Reactivity Studies of the Fe(III) and Fe(II)NO Forms of Human Neuroglobin Reveal a Potential Role against Oxidative Stress*

Neuroglobin, recently discovered in the brain and in the retina of vertebrates, belongs to the class of hexacoordinate globins, in which the distal histidine coordi-nates the iron center in both the Fe(II) and Fe(III) forms. As for most other hexacoordinate globins, the physiological function of neuroglobin is still unclear, but seems to be related to neuronal survival following acute hypoxia. In this study, we have addressed the question whether human neuroglobin could act as a scavenger of toxic species, such as nitrogen monoxide, peroxynitrite, and hydrogen peroxide, which are generated at high levels in the brain during hypoxia; we have also investigated the kinetics of the reactions of its Fe(III) (metNGB) and Fe(II)NO forms with several reagents. Binding of cyanide or NO (cid:1) to metNGB follows bi-exponential kinetics, showing the existence of two different protein conformations. In the presence of excess NO (cid:1) , metNGB is con-verted into NGBFe(II)NO by reductive nitrosylation, Biosciences) by isoelectrofocusing on polyacrylamide gels in the 3–9 pH range and by SDS 15% polyacrylamide gels, both showing a single band. In contrast to other studies 35), we had no evidence of partial formation of either intra- or inter-molecular S-S bonds, as judged from isoelectrofocusing or non-reducing SDS gels, respectively. UV / Visible Spectroscopy— Absorption spectra were collected in 1-cm cells on a UVIKON 820 spectrophotometer. Kinetic studies of cyanide binding to metNGB were carried out with an Analytik Jena Specord 200. These studies were performed under aerobic conditions by adding a small volume of a concentrated cyanide solution a of the reductive nitrosylation of metNGB were car- ried out in 0.1 M phosphate buffer, 7.0, in sealable cells for applications at The metNGB solution was first de- gassed by keeping it at at transferred by using a gas-tight syringe into sealed cell The amount of NO was finally added from a solution M ) using a SampleLock Hamilton syringe, and the measurement was started within 10 s. For comparison, the spectrum of NGBFe(II)NO was also obtained under anaerobic conditions by first reducing metNGB 1 mg/ml dithionite and then adding NO (cid:1) . Kinetic studies of the oxidation of NGBFe(II)NO by O 2 were carried out in sealed cells by first degassing the NGBFe(II)NO solution pre- pared by reductive nitrosylation of metNGB (see with Ar for at least 30 min to eliminate excess NO (cid:1) . Different amounts of oxygen were added with a gas-tight syringe from aqueous solutions equilibrated with

gesting that Ngb is a phylogenetically ancient globin that has not changed significantly during evolution (3). In contrast, Ngbs share low sequence similarity with the two major vertebrate globins, myoglobin (Mb, ϳ21%) and hemoglobin (Hb, ϳ25%) (1). Nevertheless, human neuroglobin (NGB) displays the same characteristic ␣-helical three-dimensional fold typical for oxygen-binding heme globins (5). In addition, the functionally important amino acid residues of Mb and Hb, His(F8) and His(E7) on the proximal and distal sides of the heme, respectively, are conserved in Ngb (1). However, in contrast to Mb and Hb, which display a pentacoordinate Fe(II) heme in the absence of ligands, the heme in Ngb is hexacoordinate in both Fe(II) and Fe(III) forms (5)(6)(7)(8). Site-directed mutagenesis studies have shown that the sixth coordination site is occupied by the distal His(E7), which has to be displaced before an external ligand can bind (6). Competition between external ligands and His(E7) is responsible for the control of ligand affinity to the reduced forms of Ngb and other hexacoordinate globins, such as cytoglobin, invertebrate nerve globins, and non-symbiotic plant hemoglobins (3,9).
