Formation of peroxynitrite from reaction of nitroxyl anion with molecular oxygen.

Peroxynitrite (ONOO(-)/ONOOH) is generally expected to be formed in vivo from the diffusion-controlled reaction between superoxide (O(2)) and nitric oxide ((*)NO). In the present paper we show that under aerobic conditions the nitroxyl anion (NO(-)), released from Angeli's salt (disodium diazen-1-ium-1,2,2-triolate, (-)ON=NO(2)(-)), generated peroxynitrite with a yield of about 65%. Simultaneously, hydroxyl radicals are formed from the nitroxyl anion with a yield of about 3% via a minor, peroxynitrite-independent pathway. Further experiments clearly underline that the chemistry of NO(-) in the presence of oxygen is mainly characterized by peroxynitrite and not by HO( small middle dot) radicals. Quantum-chemical calculations predict that peroxynitrite formation should proceed via intermediary formation of (*)NO and O(2), probably by an electron-transfer mechanism. This prediction is supported by the fact that H(2)O(2) is formed during the decay of NO(-) in the presence of superoxide dismutase (Cu(II),Zn-SOD). Since the nitroxyl anion may be released endogenously by a variety of biomolecules, substantial amounts of peroxynitrite might be formed in vivo via NO(-) in addition to the "classical" ( small middle dot)NO + O(2)() pathway.

the present paper we show that under aerobic conditions the nitroxyl anion (NO ؊ ), released from Angeli's salt (disodium diazen-1-ium-1,2,2-triolate, ؊ ON‫؍‬NO 2 ؊ ), generated peroxynitrite with a yield of about 65%. Simultaneously, hydroxyl radicals are formed from the nitroxyl anion with a yield of about 3% via a minor, peroxynitrite-independent pathway. Further experiments clearly underline that the chemistry of NO ؊ in the presence of oxygen is mainly characterized by peroxynitrite and not by HO ⅐ radicals. Quantum-chemical calculations predict that peroxynitrite formation should proceed via intermediary formation of ⅐ NO and O 2 . , probably by an electron-transfer mechanism. This prediction is supported by the fact that H 2 O 2 is formed during the decay of NO ؊ in the presence of superoxide dismutase (Cu(II),Zn-SOD). Since the nitroxyl anion may be released endogenously by a variety of biomolecules, substantial amounts of peroxynitrite might be formed in vivo via NO ؊ in addition to the "classical" ⅐ NO ؉ O 2 .
A further source of endogenous peroxynitrite may be the nitroxyl anion (NO Ϫ ). This anion has been reported to be generated in vivo from reduction of ⅐ NO by Cu(I),Zn-SOD (24), hemoglobin (25), and cytochrome c 2ϩ (26), respectively. There are also reports that the NOS-catalyzed oxidation of L-arginine leads to initial formation of NO Ϫ and not ⅐ NO radical (27), however, this finding is subject to controversial discussion (28). Nitroxyl anion might be further formed from reaction of Snitrosothiols with thiols (29 -31), although this reaction is less well understood. Donald et al. (32) have proven that the photochemical decomposition of Angeli's salt, a chemical NO Ϫ donor compound (33), in fact yields peroxynitrite under aerobic conditions (Reaction 1). This photochemical process is probably the reason why NO Ϫ has often been referred to as a peroxynitrite-yielding compound, even in textbooks of inorganic chemistry (34).
Only very low yields of nitrated products have been observed from NO Ϫ -induced reactions (35,36). From these facts it was concluded that during thermal decomposition of Angeli's salt only a small amount of peroxynitrite is generated (36). Recently, two research groups (37,38) apparently disproved the capability of NO Ϫ to generate peroxynitrite under more physiological conditions. They reported, for example, that typical peroxynitrite-mediated reactions, e.g. nitrosation reactions, could not be observed (37) and that NADPH could be oxidized by NO Ϫ under hypoxic conditions (38). Unfortunately, these experiments were performed in the presence of organic buffer compounds (Good's buffer) which are known to effectively react with peroxynitrite (39,40). Thus, the formation of peroxynitrite by NO Ϫ very easily may have been masked.
