Concerted Nitric Oxide Formation and Release from the Simultaneous Reactions of Nitrite with Deoxy- and Oxyhemoglobin*

Recent studies reveal a novel role for hemoglobin as an allosterically regulated nitrite reductase that may mediate nitric oxide (NO)-dependent signaling along the physiological oxygen gradient. Nitrite reacts with deoxyhemoglobin in an allosteric reaction that generates NO and oxidizes deoxyhemoglobin to methemoglobin. NO then reacts at a nearly diffusion-limited rate with deoxyhemoglobin to form iron-nitrosyl-hemoglobin, which to date has been considered a highly stable adduct and, thus, not a source of bioavailable NO. However, under physiological conditions of partial oxygen saturation, nitrite will also react with oxyhemoglobin, and although this complex autocatalytic reaction has been studied for a century, the interaction of the oxy- and deoxy-reactions and the effects on NO disposition have never been explored. We have now characterized the kinetics of hemoglobin oxidation and NO generation at a range of oxygen partial pressures and found that the deoxy-reaction runs in parallel with and partially inhibits the oxy-reaction. In fact, intermediates in the oxy-reaction oxidize the heme iron of ironnitrosyl-hemoglobin, a product of the deoxy-reaction, which releases NO from the iron-nitrosyl. This oxidative denitrosylation is particularly striking during cycles of hemoglobin deoxygenation and oxygenation in the presence of nitrite. These chemistries may contribute to the oxygen-dependent disposition of nitrite in red cells by limiting oxidative inactivation of nitrite by oxyhemoglobin, promoting nitrite reduction to NO by deoxyhemoglobin, and releasing free NO from iron-nitrosyl-hemoglobin.

Recent studies reveal a novel role for hemoglobin as an allosterically regulated nitrite reductase that may mediate nitric oxide (NO)-dependent signaling along the physiological oxygen gradient. Nitrite reacts with deoxyhemoglobin in an allosteric reaction that generates NO and oxidizes deoxyhemoglobin to methemoglobin. NO then reacts at a nearly diffusion-limited rate with deoxyhemoglobin to form iron-nitrosyl-hemoglobin, which to date has been considered a highly stable adduct and, thus, not a source of bioavailable NO. However, under physiological conditions of partial oxygen saturation, nitrite will also react with oxyhemoglobin, and although this complex autocatalytic reaction has been studied for a century, the interaction of the oxy-and deoxy-reactions and the effects on NO disposition have never been explored. We have now characterized the kinetics of hemoglobin oxidation and NO generation at a range of oxygen partial pressures and found that the deoxy-reaction runs in parallel with and partially inhibits the oxy-reaction. In fact, intermediates in the oxy-reaction oxidize the heme iron of ironnitrosyl-hemoglobin, a product of the deoxy-reaction, which releases NO from the iron-nitrosyl. This oxidative denitrosylation is particularly striking during cycles of hemoglobin deoxygenation and oxygenation in the presence of nitrite. These chemistries may contribute to the oxygen-dependent disposition of nitrite in red cells by limiting oxidative inactivation of nitrite by oxyhemoglobin, promoting nitrite reduction to NO by deoxyhemoglobin, and releasing free NO from iron-nitrosyl-hemoglobin.
The reaction of nitrite with deoxyhemoglobin (referred to as the "deoxy-reaction") was first fully characterized by Brooks in 1937 (26) and by Doyle et al. in 1981 (27). As shown in Equation 1, in an anaerobic environment nitrite is reduced to NO, whereas deoxyhemoglobin is oxidized to methemoglobin. This mechanism provides for effective pH and oxygen sensing, as both deoxyhemoglobin and a proton are required for the reaction (17,28).
We have also shown that the deoxy-reaction is allosterically regulated with maximal rates of nitrite reduction around the P 50 of hemoglobin (7,17). This occurs secondary to the higher maximal nitrite reductase activity of R-state hemoglobin (the lower redox potential of the heme tetramer in the R-state is associated with a higher bimolecular rate constant for nitrite reduction) and the amount of T-state deoxyhemoglobin available for nitrite binding (17).
Notably, NO has very high affinity for the ferrous heme and binds deoxyhemoglobin with a k a of 2 ϫ 10 7 M Ϫ1 s Ϫ1 (29). NO, therefore, reacts with red cell deoxyhemoglobin at a nearly diffusion-limited rate to yield iron-nitrosyl-hemoglobin (Equation 2) (30), which has always been considered a highly stable adduct with a k d of 3 ϫ 10 Ϫ3 s Ϫ1 (29). * This work was supported part by National Institutes of Health Grants HL58091 (DK-S) and HL078706 (DK-S). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. □ S The on-line version of this article (available at http://www.jbc.org) contains supplemental Fig. 1 A major challenge to the hypothesis that deoxyhemoglobin is a physiological nitrite reductase is that the scavenging of NO by deoxyhemoglobin (Equation 2) should sequester any generated NO (31,32). If deoxyhemoglobin irreversibly binds NO, then iron-nitrosyl-hemoglobin produced by this mechanism cannot contribute to nitrite-mediated hypoxic vasodilation. Yet, studies of NO gas inhalation in humans have demonstrated increases in red cell iron-nitrosyl-hemoglobin and plasma nitrite concentrations associated with peripheral NO-dependent vasodilation (3, 20 -22). Moreover, despite the apparent biochemical stability of iron-nitrosyl heme, artery-to-vein gradients of iron-nitrosyl-hemoglobin were observed by reductive chemiluminescence and electron paramagnetic resonance (EPR) spectroscopy in humans breathing NO gas, suggesting rapid in vivo dissociation of NO from the heme under these conditions (2,33). Indeed, because NO affinity for heme is 100,000-fold greater than that of oxygen and 1,000-fold greater than that of CO, physiological mechanisms should exist to remove NO from deoxyheme and limit NO poisoning during conditions of high NO output such as septic shock or cigarette smoking.