The physiological function of Ngb and most other hexacoordinate globins is still unknown. It has recently been reported that Ngb is up-regulated and protects neurons during the acute phase of hypoxic episodes, suggesting a role in the neuronal response to oxygen depletion (10,11). A plausible hypothesis is that Ngb functions as a temporary oxygen-storage protein, in analogy to nerve Hbs from invertebrates (12), which share a relatively high sequence similarity with Ngb (1). This hypothesis is further supported by the relatively high oxygen affinity reported for NGB (ϳ1 torr at 37°C) (6) and by its high concentration in the mitochondria-rich segments of the retina, one of the body's tissues with the highest metabolic rate (4). However, in vitro the oxygenated form of Ngb is unstable and rapidly autoxidizes to the Fe(III) form (6), which is incapable of oxygen binding. Thus, to be able to function as an oxygen storage protein in vivo, Ngb must be reduced by a not-yet-identified enzymatic system operating in brain and retinal cells, analogous to metHb reductase in red blood cells.
In this study, we have investigated the reactivity of NGB toward nitrogen monoxide, peroxynitrite, and hydrogen peroxide to evaluate an alternative hypothesis: that NGB functions as a scavenger of these toxic compounds, which are produced during hypoxia and mitochondrial respiration. In the brain, nitrogen monoxide is generated by the neuronal isoform of nitric oxide synthase (nNOS) and modulates neurotransmission. Remarkably, neuronal nitric oxide synthase is co-expressed in the same neurons as Ngb in several nuclei and regions of the brain important for stress response (13), whereas other studies have evidence for a more widespread localization (14). Moreover, NO ⅐ is produced by the mitochondrial isoform of nitric oxide synthase and regulates mitochondrial respiration by reversibly binding to cytochrome c oxidase (15,16). An important physiological function of Mb in muscle cells, in addition to secure the oxygen supply to mitochondria, is to protect cellular respiration by rapidly converting NO ⅐ to nitrate, thereby preventing inhibition of cytochrome c oxidase (17)(18)(19)(20). Hence, the physiological function of Ngb may also conceivably be related to the metabolism of NO ⅐ .
In response to hypoxia, NO ⅐ production in the brain increases dramatically because of the activation of the neuronal and inducible isoforms of nitric oxide synthases (21,22). In addition, superoxide and the product of its rapid dismutation, hydrogen peroxide, are continuously generated from the incomplete reduction of 0.15-2% of oxygen during mitochondrial respiration, particularly when cytochrome c oxidase is inhibited (23)(24)(25). As a consequence, the product of the diffusionlimited reaction between NO ⅐ and superoxide, peroxynitrite (26), a potent oxidizing and nitrating agent, is generated at high levels during hypoxia and contributes severely to brain ischemic damage and membrane lipid peroxidation (21,27) and ultimately leads to inactivation of mitochondrial respiration (28). Taken together, these factors suggest that the protective effect exerted by Ngb during hypoxic episodes may be due to the ability of Ngb to scavenge toxic compounds generated under these conditions, among which are NO ⅐ , peroxynitrite, and hydrogen peroxide. All of these species are known to interact with hemoproteins. In particular, Hb and Mb have been shown to scavenge both NO ⅐ and peroxynitrite (20, 29 -31) and thus to protect cells against oxidative stress.
In this work, we have focused on the reactivity of metNGB, which seems to be the thermodynamically most stable form under aerobic conditions. Specifically, we have performed kinetic studies of the reactions of metNGB with cyanide, NO ⅐ , peroxynitrite, and hydrogen peroxide. Moreover, we have investigated the kinetics of the reactions of dioxygen and peroxynitrite with nitrosyl NGB (NGBFe(II)NO), a species that has been detected by EPR in bacteria overexpressing recombinant Ngb (32) and that is likely to be formed in vivo in intact NO ⅐ -producing neurons. Our data show that the reactivity of NGB is significantly different from that of Mb and Hb.

EXPERIMENTAL PROCEDURES
Reagents-Buffer solutions were prepared from K 2 HPO 4 /KH 2 PO 4 (Fluka) with deionized Milli-Q water. Sodium dithionite, potassium superoxide, hydrogen peroxide, and potassium cyanide were obtained from Fluka. Nitrogen monoxide (Linde) was passed through a NaOH solution and a column of NaOH pellets before use to remove higher nitrogen oxides. Nitrogen monoxide solutions were prepared as described previously (33). Peroxynitrite was prepared from KO 2 and gaseous nitrogen monoxide according to Koppenol et al. (34) and stored in small aliquots at Ϫ80°C. The peroxynitrite solutions contained variable amounts of nitrite (maximally 50% relative to the peroxynitrite concentration) and no hydrogen peroxide. The stock solution was diluted with 0.01 M NaOH. For all of the experiments with peroxynitrite, the buffers and the 0.01 M NaOH solutions were prepared fresh daily and thoroughly degassed.
