Copper,Zinc Superoxide Dismutase as a Univalent NO−Oxidoreductase and as a Dichlorofluorescin Peroxidase*

Nitroxyl (NO−) may be produced by nitric-oxide synthase and by the reduction of NO by reduced Cu,Zn-SOD. The ability of NO− to cause oxidations and of SOD to inhibit such oxidations was therefore explored. The decomposition of Angeli's salt (AS) produces NO− and that in turn caused the aerobic oxidation of NADPH, directly or indirectly. O⨪2 was produced concomitant with the aerobic oxidation of NADPH by AS, as evidenced by the SOD-inhibitable reduction of cytochrome c. Both Cu,Zn-SOD and Mn-SOD inhibited the aerobic oxidation of NADPH by AS, but the amounts required were ∼100-fold greater than those needed to inhibit the reduction of cytochrome c. This inhibition was not due to a nonspecific protein effect or to an effect of those large amounts of the SODs on the rate of decomposition of AS. NO− caused the reduction of the Cu(II) of Cu,Zn-SOD, and in the presence of O⨪2, SOD could catalyze the oxidation of NO− to NO. The reverse reaction, i.e. the reduction of NO to NO− by Cu(I),Zn-SOD, followed by the reaction of NO− with O2 would yield ONOO− and that could explain the oxidation of dichlorofluorescin (DCF) by Cu(I),Zn-SOD plus NO. Cu,Zn-SOD plus H2O2 caused the HCO 3 − -dependent oxidation of DCF, casting doubt on the validity of using DCF oxidation as a reliable measure of intracellular H2O2 production.

with the aerobic oxidation of NADPH by AS, as evidenced by the SOD-inhibitable reduction of cytochrome c. Both Cu,Zn-SOD and Mn-SOD inhibited the aerobic oxidation of NADPH by AS, but the amounts required were ϳ100-fold greater than those needed to inhibit the reduction of cytochrome c. This inhibition was not due to a nonspecific protein effect or to an effect of those large amounts of the SODs on the rate of decomposition of AS. NO  . , by catalyzing its dismutation (1). However, additional properties have been ascribed to these enzymes, including peroxidase (2)(3)(4), superoxide reductase (5), superoxide oxidase (5), and reversible NO/NO Ϫ oxidoreductase (6 -9) activities. The biological significance of these latter activities remain to be solidified, but interest in them is increased by the finding that point mutations in Cu,Zn-SOD, most of which do not effect the SOD activity, have been associated with ϳ20% of the cases of familial amyotrophic lateral sclerosis (FALS) (10,11). We will now present data supporting the ability of SODs to catalyze the oxidation of NO Ϫ , utilizing O 2 . as the oxidant; and of Cu,Zn-SOD to catalyze the HCO 3 Ϫ -dependent oxidation of DCF by H 2 O 2 . These findings could provide an alternative explanation for the observations of Estevez et al. (12). These findings will also shed ambiguity on the use of DCF oxidation as a measure of intracellular H 2 O 2 and ONOO Ϫ production, in agreement with Rota et al. (13) who criticized this use of DCF.
Estevez et al. (12) have reported that Zn-depleted Cu,Zn-SOD (Cu(II)-SOD) is reduced much more rapidly by ascorbate than is the Zn-replete enzyme and that other reductants, such as urate and GSH, act similarly but less effectively. Since DCF was oxidized by aerobic ascorbate ϩ Cu(II)-SOD ϩ NO, and is known to be oxidized by ONOO Ϫ , but not by NO (14,15) and that Cu(II)-SODs, whether mutant or wild type, were toxic to motor neurons (12). An alternative explanation for the results of Estevez et al. (12) would entail the reduction of NO to NO Ϫ by the Cu(I)-SOD, followed by production of ONOO Ϫ from the reaction of NO Ϫ with O 2 . The ONOO Ϫ thus produced would oxidize DCF. That reduced SODs can act as univalent reductants for NO was shown by Murphy and Sies (6) and by Kim et al. (9) and might account for the observations of McBride et al. (16) in which case H 2 O 2 was the reductant of Cu,Zn-SOD, in place of the ascorbate used by Estevez et al. (12).

