Metabolic fate of peroxynitrite in aqueous solution. Reaction with nitric oxide and pH-dependent decomposition to nitrite and oxygen in a 2:1 stoichiometry.

Peroxynitrite, the reaction product of nitric oxide (NO) and superoxide (O2) is assumed to decompose upon protonation in a first order process via intramolecular rearrangement to NO3−. The present study was carried out to elucidate the origin of NO2− found in decomposed peroxynitrite solutions. As revealed by stopped-flow spectroscopy, the decay of peroxynitrite followed first-order kinetics and exhibited a pKa of 6.8 ± 0.1. The reaction of peroxynitrite with NO was considered as one possible source of NO2−, but the calculated second order rate constant of 9.1 × 104 M−1 s−1 is probably too small to explain NO2− formation under physiological conditions. Moreover, pure peroxynitrite decomposed to NO2− without apparent release of NO. Determination of NO2− and NO3− in solutions of decomposed peroxynitrite showed that the relative amount of NO2− increased with increasing pH, with NO2− accounting for about 30% of decomposition products at pH 7.5 and NO3− being the sole metabolite at pH 3.0. Formation of NO2− was accompanied by release of stoichiometric amounts of O2 (0.495 mol/mol of NO2−). The two reactions yielding NO2− and NO3− showed distinct temperature dependences from which a difference in Eact of 26.2 ± 0.9 kJ mol−1 was calculated. The present results demonstrate that peroxynitrite decomposes with significant rates to NO2− plus O2 at physiological pH. Through formation of biologically active intermediates, this novel pathway of peroxynitrite decomposition may contribute to the physiology and/or cytotoxicity of NO and superoxide.

The reaction between nitric oxide (NO) and superoxide anion (O 2 . ) yields peroxynitrite with a second order rate constant near the diffusion-controlled limit (k ϭ 4.3-6.7 ϫ 10 9 M Ϫ1 s Ϫ1 ) (1,2). The reaction constitutes an important sink for O 2 . because it is about twice as fast as the maximum velocity of SOD. 1 Consequently, peroxynitrite has been implicated in many patho-logical conditions including stroke (3), heart disease (4), and atherosclerosis (5,6). The potential cellular targets for peroxynitrite cytotoxicity include the antioxidants ascorbate, ␣-tocopherol, and uric acid (7)(8)(9)(10), protein and non-protein sulfhydryls (11), DNA (12), and membrane phospholipids (13). Decomposition of peroxynitrite is complex (14,15). The anion is rather stable in alkaline solutions but decomposes rapidly (t 1/2 ϭ 1 s at pH 7.4, 37°C) upon protonation to peroxynitrous acid (ONOOH) (pK a ϭ 6.8) (16). Two pathways of ONOOH decomposition have been proposed. Some studies have argued that ONOOH is cleaved homolytically to generate hydroxyl and NO 2 radicals. This hypothesis is based on the sensitivity to hydroxyl radical scavengers of certain peroxynitrite-induced reactions, including the formation of malondialdehyde from deoxyribose and the hydroxylation on the benzene ring of sodium benzoate, phenylalanine, and tyrosine (16,17). Studies on decomposition of peroxynitrite by electron paramagnetic resonance spectroscopy with the spin traps 5,5-dimethyl-1-pyrroline N-oxide and 4-pyridyl-1-oxide-N-tert-butylnitrone also provided evidence for the formation of free hydroxyl radicals (18,19). Against this, Koppenol et al. (15) concluded from molecular dynamic calculations that homolytic cleavage of ONOOH is highly improbable. This was reinforced by the independence of the rate of ONOOH decomposition on solvent viscosity (20). Based on these results, it was suggested that decomposition of ONOOH to NO 3 Ϫ involves formation of an activated intermediate (ONOOH*), which might account for the hydroxyl radicallike properties of peroxynitrite (15,21).
