The Reaction between Nitrite and Oxyhemoglobin

Abstract

The nitrite anion (NO^–₂) has recently received much attention as an endogenous nitric oxide source that has the potential to be supplemented for therapeutic benefit. One major mechanism of nitrite reduction is the direct reaction between this anion and the ferrous heme group of deoxygenated hemoglobin. However, the reaction of nitrite with oxyhemoglobin (oxyHb) is well established and generates nitrate and methemoglobin (metHb). Several mechanisms have been proposed that involve the intermediacy of protein-free radicals, ferryl heme, nitrogen dioxide (NO₂), and hydrogen peroxide (H₂O₂) in an autocatalytic free radical chain reaction, which could potentially limit the usefulness of nitrite therapy. In this study we show that none of the previously published mechanisms is sufficient to fully explain the kinetics of the reaction of nitrite with oxyHb. Based on experimental data and kinetic simulation, we have modified previous models for this reaction mechanism and show that the new model proposed here is consistent with experimental data. The important feature of this model is that, whereas previously both H₂O₂ and NO₂ were thought to be integral to both the initiation and propagation steps, H₂O₂ now only plays a role as an initiator species, and NO₂ only plays a role as an autocatalytic propagatory species. The consequences of uncoupling the roles of H₂O₂ and NO₂ in the reaction mechanism for the in vivo reactivity of nitrite are discussed.

Elevated levels of nitrite in blood can trigger the oxidation of hemoglobin, leading to methemoglobinemia (1, 2). The mechanism of nitrite-dependent oxidation of deoxyhemoglobin (deoxyHb)2 under anaerobic conditions, and its physiological potential in hypoxic vasodilation has recently been established (3–5). However, the mechanism of reaction between oxyhemoglobin (oxyHb) and nitrite has not yet been fully elucidated. Existing mechanistic approaches, starting with the earliest observations of Gamgee in 1868 (6), were recently reviewed (7). The end products of the reaction are methemoglobin (metHb) and nitrate. The kinetics of this reaction, when nitrite is in excess, are complex and exhibit an initial slow phase (often referred to as the lag phase) that accelerates into a rapid phase (referred to as the propagation phase) of oxidation. This reaction profile has been modeled as an autocatalytic, free radical chain reaction. The chain carrier radical and the autocatalytic steps have been variously hypothesized. Doyle et al. (8) found that hydrogen peroxide (H₂O₂) accelerated oxyHb decay, whereas both catalase and superoxide dismutase elongated the half-time of the reaction. Based on these observations they suggested a mechanism according to which H₂O₂ and superoxide played crucial roles. At the same time, Kosaka et al. (9) observed that catalase could extend the slow phase, but observed no effect of superoxide dismutase. In addition, they detected a transient protein free radical by EPR spectroscopy, and proposed a model with the intermediacy of H₂O₂, ferryl forms of hemoglobin (ferrylHb and ferrylHb-radical, analogs of peroxidase compounds II and I, respectively), and a chain propagator function of nitrogen dioxide free radical (NO₂). In agreement with this model, we have recently confirmed the formation of ferrylHb as a reaction intermediate and shown that a diffusible oxidant is formed in the nitrite/oxyHb reaction consistent with the formation of nitrogen dioxide (10). Later, Lissi (11) proposed a scheme with an autocatalytic step allowing the acceleration of the reaction as a function of time. However, whereas the Kosaka/Lissi mechanism exhibits the qualitative aspects of nitrite-mediated oxyHb oxidation, its ability to quantitatively describe the kinetics of the nitrite/oxyHb reaction have not previously been examined.

In the current study we show, using kinetic simulation that the previous mechanism cannot quantitatively describe the reaction profile and so must be regarded as deficient. We have analyzed this deficiency, both by experimental and kinetic simulation approaches to develop a model for the reaction that is consistent with experimental data. This model is based on the original work of Doyle et al. (8), Kosaka et al. (9), and Lissi (11), a recent publication by Goldstein et al. (12), original experimentation, and a modicum of speculation. The salient feature of the new model is that H₂O₂ is only generated in the initiation step of the reaction and NO₂ is only generated during the propagation/autocatalytic phase of the reaction. Such uncoupling of the roles of H₂O₂ and NO₂ in the reaction mechanism is extremely important in vivo, where the low nitrite to Hb ratio makes it highly unlikely that the reaction follows autocatalytic kinetics.3 Consequently, the major products of the nitrite/oxyHb reaction in vivo will be nitrate, metHb, and H₂O₂, the former of which is benign, and the latter two of which can be reduced by metHb reductase and dismutated by catalase, respectively. We conclude that the oxidation of oxyHb by nitrite occurs through an autocatalytic mechanism that relies exclusively on NO₂ as the autocatalyst.