In the present study, we demonstrate that in the presence of molecular oxygen NO Ϫ indeed mainly yields peroxynitrite and that NO Ϫ additionally produces HO ⅐ radicals via a minor, peroxynitrite-independent pathway. Furthermore, we present a key experiment which suggests the intermediacy of O 2 . during the NO Ϫ -mediated formation of peroxynitrite.
Solutions-Care was taken to exclude possible contamination by both bicarbonate/carbon dioxide and transition metals. Doubly distilled water was bubbled (2 liters/min) with synthetic air at room temperature for 20 min. This water was used for synthesis of oxoperoxonitrate(1-), NaOH (0.01-0.5 N) and for all other solutions. Phosphate buffer solutions (50 mM) were treated with the heavy metal scavenger resin Chelex 100 (0.3/0.5 g in 10 ml) by gently shaking for 18 h in the dark. After low-speed centrifugation for 5 min, the solutions were carefully decanted from the resin. The resin treatment resulted in an increase in pH by about 0.25 units. Various additives (DHR, NADH, and benzoic acid) were then added. The pH was adjusted to 7.5 at 37°C and the solutions were again bubbled (2 liters/min) with synthetic air or with the CO 2 mixture for 20 min. In the case of CO 2 bubbling, the pH had to be readjusted to 7.5. SIN-1 and Angeli's salt solutions were prepared as ϫ100 stock solutions at 4°C in 50 mM KH 2 PO 4 and in 10 mM NaOH, respectively, and used within 15 min.
Experimental Conditions-SIN-1 was added to 1 ml of phosphate buffer and incubated in 12-well cell culture plates (volume of each well 7 ml, Falcon, Heidelberg, Germany). Under HCO 3 Ϫ /CO 2 -free conditions, these plates were placed in an air-tight vessel (10 liters). During the first 15 min of each experiment, these vessels were flushed (5 liters/ min) with synthetic air in a warming incubator (Heraeus, Hanau, Germany). In the presence of HCO 3 Ϫ /CO 2 the plates were placed in an incubator for cell culture (37°C, humidified atmosphere of 95% authentic air and 5% CO 2 , Labotect, Göttingen, Germany). The experiments with authentic peroxynitrite (2 l of 25-125 mM ONOO Ϫ in 0.5 N NaOH was added to 1 ml of reaction solution) and with Angeli's salt (1 ml of reaction solution) were performed in reaction tubes (2.0 ml, Eppendorf, Hamburg, Germany) by using the drop-tube Vortex mixer technique as described previously (40). Under HCO 3 Ϫ /CO 2 -free conditions, the experiments with authentic peroxynitrite and Angeli's salt were performed in a glove-bag (Roth, Karlsruhe, Germany) under synthetic air.
Determination of Peroxynitrite/SIN-1/Angeli's Salt-driven Hydroxylation of Benzoic Acid-Peroxynitrite, SIN-1, and Angeli's salt (each 50 -600 M)-dependent hydroxylation of BA (5 mM) were employed. After vortexing, the samples were kept for 2 min (in the case of peroxynitrite), 4 h (in the case of SIN-1), and 30 min (in the case of Angeli's salt) at 37°C, respectively. The product formed was measured by reading its fluorescence with excitation at 290 nm and emission at 410 nm (37).
Determination of NADH-NADH was quantified by reading its fluorescence with excitation at 339 nm and emission at 460 nm (45). Standard calibration curves were prepared from known amounts of NAD(P)H. Additionally, the oxidation of NAD(P)H was followed photometrically at 340 nm using ⌬⑀ 340 ϭ 6200 M Ϫ1 cm Ϫ1 (45). Both methods gave identical results, therefore, only one parameter, the decrease of fluorescence, will be shown here.
Determination of H 2 O 2 and of O 2 -Hydrogen peroxide was quantified by two techniques. In peroxidase assays, horseradish peroxidase-catalyzed formation of a colored product was measured. 4-Aminoantipyrine and 3,5-dichloro-2-hydroxybenzenesulfonic acid were used as peroxidase substrates. The quinoneimine dye formed from these substrates was measured spectrophotometrically at 546 nm (46) (peroxidase assay). Alternatively, H 2 O 2 was quantified by the amount of O 2 released upon addition of catalase (1,000 units/ml) (catalase assay). O 2 was determined polarographically with a Clark-type oxygen electrode (Saur, Reutlingen, Germany). Both methods gave identical results, therefore, only one parameter, the peroxidase assay, will be shown here.