There are several methods for the detection of peroxynitrite in biological systems. Since ONOOH decomposition yields an intermediate that nitrates phenolic compounds (22,23), presence of nitrotyrosine in proteins was proposed to be evidence of peroxynitrite production in tissues (24). However, using both a monoclonal antibody specifically recognizing peroxynitritemodified proteins (24) as well as a published HPLC method (17), we failed to detect tyrosine nitration by authentic peroxynitrite at concentrations Ͻ0.1 mM. 2 Spectrophotometric determination of dihydrorhodamine 123 oxidation was described as another sensitive assay for the specific detection of peroxynitrite at submicromolar concentrations (25), but in our hands, interference of several redox-active compounds precluded application of this method in cell-free assay systems. 3 Under certain experimental conditions, indirect evidence for peroxynitrite production can be obtained by comparing NO release in the absence and presence of SOD. The peroxynitrite donor compound SIN-1, for example, does not release detectable amounts of free NO unless SOD is present in amounts sufficient to outcompete the reaction with concomitantly produced O 2 . (26). Based on similar results obtained with purified neuronal NO synthase, we suggested that the enzyme generates NO and O 2 . simultaneously and hence functions as peroxynitrite synthase if incubated in vitro (27). However, in contrast with the widely held view that peroxynitrite decomposes exclusively to NO 3 Ϫ , considerable amounts of NO 2 Ϫ were also found as a major stable product of SIN-1 or NO synthase under physiological conditions. 2 Similarly, excess NO 2 Ϫ formation was observed in peroxynitrite producing cells (28), suggesting that additional as yet unidentified reactions contribute to peroxynitrite decomposition.
The present study was done to elucidate the fate of peroxynitrite in aqueous solution. Studies with the authentic compound, prepared in two different ways, identified a reaction leading to release of NO 2 Ϫ and O 2 in a 2:1 stoichiometry as a route of peroxynitrite decomposition at pH Ն 7.5.

EXPERIMENTAL PROCEDURES
Materials-NO solutions were prepared by dissolving NO gas (Linde Mü nchen, Germany, 99% pure) in deoxygenated water as described previously (29). All solutions were prepared freshly each day with Nano-pure water (Barnstead ultrafiltered type I, resistance Ͼ18 megaohms cm Ϫ1 ). Sulfanilamide, sodium nitrite, cadmium, and the Griess-Ilosvay reagent for postcolumn derivatization were from Merck, Darmstadt, Germany. All other chemicals were from Sigma, Vienna, Austria.
Synthesis of Peroxynitrite-Alkaline solutions of peroxynitrite (80 -100 mM) were prepared from acidified NO 2 Ϫ and H 2 O 2 according to the Baeyer-Villinger reaction (30) and quantified spectrophotometrically using an extinction coefficient of 1670 M Ϫ1 cm Ϫ1 (26,30,31 Kinetic Experiments-Peroxynitrite decomposition was studied by stopped-flow absorbance spectroscopy at 302 nm (Bio-Sequential SX-17MV stopped-flow ASVD spectrofluorimeter, Applied Photophysics, Leatherhead, U. K.). For simple decomposition experiments, reservoir 1 contained peroxynitrite in 0.01 M NaOH, and reservoir 2 contained the buffer solution (at pH 3.0 -6.0, 1 M acetate buffer; at pH 5.0 -9.0, 1 M phosphate buffer; at pH 8.0 -10.0, 1 M Tris/HCl; at pH Ն 10, solutions of NaOH). The NaOH concentration in reservoir 1 was, in some cases, adapted to the requirements of the experiment: non-buffered experiments at pH 3.0, 10.0, and 11.0 were carried out with sufficiently low concentrations of NaOH.
The reaction of peroxynitrite with NO was studied by sequential stopped-flow, i.e. reservoirs 1 and 2 were premixed followed by mixing with contents of reservoir 3 with short delay time (10 ms). Reservoir 4 was used to push the mixed contents of reservoirs 1 and 2 forward into the main mixing chamber. Reservoir 1 contained buffer (pH 3.0 -11.0; 4 ϫ final concentration), reservoir 2 contained a solution of peroxynitrite in NaOH (4 ϫ final concentration; typical final [NaOH] 5 mM), reservoir 3 contained a saturated solution of NO (ϳ2 mM giving ϳ1 mM final concentration), and reservoir 4 contained buffer (2 ϫ final concentration). To vary NO concentrations, experiments were also done with 2-fold diluted peroxynitrite in reservoir 3 and NO in reservoir 2. This yields the same final concentration of peroxynitrite but a 2-fold lower final concentration of NO (ϳ0.5 mM). Samples of the NO solution were taken with a plastic syringe under helium gas and transferred directly into the stopped-flow reservoir. Experiments were carried out both with air-containing buffers and with buffers that had been thoroughly degassed. Degassing made no difference.