EXPERIMENTAL PROCEDURES

Materials—OxyHb was prepared from fresh human blood according to a published method (13). After preparation, hemoglobin aliquots were frozen in liquid nitrogen and stored at –80 °C. All chemicals were purchased from Sigma, except 2-(carboxyphenyl)-4,5-dihydro-4,4,5,5-tetramethyl-1H-imidazolyl-1-oxy-3-oxide (CPTIO), which was obtained from Cayman Chemical (Ann Arbor, MI). Human serum albumin (99% pure), catalase from bovine liver, and superoxide dismutase from bovine erythrocytes were used without any further purification. All experiments were carried out in 10 mm sodium phosphate buffer (PB), at pH 7.40. The metal ion chelator diethylenetriamine pentaacetic acid was added to all buffers at a concentration of 1 mm.

Sample Preparation—Samples were prepared and examined under either fully aerobic or partially deoxygenated conditions. Aerobic measurements were done in open cuvettes lightly covered by parafilm to allow gas exchange between solution and air. For partial deoxygenation, the solution of oxyHb was placed into a long-neck quartz cuvette closed with an air-tight rubber stopper, thin input and exhaust needles were pierced through the stopper, and sample was deoxygenated using gentle stream of either helium or a mixture of 10% H₂, 90% N₂. Deoxygenation was stopped and needles were withdrawn after the first slight changes in the oxyHb spectrum. The solution of NaNO₂ in buffer was deoxygenated separately and an appropriate amount was added to the deoxygenated oxyHb solution using a gas-tight Hamilton syringe. For some experiments, hemoglobin was fully deoxygenated, and known amounts of pure oxygen were added back to the sealed cuvette.

Reaction Conditions—The oxyHb concentration in all experiments was kept constant at 30 μm. Appropriate amounts of stock solution of NaNO₂ (100 mm, in PB, aerobic or deoxygenated) were injected into oxyHb solution and UV-visible spectra (HP 8453) were continuously acquired between 450 and 700 nm. All experiments were performed at 37 °C.

Multilinear Regression Analysis (MLR)—Deconvolution of spectra into individual species was accomplished with MLR, using a set of pure spectra of all components as a basis.

Electron Paramagnetic Resonance Spectroscopy (EPR)—EPR spectra of ferryl hemoglobin radical were recorded from the reaction of 2.3 mm oxyHb and 1 mm NaNO₂ at room temperature. The reaction mixture was transferred to a quartz capillary and the EPR spectrum was recorded using a Bruker EMX spectrometer with the following conditions: microwave power, 20 milliwatts; microwave frequency, 9.8 GHz; modulation frequency, 100 KHz; modulation amplitude, 5 G; time constant, 81.9 ms; and conversion time, 10.2 ms.

Kinetic Simulation—Kinetic models were fitted to the experimental data of oxyHb consumption using the Gepasi 3.0 program.

RESULTS

Effect of Nitrite and OxyHb Concentration on Oxidation Kinetics—Fig. 1A illustrates typical changes in the UV-visible spectra of oxyHb during oxidation by nitrite. The decrease in absorbance between 520 and 590 nm and the increases at 500 and 630 nm are indicative of the conversion of oxyHb to metHb. To fully characterize these changes, data such as that in Fig. 1A were analyzed by MLR using the basis spectra shown in Fig. 1B. These spectra include those of metHb, oxyHb, and ferrylHb, as well as metHb-NO^–₂ that may form due to the weak association of nitrite with metHb. The results of the regression analysis for the data in Fig. 1A, at all time points, are shown in Fig. 1C and illustrate the well established strongly biphasic kinetics of oxyHb oxidation, consisting of a slow “lag” phase that abruptly accelerates to a rapid “propagation” phase in which the reaction obtains its maximum velocity. This kinetic profile is strongly indicative of autocatalysis. In addition, Fig. 1C illustrates that the spectral changes in Fig. 1A are consistent with the intermediate formation of ferrylHb that achieves a measurable level during the rapid propagation phase. This is consistent with earlier observations using oxymyoglobin (14) and oxyhemoglobin (15). Finally, Fig. 1C also illustrates the formation of metHb-NO^–₂, contemporary with the formation of metHb, due to the equilibrium association of excess nitrite with the metHb product. Fig. 1D shows the residual of the fit, showing very little spectral intensity that is not associated with the four species studied.

FIGURE 1.

Multilinear regression analysis of experimental data. OxyHb (30 μm), was incubated with NaNO₂ (0.6 mm), in phosphate buffer, pH 7.40, at 37 °C and UV-visible spectra were obtained between 450 and 700 nm. A, changes in the absorption spectrum of oxyhemoglobin after addition of NaNO₂. Arrows show the direction of change of the 500-, 542-, 577-, and 630-nm peaks. B, complete set of basis spectra of individual hemoglobin species used in MLR. C, results of MLR analysis of the data in A using the basis spectra in B. D, residual obtained from the MLR analysis of the data in A.

