INTRODUCTION

Nitric oxide (NO) is a widespread intracellular and intercellular signaling molecule involved in the regulation of diverse physiological and pathophysiological mechanisms in the cardiovascular system and the central and peripheral nervous systems and in immunological reactions (for review, see Refs. 1-3). In contrast to the current view that NO is extremely unstable in vivo, but in agreement with its functioning as a paracrine mediator, NO can travel significant distances to reach target cells neighboring the NO-generating cell (4, 5). Along this migration, in particular at higher concentration, NO can interact with molecular oxygen to form higher nitrogen oxides (e.g. NO₂ and N₂O₃), which can either react with other biomolecules such as thiols and amines or simply hydrolyze to form nitrite (NO₂) and nitrate (NO₃). Furthermore, NO rapidly reacts with reduced hemoproteins and oxygen-derived radicals. The extent of either of these reactions largely depends on the microenvironmental conditions under which NO is released, most important, the concentration of other bioreactants (6, 7).

There is increasing evidence that the effective concentration of NO in a given biological tissue is not solely determined by its rate of enzymatic formation, but also by its rate of degradation and scavenging by other biomolecules. For example, the progression of atherosclerotic lesions has been attributed to an increased oxidative stress within the vascular wall and, consequently, to an accelerated breakdown of NO (8, 9). One important oxygen-derived metabolite capable of inactivating NO is superoxide (O₂), which reacts with NO at an almost diffusion-controlled rate (10) to form peroxynitrite (ONOO) (Equation 1).

(Eq. 1)

Peroxynitrite can isomerize, via its conjugated acid, to nitrate, which has little biological activity in its own right (Equation 2).

(Eq. 2)

However, ONOO itself forms a strong oxidant of potential pathophysiological significance that can react with a vast number of other biomolecules and can cause cell damage (11, 12). Furthermore, ONOO can react with the remaining NO to form nitrogen dioxide (NO₂), which leads to the formation of the nitrosating agent N₂O₃ (13) (Equations 3 and 4).

(Eq. 3)

(Eq. 4)

A variety of analytical techniques have been described for the quantification of either NO or O₂, some of which have been successfully applied to estimate the production of either radical in biological systems (for review, see Refs. 14-16). In contrast, only little is known about the formation of ONOO in tissues and cells, and to the best of our knowledge, to date only qualitative data based on the detection of nonspecific secondary reaction products are available. Furthermore, under conditions where both NO and O₂ are formed, almost all methods for measurement of either molecule can detect only that amount of radical that escaped from chemical interaction with the respective other reaction partner (and, of course, other biomolecules), leading to significant underestimation of their true rates of formation. To better understand the individual reactivity of NO and O₂ in a biological system and to appreciate the pathophysiological consequences of their interaction, reliable information on individual flux rates is required.

The simultaneous monitoring of NO and O₂, although highly desirable, was believed to be technically unfeasible due to the extremely rapid reaction between both radicals. We describe here an analytical technique that allows, for the first time, the simultaneous and sensitive quantification of both NO and O₂ and, under certain conditions, additionally helps to unmask the proportion of peroxynitrite and other reactive nitrogen oxide species formed. The results obtained with this new assay could be successfully simulated by a computer model that predicts the reaction of both radicals with hemoproteins and thus may give valuable new insights into the reaction between NO and O₂ in biological systems.

EXPERIMENTAL PROCEDURES

One of the few spectrophotometric techniques that allows the quantification of NO in aerobic aqueous solutions (the oxyhemoglobin assay) is based on the stoichiometric conversion by NO of oxyhemoglobin (oxyHb¹; Fe²⁺) to methemoglobin (metHb; Fe³⁺) (17) (Equation 5).

(Eq. 5)

The formation of NO can be continuously monitored by time-dependent recording of the absorbance changes in the Soret band associated with oxyHb oxidation. Maximal sensitivity is achieved by choosing a wavelength at which the extinction difference between oxyHb and metHb is maximal (e.g. 401 nm) and an isosbestic point (e.g. 411 nm) as an internal reference (dual-wavelength measurement) (Fig. 1A).

One of the most commonly used spectrophotometric assays for the quantification of O₂ is based on the rapid reduction by O₂ of ferricytochrome c (Fe³⁺) to ferrocytochrome c (Fe²⁺) (Equation 6),

(Eq. 6)

which can be followed by monitoring the increase in absorbance at 550 nm (Fig. 1B) (18). Separate measurements of either NO or O₂ formation for validation of the nitric oxide/superoxide assay were carried out in 100 mM phosphate buffer, pH 7.40 (37.0 ± 0.05 °C), using a conventional dual-wavelength double-beam spectrophotometer (UV-3000, Shimadzu, Duisburg, FRG) set to the classical detection wavelengths of the oxyHb and cytochrome c assays, respectively.