Interestingly, certain heme oxidants such as ferricyanide and peroxynitrite accelerate NO release from iron-nitrosyl-hemoglobin via the oxidation of the ferrous nitrosyl (Fe II -NO) to the ferric nitrosyl (Fe III -NO) heme (2,34,35). NO is released from the ferric heme with a rate constant of 1 s Ϫ1 , which is comparable with the rate of oxygen release from hemoglobin (29,36). Moreover, work by Stamler and co-workers (38,39) has revealed that iron-nitrosyl-hemoglobin rapidly disappears during oxygenation in the presence of high nitrite concentrations, which they ascribe to allosteric NO transfer from heme to cysteine 93, yet this effect is never observed in the absence of nitrite (40,41). We, therefore, considered the possibility that the oxidative reaction of nitrite and oxyhemoglobin could facilitate the release of NO from iron-nitrosyl-hemoglobin.
The reaction of nitrite with oxyhemoglobin (referred to as the "oxy-reaction") has been studied for over a century. It is a complex autocatalytic reaction that ultimately oxidizes oxyhemoglobin to methemoglobin and nitrite to nitrate (42,43). This reaction demonstrates unique autocatalytic kinetics, as it is initially slow (lag or induction phase) but ultimately enters a rapid autocatalytic (or propagation) phase involving radical-mediated chain reactions and branching steps. Although the mechanism of the oxy-reaction is not fully understood, it proceeds via key catalytic intermediates, including nitrogen dioxide (NO 2 ⅐) and ferrylhemoglobin (Fe IV ϭO) (44). Although these intermediates clearly oxidize hemoglobin and may similarly oxidize Fe II -NO to Fe III -NO, the disposition of iron-nitrosylhemoglobin when the oxy-reaction (generating oxidative intermediates) and the deoxy-reaction (producing iron-nitrosyl-hemoglobin) occur in parallel has never been explored.
We, therefore, studied the reaction of hemoglobin with nitrite at a range of oxygen partial pressures and found that intermediates of the oxy-reaction oxidize the ferrous heme of iron-nitrosyl-hemoglobin and release NO from the ferric heme. This reaction also quenches oxy-reaction intermediates, thereby limiting autocatalysis. These results help explain numerous paradoxes in the field of NO-hemoglobin chemistry, such as the short half-life of iron-nitrosyl-hemoglobin in human circulation during NO inhalation (2,45,46) and the apparent "transfer" of NO from the heme-iron during oxygenation in the presence of excess nitrite (32,38,39,47,48). These results may also provide insight into the mechanism of NO escape from the red cell after nitrite reduction (3,7).

EXPERIMENTAL PROCEDURES
Reagents-All chemicals were purchased from Sigma unless specified otherwise.
Preparation of Hemoglobin and Myoglobin Reagents-All reagents were prepared at 25°C in 0.1 M phosphate buffer, pH 7.4. Purified horse heart myoglobin was purchased from Sigma and reduced by incubation of an anaerobic myoglobin stock solution with 500 mM sodium hydrosulfite. Excess sodium hydrosulfite was removed by passage through two sequential Sephadex G-25 columns (Amersham Biosciences). Human hemoglobin was prepared by hypotonically lysing erythrocytes, discarding membrane fractions after centrifugation, and dialyzing against 0.1 M phosphate buffer, pH 7.4, with storage of the hemolysate at Ϫ80°C as previously described (17). Iron-nitrosylhemoglobin and iron-nitrosyl-myoglobin were synthesized by anaerobic incubation of 150 mM deoxyhemoglobin or deoxymyoglobin with 450 mM PROLI NONOate (Alexis Biochemicals). Excess PROLI NONOate was removed by passage through a Sephadex G-25 column. Hemoglobin was oxidized to methemoglobin by the reaction of hemoglobin with 20 mM potassium hexacyanoferrate(III), which was subsequently removed by passage through two sequential Sephadex G-25 columns. Nitritemethemoglobin and nitrite-metmyoglobin were prepared by incubation of methemoglobin or metmyoglobin with 150 mM sodium nitrite. Ferrylhemoglobin and ferrylmyoglobin were prepared by incubation of methemoglobin with a 100-fold excess hydrogen peroxide and metmyoglobin with a 2.5-fold excess hydrogen peroxide. Carboxyhemoglobin was prepared by incubation of deoxyhemoglobin under positive CO pressure with a channel for gas escape. Oxygen saturation and concentration of heme species were measured by visible absorption spectroscopy (HP8453 UV-visible spectrophotometer; Hewlett-Packard) followed by deconvolution of the spectrum into components from standard spectra of hemoglobin or myoglobin using least squares analysis as previously described (17). Standard reference spectra were composed of oxyhemoglobin (Ϫmyoglobin), deoxyhemoglobin (Ϫmyoglobin), iron-nitrosyl-hemoglobin (Ϫmyoglobin), methemoglobin (Ϫmyoglobin), nitrite-methemoglobin (Ϫmyoglobin), and ferrylhemoglobin (Ϫmyoglobin). Carboxyhemoglobin was included only when reaction solution was exposed to CO gas. Each reaction was deconvoluted at greater than 70 time points (or with a minimum interval of 0.5 s between time points for shorter reactions).
Reactions of Nitrite with Hemoglobin-All reactions were run at 37°C in 0.1 M phosphate buffer, pH 7.4. The anaerobic reaction was carried out with 50 M deoxyhemoglobin and 10 mM sodium nitrite in anaerobic buffer. The aerobic reaction was carried out with 50 M oxyhemoglobin and 10 mM sodium nitrite while open to room air. Reactions with partially oxygenated hemoglobin were run using a 50 M hemoglobin solution that was partially deoxygenated under positive helium pressure to the desired oxyhemoglobin concentration and immediately reacted with 10 mM sodium nitrite. During the anaerobic and partially oxygenated reactions, oxygen leak into the system was prevented by application of positive helium pressure without a channel for gas escape. Reaction kinetics were monitored by absorption spectroscopy in a glass cuvette with a path length of 1 cm. Experiments with greater than 0.05% methemoglobin or oxyhemoglobin (for anaerobic reactions) before the addition of nitrite were always discarded. The concentrations of heme species at each time point were determined by least squares spectral deconvolution using standard reference spectra as described above. The instantaneous reaction rates were measured as the instantaneous rate of deoxy-or oxyhemoglobin consumption (negative change of deoxy-or oxyhemoglobin concentration with respect to the time interval). Aerobic

Oxyhemoglobin-Nitrite Reaction in the Presence of Iron-Nitrosyl-Hemoglobin Monitored by Absorption Spectroscopy and
Chemiluminescence-25 M oxyhemoglobin, 3 M iron-nitrosyl-hemoglobin, and 0.0134% antifoam B emulsion were incubated in 0.1 M phosphate buffer, pH 7.4, at 24°C for 30 s, and 5 mM sodium nitrite was added to initiate the reaction. One reaction was carried out open to room air and was monitored spectrophotometrically in a glass cuvette with a path length of 1 cm. A parallel reaction was carried out in a purge vessel in line with a chemiluminescent NO gas analyzer (Sievers NO analyzer; GE Analytical Instruments) for detection of NO; the vessel was purged with oxygen.