Recombinant NGB was cloned and expressed in Escherichia coli as described (6). Purification was achieved by ammonium sulfate precipitation, anion-exchange chromatography, and gel filtration (6). Samples were concentrated by ultrafiltration (Amicon) and stored in liquid nitrogen as metNGB. Once thawed, the protein was kept on ice until use. Purity was checked using a PhastSystem (Amersham Biosciences) by isoelectrofocusing on polyacrylamide gels in the 3-9 pH range and by SDS 15% polyacrylamide gels, both showing a single band. In contrast to other studies (6,35), we had no evidence of partial formation of either intra-or inter-molecular S-S bonds, as judged from isoelectrofocusing or non-reducing SDS gels, respectively.
UV/Visible Spectroscopy-Absorption spectra were collected in 1-cm cells on a UVIKON 820 spectrophotometer. Kinetic studies of cyanide binding to metNGB were carried out with an Analytik Jena Specord 200. These studies were performed under aerobic conditions by adding a small volume of a concentrated cyanide solution (20 -200 to a metNGB solution in 0.1 M phosphate buffer, pH 7.0 at room temperature. Kinetic studies of the reductive nitrosylation of metNGB were carried out in 0.1 M phosphate buffer, pH 7.0, in sealable cells for anaerobic applications at room temperature. The metNGB solution was first degassed by keeping it at 0°C under an Ar atmosphere for at least 1 h and then transferred by using a Hamilton gas-tight syringe inserted into the sealed cell previously flushed with Ar for at least 30 min. The required amount of NO ⅐ was finally added from a saturated solution (2 mM) by using a SampleLock Hamilton syringe, and the measurement was started within 10 s. For comparison, the spectrum of NGBFe(II)NO was also obtained under anaerobic conditions by first reducing metNGB with 1 mg/ml sodium dithionite and then adding NO ⅐ .
Kinetic studies of the oxidation of NGBFe(II)NO by O 2 were carried out in sealed cells by first degassing the NGBFe(II)NO solution prepared by reductive nitrosylation of metNGB (see above) with Ar for at least 30 min to eliminate excess NO ⅐ . Different amounts of oxygen were added with a gas-tight syringe from aqueous solutions equilibrated with oxygen (1.3 mM) or air (0.22 mM), and the measurement was started within 10 s. Spectra were measured at 25°C either in the Soret (400 -450 nm) or the visible (500 -650 nm) region by using an HP8543 diode array spectrophotometer (cell width ϭ 1 cm).
Stopped-flow Kinetic Analysis-Kinetic studies were carried out with an Applied Photophysics SX18MV-R single-wavelength stopped-flow instrument and an On-Line Instrument Systems Inc. stopped-flow instrument equipped with an OLIS RSM 1000 rapid scanning monochromator. The width of the cells in the two spectrophotometers was 1 cm. The mixing time of the instruments was 2-4 ms. If not specified, measurements were carried out at 20°C. Kinetic studies of the reaction between NGBFe(II)NO and peroxynitrite were carried out with the Applied Photophysics apparatus. The traces were collected at different wavelengths between 400 -440 nm, and the data were analyzed with Kaleidagraph version 3.0.8. For the determination of the second-order rate constants, the traces were collected mostly at 536 nm, whereas dissociation of NO ⅐ from NGBFe(III)NO was followed at 418 nm. The results of the fits of the traces from at least five experiments were averaged to obtain each observed rate constant, giving the corresponding standard deviation. The pH was measured at the end of the reactions for control.
NGBFe(II)NO solutions were prepared by reductive nitrosylation of metNGB and then degassed under an Ar atmosphere (see above). The degassed solutions were transferred to the stopped-flow apparatus by using SampleLock Hamilton syringes. The peroxynitrite solutions were prepared by adding a small volume of a concentrated peroxynitrite solution (50 -80 mM) to thoroughly degassed 0.01 M NaOH, directly in the SampleLock syringe. The peroxynitrite concentration of the resulting solutions was determined spectrophotometrically prior to each experiment by measuring the absorbance at 302 nm (⑀ 302 ϭ 1705 M Ϫ1 cm Ϫ1 ) (36).