The photometric assay for the simultaneous determination of NO and O₂ described in this paper is based on a combination of the above-mentioned detection principles: the NO-mediated oxidation of oxyHb and the O₂-induced reduction of ferricytochrome c. For the separate recording of either signal, a programmable diode array spectrophotometer (DU-7500i, Beckman, Munich, FRG) was used. In contrast to classical dual-wavelength spectrophotometers, the diode array technique has the advantage of allowing parallel monitoring of multiple wavelengths almost at the same time (e.g. an entire absorption spectrum can be derived within 0.1 s). As the internal memory of these machines is usually limited in capacity to the storage of a couple of full spectra or a few hundred data points only, a series of programs were developed for the continuous visualization of a large number of data points on the photometer screen and for the on-line transfer of raw data between photometer and personal computer to allow unlimited data sampling and post-processing of photometric recordings. Measurements were performed as either end point or kinetic determinations using glass or disposable plastic cuvettes or under conditions of continuous flow using a flow-through cuvette with a total volume of 80 µl (QS 178-010, Hellma, Mühlheim, FRG). If not otherwise indicated, all measurements were carried out in 100 mM phosphate buffer, pH 7.4, at 37 ± 0.02 °C using a thermostatted cuvette holder connected to a D 8-L water bath (Haake, Karlsruhe, FRG).

Consecutive spectral scans between 350 and 600 nm were recorded on oxidation of oxyHb to metHb using either aqueous solutions of authentic NO or spermine NONOate as NO donor. Changes in absolute spectra obtained with the diode array technique were virtually identical to those observed using a conventional dual-wavelength spectrophotometer (17). Difference spectra were derived by subtraction of individual spectral scans from the spectrum obtained with pure oxyHb at time 0 (t_0 min). Increasing concentrations of oxyHb (0.1-7.0 µM) were reacted with excess NO or SIN-1, and the difference in absorbance between 401 and 411.5 nm (isosbestic point for oxyHb to metHb conversion) was recorded. Photometric readings were converted to NO concentrations using a molar extinction coefficient of _401-411.5 = 49.5 mM¹ cm¹ as determined under the conditions of this study. Reproducibility for end point determinations (repeated measurements of NO standards) was better than 4% (coefficient of variance = 3.4; n = 12). The molar extinction coefficient for the cytochrome c assay (₅₅₀ = 19.5 mM¹ cm¹) was determined by reacting increasing concentrations (5-30 µM) of ferricytochrome c with a 3-fold molar excess of ascorbic acid or the O₂-generating hypoxanthine/xanthine oxidase system (n = 3 each) and recording changes in absorbance at 550 nm. The extinction coefficients for the NO-induced conversion of oxyHb to metHb (_525-541.2) and for the O₂-mediated conversion of ferri- to ferrocytochrome c (_465-525) in the combined nitric oxide/superoxide assay were determined basically in the same manner, but in the presence of the respective other hemoprotein. Determinations were performed under conditions identical to those of the kinetic investigations. In those experiments where O₂ was enzymatically generated using the hypoxanthine/xanthine oxidase system, the buffer was additionally supplemented with 25 µM EDTA.