Determination of Nitrate-The nitrate yields from decomposition of Angeli's salt (100 and 200 M) were quantified by the use of nitrate reductase in conjunction with the Griess assay. The Griess assay was carried out as described elsewhere (46).
Quantum-chemical Calculations-Density functional theory and ab initio calculations were performed with the Gaussian 98W (Revision A.9) suite of programs (47). Geometries were fully optimized to stationary points, using the CBS-QB3 methodology in the density functional theory calculations and single-excitation CI calculations and secondorder Møller-Plesset (48) (MP2) calculations with the 6 -311ϩG(d) basis set on the ab initio level. Calculation of UV-VIS absorption spectra was performed by single point energy calculations on the CBS-QB3-optimized structure using the protocol of the time-dependent density functional theory method (49). Aqueous solvent interactions were evaluated with the PCM (50) procedure incorporated in Gaussian 98W. To verify whether an electron transfer between NO Ϫ and O 2 would be thermodynamically feasible, geometry optimizations and frequency calculations were done using the MP2 approximation. Molecular interactions were then evaluated with the IPCM (51) procedure. Eight samples of Angeli's salt (5 mM) prepared from the same stock solution were incubated in the dark in K 3 PO 4 (50 mM, pH 12.25) for 0 -24 h at 37°C in the presence of 100% oxygen. At selected time points an UV-visible spectrum was read from one sample by using a diode array spectrophotometer (scanning time 1 s). Afterward, the sample was discarded. A, kinetic of the absorption at 302 nm. Each value represents the mean Ϯ S.D. of six experiments. B, generation of a representative difference spectrum, obtained by subtraction of the UV-visual spectrum which was observed after 0 h of incubation from this spectrum observed after 4 h of incubation. C, comparison between the UV-visual spectrum of authentic peroxynitrite and the difference UV-visual spectrum from B.

Formation of Peroxynitrite from Nitroxyl Anion-To
Angeli's salt from the same stock solution were incubated in parallel runs in the dark at 37°C at pH 12.25 (Fig. 1A). The decay of Angeli's salt is slow at these experimental conditions (t 1/2 ϳ12 h), thus, reaction times of several hours were necessary to monitor the significant changes of the optical density. The initial absorbance at the beginning of the experiment was 0.35 Ϯ 0.05 at 302 nm. With increasing reaction time, the optical density at 302 nm increased continuously to reach a maximum value of 0.88 Ϯ 0.08 after 6 h, followed by a further decrease of the absorption at longer reaction periods. After 24 h the extinction value had dropped to 0.4 Ϯ 0.05, clearly showing that a relatively long-lived intermediate has been formed during the decay of Angeli's salt. The absorption at 302 nm had decayed completely when the reaction solution was briefly (20 -30 s) bubbled with CO 2 after 5 h of incubation (data not shown). These observations strongly indicated the intermediary formation of ONOO Ϫ . In fact, when the initial UV spectrum was subtracted from the one observed after 4 h of incubation, the resulting difference spectrum exhibited an absorption spectrum with a maximum at 302 nm ( Fig. 1, B and C), similar to what has been reported for peroxynitrite (52). This was verified by comparison with the UV spectrum of authentic ONOO Ϫ (Fig. 1C). The scatter of the difference spectrum at shorter wavelengths (Ͻ 295 nm) derives from the strong absorption of Angeli's salt in this wavelength region. The absorbance at 302 nm has been attributed to the cis-conformer of peroxynitrite (53,54). This is excellently supported by time-dependent density functional theory calculations (Table I), which show that the trans-conformer of ONOO Ϫ should absorb at longer wavelengths ( max ϭ 374 nm). Thus, cis-ONOO Ϫ is produced from the NO Ϫdonating compound Angeli's salt during its decay in aerobic solution.