We examined the influence of nitrite concentration on the duration of the lag time and on the rate of the propagation phase (Table 1). The decay rate of oxyHb and the length of the lag time strongly correlated with the nitrite concentration. We have previously demonstrated that at lower nitrite:oxyHb ratios (4:1 or lower) the autocatalytic character of the reaction is lost.³ Consequently it appears as though the initiation reaction needs to proceed above a threshold rate before the autocatalyst builds up to a level that can catalyze the propagation reaction.

TABLE 1.

Lag time and rate of oxyHb oxidation at different experimental conditions [oxyHb]₀ = 30 μm at 37 °C, pH 7.40

	Lag timea		OxyHb decay rate in the fast phasea
	Oxygen	No oxygen	Oxygen	No oxygen
mm	s			μm s-1
0.25	937 ± 5.7		0.042 ± 0.0003
0.375	395 ± 15	675 ± 130	0.087 ± 0.0035	0.072 ± 0.0086
0.6	124 ± 14	266 ± 4	0.17 ± 0.025	0.11 ± 0.0075
1.0	69 ± 2.5	107 ± 8.2	0.45 ± 0.020	0.29 ± 0.044
2.5	19 ± 0.46	22.3 ± 0.95	1.33 ± 0.026	1.02 ± 0.15

^aValues are mean ± S.E. (n = 3).

We also examined the effect of oxyHb concentration on the reaction kinetics. Fig. 2A shows that an increase of the initial concentration of oxyHb at unchanged NaNO₂ concentration resulted in accelerated oxidation; the lag phase became shorter, whereas the rate of the fast phase increased. Earlier, Pietraforte et al. (16) showed that at a fixed nitrite/heme molar ratio the reciprocal of the time required to achieve 50% oxyHb oxidation was an exponential function of hemoglobin concentration.

FIGURE 2.

Effect of initial concentration of oxyHb on the kinetics. A, various concentrations of oxyHb were incubated with NaNO₂ (0.6 mm) in phosphate buffer, pH 7.40, at 37 °C. The oxyHb decay was followed spectrophotometrically in a 1-mm path length cuvette. B, various concentrations of oxyHb were incubated with NaNO₂ (0.06 mm) in phosphate buffer, pH 7.40, at 37 °C. The oxyHb decay was followed spectrophotometrically, and the initial rates of the decay were calculated. Values represent mean ± S.E., n = 3.

We determined a rate constant of 0.33 m^–1 s^–1 for the initial step of the reaction from initial rates at fixed NaNO₂ concentrations and at low oxyHb concentration; a regime in which autocatalysis is not apparent (Fig. 2B). This value is in fairly good agreement with the value of 0.21 m^–1 s^–1 that was obtained also under non-autocatalytic conditions, but with fixed oxyHb concentration and variable nitrite concentration.³ Nitrite to Nitrate Stoichiometry—Using ion-pair high performance liquid chromatography (17, 18) we observed that the stoichiometry of the overall reaction conforms to a 1:1 conversion of nitrite to nitrate throughout the time course of the reaction (data not shown). This is consistent with the earlier report of Kosaka et al. (19).

Effect of pH—The nitrite-induced oxidation of oxyHb was previously found to be susceptible to pH (19), but the dependence has not been fully investigated. We examined the effect of pH on the reaction profile in a narrow range centered on pH 7.40 (Fig. 3A). A pH change as small as 0.04 caused a remarkable difference in lag time (Fig. 3A, inset). The dramatic effect of pH on the lag time of oxidation suggest that nitrous acid rather than nitrite is the major oxidizing species (20).

FIGURE 3.

Effect of pH and solution oxygen concentration on the decay of oxyHb. A, oxyHb (30 μm) was incubated with NaNO₂ (0.3 mm) in phosphate buffer, pH 7.29–7.44 (pH values are marked in the plot), at 37 °C. Inset, lag time dependence as a function of pH. B, oxyHb (30 μm) was incubated with NaNO₂ (1 mm) in phosphate buffer, pH 7.40, at 37 °C in the presence of oxygen partial pressures 84, 127, 152, and 760 mm Hg as indicated. For both panels, UVvisible spectra were collected and subjected to MLR analysis as described in the legend to Fig. 1. For clarity, only oxyHb decay is shown.