All compounds used in this study were either analytical grade or otherwise of the highest purity available. ABTS (2,2-azinobis(3-ethylbenzthiazoline-6-sulfonic acid, diammonium salt), catalase (from bovine liver, thymol-free; 20,000 units/mg), 18-crown-6 (1,4,7,10,13,16-hexaoxacyclooctadecane), ferricytochrome c (from horse heart, containing 2% ferrocytochrome c and 2.7% water), hemoglobin (human, lyophilized), hypoxanthine, manganese dioxide (MnO₂; activated), L-methionine, potassium superoxide (KO₂), and superoxide dismutase (from bovine erythrocytes; 4400 units/mg of protein) were purchased from Sigma (Deisenhofen, FRG). Xanthine oxidase (from bovine milk, phosphate-free lyophilisate; 0.18 unit/mg) was obtained from Boehringer (Mannheim, FRG). Sodium nitrite (NaNO₂), hydrogen peroxide (H₂O₂; 30%), and pyrogallol (1,2,3-trihydroxybenzene) were from Merck (Darmstadt, FRG). Spermine NONOate (Cayman Chemical Co., Inc., Ann Arbor, MI) was dissolved in argon-gassed 0.01 M NaOH. SIN-1 (kindly donated by Dr. K. Schönafinger, Hoechst Marion Roussel, Frankfurt/Main, FRG) was dissolved in distilled water adjusted to pH 5.0. S-Nitrosoglutathione was synthesized according to Hart (19) and dissolved in de-aerated citrate buffer, pH 2.0. NO donor stock solutions were kept on ice in the dark for up to 3 h. Aqueous solutions of authentic NO were prepared essentially as described (20). Briefly, argon (quality = 5.0, >99.99% argon; Linde AG, Unterschleissheim, FRG) was passed through a closed all-glass system comprising two scrubbing bottles, one containing an alkaline pyrogallol (5%, w/v) solution for removal of traces of oxygen and the other containing potassium hydroxide (10%, w/v) to scavenge higher oxides of nitrogen, connected in series with a three-necked beaker containing saline (0.9% NaCl) for the dissolution of NO. After flushing with argon for 30 min, the gas flow was switched to NO (quality = 3.0, >99.9% NO; AGA GAS GmbH, Hamburg, FRG) and maintained for a further 45 min. Aliquots were transferred to air-tight syringes via a septum with the system kept under positive pressure with NO to avoid changes in NO concentration due to re-equilibration between the aqueous and gas phases. Concentrations ranged between 1.7 and 2 mM NO, depending on the ambient pressure and temperature. Dilutions were made in deoxygenated and argon-flushed saline, and concentrations were determined immediately before application using the gas-phase chemiluminescence reaction with ozone (21). Stock solutions of O₂ (0.1-10 mM) were prepared by dissolving solid KO₂ in water-free dimethyl sulfoxide containing 0.5% crown ether (22). Aqueous stock solutions of ONOO were prepared by mixing NaNO₂ and H₂O₂ in a quenched-flow reactor as described (12). Peroxide contamination after treatment with MnO₂ was typically below 0.05% as assessed using the horseradish peroxidase-catalyzed ABTS cation radical formation as described (23). Frozen aliquots (final concentration = 250-350 mM) were stored at 80 °C for up to 4 weeks with only minor decomposition. Dilutions were made in 0.01 M NaOH and used immediately. Decomposed ONOO was prepared by dilution of the stock solution in 1 M hydrochloric acid and readjustment with NaOH to pH 13 after 5 min. Oxyhemoglobin was prepared as described in detail elsewhere (17). Aliquots of the stock solution (1.9-2.5 mM, expressed as concentration in heme) were stored at 80 °C for up to 6 months. Ferricytochrome c was used without further purification.

Biological Sources of NO and O₂

Endothelial cells (ECs) were harvested enzymatically from porcine aorta using collagenase and cultured in M199 medium (Boehringer Mannheim) containing 20% newborn calf serum. Antibiotics were omitted from the medium from day 5 after isolation. The cells were passaged twice using trypsin/EDTA, grown to confluence on plastic dishes (Primaria, Falcon, Heidelberg, FRG), and characterized morphologically by their typical cobblestone-like growth pattern as well as by uptake of acetylated low density lipoproteins. Absence of contamination with smooth muscle cells >2% was verified by counterstaining of cell nuclei with bisbenzimide and with an antibody against smooth muscle -actin. ECs were collected from a single Petri dish using gentle trypsinization beginning 10 ± 1 min before the start of incubation. The resulting cell suspension was subjected to gentle centrifugation at 200 × g, and cells were washed twice with Krebs-Henseleit buffer, resuspended in 5 ml of Krebs buffer containing 10% newborn calf serum, and kept at 37 °C in a culture tube. ECs were added to the incubation cuvette immediately before the start of the experiment, and the cell suspension was subjected to continuous low speed stirring using an electromagnetic cuvette-stirring unit.

Anesthetized male Wistar rats (300-350 g) were killed, and the thoracic aortas were carefully removed, cleaned up of fat and connective tissue, and cut into 4-5-mm rings. The endothelial integrity of these vascular rings was verified by their ability to relax, in an organ bath, on addition of 1 µM acetylcholine by 70% of the precontraction level achieved with phenylephrine. Vascular rings were kept in phosphate-buffered saline at 37 °C for up to 30 min before addition to the incubation cuvette. Vascular tissue was kept at the bottom of a 3-ml cuvette without stirring.

Simulations for development of a mathematical model to explain the observed results were run using Stella II (Version 1.02; High Performance Systems, Hanover, NH) on a Power Macintosh 7200/90 (Apple Computers) with numerical regression of the Runge-Kutta fourth derivative (24) (for the rate equations used, see "Results"). Calculated metHb and ferrocytochrome c concentrations were plotted as a function of time and converted into absorbance units using the molar extinction coefficients for the NO/oxyHb and O₂/ ferricytochrome c reactions determined for the combined assay (see "Results").