Oxygenation of the Deoxyhemoglobin-Nitrite Reaction Monitored by Absorption Spectroscopy and Chemiluminescence-25
M deoxyhemoglobin was reacted with 5 mM sodium nitrite and 0.0134% antifoam B emulsion in anaerobic 0.1 M phosphate buffer, pH 7.4, at 24°C. One reaction was monitored spectrophotometrically in a glass cuvette with a path length of 1 cm. After 3 min the reaction vessel was rapidly opened to room air and mixed with oxygen for 5 s to achieve full oxygenation. A parallel reaction was carried out in a purge vessel in line with a chemiluminescent NO gas analyzer for detection of NO. The anaerobic reaction was purged with helium for 3 min and was then rapidly oxygenated by switching the purging gas to oxygen; oxygenation was confirmed by detecting the color change of the hemoglobin solution from burgundy to bright red.
EPR Spectroscopy with Spectral Deconvolution of the Oxyhemoglobin-Nitrite Reaction in the Presence of Iron-Nitrosyl-Hemoglobin-65 M oxyhemoglobin and 15 M iron-nitrosylhemoglobin were incubated in 0.1 M phosphate buffer, pH 7.4, at 26°C for 3 min 17 s, when 1 mM sodium nitrite was added to initiate the reaction. The reaction was monitored by absorption spectroscopy in a 1-cm path length glass cuvette, with 500-l fractions of the reaction periodically withdrawn, frozen in liquid nitrogen in quartz EPR tubes (Wilmad LabGlass), and analyzed by EPR spectroscopy. Iron-nitrosyl-hemoglobin was measured using a Bruker EMX 10/12 spectrometer operating at 9.4 GHz, 5-G modulation, 10.1-milliwatt power, 655.36-ms time constant, and 167.77-s scan or 327.68-ms time constant and 83.89-s scan over 600 G at 110 K. The concentration of iron-nitrosyl-hemoglobin measured by EPR was determined by double integral calculation and comparison to standard samples. The concentrations of heme species measured by absorption spectroscopy were determined by least squares spectral deconvolution using standard spectra as described above.
EPR Spectroscopy with Spectral Deconvolution of Deoxy-to-Oxy Cycling of the Hemoglobin-Nitrite Reaction-80 M deoxyhemoglobin was reacted with 1 mM sodium nitrite in anaerobic 0.1 M phosphate buffer, pH 7.4, at 26°C. After 19 min the reaction vessel was mixed with oxygen for 10 s to achieve full oxygenation and was subsequently maintained under positive oxygen pressure. The reaction was monitored by absorption spectroscopy in a 1-cm path length glass cuvette, with 500-l fractions of the reaction periodically withdrawn by gas-tight syringes, placed into helium-purged anaerobic quartz EPR tubes, frozen in liquid nitrogen, and subjected to analysis by EPR spectroscopy as described above.
Effect of the T-to-R Allosteric Transition on Iron-Nitrosyl-Hemoglobin-Effect of oxygen was examined by reacting 1 mM deoxyhemoglobin with 300 M sodium nitrite for 5 min at 25°C in anaerobic phosphate-buffered saline buffer, pH 7.4, followed by rapid oxygenation. Samples were frozen before and after oxygenation and were subjected to analysis by EPR spectroscopy as described above. Effect of CO was examined by reacting 1 mM deoxyhemoglobin with 250 M nitrite for 90 min at 25°C in anaerobic phosphate-buffered saline (PBS) buffer, pH 7.4, followed by dilution into deoxygenated, air-equilibrated, or CO-saturated PBS buffer. Samples were frozen before and after dilution and were subjected to analysis by EPR spectroscopy as described above. The effect of CO was also examined by absorption spectroscopy by reacting 25 M deoxyhemoglobin with 5 mM sodium nitrite for 3 min 27 s at 25°C in anaerobic 0.1 M phosphate buffer, pH 7.4. Approximately 20 mM CO gas was then injected into the solution. The reaction was monitored spectrophotometrically in a glass cuvette with a path length of 1 cm, and the concentrations of heme species were determined by least squares deconvolution using standard reference spectra as described above.

RESULTS
Autocatalytic Behavior of Nitrite Reactions with Pure Deoxyhemoglobin and Oxyhemoglobin-Under strictly anaerobic conditions nitrite reacts with deoxyhemoglobin in an allosteric autocatalytic reaction (referred to as the deoxy-reaction) that consists of an initial slow phase followed by a rapid catalytic phase (Fig. 1A). The slow lag phase corresponds to the reaction of nitrite with T-state deoxyhemoglobin, which forms equimolar quantities of methemoglobin and iron-nitrosyl-hemoglobin (Equations 1 and 2). As these species form, they stabilize the hemoglobin tetramer in the R-state, which lowers the redox potential of unreacted hemes, effectively increasing the bimolecular rate constant for their reaction with nitrite (17). This allosteric shift increases the deoxyhemoglobin-nitrite reaction rate, producing R-state catalysis. The reaction shown in Fig. 1A has an initial rate of 0.17 M/s and reaches its peak rate of 0.47 M/s when the concentration of deoxyhemoglobin (27 M) approaches the sum of iron-nitrosyl-hemoglobin (11 M) and methemoglobin (12 M). Because reaction rate is a function of both reactant concentration (T-state deoxyhemoglobin) and heme reactivity (increased by methemoglobin and iron-nitrosyl-hemoglobin, which promote R-state transition), the deoxy-reaction rate is maximal when the concentrations of these species are maximized. As a result, the instantaneous deoxyhemoglobin-nitrite reaction rate initially increases due to the T-to-R allosteric transition and then decreases as deoxyheme is depleted (Fig. 1B). This reaction is, therefore, sigmoidal, consistent with allosteric modulation of the reaction rate.