Yield of Peroxynitrite Derived from Nitroxyl Anion-The above observations qualitatively prove the formation of peroxynitrite from NO Ϫ ; the yield of this reaction remained to be established. To this end, the potential of Angeli Ϫ /CO 2 (25 mM/5%), 37°C) were incubated for 1 h to oxidize either DHR or NADH (each 50 mM), respectively. A, oxidation of DHR. B, oxidation of NADH. Residual NADH was quantified by reading its fluorescence with excitation at 339 nm and emission at 460 nm. These values were corrected for autoxidation of NADH. In the absence of SIN-1/Angeli's salt, about 97% of the initial NADH concentration (50 M) could still be detected after 1 h of incubation. C, Angeli's salt (100 M) was incubated for 1 h in potassium phosphate buffer (50 mM, 37°C) at various pH values (6 -8) in the absence and presence HCO 3 Ϫ /CO 2 (25 mM/5%). Formation of nitrate was quantified by using nitrate reductase in conjunction with the Griess assay. Each value represents the mean Ϯ S.D. of three experiments performed in duplicate.  (40).
Peroxynitrite-independent Formation of Hydroxyl Radicals from the Nitroxyl Anion-Since only ϳ65% of the generated NO Ϫ was converted into peroxynitrite, the question arose whether other (reactive) intermediates were formed from the remaining 35% of NO Ϫ . Recently, two research groups (55,56) reported that NO Ϫ released from Angeli's salt should generate hydroxyl radicals. As peroxynitrite is known to produce HO ⅐ radicals (57,58) with an efficiency of about 28% (9,59), it is unclear whether the HO ⅐ radicals detected by these groups may have derived exclusively from peroxynitrite. To check on this important point, the Angeli's salt (0 -500 M)-induced hydroxylation of benzoic acid (5 mM) was studied and compared with that of authentic peroxynitrite (0 -500 M) (Fig. 3, A and B). In the absence of CO 2 peroxynitrite-dependent hydroxylation of BA increased in a linear manner with increasing concentrations of peroxynitrite. In the presence of CO 2 , however, peroxynitrite-mediated hydroxylation of BA was inhibited by about 99%. The effect of CO 2 to strongly suppress peroxynitrite-derived formation of hydroxyl radicals is in full agreement with recent reports (57,60). Moreover, this effect of CO 2 offers the possibility to distinguish between HO ⅐ released from peroxynitrite and HO ⅐ radicals released from other sources. Similar to peroxynitrite, Angeli's salt-mediated hydroxylation of BA increased with increasing concentration of Angeli's salt in the absence of CO 2 although not in a strictly linear fashion (Fig. 3B). The efficacy of Angeli's salt to hydroxylate BA decreased with increasing concentration compared with authentic peroxynitrite. This result again is in disagreement with observations by Miranda et al. (37) who found that Angeli's salt was much more effective in hydroxylating BA than authentic peroxynitrite. Again, the usage of HEPES as buffer compound and the fact that no attention was given to exclude traces of CO 2 may explain this discrepancy. Most interestingly, however, the presence of CO 2 did not completely inhibit the hydroxylation of BA by Angeli's salt. Depending on the initial concentration of Angeli's salt, about 14 -29% of the amount of the hydroxylated product that was found in the absence of CO 2 was still formed in its presence. In comparison to peroxynitrite generated in situ from SIN-1 (Fig. 3C), Angeli's salt was only slightly more effective in hydroxylating BA. The observation that in the absence of CO 2 authentic peroxynitrite hydroxylates BA in a linear manner, whereas peroxynitrite generated in situ from both SIN-1 and Angeli's salt hydroxylates BA in an almost identical, nonlinear manner implies the intermediary of the same reactive species formed from these peroxynitrite generating systems. For instance, as N 2 O 3 is known to effectively react with ONOO Ϫ (61), one might speculate that such an intermediate decreases the hydroxylating capabilities of in situ generated peroxynitrite. To further verify that in fact HO ⅐ radicals were produced independently from peroxynitrite, competition experiments with Angeli's salt (500 M), BA (5 mM), and the hydroxyl radical scavenger Me 2 SO (0 -10 mM) were performed in the presence of CO 2 (Fig.  3D). These