plings from the metal ion and the ligand nuclei (e.g. nitrogen) linked to the metal ion. In such cases, electron nuclear double resonance (ENDOR) can be used to probe the NMR-active nuclei that are coordinated to the paramagnetic center (16 -21). Low temperature ENDOR has been used previously to determine the copper-histidine coordination in Cu,Zn-SOD (18,21).
In this study, using the low temperature X-band EPR and ENDOR spectroscopy, we investigated the geometry and coordination of the Cu(II) active site as well as the structure of the free radical generated during SOD1 peroxidase activity. Results from this study show that H 2 O 2 drastically alters the copper-histidine coordination at the active site and is inhibited partially by HCO 3 Ϫ . During HCO 3 Ϫ -dependent peroxidase activity, the Trp-32 located on the surface of hSOD1 was oxidized to the corresponding Trp-32 radical. The copper-bound histidines His-44 and His-46 in bSOD1 and His-46 and His-48 in WT hSOD1 are affected predominantly due to oxidation by H 2 O 2 only in the absence of HCO 3 Ϫ . These findings are relevant to understanding oxidant-mediated site-specific damage, oxidative inactivation of the enzyme, and the covalent aggregations of WT hSOD1 and amyotrophic lateral sclerosis SOD1 mutants.

EXPERIMENTAL PROCEDURES
Materials-Bovine Cu,Zn-superoxide dismutase was obtained from Roche Diagnostics. Tyrosine, tryptophan, hydrogen peroxide, sodium bicarbonate, sodium hydrogen phosphate, and DTPA were purchased from Sigma.
Expression and Purification of Human SOD1-The wild type human and W32F mutant Cu,Zn-SOD (Cu(II) 97%; Zn(II) 108%) were expressed and purified as described previously (14). The SOD1 mutations were created by a two-cycle polymerase chain reaction-based mutagenesis protocol described by Zhao et al. (22) using mutagenic primers to introduce the desired mutants and cloned subsequently into a pET-3d expression system (Novagen, Madison, WI).
EPR Measurements-A typical reaction mixture (200 l) for EPR measurements consisted of SOD1 (30 M), H 2 O 2 (1 mM), and HCO 3 Ϫ (25 mM) in a phosphate buffer (100 mM) containing DTPA (100 M). The X-band EPR of Cu(II) and free radical were recorded at liquid helium and liquid nitrogen temperatures on a Bruker E500 ELEXYS spectrometer with a 100-kHz field modulation equipped with an Oxford Instruments ESR-9 helium flow cryostat and a DM_0101 cavity. Incubation mixtures (200 l) containing SOD1 and H 2 O 2 were transferred to a 4-mm quartz EPR tube (Wilmad) and frozen immediately in liquid nitrogen for EPR analysis. EPR spectral parameters were extracted from the spectra using the QPOWA program (23) based on the successful fitting to the following orthorhombic spin Hamiltonian as shown in Equation 4 (16), where ␤ is the Bohr magneton; g x , g y , and g z and A x , A y , and A z are the components with respect to the principal axes of the g-tensor and the hyperfine tensor. Spectrometer conditions were as follows: microwave frequency, 9.63 GHz; modulation frequency, 100 kHz; modulation amplitude, 5 G; receiver gain, 85 dB; time constant, 0.01 s; conversion time, 0.08 s; sweep time, 83.9 s ;and microwave power, 100 W.
For UV irradiation studies, bovine or human SOD1 (30 M) in a phosphate buffer (100 mM, pH 7.4) containing DTPA (0.1 mM) was irradiated at indicated time intervals with UV light using an EiMac VIX 300 UV 300 X Xenon arc source and water as filter.