Under fully oxygenated conditions, the nitrite reaction with oxyhemoglobin (referred to as the oxyreaction) is characterized by a lag phase followed by a rapid autocatalytic, or propagation, phase (Fig. 1C) (49). In this case the autocatalysts are ferrylhemoglobin (Fe IV ϭO) and nitrogen dioxide (NO 2 ⅐), which build up slowly during the lag phase. Because the final product of this reaction is methemoglobin, the autocatalytic phase of the oxy-reaction is associated with a transient spike of ferrylhemoglobin and rapid conversion of oxyhemoglobin to methemoglobin (Fig. 1C). The instantaneous reaction rate consequently starts low at 0.04 M/s (lag phase), increases exponentially to 8.8 M/s at 10.4 s (propagation phase), and then decreases as oxyhemoglobin is oxidized to methemoglobin (Fig. 1D). As a result, the oxy-reaction is substantially faster than the corresponding deoxy-reaction, such that the oxy-reaction goes to completion in just 17 s, whereas the analogous anaerobic deoxy-reaction goes to completion in 200 s.
In a Partially Oxygenated Environment, Nitrite Simultaneously Reacts with Deoxy-and Oxyhemoglobin-Although the reactions of nitrite with fully deoxygenated and fully oxygenated hemoglobin have been studied in detail and are compared in Fig. 1, A-D, the more physiologically relevant reaction of partially oxygenated hemoglobin has not been characterized. We, therefore, reacted nitrite with hemoglobin at a range of oxygen partial pressures and found that the oxy-and deoxy-reactions proceed in parallel. As shown in Fig. 1E, the concentrations of both oxy-and deoxyhemoglobin fall as they are consumed in the concurrent oxy-and deoxy-reactions, whereas the concentration of methemoglobin increases at a more rapid rate because it is produced in both reactions. Notably, at mixed hemoglobin oxygen saturations the oxy-reaction does not dominate, as may have been predicted by its significantly faster maximal reaction rate (8.8 M/s for oxyhemoglobin versus 0.47 M/s for deoxyhemoglobin), and the two reactions go to completion in comparable time periods. Moreover, the initial rate of deoxyhemoglobin decay in a partially oxygenated environment (0.41 M/s; Fig. 1F) is much higher than the initial rate in the fully deoxygenated reaction (0.17 M/s; Fig. 1B) but is similar to the maximal rate attained during deoxy-reaction R-state catalysis (0.47 M/s; Fig. 1B). This is expected, as partial oxygenation imparts an R-state conformation on the unliganded hemes and, thus, increases the rate of nitrite reduction. The rate of deoxyhemoglobin consumption also peaks during oxy-reaction autocatalysis, most likely due to oxidation of deoxyhemoglobin to methemoglobin by reactive intermediates of the oxyhemoglobin-nitrite reaction. It is interesting to note that deoxyhemoglobin does not decay exponentially, as would be expected if the deoxy-reaction occurred in isolation at its maximal rate, as in the reaction of nitrite with deoxymyoglobin (50). Furthermore, deoxy-and oxyhemoglobin do not decay in parallel, as would be expected with rapid re-equilibration of oxygen among the available hemes. In fact, oxyhemoglobin decay is retarded with respect to deoxyhemoglobin decay, such that the fractional saturation of hemoglobin increases with time. This suggests that both oxygen affinity and nitrite reduction rate increase as a function of time.
In contrast to the increased initial rate of the deoxy-reaction, the lag phase of the oxy-reaction in a partially oxygenated environment is more apparent and prolonged (Fig. 1F). At 60% hemoglobin oxygen saturation, oxy-reaction autocatalysis occurs at 56 s with a maximal rate of 2.79 M/s, compared with 10.4 s with a rate of 8.8 M/s for the pure oxy-reaction (Fig. 1D). This suggests a possible interaction between the deoxy-and oxy-reactions that extends the oxy-reaction lag phase and decreases its rate during autocatalysis.
The Reaction of Nitrite with Oxyhemoglobin in a Partially Oxygenated Environment Is Modulated by Its Parallel Reaction with Deoxyhemoglobin-Surprisingly, at hemoglobin oxygen saturations below 43%, the oxyhemoglobin-nitrite reaction no longer exhibits overt autocatalysis. As shown in Fig. 2, A and B, at 43% oxyhemoglobin the oxy-reaction has a maximum rate of 0.19 M/s and does not transition into the propagation phase demonstrated in Fig. 1, C and D. The concurrent deoxy-reaction does not deviate from its R-state catalysis rate of 0.41 M/s. At even lower oxygen saturation levels the deoxyhemoglobinnitrite reaction rate remains fairly constant, whereas the maximum rate of the parallel oxyhemoglobin-nitrite reaction decreases with falling oxygen saturations (Fig. 2G). Notably, the reductive deoxyhemoglobin-nitrite reaction dominates at low oxygen saturations, the oxidative oxyhemoglobin-nitrite reaction dominates at high oxygen saturations, and the two reactions proceed with comparable maximum reaction rates when ϳ50% of hemoglobin is oxygenated (Fig. 2G). These results indicate that at physiologically low hemoglobin oxygen saturations nitrite will preferentially react with deoxyheme to generate NO.
Increasing the concentration of oxyhemoglobin above 43% results in partial restoration of oxy-reaction autocatalysis. At 48% oxyhemoglobin, the oxyhemoglobin-nitrite reaction enters the rapid propagation phase at 79 s and attains a maximum rate of 0.92 M/s (Fig. 2, C and D). Notably, at 60% oxy-hemoglobin, autocatalysis is observed earlier, at 56 s, and is more pronounced, with a maximum rate of 2.79 M/s (Fig. 2, E  and F). At increasingly higher oxygen partial pressures, the oxyreaction rate during autocatalysis increases further (Fig. 2G) and occurs earlier in the course of the reaction (Fig. 2H). Moreover, these changes in oxy-reaction kinetics cannot be accounted for by the higher total concentration of oxyhemoglobin at higher oxygen saturation levels (data not shown). We, therefore, hypothesized that a product of the deoxyhemoglobin-nitrite reaction partially inhibits the oxyhemoglobin-nitrite reaction by reacting with and scavenging catalytic intermediate(s) of the oxy-reaction.