Effect of Oxygen—As physiological oxygen levels are significantly lower than the concentration of oxygen in air-exposed solutions, we examined the effect of oxygen concentration on nitrite-dependent oxyHb oxidation. At the lowest oxygen concentration used, oxyHb was still >99% oxygen saturated and deoxyHb could not be detected by spectral deconvolution. As shown in Fig. 3B, the decrease in oxygen concentration resulted in an increase in the length of the lag phase of the reaction. However, if oxygen concentration was increased above atmospheric levels (100% oxygen) the rate of reaction was not further increased. This suggests that atmospheric oxygen pressure is high enough to saturate this effect. These results demonstrate that the concentration of oxygen in the solution surrounding the hemoglobin plays an important role in determining the length of the lag phase during the oxidation process. The effect of reducing oxygen concentration on the lag time and propagation is shown in Table 1.

Influence of Catalase and Superoxide Dismutase—It has previously been demonstrated that catalase, added at the onset of the reaction, slows the rate of oxyHb oxidation by nitrite (8, 9). This is confirmed in Fig. 4A, in which catalase inhibits the reaction rate in a concentration-dependent manner. Interestingly, human serum albumin (1 mg/ml, an equivalent protein level to 50 kilounits of catalase) also had a slight inhibitory effect, suggesting a small nonspecific effect from the presence of protein. In contrast, when catalase is added toward the end of the lag phase (Fig. 4B) there is little to no effect on the progression of oxidation and the acceleration into the propagation phase is not hindered. Only at the highest concentration of catalase is a slight diminution of the rate observed. The values of lag time and rates of the fast phase (mean ± S.E.) are reported in Table 2. Incubation of nitrite with catalase had no effect of its activity as measured by following oxygen evolution from hydrogen peroxide using an oxygen electrode (data not shown). These data suggest that H₂O₂ formation is essential for the initiation of oxyHb oxidation, but is not a critical intermediate in the autocatalytic chain reaction. Superoxide dismutase, at concentrations similar to those used by Doyle et al. (8), had little effect on hemoglobin oxidation (Fig. 4C) when added at the start of the reaction. This is in agreement with the study of Kosaka et al. (9) but not Doyle et al. (8). In addition, superoxide dismutase had no effect when added at the end of the lag phase (Fig. 4C), suggesting that superoxide plays no role in either the initiation or propagation steps of this reaction.

FIGURE 4.

Effect of catalase and superoxide dismutase on the consumption of oxyHb. OxyHb (30 μm) was incubated with NaNO₂ (1 mm), in phosphate buffer at pH 7.40 and 37 °C. A, catalase (•, 0; ○, 1; ▾, 10; Δ, 50 kilounits/ml), or human serum albumin (•, 1 mg/ml) was added at the beginning of the reaction. B, catalase (•, 0; ○, 1; ▾, 10; Δ, 50 kilounits/ml) is indicated by the arrow. C, superoxide dismutase (•, 0; ○, 1; ▾, 5 kilounits/ml) was added at the beginning of the reaction, or at the point indicated by the arrow (Δ, 1; □, 5 kilounits/ml). For all panels, UV-visible spectra were collected and subjected to MLR analysis as described in the legend to Fig. 1. For clarity, only oxyHb decay is shown.

TABLE 2.

Effect of catalase and superoxide dismutase (SOD) on the lag time and the rate of oxyHb decay [oxyHb]₀ = 30 μm, mm at 37 °C, in PB, pH 7.40

Enzyme	Lag timea	OxyHb decay rate in the fast phasea
kU ml-1	s	μm s-1
No additive	66 ± 2.8	0.33 ± 0.024
HSA	120 ± 28	0.25 ± 0.015
Catalase added at the beginning
1	69 ± 5.4	0.35 ± 0.028
10	94 ± 5.4	0.32 ± 0.035
50	325 ± 75	0.11 ± 0.028
Catalase added to the fast phase
1	71 ± 4.5	0.36 ± 0.012
10	59 ± 4.1	0.34 ± 0.018
50	65 ± 3.2	0.30 ± 0.059
SOD added at the beginning
1	63 ± 4.7	0.36 ± 0.012
2	67 ± 4.4	0.38 ± 0.0026
5	65 ± 3.9	0.34 ± 0.0017
SOD added to fast phase
1	62 ± 2.1	0.37 ± 0.014
2	58 ± 4.2	0.38 ± 0.021
5	54 ± 4.2	0.38 ± 0.022

^aValues are mean ± S.E. (n = 3).