experiments demonstrated that low concentrations of Me 2 SO effectively inhibited in an apparently exponential manner the NO Ϫ -derived hydroxylation of BA. Thus, NO Ϫ very likely generates HO ⅐ radicals via a peroxynitrite-independent pathway. However, as the Angeli's salt-induced hydroxylation of BA was only to ϳ71% inhibited by Me 2 SO, a hydroxylating species other than the HO ⅐ radical may be additionally generated with low yields by the nitroxyl anion. Since the yield of HO ⅐ radicals from peroxynitrite is ϳ28% (9, 59), and because NO Ϫ generates peroxynitrite with a yield of ϳ65%, and because the hydroxyl radicals from this in situ generated peroxynitrite hydroxylated BA in a range from ϳ86 to 71%, and because the peroxynitriteindependent hydroxylation of BA is only to 71% induced by HO ⅐ radicals, it can be estimated that the NO Ϫ -derived yield of HO ⅐ radicals generated via the peroxynitrite-independent pathway is about 3 Ϯ 1.5%. To clarify whether this pathway is additionally oxygen dependent, Angeli's salt was decomposed at various pH values in the absence (hypoxia) or presence of oxygen (normoxia) and CO 2 (Fig. 3E). In the absence of O 2 the yield of HO ⅐ radicals from Angeli's salt increased with increasing H ϩ concentration, in agreement with findings of Stoyanovsky et al. (55). Furthermore, at atmospheric O 2 levels and in the presence of CO 2 , Angeli's salt was in regard to the O 2 -free solution only about half as effective to hydroxylate BA. Thus, under conditions where the nitroxyl anion cannot react with oxygen, the NO Ϫ -derived yield of HO ⅐ radicals increased to ϳ8% at physiological pH values.
Angeli's Salt-derived Nitrosation of 2,3-Diaminonaphthalene-Since there is a peroxynitrite-independent production of HO ⅐ radicals, one may question that in the presence of oxygen the chemistry of NO Ϫ is generally dominated by peroxynitrite. A NO Ϫ -mediated nitration of tyrosine would, for instance, indicate that peroxynitrite is the attacking species, but unfortunately, in situ generated peroxynitrite does not effectively nitrate tyrosine (62,63). On the other hand, as in situ generated peroxynitrite is able to induce nitrosation reactions (16,17,21), the nitroxyl anion released from Angeli's salt is expected to provoke nitrosation reactions when its chemistry is indeed governed by peroxynitrite. To verify this assumption, the reaction between Angeli's salt (500 M) and DAN (200 M) was  studied in the presence of CO 2 . The product of this reaction was identified as 2,3-naphthotriazole (NAT) because its emission spectrum (Fig. 4A) was found to be identical with that of authentic NAT (Fig. 4B) (44). Thus, peroxynitrite in situ generated from the nitroxyl anion is able to induce nitrosation reactions. To rule out the possibility that only minor amounts of NAT were formed from reaction of Angeli's salt with DAN, the yield of this reaction was also determined (Table III). Angeli's salt concentrations in the range from 50 to 600 M produced NAT with yields of about 7 and 10% in the absence and presence of CO 2 , respectively. Since such yields are very typical for peroxynitrite-induced reactions, one must conclude that the chemistry of the nitroxyl anion is mainly characterized by peroxynitrite and with a minor contribution by independently generated HO ⅐ radicals.

Putative Mechanism by Which Nitroxyl Anion Generates Peroxynitrite-The question arises how the formation of cis-
ONOO Ϫ from NO Ϫ and O 2 may proceed. It has often been suggested that these two molecules, both having a triplet ground state (64), react with each other, thereby directly generating ONOO Ϫ (24,30,32,36,56,65).
In fact, the reaction between 3 NO Ϫ and 3 O 2 appears to be thermodynamically feasible, because MP2 ab initio calculations as well as CBS-QB3 calculations 2 in conjunction with the Isodensity Polarized Continuum Model (IPCM) for solvation predicted both an exergonic reaction in the gas phase as well as in aqueous solution (Table IV). These MP2/IPCM calculations also support an outer sphere electron transfer between 3 NO Ϫ and 3 O 2 , i.e. causing solvent-separated ⅐ NO ϩ O 2 . , to be a thermodynamically feasible process.

REACTION 3
The subsequent formation of cis-ONOO Ϫ is, of course, exothermic (Table IV).