Decomposition of Peroxynitrite and Determination of NO 2 Ϫ and NO 3 Ϫ-Unless indicated otherwise, peroxynitrite (1 mM or 0.5 mM) was decomposed by incubation in 0.1 M phosphate buffer for 1 h at pH 3.0 -9.0. [Me 4 N][ONOO] (0.25 mM or 0.1 mM) was decomposed in 0.5 M phosphate buffer under the same conditions. NO 2 Ϫ was determined by the Griess assay. The samples (0.1 ml) were mixed with 10 l of H 2 O and 10 l of an EDTA solution (0.5 M, pH 8.0), followed by addition of 0.12 ml of freshly prepared Griess reagent (20 mg N-(1-naphthyl)ethylenediamine and 0.2 g sulfanilamide dissolved in 20 ml of 5% (w/v) phosphoric acid) and measurement of the absorbance at 546 nm. For determination of NO 2 Ϫ ϩ NO 3 Ϫ , samples (0.2 ml) were adjusted to pH ϳ7.5 and mixed with 20 l of an aqueous zinc suspension (100 mg/ml) and 20 l of an EDTA solution (0.5 M, pH 8.0). Samples were spun down for 5 min, and 0.12 ml of the supernatant were mixed with 0.12 ml of the Griess reagent, followed by determination of the absorbance at 546 nm. Calibration curves were established with NO 2 Ϫ and NO 3 Ϫ (10 -50 M each). The calculated amount of NO 2 Ϫ present in stock solutions of conventionally prepared peroxynitrite agreed well with NO 2 Ϫ measured after decomposition at pH 3.0. This amount was subtracted from the measured values.
The NO 2 Ϫ /NO 3 Ϫ data were confirmed by HPLC analysis according to published protocols (33,34). 50 l samples were injected onto a 250 ϫ 4 mm C18 reversed phase column (LiChrospher 100 RP-18, 5 m particle size, Merck, Vienna, Austria) and eluted with 5% (w/v) NH 4 Cl, pH 7.0, at a flow rate of 0.7 ml/min. NO 2 Ϫ was detected by postcolumn derivatization with the stable Griess-Ilosvay reagent (Merck) (0.7 ml/ min), heating to 60°C, and measurement of the absorbance at 546 nm. To study the reaction of peroxynitrite with NO, 4-l aliquots of an ϳ2 mM NO solution were injected into 1.8-ml glass vials completely filled with 0.1 M phosphate buffer, pH 7.4, and sealed with a septum. At the indicated time points, 1.8 -3.6 l of peroxynitrite solution (0.5 mM) were applied to give concentrations of 0.25-1 M. The output current was recorded at 1.66 Hz under constant stirring. The sensor was calibrated with NO 2 Ϫ standards according to manufacturer recommendations.

RESULTS
Decomposition of peroxynitrite was monitored as decrease in absorbance at 302 nm at 20°C. As expected, decomposition at pH 3 was very fast and followed first order kinetics with a calculated rate constant (k calc ) of 0.86 Ϯ 0.05 s Ϫ1 but slowed down at increasing pH. The k calc values and corresponding Hill coefficients summarized in Table I demonstrate that peroxynitrite decay was first order under most conditions although Hill coefficients smaller than 1.0 were obtained at pH 8.0 (0.67 Ϯ 0.02) and pH 11.0 (0.5 Ϯ 0.1). Using the Hill equation for overall kinetic analysis of decomposition at pH 3-11, we calculated a pK a of 6.8 Ϯ 0.1, which agrees well with published data (35). The possible contribution of transition metals to peroxynitrite decomposition was studied with 0.6 mM peroxynitrite in 0.  1 and 1 mM). At a concentration of 2.5 mM DTPA, the peroxynitrite decay rate was enhanced 10-fold.