Influence of CPTIO—Our experimental results on the effect of catalase and superoxide dismutase indicate that superoxide is not a major propagating species and that hydrogen peroxide is mainly involved in the initiation steps of the reaction. To examine if NO₂ radical is an important intermediate in the reaction mechanism, we examined the effect of CPTIO. Goldstein and co-workers (21) have recently shown that the nitronyl nitroxide CPTIO reacts with NO₂ with a rate constant of 1.5 × 10⁷ m^–1 s^–1 yielding nitrite and CPTIO⁺ cation. The reaction is reversible with an equilibrium constant of 145. We observed that when CPTIO is added at the beginning to the oxyHb-nitrite reaction, the autocatalytic phase was suppressed (dotted line in Fig. 5A). In addition, when CPTIO was added during the autocatalytic decay of oxyHb, the oxidation was immediately stopped (Fig. 5A, dashed line). This indicates that CPTIO can scavenge intermediates in both the initiation and propagation phases of the reaction, which include NO₂, ferryl heme, and protein radical species. Possible interactions between CPTIO and ferryl heme species were examined. Fig. 5B demonstrates that CPTIO addition into a solution of ferrylHb did not result in a loss of the ferryl species as determined by spectrophotometry. Also, the decay of the EPR signal of the hemoglobin protein radical formed from the nitrite-oxyHb reaction did not change in the presence of CPTIO (Fig. 5C), although the EPR spectrum of CPTIO was lost during this reaction. This suggests that CPTIO is oxidized during the reaction, but that it does not directly affect the protein radical signal. The above results diminish the potential role of CPTIO as a scavenger of ferryl heme and protein radical, and favor the pivotal importance of NO₂ in autocatalysis.

FIGURE 5.

Effect of CTPIO on the kinetics of oxyHb. A, oxyHb (30 μm) was incubated with NaNO₂ (1 mm)in phosphate buffer at pH 7.40 and 37 °C. CPTIO (60 μm) was added at the start of the reaction (dotted line)or during the fast phase (dashed line, insertion of CPTIO is denoted with an arrow). Control experiment without CPTIO (solid line). B, spectrum of ferryl species derived from metHb and excess H₂O₂ (solid line), and the remaining spectrum after addition of CPTIO (80 μm) and subtraction of CPTIO spectrum. C, EPR spectrum of ferryl radical recorded 5 min after addition of 2 mm NaNO₂ to 2.3 mm oxyHb (solid line), after mixing 2 mm NaNO₂ with 2.3 mm oxyHb and CTPIO (3 μm)(dashed line), 5 min later (dotted line).

Kinetic Simulation—Complex kinetic reaction schemes are somewhat intractable to both analytical solution, and intuitive understanding, therefore we have employed kinetic simulation methods to examine if the experimental reaction profile can be explained by previously published models. We began by taking the model of Kosaka et al. (9) as the basic framework for the simulation with the inclusion of the autocatalytic step of Lissi (11), as this construction best qualitatively describe the experimental observations made here and those described in the literature. The initiation reaction (Equation 1)

(Eq. 1)

(Eq. 2)

(Eq. 3)

(Eq. 4)

(Eq. 5)

(Eq. 6)

(Eq. 7)

involves the two-electron reduction of bound oxygen to form hydrogen peroxide. One electron is ultimately derived from the ferrous heme and the other from nitrite, forming ferric heme and nitrogen dioxide, respectively. The rate constant for this reaction has been estimated by us to be 0.33 m^–1 s^–1. Equation 2, the reaction between hydrogen peroxide and ferric hemoglobin, is part of both the initiation and the propagation processes. This reaction forms ferrylHb-radical, which is proposed to undergo two sequential one-electron reductions, by nitrite, to reform ferric hemoglobin. These reactions, shown in Equations 3 and 4, each generate nitrogen dioxide and are branching steps in which one ferryl-radical hemoglobin molecule gives rise to two nitrogen dioxide radicals. The reaction in Equation 5 regenerates ferric Hb and hydrogen peroxide from the reaction between nitrogen dioxide and oxyHb, additionally forming the stable end-product, nitrate. It is this reaction that completes the cycle, allowing one nitrogen dioxide to ultimately generate two nitrogen dioxides and thus provide a mechanism for autocatalysis. The radical chain reaction is terminated by the dimerization of nitrogen dioxide to form dinitrogen tetraoxide, and the subsequent hydrolysis of this intermediate to nitrite and nitrate (Equations 6 and 7).

This series of equations has hydrogen peroxide as an essential component of the initiation process, but its role in the propagation phase depends largely upon the fate of the hypothesized “Hb³⁺ HOOH” product from Equation 5. If this product were to dissociate then H₂O₂ would be an essential component of the propagation and autocatalytic elements of the reaction scheme. Kinetic simulations of this model (Model 1 in Table 3) provided a reasonable fit to the data (Fig. 4A), but only when the rate constant for Equation 2 was allowed to vary in the fitting procedure. The best fit to the data were obtained with a value for this rate constant that was at least 3 orders magnitude larger than the published value of ∼500 m^–1 s^–1 (22). It became clear from these simulations that H₂O₂ could play no role in the propagation reaction as reaction 2 is simply too slow. However, if reaction 5 occurs through a mechanism that does not involve the release of H₂O₂ it would be possible for the propagation cycle to bypass reaction 2 and therefore to not be limited by this slow step.

TABLE 3.