Surprisingly, iron-nitrosyl-hemoglobin did not accumulate to predicted levels in a partially oxygenated environment. In Fig. 2A, significantly less iron-nitrosyl-hemoglobin was detected than was expected if the oxy-and deoxy-reactions proceeded in parallel without any interaction. In fact, most of the generated iron-nitrosyl-hemoglobin disappeared during the course of reactions shown in Fig. 2, C and E. Because ironnitrosyl-hemoglobin was apparently consumed precisely during oxy-reaction autocatalysis, we hypothesized that iron-nitrosyl-hemoglobin did not accumulate because of its reaction with catalytic intermediates of the oxyhemoglobin-nitrite reaction.
Iron-Nitrosyl-Hemoglobin Delays and Reduces Oxyhemoglobin-Nitrite Reaction Autocatalysis-For both hemoglobin and myoglobin, the reaction of deoxyheme with nitrite yields two products, met-heme and iron-nitrosyl-heme (Equations 1 and 2). To determine whether one or both of these products interact with the oxy-reaction and dampen its autocatalytic phase, we added a range met-heme and iron-nitrosyl-heme concentrations to the aerobic oxyheme-nitrite reaction and compared their effects on autocatalysis.
Because the monomeric structure of myoglobin eliminates possible allosteric confounding by the other hemes of the hemoglobin tetramer, we began by first studying the effects of metmyoglobin and iron-nitrosyl-myoglobin on the oxymyoglobin-nitrite reaction (Fig. 3, A-C). The addition of 30% metmyoglobin to the reaction of oxymyoglobin with nitrite had no effect on the magnitude or timing of autocatalysis (Fig. 3B) compared with the control reaction (Fig. 3A). The control reaction reached autocatalysis at 31 s with a rate of 1.05 M/s, whereas the reaction with exogenous metmyoglobin became autocatalytic at 29 s with a rate of 1.1 M/s (see insets in Fig. 3). The differences between these times and rates are not significant, as they represent normal and expected variation between experiments. In contrast, the addition of the same amount of iron-nitrosyl-myoglobin to the reaction delayed autocatalysis to 184 s and decreased the maximum reaction rate to 0.38 M/s (Fig. 3C). Notably, the inhibitory effect of iron-nitrosyl-myoglobin was dose-dependent, such that autocatalysis was delayed longer and the rate was diminished more significantly with higher concentrations of exogenous iron-nitrosyl-myoglobin (data not shown).
To confirm and generalize the inhibitory effect of iron-nitrosyl-heme on the oxy-reaction, we examined the impact of iron-nitrosyl-hemoglobin and methemoglobin on the oxyhemoglobin reaction with nitrite (Fig. 3, D-F). As with the corre-sponding myoglobin reactions, the addition of iron-nitrosylhemoglobin to the oxyhemoglobin-nitrite reaction led to a dose-dependent delayed onset and diminished rate of autoca-talysis. Compared with the control rate of 4.63 M/s at 25 s (Fig. 3D), the addition of 30% iron-nitrosylhemoglobin to the oxy-reaction delayed and decreased the rate during autocatalysis to 2.98 M/s at 34.4 s (Fig. 3F). However, although the addition of metmyoglobin had no effect on the oxymyoglobin-nitrite reaction, 30% exogenous methemoglobin decreased the lag phase to oxyhemoglobin-nitrite autocatalysis to 15.4 s and raised its rate to 5.39 M/s (Fig. 3E).
Thus, increasing the amount of iron-nitrosyl-hemoglobin led to a dose-dependent decrease in the rate of autocatalysis and increase in the lag phase of the oxyhemoglobinnitrite reaction, whereas increasing the amount of methemoglobin had the opposite dose-dependent effect (data not shown). The decreased rate and prolonged lag phase of the oxy-reaction observed in a partially oxygenated environment is, therefore, at least partly due to the presence of iron-nitrosyl-hemoglobin from the concurrent deoxy-reaction.
Oxyhemoglobin-Nitrite Autocatalysis Results in the Consumption of Iron-Nitrosyl-Hemoglobin-As demonstrated in the reaction of nitrite with partially oxygenated hemoglobin, oxy-reaction autocatalysis is associated with the depletion of iron-nitrosyl-hemoglobin produced in the concurrent deoxy-reaction (Fig. 2, C and E). Similarly, iron-nitrosyl heme added to aerobic reactions of nitrite with oxyhemoglobin or oxymyoglobin disappeared when these reactions became autocatalytic (Fig. 3, C and F). Quantitative consumption of iron-nitrosyl-hemoglobin was subsequently confirmed by electron paramagnetic resonance spectroscopy (EPR). In the reaction shown in Fig. 4, A-C, nitrite was added to a solution of iron-nitrosyl-hemoglobin and oxyhemoglobin at 3 min 17 s to initiate the oxy-reaction, and the iron-nitrosyl EPR signature at g ϭ 2 decreased precisely during autocatalysis (Fig. 4, A and B). The fall in iron-nitrosyl-hemoglobin concentration measured by EPR (Fig. 4C) confirmed the accuracy of iron-nitrosyl-hemo-  A and B), 48% oxyhemoglobin (C and D), and 60% oxyhemoglobin (E and F). Data shown in panels E and F are identical to that presented in Fig. 1, E and F; however, the axes are adjusted for ease of comparison to reactions carried out at other oxygen saturation levels (panels A-D). G, maximum reaction rates of the oxyhemoglobin-nitrite and deoxyhemoglobin-nitrite reactions as a function of percent oxyhemoglobin and deoxyhemoglobin present at the beginning of the reaction. H, duration of oxyhemoglobin-nitrite reaction lag phase (time to maximal reaction rate reached at autocatalysis) as a function of percent oxyhemoglobin and deoxyhemoglobin present at the beginning of the reaction. APRIL 27, 2007 • VOLUME 282 • NUMBER 17 globin detection by absorption spectroscopy and reiterated its disappearance during oxy-reaction autocatalysis. Thus, ironnitrosyl-heme not only delays and decreases oxy-reaction autocatalysis but is also consumed in the process.