Stopped-flow data showed that peroxynitrite decomposition was faster in the presence of ϳ1 mM NO and that the increase in rate was dependent on the NO concentration. However, calculation of rate constants was difficult because the exact NO concentrations in these experiments were not known and the effect of NO was observed only as a relatively small increase of an already fast reaction. Therefore, we used an NO-sensitive electrode to measure the consumption of NO by known amounts of peroxynitrite. Fig. 1 (Table II), our results do not support previous suggestions according to which formation of NO 2 Ϫ is due to contamination of peroxynitrite solutions with trace metals (36) but indicate that NO 2 Ϫ release results from an as yet unrecognized pathway of peroxynitrite decomposition. To address this issue, we measured NO 2 Ϫ and NO 3 Ϫ after peroxynitrite decomposition at pH 3-9 and found that the relative amount of NO 2 Ϫ increased with increasing pH (Fig. 2A). Assum-ing that these results were not due to a reaction of peroxynitrite with contaminants in the stock solutions, our findings led us to speculate that 2 mol of peroxynitrite decomposed to 2 mol of NO 2 Ϫ and 1 mol of O 2 . Indeed, using a Clark-type O 2 sensor, we found that the pH-dependent formation of NO 2 Ϫ was accompanied by release of stoichiometric amounts of O 2 (Fig. 2A). The replot of the data (Fig. 2B) Fig. 3B yielded a slope of 0.657 and a correlation coefficient of 0.983. Although these data nicely confirmed the results obtained with the conventional preparation, two interesting differences were observed. First, while release of NO 2 Ϫ and O 2 was negligible when the Baeyer-Villinger preparation of peroxynitrite was decomposed at pH   Յ6.5 (cf. Fig. 2A The assumption that two different pathways of decomposition account for the formation of NO 2 Ϫ and NO 3 Ϫ was supported by a pronounced temperature sensitivity of the NO 2 Ϫ /NO 3 Ϫ ratio. As shown in Table III Ϫ at 5 and 56°C, respectively. A similar increase of the NO 2 Ϫ /NO 3 Ϫ ratio was observed with three concentrations (0.1, 0.5, and 1 mM) of the Baeyer-Villinger preparation (not shown). We also determined the temperature dependence for the overall peroxynitrite decomposition rate between 5 and 50°C by stopped-flow spectroscopy. The Arrhenius plots showed a strictly linear relationship between ln k obs and T Ϫ1 at pH 5.0 and 7.4 (Fig. 4A). From the slope of the plots, values for E act of 92.0 Ϯ 2 kJ mol Ϫ1 and 90.0 Ϯ 0.8 kJ mol Ϫ1 were calculated for decomposition at pH 5.0 and 7.4, respectively. Assuming that the NO 2 Ϫ /NO 3 Ϫ ratios reflect the kinetic partitioning of the two pathways leading to NO 2 Ϫ and NO 3 Ϫ formation, the difference in E act of the reations (⌬E act ) was estimated as 26.2 Ϯ 0.9 kJ mol Ϫ1 (Fig. 4B).

DISCUSSION
The present study was carried out to identify the pathways of formation of NO 2 Ϫ in the course of peroxynitrite decomposition. Stopped-flow kinetic experiments confirmed that peroxynitrite decomposed rapidly upon protonation with a pK a of 6.8. The first order rate constants calculated for peroxynitrite decomposition at different pH values agreed well with previously published data (11,15,37). Under physiological conditions (pH 7.4 and 37°C), decomposition consistently yielded about 30% NO 2 Ϫ , whereas NO 3 Ϫ was the sole product at pH Ͻ5.0. Some studies indicated that under certain experimental conditions, peroxynitrite does indeed decompose to NO 2 Ϫ , but this was attributed to minor side reactions catalyzed by contaminating trace metals (36,38). In our hands, NO 2 Ϫ release was not metalcatalyzed because it was affected neither by chelators nor by metal ions (0.1 mM).
Previous studies showed that small amounts of NO are released upon peroxynitrite decomposition under certain conditions (39) and that peroxynitrite reacts with NO according to Equation 1 (40,41).
Lewis et al. (28) observed that activated macrophages release more NO 2 Ϫ than expected and considered Equation 1 as a pos-   Table I were replotted as ln((k 1 )/(k 2 )) with ((k 1 )/(k 2 )) ϭ ((%NO 2 Ϫ )/(%NO 3 Ϫ )) versus (1/ T). From the slope of the linear plot, the difference between the activation energies (⌬E act ) of the two reactions yielding NO 2 Ϫ and NO 3 Ϫ was calculated. Data are the mean values of three experiments. sible source for excess NO 2 Ϫ . To judge whether Equation 1 could account for NO 2 Ϫ formation under physiological conditions, we studied the reaction of peroxynitrite with NO using both stopped-flow kinetic spectroscopy and an NO-sensitive electrode. These experiments yielded an apparent second order rate constant of 9.1 ϫ 10 4 M Ϫ1 s Ϫ1 , which is in the same order of magnitude as the rate constant determined for peroxynitrite reacting with CO 2 (42). However, the concentration of NO that would be required to render the reaction as fast as peroxynitrite decomposition (0.69 s Ϫ1 at pH 7.4 and 37°C) is so high (7.6 M) that the reaction with NO is probably insignificant under most physiological conditions (43).