The models used for kinetic simulation of experimental data

Model	Equations
1	1, 2, 3, 4, 5, 6, 7
2	1, 2, 3, 4, 8, 9 and 6, 7
3	11, 12, 13, 14, 2, 3, 4, 8, 9, 6, 7

The reaction of NO₂ with hemeproteins has been recently addressed by Goldstein et al. (12), who showed that NO₂ reacts with oxyHb through the intermediate formation of a ferricperoxynitrate complex to form the ferrylHb-radical without either the formation or release of H₂O₂ (Equations 8 and 9).

(Eq. 8)

(Eq. 9)

Incorporation of these equations into the original model therefore provides a way for the propagation step to be no longer limited by the rate of Equation 2.

Kinetic simulations of the reaction scheme in which Equations 8 and 9 replace Equation 5 (Model 2 in Table 3) resulted in a model that more closely approximated the data (Fig. 4A) using rate constants that did not grossly violate previously published values (data not shown), although it was unable to accurately simulate the abrupt acceleration of the reaction into the propagation phase. However, this model is fundamentally inconsistent with experimental data, as it predicts that autocatalysis will occur even in the absence of H₂O₂. If the rate constant of Equation 2 is set to zero in the simulation, or an additional step is added to represent the presence of catalase (i.e. rapid consumption of H₂O₂) it had very little effect on the time course. The reason for this is that the NO₂ generated in Equation 1 can itself initiate the radical chain reaction through Equations 8 and 9 much more rapidly than H₂O₂ through Equation 2. Consequently, Equation 2 is bypassed completely and the reaction becomes H₂O₂-independent.

As neither the original nor the modified scheme was able to describe the experimental data with much fidelity a more serious reappraisal of the mechanism was necessary. Our simulations indicated that any mechanism that generates NO₂ in the initiation step of the reaction is fundamentally incompatible with the experimental data. The initiation step (Equation 1) is clearly not a simple bi-molecular step as it involves two one-electron transfers to oxygen to generate H₂O₂. We hypothesize an alternative equation for the initiation reaction (Equation 10) that forms H₂O₂ but not NO₂.

(Eq. 10)

Whereas again, this is clearly not a single fundamental step, it is possible to decompose this into the four elementary reactions shown in Equations 11, 12, 13, 14.

(Eq. 11)

(Eq. 12)

(Eq. 13)

(Eq. 14)

This hypothetical scheme postulates that the initial reaction between nitrite and oxyHb is an addition reaction of nitrite to bound oxygen to form a ferrous-peroxynitrate intermediate (Equation 11). This complex oxidizes nitrite to form nitrate (Equation 12), forming a ferrous peroxynitrite intermediate that we postulate to be reduced by the heme iron to form H₂O₂ and NO (Equation 13). The nascent NO is than rapidly dioxygenated to nitrate by oxyHb (Equation 14). Whereas this scheme is currently hypothetical, its incorporation into the kinetic model in place of Equation 1 (Model 3 in Table 3) now provides a scheme that corresponds to experimental traces with high fidelity (Fig. 6A). In Fig. 6B we show kinetic simulations of experimental data using Model 3 at various nitrite concentrations. Table 4 details the rate constants used in the simulations and the derived values generated from the fit of the simulations to the experimental data. The rate-limiting step of the initiation reaction (Equation 11) has a rate constant slightly higher than the value of 0.33 m^–1 s^–1 that we have determined from low-nitrite (nonautocatalytic) experiments. The simulated rate constant of the reaction of metHb and H₂O₂ (101 m^–1 s^–1) is close to published values of 200 m^–1 s^–1 for hemoglobin (25), and 510 m^–1 s^–1 for myoglobin (26). The rate constant of 6.9 × 10⁵ m^–1 s^–1 of the chain branching step (Equation 8) was lower than the value of Goldstein et al. (12) found for myoglobin (4.5 × 10⁷ m^–1 s^–1).

FIGURE 6.

Simulation of the reaction between oxyHb and nitrite using three different models. A, the three kinetic models described in Table 3 were fitted against experimental data obtained from the incubation of oxyHb (30 μm) with nitrite (1 mm) in phosphate buffer at 37 °C. The curves represent the change in oxyHb concentration as a function of time for experimental data (solid line), Model 1 (-·· -·· -), model 2 (- -), and model 3 (dotted line). B, Model 3 was fitted against experimental data obtained after incubating oxyHb (30 μm) in phosphate buffer at 37 °C, with a range of initial nitrite concentrations (0.375, 0.6, 0.8, 1.0, 1.5, and 2.0 mm as shown on the figure). Curves represent experimental data (solid line) and fit (dotted line). The rate constants used for or derived from these fits are given in Table 4.

TABLE 4.