RESULTS
Binding of Cyanide to metNGB-The reaction between met-NGB and cyanide was studied by UV/visible spectroscopy in the wavelength range 350 -480 nm at pH 7.0 and room temperature. As shown in Fig. 1, upon mixing the absorbance maximum of the Soret band of NGB shifted from 414 (metNGB) to 416 nm (NGBFe(III)CN). This process did not give rise to a single isosbestic point: specifically, in the first 80 s, an isosbestic point was at 418 nm, whereas in the time range 200 -600 s, it was shifted at 422 nm. The approximate extinction coefficient determined for the new absorbance band at 416 nm was 96 mM Ϫ1 cm Ϫ1 . The kinetics of the reaction were obtained by fitting the traces extracted at 400 and/or 407 nm. As shown in Fig. 1B, Inset, the traces had to be fitted to a bi-exponential equation, confirming the absence of a unique isosbestic point. Because both observed rate constants depended linearly on the cyanide concentration, this result suggests the presence of two distinct conformations of the protein that bind this ligand at different rates. The second-order rate constants obtained from the linear fits of the plots of the observed rate constants of the fast and the slow components are 1.
The distal histidine coordinated to the Fe(III) center in met-Ngb has to dissociate before an external ligand can bind to the heme. To obtain information on the rate of this dissociation process, we studied the binding of CN Ϫ to metNGB in the presence of a very large excess of this ligand (300 mM) by stopped-flow spectroscopy. Because of the basicity of cyanide, the pH measured at the end of this reaction was strongly alkaline. Under alkaline conditions, the rate of dissociation of His(E7) from Fe(III) is expected to be lower than under neutral conditions, given that the partly deprotonated coordinate histidine is a better ligand. Nevertheless, the rate constant determined for the fast component of this reaction, 0.45 s Ϫ1 , matched that expected from the second-order rate constant measured at pH 7.2 (Fig. 1B). This result shows that dissociation of the coordinated distal histidine, even when it is partly deprotonated, takes place at a rate higher than 0.45 s Ϫ1 and suggests that binding of cyanide to metNGB is not pH-dependent in the neutral-alkaline range.
Reductive Nitrosylation of metNGB-The reaction between metNGB and NO ⅐ was studied by UV-visible spectroscopy in the wavelength range 350 -650 nm at pH 7.0 and room temperature. As depicted in Fig. 2A, addition of NO ⅐ to a metNGB solution caused a shift of the absorbance maximum of the Soret band from 414 (metNGB) to 416 nm (NGBFe(II)NO). In the visible part of the spectrum (Fig. 2B), a new absorbance max-imum was observed around 550 nm, with a shoulder at 572 nm. This spectrum is practically identical to that obtained under anaerobic conditions by adding NO ⅐ to the reduced protein ( Fig.  2B, Inset) and to that recently obtained for mouse NgbFe(II)NO (see supplemental data in Ref. 32). The approximate extinction coefficients determined for the two absorbance maxima of human NGBFe(II)NO are ⑀ 416 ϭ 118 mM Ϫ1 cm Ϫ1 and ⑀ 550 ϭ 11.1 mM Ϫ1 cm Ϫ1 . These observations demonstrate that, in analogy to Mb and Hb (37), reaction of metNGB with an excess NO ⅐ leads to reduction of the iron center and generation of the thermodynamically stable species NGBFe(II)NO.
The kinetics of the reaction were obtained by fitting the traces extracted at 400, 405, and/or 531 nm. As shown in Fig. 3 Inset, also for this reaction the traces had to be fitted to a bi-exponential equation. In analogy to cyanide binding to met-NGB, the finding that both observed rate constants depended linearly upon the NO ⅐ concentration suggests that this ligand binds at different rates to two dissimilar conformations of the protein, and that dissociation of His(E7) occurs at a significantly higher rate. The second-order rate constants obtained from the linear fits of the plots of the observed rate constants of the fast and the slow components are 21 Ϯ 1 M Ϫ1 s Ϫ1 and 2.9 Ϯ 0.4 M Ϫ1 s Ϫ1 , respectively.