NO Formation and Release by the Nitrite-Hemoglobin Reactions
Consumption of Iron-Nitrosyl-Hemoglobin during Oxy-Reaction Autocatalysis Leads to the Release of NO Gas-We then considered whether oxyhemoglobin-nitrite autocatalysis can oxidize the iron-nitrosyl ferrous heme (Fe II -NO) to ferric heme-NO (Fe III -NO) and subsequently release NO into the gas phase, analogous to the effect of ferricyanide on iron-nitrosylhemoglobin (34). To determine whether free NO is released, we ran the oxyhemoglobin-nitrite reaction with 12% exogenous iron-nitrosyl-hemoglobin in a purge vessel in line with a nitric oxide analyzer (NOA) 3 (Fig. 4D). To accurately correlate NO gas release with iron-nitrosyl-hemoglobin disappearance and oxyhemoglobin-nitrite autocatalysis, an identical reaction was carried out in a cuvette, measured by absorption spectroscopy, and analyzed by least squares deconvolution (Fig. 4E). Because hemoglobin produces significant foaming when purged with inert gas, antifoam solution was added to the reaction. Antifoam oxidized a small fraction of oxyhemoglobin to methemoglobin before the start of the reaction but did not lead to continued oxidation that would interfere with accurate interpretation of reaction results (data not shown). This accounts for the uncharacteristic presence of methemoglobin at the onset of reactions depicted in Figs. 4E and 5A.
Comparison of the timing of NO gas release as detected by the NOA with the progress of the reaction monitored by absorption spectroscopy indicated that iron-nitrosylhemoglobin is the source of NO. According to Fig. 4E, the oxyhemoglobin-nitrite reaction entered autocatalysis, and iron-nitrosyl-hemoglobin began to disappear at 51 s. NO release was appreciated by the NOA starting at ϳ55 s (Fig. 4D), suggesting that oxidation of iron-nitrosyl-hemoglobin by intermediates of the nitrite-oxyhemoglobin reaction generates free NO. The yield of gas phase NO from iron-nitrosylhemoglobin was calculated from the area under the curve of the NO signal over time to be 2.7%, because 0.408 nmol of NO was freed from 15 nmol of iron-nitrosyl (Fig. 4D) and ranged up to 4.2% in other experiments (data not shown).

NO Generated in the Deoxyhemoglobin-Nitrite Reaction Is
Released after Oxygenation-Finally, we wondered whether oxyhemoglobin-nitrite autocatalysis seen in partially oxygenated solutions can release NO from iron-nitrosyl-hemoglobin produced in the deoxyhemoglobin-nitrite reaction under hypoxia. We already had evidence supporting this hypothesis, since oxy-reaction autocatalysis in partially oxygenated conditions correlated with the disappearance of iron-nitrosyl-hemoglobin (Fig. 2, C and E). We now explored whether this loss of iron-nitrosyl-heme was also associated with the release of gas phase NO.
In this series of experiments we subjected hemoglobin to a deoxygenation/oxygenation cycle in the presence of excess nitrite. As shown in Fig. 5A, the reaction was initiated in an anaerobic environment but was rapidly oxygenated at 3 min, converting unreacted deoxyhemoglobin to oxyhemoglobin. During the anaerobic segment of the reaction, deoxyhemoglobin was slowly oxidized to methemoglobin as nitrite was reduced to NO, generating iron-nitrosyl-hemoglobin. Upon oxygenation, the remaining deoxyhemoglobin was replaced by 3 The abbreviation used is: NOA, nitric oxide analyzer. oxyhemoglobin, which immediately began to react with nitrite. The oxyhemoglobin reaction entered the autocatalytic phase at 3 min 48 s, and iron-nitrosyl-hemoglobin that had accumulated during the anaerobic segment of the reaction completely disappeared.
NO release was monitored in a parallel reaction carried out in the NOA purge vessel (Fig. 5B). There was a low level of NO release during the anaerobic reaction, but subsequent to oxygenation at 3 min and initiation of oxyhemoglobin-nitrite autocatalysis, the NOA documented a significant and rapid rise in gas phase NO starting at ϳ3 min 45 s. It, therefore, took 45 s for NO to be released after reaction oxygenation, consistent with the 48 s that elapsed before iron-nitrosyl-hemoglobin disappearance as monitored by the spectrophotometer (Fig. 5, A and B). Analogous to oxyhemoglobin-nitrite reactions in the presence of exogenous iron-nitrosyl-hemoglobin, not all NO released from iron-nitrosyl-hemoglobin during oxy-reaction autocatalysis was detected as gas phase NO. The efficiency of NO release in this reaction was 5.4%, as 0.49 nmol of NO was released from ϳ9 nmol of iron-nitrosyl-hemoglobin (Fig.  5B).
These experiments were repeated using freeze-quench EPR and revealed the formation of iron-nitrosylhemoglobin during the anaerobic phase of the hemoglobin-nitrite reaction followed by its decay during the autocatalytic phase of the oxyhemoglobin-nitrite reaction subsequent to oxygenation (Fig. 5, C and D). Iron-nitrosyl-hemoglobin is, therefore, not an irreversible trap for NO generated in the deoxy-reaction but, rather, promotes the release of NO gas in a partially oxygenated environment.
Other groups have suggested that the observed decrease in the EPR iron-nitrosyl complex during hemoglobin oxygenation in the presence of nitrite occurs secondary to an intramolecular transfer of NO from the heme to the ␤-93 cysteine upon a change in Hb quaternary state from T to R during oxygenation (32,39,47,48,52). If this is indeed the mechanism, then the transfer of NO from the heme of ironnitrosyl-hemoglobin should occur within microseconds of oxygenation, during the allosteric structural transition of iron-nitrosyl-hemoglobin from T to R and before oxyhemoglobin-nitrite autocatalysis, as it is limited by the rate of oxygen binding itself. In contrast to previously published results (32, 39), we did not observe such a decrease upon oxygenation using freezequench EPR, despite a clear transition from a five-coordinate (T-state) to a six-coordinate (R-state) iron-nitrosyl complex when we froze the samples immediately upon oxygenation (Fig. 6A). Thus, after 5 min of anaerobic reaction time, the reaction of deoxyhemoglobin (1 mM) with nitrite (300 M) yielded 13 M iron-nitrosyl-hemoglobin before oxygenation and 13.1 M iron-nitrosylhemoglobin after oxygenation. These results were consistent over several experiments, such that 13.5 Ϯ 0.8 M iron-nitrosyl-hemoglobin was detected before reaction oxygenation and 12.3 Ϯ 0.8 M after oxygenation (n ϭ 3, p ϭ 0.31).