Since pure peroxynitrite decomposed to NO 2 Ϫ without detectable release of free NO, we considered additional pathways that could account for NO 2 Ϫ formation. Assuming that there were no other redox-active reaction partners of peroxynitrite present, we speculated that two peroxynitrite molecules might combine to release 2 molecules of NO 2 Ϫ and 1 molecule of O 2 according to Equation 2.
This hypothesis was corroborated by determination of NO 2 Ϫ and O 2 formed upon decomposition of two different peroxynitrite preparations. With both products, we obtained linear correlations between NO 2 Ϫ and O 2 release with slopes close to the theoretical value of 0.50. At pH 3.0 -11.0, peroxynitrite decomposition was first order at most. Moreover, even though the temperature dependence of the NO 2 Ϫ /NO 3 Ϫ ratio clearly indicated that the two pathways have different activation energies (⌬E act ϭ 26.2 Ϯ 0.9 kJ mol Ϫ1 ), the Arrhenius plot for overall peroxynitrite decomposition at pH 7.4 (ϳ30% NO 2 Ϫ ) was strictly linear and yielded an E act value that was virtually identical to that observed at pH 5.0 (ϳ10% NO 2 Ϫ ). The E act of overall peroxynitrite decomposition was found to be rather high (92 Ϯ 2 and 90.0 Ϯ 0.8 kJ mol Ϫ1 at pH 5.0 and 7.4, respectively). A similar value (77.5 kJ mol Ϫ1 at pH 5.0) was reported by Koppenol et al. (15).
From these observations, we conclude that the rate-limiting step in both reactions is the same, a conformational change of ONOOH to an activated intermediate that either rearranges to HNO 3 (15,35) or undergoes a reaction with peroxynitrite anion to yield NO 2 Ϫ and O 2 (this study). Any potential model must account for the thoroughly characterized kinetics of peroxynitrite decomposition as well as the stoichiometries of the end products. Further, a bimolecular rate law for either of the product determining steps is excluded because the partitioning of the two pathways does not depend on the concentration of peroxynitrite. Fig. 5 shows a hypothetical pathway of peroxynitrite decomposition that appears to be most consistent with the data presented both here and in the literature. According to this scheme, activated ONOOH can either isomerize to NO 3 Ϫ or decompose to HO and NO 2 radicals. At alkaline pH, the OH radical may react with peroxynitrite anion yielding O 2 , NO, and OH Ϫ , and NO could react with NO 2 radicals to yield N 2 O 3 and finally nitrite.
The novel pathway of peroxynitrite decomposition described here could have important physiological consequences, as it possibly involves generation of intermediates with biological activities not attributed so far to peroxynitrite. In a recent paper, it was reported that peroxynitrite decomposition could lead to release of singlet O 2 (44). If that observation were due to the novel reaction proposed here, peroxynitrite-dependent toxicity might be mediated by singlet O 2 toxicity under certain pathophysiological conditions. Alternatively, decomposition to H NO 2 Ϫ and O 2 may be responsible for the observed NO-like biological activity of peroxynitrite. At pH 7.4, peroxynitrite oxidizes hemoglobin to methemoglobin with an efficiency of about 20% (26), and it is tempting to speculate that this reaction represents scavenging by hemoglobin of the NO that is formed as intermediate during decomposition to NO 2 Ϫ and O 2 (Fig. 5). Also, our working hypothesis involves intermediary formation of N 2 O 3 , a potent nitrosating agent that could account for the observed peroxynitrite-induced nitrosation of GSH, especially in light of our findings that the nitrosation reaction has a pronounced pH dependence and does not occur at significant rates below pH 7.5 (45). Accordingly, reactive intermediates formed in the course decomposition to NO 2 Ϫ and O 2 could be responsible for stimulation of soluble guanylyl cyclase by peroxynitrite (45), resulting in cyclic GMP-mediated biological effects such as vascular smooth muscle relaxation and inhibition of platelet aggregation (46,47).