The raw data output of the photodiode array spectrophotometer was transferred via the serial interface to a personal computer and stored in individual files for later processing by a commercial graphics and data analysis software (Origin 4.0; MicroCal Inc., Northampton, MA). Base-line smoothing was achieved by applying the method of moving average. With an increasing number of consecutive data points set in relation to one original data point, base-line noise decreases, and thus, sensitivity increases provided the length of the averaging interval does not exceed the length of the signal to be monitored (for the effect of this procedure on base-line noise, see Fig. 2). In contrast, assay sensitivity for fast kinetics is improved, and changes in peak shape are avoided by selecting rather short intervals. Usually, an averaging interval of 5-10 s was used for rapid kinetics and 30-240 s for slower reactions. This flexibility in choice of the most appropriate averaging interval adds to the general advantage of the diode array technique over conventional spectrophotometry, i.e. the possibility of adjusting the data processing with respect to the signal without loss of original data points after the actual measurement has taken place.

Results are presented as means ± S.D. The rates of formation in either assay system are reported as net values corrected for the blank determined under the same conditions, but in the absence of an NO- or O₂-generating system. As pilot tests revealed that the presence of oxyHb causes slow, superoxide dismutase-independent reduction of ferricytochrome c over time, all measurements for direct comparison of the "classical" cytochrome c assay with the combined NO/O₂ assay were performed in the presence of 5 µM oxyHb. Intra- and interassay comparisons were made applying the Mann-Whitney U test for unpaired variables, and a p value of <0.05 was accepted to denote statistical significance.

RESULTS

To select the optimal wavelengths that would allow the simultaneous and independent monitoring of the redox conversion of either hemoprotein, spectral changes of the individual reactions were compared. The classical wavelengths of either assay (i.e. 401 and 411.5 nm for the oxyHb assay and 550 nm for the cytochrome c assay) could not be used for a combined assay as they overlap with opposite spectral changes of the respective other reaction (see difference spectra depicted in Fig. 1, A and B). However, an isosbestic point was found at 525 nm, a wavelength at which absorbance changes of either reaction were negligible. 541.2 nm, which represents an isosbestic point for the ferri- to ferrocytochrome c conversion, was chosen for recording of the changes in oxyHb concentration, and 465 nm, a point at which absorbance changes for the conversion of oxyHb to metHb are negligible, was chosen for monitoring the ferri- to ferrocytochrome c conversion (Fig. 3, A-C).

H₂O₂, which can be formed by the spontaneous or enzymatic (superoxide dismutase-mediated) dismutation of O₂, has the potential to interfere with either hemoprotein (25, 26), in particular at higher rates of O₂ generation. Therefore, catalase was added, and a final concentration of 100 units/ml was found to be sufficient to completely prevent erroneous absorbance changes (false positive oxyHb oxidation and partial reversal of ferricytochrome c reduction) caused by this oxidant. Although a hemoprotein as well, catalase did not affect the absorbance spectrum of the ferricytochrome c/oxyHb mixture at this low concentration. Likewise, the measured rates of NO formation generated by 20 µM spermine NONOate were virtually identical in the absence and presence of this concentration of catalase using either the combined assay or the classical oxyHb technique (n = 2 each) (data not shown).

Initial experiments were carried out with equimolar concentrations of ferricytochrome c and oxyHb (5 µM each), which required slightly different detection wavelengths and molar extinction coefficients (data not shown). By systematic variation of the concentration ratio of either hemoprotein (5:5, 5:10, 5:15, and 5:20 and 10:5, 15:5, and 20:5 for ferricytochrome c and oxyHb, respectively; n = 2-3 for each concentration ratio) and the use of the hypoxanthine/xanthine oxidase system for generation of O₂ and spermine NONOate for generation of NO, it was found that the trapping efficacy O₂ approached an optimum at a ferricytochrome c/oxyHb molar ratio of 4:1. The recovery of NO at a ferricytochrome c/oxyHb ratio of 4:1 was not significantly different from that at a ratio of 1:1 as evidenced by comparison of the rates of NO generation from 20 µM spermine NONOate under both conditions (586.0 ± 12.5 versus 594.6 ± 38.1 nM NO/min at 4:1 and 1:1 ferricytochrome c/oxyHb, respectively, at 37 °C and pH 7.4; n = 5). However, O₂ recovery increased by almost 50% at higher ferricytochrome c concentrations, both in the absence and presence of simultaneous NO generation (901.2 ± 136.6 versus 604.3 ± 93.7 nM O₂/min, respectively, with 25 µM hypoxanthine and 0.75 milliunit of xanthine oxidase at 37 °C and pH 7.4; p < 0.05, n = 5). There was no significant difference in measured O₂ rates between 15 and 20 µM ferricytochrome c. Concentrations higher than 20 µM ferricytochrome c could not be tested in the presence of 5 µM oxyHb due to technical limitations (too high an absorbance for auto-zeroing of the diode array photometer). Final concentrations of 20 µM ferricytochrome c and 5 µM oxyHb were thus chosen for all further experiments. The molar extinction coefficients under these conditions were determined to be _465-525 = 7.3 mM¹ cm¹ for the conversion of ferri- to ferrocytochrome c using ascorbate as reductant and _541.2/525 = 6.6 mM¹ cm¹ for the oxyHb to metHb conversion using spermine NONOate as NO donor. Assay reproducibility was better than 4% (coefficient of variance = 3.7; n = 12).