Moreover, when CO gas was used instead of oxygen to promote the T-to-R transition without initiating the oxyhemoglobin-nitrite reaction, iron-nitrosyl-hemoglobin was not consumed (Fig. 6B). In this series of experiments, 1 mM Ϫ ) was added after oxyhemoglobin and iron-nitrosyl-hemoglobin had equilibrated for 3 min 17 s. A, reaction progress was monitored by absorption spectroscopy. B, EPR confirmed the consumption of iron-nitrosyl-hemoglobin during oxyhemoglobinnitrite reaction autocatalysis. Consecutive spectra were taken at the indicated time points. C, the change in iron-nitrosyl-hemoglobin (Fe II -NO) concentration over time was also measured by EPR. Similar reactions of oxyhemoglobin (25 M) with nitrite (5 mM) in the presence of iron-nitrosyl-hemoglobin (3 M) were carried out in parallel and monitored by the nitric oxide analyzer (D) and visible absorption spectroscopy (E). Oxyhemoglobin and iron-nitrosyl-hemoglobin were present at the beginning of the reaction, whereas nitrite was added at 36 s. The dotted line across panels D and E indicates the point of maximal oxy-reaction autocatalysis.
deoxyhemoglobin was reacted with 250 M nitrite for 90 min, and the sample was frozen for EPR after dilution into deoxygenated, air-equilibrated, or CO-saturated buffer. The concentrations of ironnitrosyl-hemoglobin were 3.8 M when the reaction was diluted into deoxygenated buffer, 3.6 M when diluted into airsaturated buffer, and 3.6 M when diluted into CO-saturated buffer. The average percentage change in iron-nitrosyl-hemoglobin compared with solutions diluted into deoxygenated buffer was Ϫ8 Ϯ 5% when diluted into air-saturated buffer and Ϫ4 Ϯ 16% when diluted into CO-saturated buffer (n ϭ 3 for each condition). There was, therefore, no significant loss of iron-nitrosyl-hemoglobin signal by EPR upon T-to-R allosteric transition induced by heme ligation with oxygen or CO.
We further confirmed the stability of iron-nitrosyl-hemoglobin and the absence of intramolecular NO transfer upon T-to-R transition by absorption spectroscopy (Fig. 6C). In a reaction analogous to that depicted in Fig. 5A, we introduced CO gas into an ongoing deoxyhemoglobin-nitrite reaction at 3 min 27 s. All unreacted deoxyhemoglobin was converted to carboxyhemoglobin, thereby shifting hemoglobin into R-state. Importantly, there was no subsequent change in the concentration of any of the heme species during the remainder of the reaction, and NO remained bound firmly to Fe II as iron-nitrosyl (Fig. 6C). Iron-nitrosyl loss, therefore, requires the autocatalytic phase of the oxyhemoglobin-nitrite reaction (i.e. oxy-reaction catalytic intermediates) rather than a T-to-R allosteric shift, consistent with our proposed mechanism of NO release.

DISCUSSION
Recent studies reveal that the ubiquitous circulating anion nitrite (NO 2 Ϫ ) is a vasodilator and intrinsic signaling molecule (2,(3)(4)(5)16). Nitrite infusions into the human circulation increase blood flow at near physiological concentrations (3, 20 -22). The vasodilator activity of nitrite has been attributed to allosterically controlled heme-based reduction of nitrite to NO by deoxygenated hemoglobin. Consistent with this hypothesis, vasodilation during nitrite infusions is associated with increases in red cell ironnitrosyl-hemoglobin and to a lesser extent S-nitrosated hemoglobin (3). In fact, in vitro incubation of nitrite with deoxygenated red cells or hemoglobin solutions has been shown to produce vasodilation (3, 7, 20 -22), NO-dependent cGMP accumulation (7), gas phase NO generation (7,17), and NO-dependent inhibition of mitochondrial respiration (7). Nonetheless, the mechanism for NO escape from the red cell after nitrite reduction by deoxyhemoglobin remained a mystery.
Indeed, a major challenge to the hemoglobin nitrite reductase hypothesis and other erythrocyte NO export theories is explaining how NO can escape heme autocapture (45). Nitric oxide reacts with both deoxy-and oxyhemoglobin extremely fast, with bimolecular rates in the range of 10 7 -10 8 M Ϫ1 s Ϫ1 (30, 45, 54 -58). Modeling calculations have shown that only 0.1 pM NO could be present outside a red cell at steady state, even at supraphysiological nitrite levels, unless additional mechanisms exist to limit scavenging reactions of NO with hemoglobin (61). Although the formation of iron-nitrosyl-hemoglobin preserves the chemical nature of NO as a ligand on deoxyhemoglobin, its slow off rate suggests that iron-nitrosyl should be an irreversible trap for NO. Surprisingly, a number of physiological and experimental observations have suggested that iron-nitrosyl heme is in fact not stable under oxidizing conditions. For instance, iron-nitrosyl-hemoglobin and nitrite form in the pulmonary vasculature during NO gas inhalation, and both are subsequently consumed, creating an artery-to-vein gradient (2,33,46). The observation of iron-nitrosyl-hemoglobin decay by EPR, despite a transit time of less than 30 s, is difficult to reconcile with known chemistry. Stamler and co-workers have also described rapid iron-nitrosyl-hemoglobin disappearance by EPR after reaction reoxygenation in the presence of 200 M nitrite, which they attributed to an allosteric transfer of NO from the heme to the thiol (32,39). However, experiments by our group have demonstrated that this phenomenon is only observed in the presence of nitrite, suggesting that the oxyhemoglobin-nitrite reaction is required for NO release from iron-nitrosyl-hemoglobin. Moreover, we and others have found that heme oxidants such as ferricyanide and peroxynitrite rapidly oxidize the ferrous nitrosyl heme (Fe II -NO) to the ferric nitrosyl heme (Fe III -NO) with the ligand still in place (2, 34, 35).  (5 mM) was allowed to proceed to partial completion and generate ϳ3 M iron-nitrosyl-hemoglobin. After 3 min the anaerobic reaction chamber was fully oxygenated, and the oxyhemoglobin-nitrite reaction in the presence of endogenous iron-nitrosyl-hemoglobin continued to completion. Reaction progress was monitored in parallel by absorption spectroscopy (A) and the nitric oxide analyzer (B). The dotted line across panels A and B indicates the point of maximal oxy-reaction autocatalysis. C and D, generation and consumption of iron-nitrosyl-hemoglobin was confirmed by a separate reaction of deoxyhemoglobin (80 M) with nitrite (1 mM). Deoxyhemoglobin and nitrite were incubated for 19 min before complete oxygenation of the reaction solution. The consumption of iron-nitrosyl-hemoglobin (Fe II -NO) was confirmed by EPR spectroscopy. Consecutive spectra were taken at the indicated time points.