Reactions of metNGB with Peroxynitrite, Hydrogen Peroxide, and Nitrite-To better understand the reactivity of metNGB, we investigated its reactions with the two oxidizing agents, H 2 O 2 and peroxynitrite. In both cases, even after addition of a large excess of the oxidants (up to 0.3 mM H 2 O 2 and 275 M  Table I. Inset, kinetic trace at 407 nm of the reaction between 2.9 M metNGB and 13 mM cyanide. The two observed rate constants obtained from the bi-exponential fit depicted are (2.1 Ϯ 0.1) ϫ 10 Ϫ2 s Ϫ1 and (4.8 Ϯ 0.2) ϫ 10 Ϫ3 s Ϫ1 , respectively. peroxynitrite) no changes were observed in the spectra of the protein (data not shown). In addition, the decay rate of peroxynitrite determined at 302 nm was not affected by the presence of the protein. Because nitrite is always present as a contaminant of peroxynitrite and NO ⅐ solutions, we also studied its reaction with metNGB. Also in this case, no changes were observed in the spectra of metNGB, even upon addition of 1 mM nitrite (not shown).
Oxidation of NGBFe(II)NO by Dioxygen-The reaction between NGBFe(II)NO and O 2 was studied by UV-visible spectroscopy in the wavelength range 350 -650 nm at pH 7.0 and 25°C. Upon addition of oxygen, the maximum of the spectrum shifted from 550 (NGBFe(II)NO) to 531 nm (metNGB) in the visible region and from 416 to 414 nm in the Soret region. At first look, a single set of isosbestic points seemed to be present at 536 and 511 nm (visible) and 411 nm (Soret). However, detailed analysis of the spectra revealed two distinct sets of isosbestic points, only ϳ1 nm apart. The kinetics of the reaction were obtained by fitting the traces extracted at 420 and/or 575 nm. As shown in Fig. 4 Inset, also for this reaction the traces had to be fitted to a bi-exponential equation, in agreement with the absence of unique isosbestic points. The observed rate constants of the fast component depended linearly upon the O 2 concentration, and the second-order rate constant derived from the linear fit of the data was 16.0 Ϯ 0.9 M Ϫ1 s Ϫ1 . In contrast, the observed rate constants for the slow component increased only slightly with increasing O 2 concentrations. The linear fit through the points of the second step gave a value for the second-order rate constant of 0.4 Ϯ 0.2 M Ϫ1 s Ϫ1 . The average value of the observed rate constant for this slower step is Oxidation of NGBFe(II)NO by Peroxynitrite-The reaction between NGBFe(II)NO and peroxynitrite was studied by rapidscan UV-visible spectroscopy in the wavelength range 300 -650 nm at pH 7.2 and 20°C. Upon addition of peroxynitrite, the absorbance maximum of the Soret band of NGB shifted from 416 (NGBFe(II)NO) to 419 nm (NGBFe(III)NO), with two isosbestic points at 416 and 436 nm (Fig. 5). Over a longer time scale, a new absorbance band appeared with a maximum at 414 nm, which corresponds to that of metNGB (Fig. 5). This second process gave rise to a second clear isosbestic point at 415 nm. The same reaction was also followed in the visible region of the spectrum. As shown in Fig. 6A, the peroxynitrite-mediated oxidation of NGBFe(II)NO led first to the transient formation of NGBFe(III)NO, which gave rise to a spectrum with absorbance maxima at 532 and 565 nm (indicated by the bold line in Fig. 6) with isosbestic points at 515, 538, 559, and 571 nm. These maxima correspond closely to those for HbFe(III)NO (533 and 566 nm) (38). In the second part of the reaction (Fig.  6B), dissociation of NO ⅐ was accompanied by the generation of the absorbance features characteristic for metNGB (isosbestic points at 524, 540, 555, and 581 nm). The approximate extinction coefficients determined for the absorbance maxima of NGBFe(III)NO are ⑀ 419 ϭ 117 mM Ϫ1 cm Ϫ1 , ⑀ 532 ϭ 12.6 mM Ϫ1 cm Ϫ1 , and ⑀ 565 