Average rate constants derived from the fits shown in Fig. 6 and two identical sets of data

Reaction	Rate constant ± S.E.	Source
	m-1s-1 or s-1
2	101 ± 56	Derived from simulation
3	750	Ref. 23
4	6.9 × 105	Derived from simulation
6	4.5 × 108 and 6.9 × 103	Ref. 24
7	1000	Ref. 24
8	5.4 × 105 ± 1.8 × 105	Derived from simulation
9	Fast (108)	Ref. 12
11	1.01 ± 0.45	Derived from simulation
12	350 ± 220	Derived from simulation
13	Fast first order (1000)
14	8.9 × 107	Derived from simulation

DISCUSSION

The nitrite-mediated oxidation of oxyHb at high nitrite concentrations exhibits biphasic kinetics indicative of an autocatalytic reaction. It has been proposed that this reaction occurs through a complex series of steps involving the formation of ferryl oxidation states of the heme iron and protein-free radicals. The presence of protein radicals has previously been detected by direct EPR spectroscopy, spin-trapping, and immunological methods (9, 10). We show here, in agreement with our previous studies using myoglobin (14), that the spectral changes that occur during the reaction of oxyHb with nitrite are consistent with the intermediacy of ferrylHb species during the progression of the reaction. In addition, we show that at high nitrite concentrations, the product of the reaction is a mixture of both metHb and metHb-NO^–₂.

The nitrite-dependent oxidation of oxyHb is thought to occur by a free radical chain reaction mediated by NO₂ (9), superoxide radical (8), or with the participation of both. Because addition of superoxide dismutase did not affect the reaction kinetics, we conclude that superoxide does not play a crucial intermediary role. Although it is not clear why other investigators have observed an inhibitory effect of superoxide dismutase (8, 27), one possibility is that superoxide dismutase preparations may have been contaminated with catalase.

The inhibitory effects of catalase on nitrite-mediated oxyHb oxidation have been well established and are confirmed here. In these experiments we employed relatively high catalase concentrations to overcome any potential nitrite-catalase interaction (28), and also the possible competition between catalase and metHb (29). Interestingly, catalase was only effective when added at the start of the reaction but not when added toward the end of the lag period. This suggests that H₂O₂ is crucial as an element of chain initiation, but is less important in the propagation and chain branching steps of the reaction.

The other intermediate that has been proposed to play a role in both the initiation and propagation of oxyHb oxidation by nitrite is NO₂. It is much more challenging to firmly establish the role of NO₂ in this reaction mechanism, but we have previously invoked its role in the oxidation of bystander proteins that occurs during this reaction (but not during the reaction between hydrogen peroxide and hemoglobin) (10). In this study we have attempted to intercept NO₂ using the nitronyl nitroxide CPTIO. Whereas this compound is better known as a nitric oxide trap, a recent study by Goldstein et al. (21) indicates that it reacts with NO₂ with a second-order rate constant of 1.5 × 10⁷ m^–1 s^–1, more than 3 orders of magnitude greater than its reaction with nitric oxide. Whereas CPTIO, at any viable concentration, would never compete with oxyHb as a nitric oxide trap, it may compete with other NO₂ targets. Experimentally, we show that CPTIO dramatically inhibits oxyHb oxidation whether introduced at the start of the reaction or halfway through the propagation phase. This, with evidence that CPTIO does not greatly interfere with either ferrylHb or protein radical, strongly suggests an intermediary role of NO₂ in the propagation phase of the reaction.

Complex autocatalytic reaction mechanisms are difficult to intuitively understand and so kinetic simulation provides an essential tool in translating intuition into mathematical precision. We show here that the previously published kinetic models of the oxyHb/nitrite reaction, while qualitatively describing an autocatalytic mechanism, fail to describe quantitatively the reaction kinetics. A significant amount of experimental evidence exists to support some of the major features of these earlier mechanisms, suggesting that a multistep heme redox cycle involving ferrylHb and ferrylHb-radical intermediates is the central “engine” that drives the autocatalytic acceleration of the reaction kinetics. The discrepancies that kinetic simulation analysis reveal are that (i) the reaction between H₂O₂ and metHb is too slow to participate in the autocatalytic redox cycle and can only be a part of the initiation reaction and (ii) as a consequence of point (i), the production of NO₂ in the initiation step would negate any role for H₂O₂ in the entire reaction scheme. Consequently, the roles of H₂O₂ and NO₂ have to be uncoupled during the reaction time course, with H₂O₂ only playing a role during initiation and NO₂ only playing a role during the autocatalytic propagation steps of the reaction. The former can be easily rationalized by incorporating the reactions described by Goldstein et al. (Equations 8, 9, and 12), which allow the direct conversion of oxyHb to ferryl-radical Hb without the intermediacy of H₂O₂. The removal of NO₂ from the initiation step requires a re-evaluation of the initial reaction mechanism. We have proposed an alternative mechanism for the initiation reaction (Equations 11, 12, 13, 14) based on the need to eliminate NO₂ from this step. Whereas these steps are currently conjectural, their incorporation into the reaction mechanism gives a kinetic scheme that is able to quantitatively describe experimental data.