ENDOR Measurements-X-band ENDOR spectra were recorded on a Bruker E500 ELEXYS spectrometer using an ENDOR/triple accessory (Bruker) cavity. ENDOR spectra were recorded by fixing the magnetic field at an EPR resonance and by applying saturating microwave power partially while sweeping the NMR transition with a radio frequency source. 1 H and 14 N ENDOR spectra were recorded at 8 K at the field positions shown in Fig. 4A, inset (vide infra). The EPR spectrum was saturated partially with 6.33 mW (15 dB) of microwave power, and the ENDOR spectra were recorded at 9.47 GHz with a 200-kHz modulation depth, 100-W radio frequency power, and 1500 scans at field position I and 5000 scans at field positions II and III, respectively. Simulations of 14 N interactions in ENDOR spectra were done by DDPOWHE program (24) using the following spin Hamiltonian as shown in Equation 5 (16), where B is the external magnetic field and A i and Q i correspond to the hyperfine and quadrupole tensors, respectively, of nucleus i. A i is described by its three principal values (A i xx , A i yy , and A i zz ) and the Euler angles (␣ i , ␤ i , ␥ i ) that transform its principal axis system (xЉ, yЉ, zЉ) to that of g(xЈ,yЈ,zЈ). To simulate the resulting spectra, each nitrogen spectrum was multiplied by a factor chosen such that their sum gave the best fit to the experimental data (25). For protons with a spin of I ϭ 1 ⁄2, the ENDOR transition frequencies are given to first order by Equation 6 (16), where A H is the orientation-dependent nuclear hyperfine coupling constant and H is the nuclear Larmor frequency. If H Ͼ A H /2 as is usually the case for protons in biological systems, the ENDOR spectrum consists of a hyperfine-split doublet centered around the Larmor frequency. For H Ͻ A H /2 as is the case for nuclei with small nuclear Larmor frequencies, the ENDOR spectrum consists of a Larmor-split doublet centered at A H /2. For a quadrupolar nucleus such as 14 N (I ϭ 1) where the ligand nuclei have a spin greater than 1 ⁄2, the ENDOR pattern is further split into 2I lines given by the first order resonance frequencies as shown in Equation 7 (16), where m is the spin state of the lower two energy levels, which are involved in the transition given by ϪI ϩ 1 Յ m Յ I. This hyperfine splitting is governed by P N , the orientation-dependent quadrupole coupling constant (26).

The Effect of H 2 O 2 and Bicarbonate on the EPR Spectra of the Active Site Cu(II) in Human and Bovine SOD1-
The EPR spectra of bovine SOD1, WT hSOD1, and mutant W32F hSOD1 along with the computer simulations (dotted lines) are shown in Fig. 2. Based on the g-tensor (g x ϭ 2.030, g y ϭ 2.090, and g z ϭ 2.268) and the Cu(II) hyperfine splitting (A z ϭ Ϫ395.0 MHz) parameters (27,28), it is evident that the Cu(II) active site geometry is similar in all cases. The precise determination of hyperfine tensor parameters, A x and A y , was hindered due to spectral broadening in this region as a result of close g values EPR/ENDOR of HCO 3 Ϫ -dependent Peroxidase Activity of Cu,Zn-SOD and lower hyperfine coupling. The Euler angles of ␣ ϭ 85°and ␤ ϭ 20°that describe the relative orientations of g-and Atensors were obtained from simulating EPR spectra of W32F hSOD1, WT hSOD1, and bSOD1. Generally, the non-coincidence of g-and A-tensors and low anisotropy of A z -tensor element of all SOD1 indicates the presence of low symmetry at the Cu(II) active site (29,30). Large g-and A-strains indicating the distribution of g-and A-tensors due to the conformational heterogeneity of SOD1 were required to simulate the observed spectra.
The time-dependent EPR spectra obtained from incubations containing the WT hSOD1, W32F hSOD1, and bSOD1 and H 2 O 2 in the absence (left) and presence of HCO 3 Ϫ (right) along with computer simulations (dotted lines) are shown in Fig. 2. Although it is probable that a mixture of several species is formed during the reaction with H 2 O 2 , we have simulated the EPR spectra based on the spin Hamiltonian parameters of two predominating species with one corresponding to the native SOD1 and the other corresponding to the Cu(II) species, characterized by less rhombic parameters (g x ϭ 2.047, g y ϭ 2.055, g z ϭ 2.268, A z ϭ Ϫ420 MHz). In the presence of HCO 3 Ϫ , the native enzyme parameters were found to be dominant, whereas in the absence of HCO 3 Ϫ , the less rhombic parameters were predominant for simulations. In the presence of H 2 O 2 , the Cu(II) signal intensity of WT hSOD1 slowly decreased up to 15 min because of the reduction of Cu(II) to Cu(I) and then increased in intensity because of reoxidation of Cu(I) to Cu(II) (Fig. 2). The Cu(II) hyperfine lines are also broadened possibly due to oxidation of His-46 (or His-44) and His-48 (or His-46) by the enzyme-derived oxidant, E-Cu(II)-⅐ OH. However in the presence of HCO 3 Ϫ , the EPR signal intensity of Cu(II) decreased significantly at 1 min and later increased until 30 min. The Cu(II) hyperfine signals reappeared much like the native enzyme, indicating that the Cu(II) active site geometry remained In the presence of H 2 O 2 and HCO 3 Ϫ , a new EPR signal appeared in the free radical region (Fig. 2, indicated by an arrow), which grew in intensity until 30 min and then decreased. The P1 ⁄2 values were calculated to be ϳ0.072 mW at 30 K and 0.32 mW at 77 K, suggesting that the tryptophan-derived radical saturates at a relatively low microwave power. As the EPR spectra obtained in the presence of 95% N 2 and 5% CO 2 were similar (not shown), we conclude that the free radical formed probably is carbon-centered and not oxygen-centered (31).