MATERIALS AND METHODS
Angeli's Salt (Na 2 N 2 O 3 ; AS) from the Cayman Chemical Co. was generously provided by J. S. Stamler. Stock solutions of ϳ10 or 20 mM were prepared in 10 mM NaOH and were stored at Ϫ20°C until used. The extinction coefficient of AS at 250 nm in the NaOH was taken to be 8,000 M Ϫ1 cm Ϫ1 (17) and in sodium phosphate buffer, pH 7.4, to be 4,400 M Ϫ1 cm Ϫ1 . NADPH, diethylene triamine pentaacetic acid (DTPA), and bovine serum albumin were from Sigma; potassium ferricyanide was from J. T. Baker; cytochrome c was from Fluka; catalase was from Boehringer/Ingelheim; bovine Cu,Zn-SOD was from Grunenthal; and recombinant human Mn-SOD was from Biotechnology General. All reactions were performed at 23°C in 100 mM sodium phosphate, 50 M DTPA at pH 7.4. The decomposition of AS was followed at 250 nm (17,18), the oxidation of NADPH at 340 nm (19), the reduction of cytochrome c at 550 nM (20), and the oxidation of DCF at 500 nm (21).

Oxidation of NADPH by NO
Ϫ -An alkaline stock solution of AS, when diluted into neutral buffer to 0.175 mM, decomposed at an initial rate of 9 M/min as shown by line 1 in Fig. 1. When 100 M NADPH was present it was oxidized by the AS at a rate of 5 M/min (line 2). Thus the ratio of AS decomposed to NADPH oxidized was almost 2:1, in agreement with earlier reports (22,23). Since AS decomposition is known to be a source of NO Ϫ (7,18,23,24), it appears that NO Ϫ , or a species derived therefrom, under aerobic conditions, probably ONOO Ϫ , oxidizes NADPH. Both Cu,Zn-SOD and Mn-SOD were able to inhibit the oxidation of NADPH by NO Ϫ as shown in Fig. 2 and in agreement with Reif et al. (22). This was not due to an effect on the rate of decomposition of AS, which was not influenced by Cu,Zn-SOD (not shown). Moreover it was not due to a nonspecific protein effect, since comparable levels of serum albumin caused only a small and transient effect (Fig. 3) that was probably due to the consumption of NO Ϫ by oxidation of albumin cysteine and methionine residues. Thus GSH was also able to inhibit the oxidation of NADPH by NO Ϫ in a way that suggested consumption of the GSH (Fig. 4).
The inhibitory effect of the SODs did not appear to be due to the dismutation of O 2 . , since the levels of SOD needed to cause the inhibitions seen in Fig. 2 are ϳ100-fold greater than would be required to cause comparable inhibition of the reduction of cytochrome c by O 2 . . An interesting possibility is that NO Ϫ can reduce the redox-active metals (Cu(II) or Mn(III)) of the SODs and that O 2 . then reoxidizes these metal centers, thus providing for catalytic consumption of NO Ϫ . The reactions envisioned are as follows.
Exploration of this possibility required demonstration that NO Ϫ could reduce the redox-active metals at the active sites of the SODs and that O 2 . was produced during the oxidation of NADPH by NO Ϫ . NO Ϫ Reduces Cu,Zn-SOD-Addition of AS to 0.2 mM to a buffered solution of 4 mg/ml of Cu,Zn-SOD resulted in bleaching of the enzyme (Fig. 5). That this bleaching was due to reduction of the Cu(II) to Cu(I) was shown by the ability of ferricyanide to restore the lost absorbance. Catalase did not prevent the reduction of Cu,Zn-SOD by NO Ϫ (not shown). Thus H 2 O 2 was not involved. This attests to the reality of Reaction 1. The less than complete bleaching shown in Fig. 5 probably was  followed by the autoxidation of NADP ⅐ to NADP ϩ . There was no reduction of cytochrome c by aerobic AS in the absence of NADPH.