ϭ 13.0 mM Ϫ1 cm Ϫ1 . Interestingly, the identical reaction mechanism has recently been observed for the reaction between peroxynitrite and HbFe(II)NO. 2 The kinetics of the two steps of the reaction were studied by single-wavelength stopped-flow spectroscopy under pseudofirst order conditions with peroxynitrite in excess. The first step was followed at 436 nm, whereas the rate constant for the 2 S. Herold, submitted for publication. second step was obtained by fitting the traces collected at 418 nm. In addition, the decay of peroxynitrite was determined at 302 nm. As expected, the observed rate constant for the first reaction step, the oxidation of NGBFe(II)NO to NGBFe(III)NO, depended linearly upon the peroxynitrite concentration (Fig.  7). The second-order rate constant obtained from the linear plot is (1.29 Ϯ 0.07) ϫ 10 5 M Ϫ1 s Ϫ1 (at pH 7.2 and 20°C), a value much larger than that of (6.1 Ϯ 0.3) ϫ 10 3 M Ϫ1 s Ϫ1 obtained under identical conditions for the analogous reaction with Hb. 2 As anticipated, the observed rate constant for the second reaction step, the dissociation of NO ⅐ from NGBFe(III)NO, was independent of the peroxynitrite concentration and was 0.12 Ϯ 0.01 s Ϫ1 . This value is one order of magnitude smaller than that of the corresponding dissociation rate from Hb (ϳ1 s Ϫ1 ). 2 Finally, traces collected at 302 nm showed that the decay rate of peroxynitrite was not affected by the presence of the protein. However, when peroxynitrite was not present in large excess over protein, an initial rapid decay was observed, matching approximately the rate of its reaction with NGBFe(II)NO. DISCUSSION This study shows that NGB may function as scavenger of reactive oxygen and nitrogen species generated during brain hypoxia. As discussed below, whereas NGBFe(III) does not undergo fast reactions, a feature that, moreover, prevents formation of cytotoxic ferryl heme, NGBFe(II)NO reacts rapidly with the oxidizing agent peroxynitrite. These features are compatible with the protective role attributed to this protein during periods of oxygen deficiency (10,11).
Reactivity of metNGB-Similarly to the Fe(II) form (6,32), the reactivity of the Fe(III) form of NGB is strongly decreased by the coordinated distal histidine. In contrast to metHb and metMb, metNGB does not produce the oxidizing oxoFe(IV) species (ferrylNGB) when treated with hydrogen peroxide. Thus, the Fe(III) center of NGB seems to be protected from attack by strong oxidants by the coordinated distal His. The lack of reactivity of metNGB toward H 2 O 2 may be favorable under oxidative stress. Indeed, ferrylMb and ferrylHb, generated from the reaction of H 2 O 2 with these proteins, contribute to oxidative damage caused by reperfusion of ischemic tissues (39,40).
The observed lack of reactivity of metNGB toward peroxynitrite may also be related to heme hexacoordination. As recently shown, the reactivity of metMb toward peroxynitrite is strongly influenced by the nature of the amino acid residue in position E7 (41,42). The metMb mutant in which the distal histidine is replaced by alanine catalyzes the isomerization of peroxynitrite to nitrate more efficiently than wild-type Mb, where the strong hydrogen bond between the distal histidine and the water molecule coordinated to the Fe(III) center lowers the reactivity toward peroxynitrite (at pH 7.0 and 20°C, k cat are (1.3 Ϯ 0.1) ϫ 10 7 M Ϫ1 s Ϫ1 and (7.7 Ϯ 0.1) ϫ 10 4 M Ϫ1 s Ϫ1 , respectively) (41,42). Our data show that in the extreme case of metNGB, where the access to the heme is blocked by His(E7), the reaction between the heme and peroxynitrite does not occur: specifically, neither the formation of high valent ferryl species nor acceleration of the decay rate of peroxynitrite are observed.