As with WT hSOD1, the Cu(II) signal intensity of W32F hSOD1 decreased initially but grew with time because of rapid oxidation of Cu(I) by H 2 O 2 . In the presence of HCO 3 Ϫ , the EPR signal intensity decreased but not to the same extent as did WT-SOD1. However, no free radical signal was detected from W32F SOD1 (Fig. 2, right panel). This confirms that the free radical detected from incubations containing WT hSOD1, H 2 O 2 , and HCO 3 Ϫ was derived, most probably, from Trp-32. Effect of UV on WT and W32F Human SOD1 and Bovine SOD1-Irradiation of WT hSOD1 and bSOD1 with UV light induced tryptophanyl and tyrosyl radicals. Typical EPR spectra obtained from UV-irradiated hSOD1 and bSOD1 and measured at 8 K are shown in Fig. 3. The UV irradiation of W32F hSOD1 under similar conditions did not induce the free radical signal (Fig. 3, bottom panel). This finding suggests that the free radical probably is derived from Trp. Upon expanding the free radical region, hyperfine features were resolved (Fig. 3, insert). To identify the nature and structure of the free radical, we used the data published previously (32)(33)(34)(35). Computer simulations (shown as dotted lines) were performed using the EPR parameters reported for Trp and Tyr radicals obtained through EN-DOR and high frequency EPR studies (32)(33)(34)(35) and ␤-methylene hydrogen hyperfine coupling constants (vide infra). Thus, the experimental spectra could be fitted successfully using the literature values (32)(33)(34)(35). Based on the EPR parameters shown in Tables I and II, we conclude that the doublet signal is the result of a Trp-32-derived radical in WT hSOD1. Unlike hSOD1, which has Trp-32, bSOD1 has a tyrosine residue. The corresponding signal (Fig. 3) is attributed to the tyrosyl radical.
Probing of Copper-Histidine Coordination by ENDOR during SOD1 Peroxidase Activity-The histidyl protons at the active Cu(II) site in SOD1 show ϩ and Ϫ absorptions centered around H that are split by their hyperfine coupling constants (18). Fig. 4A shows the time course of proton ( 1 H) ENDOR spectra from histidines coordinated to Cu(II) at the active site of bSOD1 in the presence of H 2 O 2 and/or HCO 3 Ϫ obtained at the field positions I, II, and III indicated in Fig. 4A (inset). The proton signal (marked a in Fig. 4A) from His-44 is absent in the presence of H 2 O 2 alone (18). However, the proton ENDOR spectra of bovine SOD1 treated with H 2 O 2 in the presence of HCO 3 Ϫ remained virtually unchanged (Fig. 4A). The time course of the 14 N ENDOR spectra from histidines liganded to Cu(II) of bovine SOD1 with H 2 O 2 and/or HCO 3 Ϫ is shown in Fig. 4B. The spectra for the field position I were simulated (Fig. 4B, dotted line) (1) and N x (2) were used for simulation as shown in Fig. 4B at field position III. In the presence of HCO 3 Ϫ and H 2 O 2 , there was an excellent fit between the experimental ENDOR and the computer simulations using the values obtained by Reinhard et al. (18). From these findings, we conclude that the copper-bound His is not affected greatly in the presence of HCO 3 Ϫ and H 2 O 2 . However, in the presence of H 2 O 2 alone, the triplet signal in the low frequency region was absent (Fig. 4B at field position I). The triplet signal has been assigned previously to the His-46(44) and His-48(46) nitrogens (18). Thus, the two coordinated nitrogens of His are altered with H 2 O 2 treatment, and the geometry is relaxed slightly, resulting in a lower Cu(II) hyperfine coupling (an A z value of 420 MHz) compared with the pure tetragonal axial symmetric systems, viz. copper tetraphenylporphyrin and copper phthalocyanin with A z values of 631 and 637 MHz, respectively (36). However, in the presence of HCO 3 Ϫ , there was only a minimal change in the spectra, indicating that the active site geometry was not affected greatly under these conditions. The lower Cu(II) hyperfine A z value coupled with an unusually large value for the 14 N quadrupole tensors (Fig. 4) used for simulation (as compared with an axial Cu(II) site (36, 37)) indicates a low symmetry for the Cu(II) active site in the presence of HCO 3 Ϫ .