Oxidation of Dichlorofluorescin-The oxidation of DCF to its fluorescent product has been widely used as a measure of intracellular production of H 2 O 2 (11,13,25), and such use of DCF has been criticized (13). Nevertheless increased DCF oxidation as a consequence of the overproduction of normal Cu,Zn-SOD, or of the FALS-associated mutants, thereof, have been so interpreted (11,25). We now present data that should decrease enthusiasm for this use of DCF. Thus Fig. 7 demonstrates that DCF is not oxidized by H 2 O 2 per se but is oxidized by Cu,Zn-SOD plus H 2 O 2 and that HCO 3 Ϫ markedly stimulates this oxidation. HCO 3 Ϫ -dependent peroxidations catalyzed by Cu,Zn-SOD have been reported previously (3, 26 -28) and have been explained on the basis of the oxidation of HCO 3 Ϫ to CO 3 . , at the active site, followed by diffusion of the carbonate radical into the bulk solvent where it can cause diverse oxidations (3,28). Thus the data in Fig. 7 indicates that increased oxidation of DCF within cells could signal increased oxidation of HCO 3 Ϫ by Cu,Zn-SOD plus H 2 O 2 due to increased Cu,Zn-SOD, or due to increased H 2 O 2 , or to increased peroxidase activities of the FALS-associated, or of the Zn-depleted, enzymes. DCF can also be oxidized by ascorbate ϩ SOD ϩ a source of NO under aerobic conditions (12). Fig. 8 (line 1) demonstrates that 10 M DCF is oxidized when exposed to 20 M AS and that 30 g/ml Cu,Zn-SOD inhibited. Catalase at 260 units/ml had no significant effect, in the presence or absence of SOD. When 10 M DCF was exposed to 50 M AS in an argon-purged buffer, a slow and rapidly decreasing rate of DCF oxidation was seen. Subsequent aeration increased this rate dramatically (line 2). Dihydrorhodamine has similarly been seen to be oxidized by AS in an oxygen-dependent manner (18). It appears that NO Ϫ ϩ O 2 yields an oxidant capable of rapid oxidation of DCF or DHR. We assume that this oxidant is ONOO Ϫ or something derived from it. It has been shown that NO Ϫ produced by reduction of NO by Cu(I),Zn-SOD is in the triplet ground state (9), which is known to most readily react with triplet ground state O 2 yielding ONOO Ϫ (24,29). It should be noted that Miranda et al. (18) proposed that the product of the AS-generated NO Ϫ ϩ O 2 reaction is somehow different from ONOO Ϫ . That could be an isomeric form of ONOO Ϫ produced by the reaction of singlet NO Ϫ with O 2 . However, when NO Ϫ is produced biologically, as in the measurements of Estevez et al. (12), it would be triplet NO Ϫ , and this would yield ONOO -, per se, when reacting with O 2 . In any case ONOO Ϫ , or the similar product proposed by Miranda et al. (18), would rapidly oxidize DCF and DHR.

DISCUSSION
The sum of Reactions 1 plus 2 is as follows. REACTION 3 In essence this is an expression of the superoxide reductase activity (5) of Cu,Zn-SOD and is in accord with Murphy and Sies (6). It should be noted that Cu,Zn-SOD was not inactivated by incubation with 0.25 mM AS for 50 min (not shown). NO Ϫ / HNO can oxidize NADPH anaerobically as well as aerobically (22,23). The anaerobic reaction presumably involves the divalent oxidation of NADPH and the concomitant reduction of HNO to NH 2 OH, while the aerobic reaction could additionally be due to univalent oxidation of NADPH by ONOO Ϫ formed from the rapid reaction of NO Ϫ with O 2 (24,29). There are reasons for believing that nitric-oxide synthase can produce both NO Ϫ and O 2 . , particularly when tetrahydrobiopterin is limiting (30 -32), and the observation that Cu,Zn-SOD increases the yield of NO produced per NADPH consumed could be partially explained by the superoxide reductase activity of Cu,Zn-SOD as expressed in Reaction 3. Nitric-oxide synthase plus NADPH and L-arginine produces more O 2 . in the absence of tetrahydrobiopterin (32). That can partially be explained on the basis of NO Ϫ production. Thus the NO Ϫ plus O 2 would yield ONOO Ϫ , which would oxidize NADPH to NADP ⅐ , and that in turn would reduce O 2 to O 2 . .
Hence the stoichiometry of NADPH consumed per NO produced, by nitric-oxide synthase, could be influenced by SOD both on the basis of the oxidation of NO Ϫ to NO and by the prevention of the nonenzymic oxidation of NADPH by NO Ϫ / ONOO Ϫ , in agreement with Reif et al. (22). It has been suggested by Stoyanovsky et al. (33) that HNO can dimerize and then decompose to N 2 ϩ 2HO ⅐ . If this were the case under our conditions, then NADPH oxidation should have been appreciably due to HO ⅐ and should have been noticeably inhibited by an HO ⅐ scavenger such as ethanol. However ethanol, added to 2% of the reaction volume, did not at all inhibit the oxidation of NADPH by NO Ϫ (not shown). Having previously noted that the oxyethyl radical does not oxidize NADPH (34) we can now exclude HO ⅐ from having a role in the oxidation NADPH by NO Ϫ under our conditions. It should be noted that Stoyanovsky et al. (33), using spin trapping, observed maximal HO ⅐ production from HNO at pH 5, but very little at neutrality, while we worked at pH 7.4.