To the best of our knowledge, Reaction 3 was first proposed in 1927 by Andrussow (66). In 1966 Fehsenfeld et al. (67) verified that Reaction 3 indeed proceeds in the gas phase at room temperature. Some years ago, the electron transfer was mentioned with little modifications by the Ignarro group (68), but its significance under physiological conditions has often been questioned by other researchers. Although Reaction 3 appears to be thermodynamically feasible, any experimental indications that it really proceeds in aqueous solution are as yet missing. Provided that in Reaction 3 ⅐ NO is really released, the addition of superoxide dismutase is expected to strongly increase the yield of ⅐ NO during the decay of Angeli's salt.   . ), and cis-ONOO Ϫ were fully optimized to stationary points using second-order Møller-Plesset pertubation theory (MP2) on the 6 -311 ϩ G(d) basis set. Frequency calculations were performed on the same level of theory. As these calculations refer only to the conditions in the gas phase, single point calculations were performed on the MP2/6 -311 ϩ G(d)//MP2/6 -311 ϩ G(d) level for water with the IPCM solvation model.
Ϫ19.7 Ϫ9.5 a Aqueous solvent correction for the gas phase values.

Nitroxyl Anion-derived Radicals
absence and presence of CO 2 , respectively. The Cu(II),Zn-SODinduced formation of H 2 O 2 during the decay of Angeli's salt was verified by two independent methods, namely peroxidase assay and catalase assay. Either replacing Cu(II),Zn-SOD by an equivalent amount of albumin (Fig. 5) or treating Cu(II),Zn-SOD with authentic peroxynitrite (500 M, data not shown) strongly decreased H 2 O 2 formation to Յ4 M. These results strongly indicate that Reaction 3 is indeed operating. Effect of Superoxide Dismutase on Nitroxyl Anion-mediated Oxidation Reactions-The above described action of Cu(II),Zn-SOD implied that nitroxyl anion-related formation of peroxynitrite is inhibited by superoxide dismutase. To further support this assumption, the inhibitory effects of Cu(II),Zn-SOD on both SIN-1-and Angeli's salt-mediated oxidation of both DHR and NADH were studied in the absence and presence of CO 2 . As expected, Cu(II),Zn-SOD inhibited these oxidations with increasing activity of superoxide dismutase (Fig. 6, A-D). Cu(II),Zn-SOD was generally somewhat more effective in inhibiting SIN-1-dependent oxidation reactions than those mediated by Angeli's salt. This might reflect either the likely differences in the kinetics of O 2 . formation or the property of the nitroxyl anion to additionally oxidize these target molecules by releasing HO ⅐ radicals via superoxide-independent pathways. Effect of Nitric Oxide on Nitroxyl Anion-mediated Oxidation Reactions-An anonymous referee mentioned that "NO Ϫ and nitric oxide would be present simultaneously" under various conditions, therefore, the nitroxyl anion chemistry presented so far might be of minor importance for biological systems because the following reaction sequence is known to rapidly proceed (69,70).
Provided that such a reaction sequence is indeed effectively operating, nitroxyl anion-derived formation of peroxynitrite, and thus of nitrate (see Fig. 2C), should be effectively inhibited by nitric oxide. To check on this, the inhibitory effects of DEA-NONOate (0 -150 M) on Angeli's salt (200 M)-mediated formation of nitrate were studied. Since both compounds have nearly the same half-life at the selected experimental conditions, nitric oxide and nitroxyl anion should be generated with approximately the same rate, providing the optimum conditions for the suggested reaction sequence. In line with the reviewer's proposal, nitric oxide inhibited in a linear manner nitroxyl anion-derived formation of nitrate (Fig. 7A) (Fig. 7C). Thus, the chemical power of the nitroxyl anion is strongly increased by nitric oxide when the [ ⅐ NO/[NO Ϫ ] ratio is Յ1. On the other hand, DEA-NONOate inhibited the Angeli's salt-mediated oxidation of NADH when the [ ⅐ NO/[NO Ϫ ] ratio is ϳ2, i.e. under conditions where Reactions 6 -8 are favored. Thus, the interplay between ⅐ NO and NO Ϫ is much more complicated than hitherto believed, but Reactions 6 -8 may contribute to the reduction of the (ϳ65%) yield of peroxynitrite from Angeli's salt at sufficiently high nitric oxide concentrations. DISCUSSION Contrary to the