FIG. 3. The effect of UV light on free radical formation from bovine SOD1, WT human SOD1, and W32F human SOD1. Top, EPR spectra of UV-irradiated frozen solutions containing bovine SOD1 (30 M) in phosphate buffer (100 mM, pH 7.4) at 8 K at zero time and 2 min after UV irradiation. Middle, same as above with the exception that the EPR spectra were obtained from UV irradiation of WT human SOD1 (30 M) at 0 and 2 min. Inset shows the expanded portion of the free radical region (g ϭ 2.0054). Bottom, same as above with the exception that the EPR spectra were obtained from UV irradiation of human SOD1 W32F. All of the spectra were recorded at 8 K. Spectrometer conditions are the same as in Fig. 1. Experimental (continuous line) and simulated (dotted lines) EPR spectra of UV light-generated tryptophanyl and tyrosyl radicals using the magnetic parameters in Tables I and  II,

EPR/ENDOR of HCO 3
Ϫ -dependent Peroxidase Activity of Cu,Zn-SOD DISCUSSION Using EPR and ENDOR spectroscopy, we investigated the copper coordination geometry of human and bovine SOD1 treated with H 2 O 2 and HCO 3 Ϫ . The results indicate that, in the presence of H 2 O 2 , the copper coordination geometry became less rhombic presumably because of oxidative modifications of two coordinated histidines. However, in the presence of H 2 O 2 and HCO 3 Ϫ , the ENDOR pattern was identical to that of native SOD1, suggesting that the active site copper geometry was affected minimally. In the presence of HCO 3 Ϫ and H 2 O 2 , a free radical signal attributable to Trp-32 was induced from WT hSOD1.
Tryptophan-derived Free Radical from Human SOD1 Peroxidase Activity-Previous studies on peroxidase activity of SOD1 and familial amyotrophic lateral sclerosis SOD1 mutants revealed that HCO 3 Ϫ is absolutely essential for SOD1/H 2 O 2 -induced hydroxylation of 5,5Ј-dimethyl-1-pyrroline N-oxide and oxidation of several substrates, viz. urates, ABTS (2,2Ј-azino-bis-[3-ethylbenzothiazoline]-6-sulfonic acid), ␥-tocopherol, dichlorodihy-drofluorescein, and others, which are not accessible to the Cu(II) site through the narrow active site channel, as shown in Fig. 1 (4 -8, 38). Early on, it was proposed by Hodgson and Fridovich (1, 2) that the reaction between H 2 O 2 and SOD1 could generate a copper-bound hydroxyl radical that reacts with HCO 3 Ϫ to form a diffusible oxidant, CO 3 . . This radical was proposed to diffuse out of the active site and cause oxidation of several substrates outside of the active site (5)(6)(7)(8)39). However in the absence of HCO 3 Ϫ , there was no significant peroxidase activity. Recently, an alternate mechanism for HCO 3 Ϫ -mediated peroxidase activity and inactivation of SOD1 via an enzyme-bound non-diffusible reactive intermediate, HCO 4 Ϫ , was proposed to explain the HCO 3 Ϫ -enhanced enzyme inactivation (15). The proposed reactive intermediate species formed at the Cu(II) active site was presumed to oxidize the coordinated histidines in the presence of HCO 3 Ϫ . The present EPR data of WT hSOD1 and bSOD1 in the presence of H 2 O 2 showed that the site-specific oxidation of active site histidines was dependent on the ratio between the bicarbonate and phosphate concentrations (data not shown). The appearance of a tryptophanderived radical formed from WT hSOD1 and not from W32F hSOD1 is consistent with the reaction between CO 3 . radical and tryptophan 32 on the surface. The CO 3 . radical reacts rapidly with Trp or Tyr residues (k Trp/CO Ϫ ⅐ 3 ϭ 7.0 ϫ 10 8 M Ϫ1 s Ϫ1 ; k Tyr/CO Ϫ ⅐ 3 ϭ 4.5 ϫ 10 7 M Ϫ1 s Ϫ1 ) than with histidine (k His/CO Ϫ ⅐ 3 ϭ 5.6 ϫ 10 6 M Ϫ1 s Ϫ1 ). This may explain in part the partial protection of the Cu(II) active site of SOD1 in the presence of H 2 O 2 and HCO 3 Ϫ as shown in Fig. 2 (right panel). Further characterization of the UV-induced radical at 8 K with resolved hyperfine features and their EPR computer simulation parameters (Tables I and II) clearly 2 supports the formation of the Trp and Tyr radicals. The hyperfine splittings in Trp and Tyr radicals arise mainly from the ␤-methylene hydrogens as shown in Equations 8 and 9 (40,41), where A ␤1 H and A ␤ 2 H denote the hyperfine splitting constants of the two methylene protons, c is the spin density at the neighboring carbon, B P and B \ are empirical constants, B P ϭ 0, B \ ϭ 5.8 millitesla (40,41), and is the dihedral angle between the H ␤ -C ␤ bond and the p-orbital axis of C 1 , perpendicular to the plane of the aromatic ring. The dihedral angles () in the Trp radical in human SOD1 (9) (from Protein Data Bank accession code 1SPD) and in the Tyr radical in bovine  SOD1 (10) (from Protein Data Bank accession code 2SOD) were obtained from the x-ray crystal structure data. Using these values, the A ␤1 H and A ␤2 H were calculated and used for the simulation. The best fit anisotropic g and the hyperfine couplings of Trp and Tyr radicals in human SOD1 and bovine SOD1 are given in Tables I and II. Direct Probing of the Cu(II) Active Site by ENDOR-As shown in Fig. 2, the time-dependent EPR spectra of the SOD1/ H 2 O 2 /HCO 3 Ϫ are broadened inhomogeneously, making it impossible to obtain hyperfine parameters from the coordinating histidine ligands. In contrast, the ENDOR spectroscopic technique enabled us to probe the Cu(II) active site and obtain additional hyperfine information from the coordinating histidines. Elam et al. (15) propose that the enzyme-associated non-diffusible oxidant, viz. HCO 4 Ϫ /CO 3 . , formed at the active site reacts rapidly with the copper-bound histidines, resulting in HCO 3 Ϫ -enhanced enzyme inactivation during peroxidase activity (15). However, the present ENDOR data show that the coupled proton signals and the triplet signal due to Cu(II)bound His in the 1 H and 14 N ENDOR spectra are present in the presence of HCO 3 Ϫ and H 2 O 2 (Fig. 4, A and B). In the absence of HCO 3 Ϫ , the ENDOR spectra obtained from solutions containing Cu,Zn-SOD and H 2 O 2 show drastic changes with regard to histidine at the active site geometry (Fig. 4, A and B). The ENDOR/EPR results reveal the following. (i) HCO 3 Ϫ preserves the copper coordination to histidine that is damaged by H 2 O 2 .
(ii) The oxidant formed at the active site in the presence of HCO 3 Ϫ /H 2 O 2 diffuses out of the active site and oxidizes Trp-32 to the Trp radical.
In conclusion, using the low temperature EPR and ENDOR, we showed that H 2 O 2 drastically alters the copper-histidine coordination at the active site that was partially inhibited by HCO 3 Ϫ . The observation of a Trp radical formed at the surface by EPR from the site-specific oxidation of Trp suggests that the active oxidant formed from HCO 3 Ϫ is diffusible, contrary to what had been proposed earlier (15), and clearly rules out the generation of "free hydroxyl radical" or an "enzyme-bound HCO 4 Ϫ " reactive intermediate in the active site. The proposed mechanism based on a site-specific inhibition of the damage of copper-bound histidine, CO 3 . radical generation, and its scavenging by antioxidants may lead to a better understanding of the oxidative hypothesis proposed in amyotrophic lateral sclerosis.