Selectivity of specific wavelengths for the measurement of NO and O₂ was assessed as follows. Repeated addition of aliquots of an aqueous solution of authentic NO (final concentration = 0.2-1 µM) to a mixture of oxyHb and ferricytochrome c in phosphate buffer, pH 7.4, produced stepwise increases in the absorbance difference between 541.2 and 525 nm, indicative of metHb formation, without inducing any changes in the absorbance difference between 465 and 525 nm, demonstrating that the cytochrome c signal was unaffected (n = 5) (data not shown). Likewise, selective gradual increases in metHb formation were seen when NO was continuously generated by the NO donor compound spermine NONOate (Fig. 4, upper panel). Measured rates of NO formation by spermine NONOate in the combined assay were identical, within experimental error, to those obtained with oxyHb alone (Table I), demonstrating that NO detection is not influenced by the presence of cytochrome c. Likewise, addition to the ferricytochrome c/oxyHb mixture of aliquots of a stock solution of KO₂ in dimethyl sulfoxide resulted in a stepwise reduction of ferricytochrome c without changes in the hemoglobin signal (n = 3) (data not shown). In the presence of the O₂-generating hypoxanthine/xanthine oxidase system, a selective increase in the absorbance difference between 465 and 525 nm was observed (Fig. 4, middle panel). In the presence of the same concentration of both spermine NONOate and hypoxanthine/xanthine oxidase, i.e. under conditions of simultaneous generation of NO and O₂, measured individual rates of radical formation were identical to the separately determined rates (Fig. 4, lower panel). This indicates that the concentrations of oxyHb and ferricytochrome c present in the combined assay are sufficient to effectively trap all of the NO and O₂ generated. As for NO, the measured rates of O₂ formation using the combined assay were not significantly different from those determined with the classical cytochrome c assay (Table I). The slightly (<10%) smaller rates of O₂ formation using the combined NO/O₂ assay compared with the classical cytochrome c assay (Table I) may be explained by reaction of O₂ with oxyHb, which, although comparatively slow (27), may partially compete with the trapping by ferricytochrome c under these conditions. Absorbance changes for the ferri- to ferrocytochrome c conversion elicited by high concentrations of hypoxanthine/xanthine oxidase (producing flux rates of ~1 µM O₂/min) were completely (>96%) abolished in the presence of 100 units/ml superoxide dismutase, confirming the specificity of the signal for O₂. In a separate set of experiments, the recovery of NO and O₂ at low (0.1 µM/min) and high (1 µM/min) flux rates was additionally tested in a crossover design under conditions of simultaneous formation of both radicals at either equimolar generation rates or at a 10-fold excess of one radical over the other (Table II). Under all conditions, i.e. at equimolar flux rates as well as at 10-fold higher generation rates of NO over O₂ and vice versa, individual rates of formation significantly differed neither from each other nor from the rates measured in the absence of the respective other radical (compare Tables I and II).

Table I. Comparison of classical oxyhemoglobin and cytochrome c assays with the combined nitric oxide/superoxide assay Nitric oxide was generated at either low () or high () rates from the NO donor spermine NONOate (4 and 40 µM); O2 was enzymatically generated at two different formation rates to approximately match the NO fluxes applied using xanthine oxidase/hypoxanthine (0.1 unit/liter and 10 µM, respectively, for the low () and 0.75 unit/liter and 25 µM, respectively, for the high () flux rates). The data presented are the mean ± S.D. of five individual incubations (initial rates were determined between 0.5 and 10 min after the start of the reaction). Conditions Classical assay systems Combined NO/O2 assay Level of significance µM/min µM/min NO () 0.085  ± 0.005 0.083  ± 0.008 NSa NO () 0.949  ± 0.013 0.941  ± 0.098 NS  O2 () 0.128  ± 0.013 0.116  ± 0.016 NS  O2 () 1.042  ± 0.047 0.974  ± 0.106 NS  a NS, not significant.