Scheme 1 presents a diagram showing the relationship between the initiation and autocatalytic propagation steps of the reaction and the role of H₂O₂ and NO₂ in these steps. The reaction between nitrite and oxyHb generates H₂O₂ and metHb, which can react to form ferrylHb-radical. These steps represent the initiation of the chain reaction. The ferrylHb-radical can then generate two molecules of NO₂ from nitrite as it is reduced back to metHb. Each NO₂ can then continue the reaction chain by oxidizing oxyHb to ferrylHb-radical thus providing for autocatalytic propagation.

SCHEME 1.

The mechanism of the nitrite-induced oxidation of oxyHb. The roles of hydrogen peroxide and nitrogen dioxide are uncoupled from each other, with hydrogen peroxide participating in the initiation and nitrogen dioxide driving the autocatalytic propagation steps. Stable end products (metHb and nitrate) are produced in both phases.

These modifications of the reaction mechanism have significant implications for the importance of the nitrite/oxyHb reaction in vivo and its relationship with the nitrite/deoxyHb reaction during cycles of oxygenation and deoxygenation. First, evidence suggests that the nitrite/oxyHb reaction will never accelerate autocatalytically in vivo due both to the very low nitrite/oxyHb ratio and also to the presence of antioxidants such as catalase and ascorbate.³ However, oxidative damage could still be caused by the concomitant production of NO₂ in the initiation step. If NO₂ is not formed during initiation, then this phase of the reaction is even more benign as only H₂O₂ is formed, which can be easily dealt with by catalase at no energetic cost to the cell. Second, it has been established that the presence of deoxyHb is able to inhibit the progress to autocatalysis (15). In this study we show that decreasing solution oxygen, to levels that show no measurable increase in deoxyHb, is able to prolong the lag time of the nitrite/oxyHb reaction. This is in agreement with a previous study by Wade and Castro (30) and newer data by Grubina et al. (15). However, performing the experiment in 100% oxygen did not affect the kinetics of reaction as compared with room oxygen levels. This result suggests that it is not oxygen per se that is accelerating the reaction, but it is the low levels of deoxyHb generated (<1% and undetectable using spectral deconvolution) as oxygen tension drops below normal atmospheric levels that is able to inhibit the kinetics of the reaction. Whereas this inhibitory action is not fully understood it is possibly related to the formation of nitric oxide from the reaction between nitrite and deoxyHb, which occurs most rapidly on deoxygenated “R”-state hemes, reacting with ferryl hemoglobin and thus preventing NO₂ formation (31).

Because the nitrite level in red blood cells and in plasma is 300–500 nm (32, 33) under physiological conditions, there is little chance of autocatalysis and NO₂ formation. Consequently the rate of metHb formed will be limited by the slow first step of the reaction scheme with a rate constant of about 0.2–0.4 m^–1 s^–1. The metHb formed from this reaction is likely a small percentage of the net metHb formation due to processes such as hemoglobin autoxidation and NO scavenging. In addition, methemoglobin reductase will assist in keeping the metHb level low. Titov and Petrenko (2) could not induce considerable hemoglobin oxidation in erythrocyte lysate by submillimolar levels of nitrite without addition of high levels of H₂O₂. On the other hand, as a consequence of poisoning, the nitrite level in plasma may reach 100–400 μm resulting in severe methemoglobinemia (1, 34). The gastrointestinal absorption of nitrite is quick and complete, following by a rapid NO₂ decay accompanied with fast development of the maximal metHb level in plasma (35). This suggests that nitrite levels in this range may overwhelm the antioxidant mechanisms of the red cell, resulting in an autocatalytic reaction. Although the major toxic effects of nitrite poisoning is metHb formation, NO₂ formed during the propagation phase may lead to protein damage through protein radical formation and tyrosine nitration (10, 36–39) as well as lipid peroxidation and the potential formation of nitrated lipids (40).

In summary, we have established that the kinetics of the nitrite-dependent oxidation of oxyHb at the same initial oxyHb concentration is determined by the extent of nitrite excess. The presence of catalase highly increases the lag time, but does not affect the reaction rate in the autocatalytic phase. Superoxide dismutase has no influence on the kinetics. Increased deoxyHb resulting from deoxygenation may trigger competitive processes, and decelerate the oxidation. We propose a mechanism, which describes the autocatalytic oxidation of oxyHb by nitrite, and according to kinetic simulations, is consistent with all our experimental data. The important novel feature of this model is that H₂O₂ is only formed during initiation and NO₂ is only formed during propagation.