The reduction of the Cu(II) at the active site of Cu,Zn-SOD by NO Ϫ is viewed as a reversible reaction, in accord with Murphy and Sies (6). That being said, the observation of Estevez et al. (12) might now be explained in terms of the reduction of NO to NO Ϫ by the ascorbate-reduced Cu-SOD, rather than by the reduction of O 2 to O 2 . . The NO Ϫ , thus produced, would lead to formation of ONOO Ϫ that would account for the oxygen-dependent DCF oxidation that they observed. Another possible explanation is based on the observations of McBride et al. (16) The differences in reactivity between singlet and triplet NO Ϫ adds complexity to the data presented herein and to that in the literature (8,9,18,23,24). Thus GSH may be oxidized by 1 NO Ϫ and not by 3 NO Ϫ (9). Hence if the biologically generated NO Ϫ is in the triplet state then GSH might not be as effective a scavenger of NO Ϫ as proposed by Miranda et al. (18). The oxidations of NADH and of DCF could have been due to NO Ϫ or to ONOO Ϫ derived therefrom. NO Ϫ is a small anion, akin to O 2 . , and should readily gain access to the active site of SOD, but the possibility of SOD reacting with ONOO Ϫ cannot be excluded. If the effect of SOD was mainly due to reaction with NO Ϫ , we can now attempt to estimate the rate constant for that interaction. The NO Ϫ produced by the decomposition of AS is said to be 1 NO Ϫ (18), which can relax to 3  [SOD] is 2 ϫ 10 Ϫ6 M and [ 3 O 2 ] in aqueous solutions equilibrated with air is ϳ2 ϫ 10 Ϫ4 M. Two values have been reported for k 2 and these are 4 ϫ 10 7 M Ϫ1 s Ϫ1 (29) and 7 ϫ 10 7 M Ϫ1 s Ϫ1 (24). If we take the lower value we get k 1 ϭ 4 ϫ 10 9 M Ϫ1 s Ϫ1 , and this is again a minimal estimate. Thus whether SOD reacts with 1 NO Ϫ or 3 NO Ϫ , or both, its rate constant must be very high to account for its observed ability to inhibit the oxidations caused by NO Ϫ . Whether this has relevance in the biological milieu is not yet known.
There is uncertainty in the literature concerning the pK a , Cu,Zn-SOD to 30 g/ml at the first arrow and catalase to 260 units/ml at the second arrow. Line 2, conditions were the same as in line 1 except that AS was added to 50 M to the reaction mixture, which had been purged with argon for 20 min, then the reaction mixture was aerated at the arrow. redox potential, and even the spin state of NO Ϫ in aqueous solution. That being the case we must explain repeatable observations as best we can and hope that, in time, the physical chemistry and theoretical treatments of this fascinating molecule will yield certain results, which in turn, will lead to refinements of those currently reasonable explanations. It should be noted that we were not necessarily dealing with free NO/ NO Ϫ but rather with these species at the active site of SOD, and binding may substantially change their physicochemical character. Thus the interaction of NO with reduced SOD could give rise to a bound nitroxyl, which can be represented by the following equilibrium: Enz-Cu(I)NO y Enz-Cu(II)NO Ϫ (where Enz indicates enzyme) and which could react with O 2 yielding ONOO Ϫ that could diffuse into the bulk solvent and there oxidize DCF or DHR. A similar scenario applies to the results of Sharpe and Cooper (36) who reported that aerobic ferrocytochrome c plus NO oxidized DHR. This recalls the proposal made by Estevez et al. (12) with respect to the putative bound O 2 . made by the autoxidation of Cu(I)-SOD.
Note Added in Proof-The oxidation of NAD(P)H by ONOO Ϫ , with subsequent production of O 2 Ϫ by autoxidation of NAD(P) ⅐ , has been described by Kirsch and de Groot (37).