Table II. Simultaneous quantification of NO and O2 formed at different rates using the nitric oxide/superoxide assay NO and O2 were generated at two different flux rates. Sources of and conditions for NO and O2 generation were identical to those given in legend to Table I Neither the low () nor the high () formation rates of NO differed statistically from each other when measured in the presence of low and high flux rates of O2 and vice versa. Data are given as the means ± S.D. of five to seven individual kinetic runs determined at 3 different days. Conditions NO O2 µM/min µM/min NO ()/O2 () 0.081  ± 0.012 0.116  ± 0.017 NO ()/O2 () 0.090  ± 0.013 0.915  ± 0.049 NO ()/O2 () 0.940  ± 0.050 0.123  ± 0.046 NO ()/O2 () 0.943  ± 0.050 0.971  ± 0.083

Maximal assay sensitivity suitable for monitoring relatively slow changes in cellular metabolic activity was calculated to approach 25 nM for NO and O₂ production, respectively, using 240 s as the time interval for data post-processing, applying the method of moving average (see "Experimental Procedures" for details). Shorter averaging intervals (10-30 s), which are recommended for routine monitoring of rapid concentration changes, decrease the detection limit to ~100-60 nM for either radical (calculated for a signal-to-noise ratio of 3:1).

Reactions to ONOO

Because NO and O₂ react at an almost diffusion-controlled rate to form ONOO and because the oxyHb assay has been reported to cross-react with ONOO (28), we sought to investigate what absorbance changes occur on addition of peroxynitrite to the combined NO/O₂ assay. Consecutive addition to the ferricytochrome c/oxyHb mixture of authentic ONOO resulted in stepwise contrary changes in the absorbance difference for each signal, indicating oxidation of both oxyHb and ferrocytochrome c (Fig. 5). Interestingly, the extent of oxidation was dependent on the concentration of ONOO as well as on the concentration of oxyHb and ferrocytochrome c, with decreasing efficacy at lower hemoprotein concentrations. The quantum yield of this reaction as estimated from the sum of oxidized ferrocytochrome c and oxyHb in relation to the concentration of ONOO applied was in the range of 3-10% at 100 and 10 µM ONOO, respectively (n = 3). Controls with solutions of decomposed ONOO at identical pH were found to have no effect (n = 2), confirming that the observed changes in the absorbance difference were not the result of an artifactual change in pH. Moreover, the difference spectrum of the reaction of ONOO with oxyHb was indistinguishable from that obtained with oxyHb and authentic NO, whereas that of ferrocytochrome c and ONOO was the mirror image of that obtained with ferricytochrome c and O₂. Thus, whereas ONOO appears to mimic NO in its reaction with oxyHb, it produces spectral changes opposite to those elicited by O₂. Methionine was found to inhibit the oxidation by ONOO of ferrocytochrome c and oxyHb in a concentration-dependent manner (tested at 1-50 mM; maximal effects at 20 mM), presumably via oxidation of its thioether group to form the corresponding sulfoxide. Other known ONOO scavengers such as ascorbate, cysteine, ebselen, and glutathione (all tested at 0.01-1 mM) could not be used in this assay because of redox interferences with one or the other hemoprotein.

In addition to the validation experiments described before, a separate set of investigations was performed using SIN-1 as model compound for the simultaneous generation of NO and O₂ (29). Addition of SIN-1 to the assay mixture resulted in a time-dependent increase in absorbance for both the NO and O₂ signals (Fig. 6). Increases in ferrocytochrome c formation by SIN-1 were completely inhibited in the presence of 5 units/ml superoxide dismutase (n = 2). The release of NO and O₂ from 100 µM SIN-1 at pH 7.40 and 37 °C was determined to be 1.24 ± 0.07 µM NO/min and 1.12 ± 0.04 µM O₂/min (n = 3), which is in good agreement with the proposed mechanism of oxidative breakdown of sydnonimines (30, 31). Interestingly, the increase in the cytochrome c-related signal turned into a decrease of about the same rate after all of the oxyHb was consumed by reaction with NO. This is explained by a reoxidation of the formed ferrocytochrome c by ONOO, which is produced when NO is no longer scavenged by oxyHb and was found to continue until all of the ferrocytochrome c was completely converted to ferricytochrome c. In agreement with this assumption, addition of fresh oxyHb to such incubations resulted in a reversal of this phenomenon, with further increases in both the NO and O₂ signals until all of the oxyHb was consumed again (n = 2) (data not shown). Addition of 20 mM L-methionine completely prevented ferrocytochrome c reoxidation without affecting the initial rates of NO and O₂ formation (Fig. 6, inset). These data suggest that, using this assay, measurements made in the presence and absence of methionine provide a straightforward means to additionally test for the presence of already formed ONOO. Identical results were obtained when NO and O₂ were generated using spermine NONOate and hypoxanthine/xanthine oxidase (data not shown).

Representative tracings of parallel on-line measurement of the co-generation of NO and O₂ during the nonenzymatic decay of SIN-1 (100 µM) using the nitric oxide/superoxide assay.