observations presented above, formation of peroxynitrite during decay of Angeli's salt in the presence of oxygen, i.e. from reaction of NO Ϫ and O 2 , has not been reported in previous studies (37,38). In the present study we avoided the use of tertiary amines as both buffer compounds and heavy metal chelators because it has been shown that oxidizing species stimulate the artificial generation of O 2 . in the presence of these amines (8,40). The so formed superoxide reacts in a diffusioncontrolled manner with the NO 2 ⅐ radicals produced from peroxynitrite, thus yielding peroxynitrate (O 2 NOO Ϫ ) (8,14,17). Peroxynitrate is at physiological pH values of low reactivity in terms of nitration/oxygenation of prototypical biomolecules (14,17,73). As a consequence, NO 2 ⅐ -mediated nitration and nitrosation reactions are largely suppressed when O 2 . is simultaneously formed. Presumably because of these chain reactions, the nitroxyl anion-derived formation of peroxynitrite was not observed in the above mentioned studies, where 10 mM HEPES or 500 mM triethanolamine (37,38), respectively, were employed. Due to spin conservation, the nitroxyl anion released from Angeli's salt is initially produced in the singlet state (32). Reaction of 1 NO Ϫ with 3 O 2 is a spin-forbidden process, thus, 1 NO Ϫ should not react fast with 3 O 2 . Furthermore, the reaction of 1 NO Ϫ with 3 O 2 would lead to triplet peroxynitrite, 3 ONOO Ϫ , which, according to ab initio calculations is not a stable molecule (energy minimum). 3 On the other hand, 1 NO Ϫ is isoelectronic to singlet oxygen. As singlet oxygen decays within a few ms to 3 O 2 (74), a similar behavior can be expected for 1 NO Ϫ (Fig. 8). We favor an electron transfer from 3 (24) is often referred to for demonstrating the direct reaction between Cu(II),Zn-SOD and NO Ϫ . These authors observed that under aerobic conditions Cu(II),Zn-SOD stimulates the formation of ⅐ NO from NO Ϫ and hypothesized "that SOD accepts an electron from NO Ϫ , converting it to ⅐ NO." Although a reaction between Cu(II),Zn-SOD and NO Ϫ seems to be chemically feasible, there is yet no convincing proof that such a reaction really takes place. According to our data, the superoxide dismutase-dependent release of ⅐ NO from NO Ϫ can satisfactorily be explained by the scavenging of O 2 . by Cu(II),Zn-SOD. Schmidt et al. (27) reported that the Cu(II),Zn-SOD to NO Ϫ ratio must be about 50 under aerobic conditions for achieving a quantitative conversion of NO Ϫ to ⅐ NO. However, such a high ratio is in fact required when superoxide dismutase would scavenge all O 2 .
formed to prevent the diffusion-controlled reaction between ⅐ NO and O 2 . .
Our data support the conclusion of Stoyanovsky et al. (55), that NO Ϫ forms hydroxyl radicals via a pathway independent of peroxynitrite. The mechanism offered by these authors appears plausible (Fig. 8). A recent theoretical study indicates that the pK a value of HNO is at about 7 (64) rather than at 4.7, as generally assumed (76). Provided that the theoretical pK a 3 M. Kirsch   value can be verified by experiment, the dimerization of two HNO molecules and the subsequent generation of HO ⅐ radicals should even occur at physiological pH values. We estimated that NO Ϫ yields HO ⅐ radicals under normoxia with an efficiency of only ϳ3% via the peroxynitrite-independent pathway. Given the fact that the reaction of 3 NO Ϫ with 3 O 2 strictly depends on the availability of O 2 , and as the O 2 concentration in various tissues is significantly lower than under our experimental conditions, the NO Ϫ -mediated production of hydroxyl radicals might be expected to be increased under hypoxic conditions (up to 8% of the NO Ϫ yield at physiological oxygen concentrations).
Although there can be no doubt that peroxynitrite and radicals derived from it effectively damage biomolecules and a variety of cell types, there is still some uncertainty about the cytotoxic significance of peroxynitrite in vivo, especially when ⅐ NO and O 2 . are generated from independent sources. There is evidence that the chemical power of in situ generated peroxynitrite is maximal, when the ratio ( . ) is about one and the damaging potential should then rise to a maximum. This is the case when the nitroxyl anion transfers an electron to molecular oxygen (Fig. 8)