Consistent with these observations, we now report that in a partially oxygenated environment the reactions of nitrite with oxyhemoglobin and deoxyhemoglobin proceed in parallel. The deoxy-reaction reduces nitrite to NO (Equation 1), which leads to the formation of iron-nitrosyl-hemoglobin (Equation 2). At the same time, the oxy-reaction oxidizes nitrite to nitrate through a series of highly reactive free radical intermediates. Because iron-nitrosyl-hemoglobin is consumed specifically during the autocatalytic phase of the oxyhemoglobin-nitrite reaction, we believe that nitrogen dioxide, or possibly another intermediate generated during oxy-reaction autocatalysis, oxi-dizes the ferrous heme of iron-nitrosyl-hemoglobin to NO-methemoglobin, which subsequently releases free NO. Indeed, NO 2 ⅐ has been shown to oxidize the iron centers of iron-nitrosyl-hemoglobin and -myoglobin to NO-methemoglobin and -myoglobin, with the subsequent release of NO (35,60). Because this catalytic intermediate is necessary for oxyreaction autocatalysis and it is consumed as it oxidizes the ironnitrosyl heme, the rate of the autocatalysis is thereby also limited. As a result, this interactive oxidative denitrosylation chemistry effectively channels nitrite through reductive pathways to NO.
The mechanism of oxidative denitrosylation helps explain a number of paradoxical observations in the NO-hemoglobin field such as the apparent instability of iron-nitrosylhemoglobin after oxygenation in the presence of nitrite (32,38,39,47,48) and the rapid dissociation of iron-nitrosylhemoglobin in vivo during NO gas inhalation (2,45,46). Such chemistry could contribute to the vasodilatory activity of deoxygenating red cells and hemoglobin solutions in the presence of nitrite (3,7).
We recognize that these experiments were performed at supraphysiological nitrite concentrations, as hemoglobin concentrations must be maintained between 25 and 150 M heme to allow for accurate analysis by absorption spectroscopy. Nonetheless, they may be physiologically relevant because of the second order nature of these reactions. In vivo, erythrocyte nitrite concentration is low (300 nM in the red cell) and hemoglobin concentration is high (20 mM); in our experimental setup the relative ratio of these reactants is reversed. However, we recognize that these reactions will require compartmentalization to increase the local concentrations of reactants. The physiological efficiency of nitrite reduction and NO release are likely to be greater in vivo if the effective concentrations of the reactants, i.e. nitrite and hemoglobin, are maximized at the erythrocyte submembrane. A putative nitrite reductase metabolon located within the red cell lipid raft composed of deoxy-and oxyhemoglobin, an anion exchange protein (for nitrite import into the cell), carbonic anhydrase, aquaporin, and Rh channels (51) would effectively localize the necessary reagents (nitrite, protons), the NO-generating deoxyhemoglobin-nitrite reaction, and the autocatalytic oxyhemoglobin-nitrite reaction necessary to oxidize iron-nitrosyl-hemoglobin and release free NO near highly hydrophobic channels at the membrane. Because NO is highly lipophilic, it could then rapidly diffuse out of the cell and avoid autocapture.
In addition to oxidative NO release, the oxyhemoglobin-nitrite reaction facilitates the metabolism of iron-nitrosyl-hemoglobin. Because NO binds hemoglobin 100,000 times more avidly than oxygen and 1,000 times more avidly than carbon monoxide, exposure to NO could lead to NO poisoning and impaired oxygen transport. Because primordial flavohemoglobins served as NO scavengers and only later evolved to bind and deliver oxygen (59), one could speculate that red cell nitrite chemistry may have originally evolved for the function of chemical oxidative denitrosylation. Once formed, iron-nitrosyl-hemoglobin would localize to the erythrocyte submembrane, since NO binding breaks the bond between the heme and the proximal histidine, making a super T-state tetramer that may FIGURE 6. There is no loss of iron-nitrosyl-hemoglobin upon T-to-R allosteric transition in the absence of the oxyhemoglobin-nitrite reaction. A, deoxyhemoglobin (5 mM) was reacted with nitrite (300 M) for 5 min, and the sample was frozen for EPR before and after oxygenation. B, deoxyhemoglobin (1 mM) was reacted with nitrite (250 M) for 90 min, and the sample was frozen for EPR after dilution into deoxygenated, air-equilibrated, or CO-saturated buffer. C, reaction of deoxyhemoglobin (25 M) with nitrite (5 mM) was allowed to proceed to partial completion while monitored by absorption spectroscopy. After 3 min 27 s the anaerobic reaction chamber was exposed to CO gas. The rugged shape of the curve immediately after introduction of CO gas is due to temporary failure to obtain clean spectra during turbulent mixing of the reaction solution.
preferentially bind to band 3 (53). At the same time, nitrite enters the red cell via anion channels and reacts with oxyhemoglobin at the submembrane, generating NO 2 ⅐, which oxidizes iron-nitrosyl-hemoglobin to NO-methemoglobin, thereby releasing free NO. Methemoglobin is then recycled to deoxyhemoglobin by methemoglobin reductase.
In conclusion, these experiments reveal fundamental and novel metal-and nitrite-catalyzed reaction pathways that limit iron-nitrosyl-hemoglobin accumulation within the erythrocyte and generate free NO. These reactions have the potential to contribute to nitrite and heme-globin dependent hypoxic signal transduction.