The on-line measurement of NO and O₂ under conditions of continuous flow was demonstrated using a perfusion system with either phosphate buffer or Krebs-Henseleit buffer supplemented with 5 µM oxyHb and 20 µM ferricytochrome c, which was passed through a flow-through cell placed in the cuvette holder of the diode array spectrophotometer, and continuous recording of absorbance changes at 465, 525, and 541.2 nm. The flow rate of the system was adjusted to values of 1.0-10.0 ml/min by means of a low pulse peristaltic pump (Minipuls 3, Gilson Medical Electronics Inc., Middleton, WI), and calibration for NO and O₂ was performed by coinfusion of small volumes (10-40 µl/min) of NO and KO₂ standards, respectively, using a high precision infusion pump (Precidor, Infors AG, Bottmingen, Switzerland). Provided dilution by the coinfusion of the total ferricytochrome c/oxyHb concentration in the buffer was kept below 0.5%, blanks with coinfused saline instead of NO or KO₂ did not produce any artifactual changes in absorbance. Linearity of the photometric response was demonstrated for NO concentrations in the range of 0.05-3 µM (r = 0.97; n = 15). Reproducibility was comparable to that for end point determinations (coefficient of variance = 5.6%; n = 24). Using this system, the flux rates of NO and O₂ formation by three different classes of NO donor compounds (spermine NONOate, S-nitrosoglutathione, and SIN-1) were investigated (Fig. 7). Whereas the spontaneous decomposition in phosphate buffer of any one of the model compounds was found to be associated with the release of small amounts of NO, only in the case of SIN-1 was NO release accompanied by a stoichiometric generation of O₂.

Having demonstrated the validity of the method for measuring NO and O₂ in relatively simple in vitro systems, it was of interest to see whether or not it would be applicable to more complex biological systems without further modification. Two different biological sources were used to demonstrate, in a first pilot investigation, the suitability of the combined NO/O₂ assay for application to intact cells and tissues: cultured aortic ECs and isolated vascular tissue. ECs in suspension were found to continuously release NO and O₂ at low, yet detectable rates (n = 3) (Fig. 8A). NO was enhanced severalfold over basal release rates in the presence of the calcium ionophore A23187, whereas the signal for O₂ was markedly suppressed (Fig. 8B). In contrast, addition of NADPH (data not shown) or NADH to ECs in suspension selectively increased the signal for O₂, leaving the NO signal virtually unaffected (Fig. 8C). In freshly isolated resting aortic tissue, the basal release of NO was below the quantifiable limit. Addition of NADH markedly increased the formation of O₂ (Fig. 8D), and this increase was fully inhibitable by superoxide dismutase (n = 5) (Fig. 8E). These data clearly demonstrate that this method can also be used for reliable kinetic measurement of NO and O₂ in biological samples.

Basal and stimulated NO and O₂ release in cultured aortic endothelial cells (A-C) and isolated aortic tissue (D and E).

In our experiments, NO and O₂ were formed simultaneously at specified rates. Under these conditions, either radical can undergo various reactions. To better understand the NO/O₂ interaction, a theoretical model that allows the prediction of experimental results via a mathematical simulation was elaborated, which is based on the following assumptions and experimental evidence.

Superoxide can react with ferricytochrome c (Equation 6; k₆ = 2 × 10⁶ M¹ s¹) (32), with NO to form ONOO (Equation 1; k₁ = 6 × 10⁹ M¹ s¹) (10), or dismutate spontaneously to give hydrogen peroxide (Equation 7; k₇ = 5 × 10⁵ M¹ s¹) (33).

(Eq. 7)

NO can react with oxyHb to form metHb (Equation 5; k₅ = 3 × 10⁷ M¹ s¹) (34, 35) and with O₂ as in Equation 1. Peroxynitrite can isomerize to yield nitrate (Equation 2) at a rate of k₂ = 0.8 s¹ (36). ONOO, presumably through the s-trans-isomer of ONOOH, which is a powerful oxidant. As seen in Figs. 6 and 9, addition of peroxynitrite does oxidize ferrocytochrome c to ferricytochrome c, suggesting that when oxyHb is consumed by NO, the resultant ONOO formed upon protonation can either oxidize ferrocytochrome c or isomerize to form NO₃. This reaction rate is unknown; however, a value can be estimated based on our observations. We have shown that the reoxidation of ferrocytochrome c in the presence of fluxes of NO/O₂ is abolished by methionine, indicating that ONOO is the oxidant responsible for this reaction. In Fig. 9, after oxyhemoglobin is exhausted (300-s exposure to spermine NONOate and hypoxanthine/xanthine oxidase), the rate of reoxidation of ferrocytochrome c is ~50% that of the oxidation of oxyHb and the reduction of ferrocytochrome c in the previous 300 s. Although it is predicted that 100% of this flux would result in the formation of ONOO, this indicates that <50% actually oxidizes ferrocytochrome c, suggesting the involvement of competing reactions such as the rearrangement of protonated ONOO to NO₃ at a rate of 0.8 s¹ at neutral pH (Equation 2). Since ~50% of the ferrocytochrome c is oxidized by NO/O₂, we can approximate a rate constant for Equation 8,

(Eq. 8)

by the following equation: 0.8 s¹/5 × 10⁶ M = 1.6 × 10⁵ M¹ s¹ = k₈. Although it is unclear at this point whether ONOO or ONOOH is the actual oxidant, this value is not out of line with ONOO interaction with other metallocomplexes. It has been reported that hemoproteins such as myeloperoxidase react with ONOO with a rate constant of 5 × 10⁵ M¹ s¹ (37) and with Fe²⁺-EDTA with a rate constant of 5700 M¹ s¹ (38), implying that such a rate constant for ferrocytochrome c is not unexpected.

The simulation was thus performed using the following rate equations: d[metHb]/dt = k₅[oxyHb][NO] (where k₅ = 3 × 10⁷ M¹ s¹), d[ferrocytochrome c]/dt = k₆[ferricytochrome c][O₂]  k₈[ONOO][ferrocytochrome c] (where k₈= 1.6 × 10⁵ M¹ s¹), d[NO]/dt = k_NO  k₅[oxyHb][NO]  k₁[O₂][NO] (where k_NO = rate of NO generation), d[O₂]/dt = k_sup  k₆[ferricytochrome c][O₂]  k₁[O₂][NO]  k₇[O₂]² (where k_sup = rate of O₂ generation), d[ONOO]/dt = k₁[O₂][NO]  k₈[ONOO][ferricytochrome c]  k₂[ONOO], d[NO₃] = k₂[ONOO], and d[H₂O₂]/dt = k₇[O₂]².

As depicted in Fig. 9, the simulated data are in good agreement with the data obtained experimentally. The most striking result from these simulations is the finding that ONOO is not formed to any appreciable extent until >90% of the oxyHb is consumed. In addition to the high efficiency of oxyHb in trapping NO, ferricytochrome c plays a significant role in maintaining low O₂ levels. When the steady-state concentrations of NO and O₂ were calculated at flux rates of ~1 µM/min for either radical in the presence of 5 µM oxyHb and 20 µM ferricytochrome c, these concentrations were maintained at <1 nM for NO and <10 nM for O₂. These data suggest that there is very little ONOO formation in the presence of micromolar amounts of both hemoproteins. However, when oxyHb was exhausted, there was significant ONOO formation as the levels of NO rose to ~80 nM, which is sufficiently high for O₂ to be scavenged by NO instead of ferricytochrome c.

Two other conditions that are important for understanding the chemical reactions presented here, i.e. the simultaneous generation of NO and O₂ in the absence of either oxyHb or ferricytochrome c, were simulated using our computer model. As seen from the experimental data, ferrocytochrome c is rapidly oxidized after oxyHb is consumed. Presumably, it is ONOO that is responsible for this reaction as ferrocytochrome c is not readily oxidized by either NO or O₂ under these conditions. When no oxyHb is present, ferrocytochrome c concentrations never rise above 10 nM as determined by simulation (assuming equimolar flux rates of 1 µM/min for NO and O₂, respectively). This is consistent with the observation that after oxyHb was exhausted, oxidation of ferrocytochrome c continued until it was completely converted into ferricytochrome c. How can ONOO formation be so dominant under these conditions (no oxyHb), while in the absence of ferricytochrome c, oxyHb is readily oxidized with little contribution from ONOO? In the absence of ferricytochrome c, the O₂ concentration rises to ~40 nM and slowly decays, which appears to be due to the dismutation kinetics (Equation 7) and reaction with NO (Equation 1) as the oxyHb concentration decreases. Under these conditions, NO is limiting; and hence, the factors that determine the kinetics are k₅[oxyHb] and k₁[O₂]. Since O₂ never rises much above 45 nM (note: in the absence of either ferricytochrome c or oxyHb, the steady-state level is ~30 nM), the NO reaction with oxyHb will dominate. Conversely, in the sole presence of ferricytochrome c (0 µM oxyHb), NO continues to rise to levels of ~80-100 nM, where the NO/O₂ reaction becomes significant (note: under these conditions, O₂ concentrations do not rise above 1 nM). Since ferricytochrome c consumes O₂, NO levels would be expected to rise gradually until levels are reached where NO can compete for O₂. Since the concentration of ferrocytochrome c in the assay is only 20 µM, with a second-order rate constant for O₂ of 2 × 10⁶ M¹ s¹, it is not unexpected that significant ONOO formation will occur at NO concentrations above 10 nM. These two different conditions clearly demonstrate the importance of oxyHb in preventing the formation of the deleterious molecule, ONOO. As the reaction between NO and O₂ is near the diffusion-controlled limit, the importance of the role of hemoglobin in controlling the formation of ONOO is often overlooked.

DISCUSSION

The most important