Our studies of the reactions of metNGB with cyanide and NO ⅐ further confirm the lower reactivity of this protein, as the rate constants for binding of these two ligands to metNGB are significantly lower than those for the corresponding reactions with metHb and metMb (Table I). The kinetic data show that metNGB displays two distinct conformations, which bind cyanide and NO ⅐ at different rates. For cyanide binding, the two metNGB conformations are present approximately in a 1:1 ratio, and the second-order rate constants differ by a factor of ϳ4.5. The ratio between the two conformers did not change significantly among different experiments and was independent of the duration of storage of the protein. Similar results were obtained in a recent NMR study showing that mouse metNgb is present in two conformers in an ϳ2:1 ratio, which differ by a factor of ϳ2 in their CN Ϫ binding rates (7). The different orientation of the heme in the two isomers has been proposed to affect the dissociation rate of His(E7) from Fe(III) and, consequently, the reactivity toward CN Ϫ (7). However, despite the slow rate of cyanide binding to metNGB, dissociation of His(E7) does not represent the rate-determining step under the conditions of our experiments. In the concentration range studied (1-300 mM), the observed rate constants increased linearly with increasing CN Ϫ concentration, and no saturation was reached (Fig. 1B). This result suggests that dissociation of His(E7) may lead to two conformations in which the heme pocket is more or less accessible to external ligands, possibly in analogy with the open and closed conformations of the CO derivative of mouse Ngb observed by Raman spectroscopy (8). On the basis of our experiments, it is possible to set only the lower limit of the rate of dissociation of His(E7) from metNGB, which is 0.45 s Ϫ1 , the highest rate observed. For comparison, dissociation of externally added imidazole from metMb proceeds at a rate of 8 Ϯ 1 s Ϫ1 (43). The structural implications of the observed functional heterogeneity in human metNGB remain to be investigated.
Formation of NGBFe(II)NO and Its Reactions with O 2 and Peroxynitrite-As observed with metMb and metHb (37), reaction of metNGB with an excess of NO ⅐ leads to the generation of the reduced nitrosyl complex, NGBFe(II)NO (Equation 1).  (37), whereas the hydrolysis of NGB-Fe(II)(NO ϩ ) must proceed at a rate higher than 0.024 s Ϫ1 , the highest observed rate constant in this system (Fig. 3). In contrast, MbFe(II)(NO ϩ ) is rather stable under neutral conditions and is hydrolyzed at a significant extent only under alkaline conditions.
It has recently been reported that at a low O 2 concentration, recombinant mouse Ngb overexpressed in intact E. coli cells is partly present in the reduced nitrosyl form, NgbFe(II)NO, and that coordination by NO ⅐ apparently lowers the oxidation rate of mouse Ngb after sonication of the cells (32). In contrast, our data show that NO ⅐ apparently does not protect human NGB against oxidation: in the presence of 220 M O 2 (air saturation), oxidation of NGBFe(II)NO (0.0035 s Ϫ1 at 25°C) is faster than autoxidation of oxygenated NGB (0.0015 s Ϫ1 at 37°C) (6).
Oxidation of NGBFe(II)NO by O 2 proceeds in two steps: the rate of the first step is linearly dependent upon the O 2 concentration, whereas that of the second step shows only a weak dependence upon the amount of O 2 added (Table I). The secondorder rate constant of this second step, calculated by assuming that it increases linearly with O 2 concentration, is ϳ40 times lower than that of the first step. This difference seems too large to justify multiple protein conformations, in particular because the reaction between NGBFe(II)NO and peroxynitrite indicates the presence of a unique NGBFe(II)NO conformation (see below).
The analogous reaction between MbFe(II)NO and O 2 also proceeds in two steps (44). Factor analysis by singular value decomposition indicated that this reaction proceeds by means of an N-bound peroxynitrite intermediate (44).
In the presence of an excess of O 2 , the observed rate constants for the two reaction steps were 1.34 ϫ 10 Ϫ3 and 2.81 ϫ 10 Ϫ3 s Ϫ1 (at pH 7.0 and 37°C), respectively (44), but unfortunately, no data are available on their O 2 -concentration dependence. Interestingly, the rates of the second steps of the reactions with Mb and NGB are similar, whereas the first step of the reaction is significantly faster for NGBFe(II)NO (even at lower temperature) than for MbFe(II)NO and HbFe(II)NO (Table I) (44).
Our kinetic data with NGBFe(II)NO may suggest a similar pathway (Equation 2) for the reaction with O 2 . However, despite the very slow rate of the second reaction step, the intermediate species NGBFe(III)N(O)OO could not be detected. The spectrum of MbFe(III)N(O)OO, calculated from the kinetic data (44), is very similar to that of MbFe(III)OONO, the O-bound peroxynitrite complex characterized as an intermediate of the reaction between MbFeO 2 and NO ⅐ (20). In conclusion, it seems unlikely that the reaction with NGB proceeds according to Equation 2. Alternatively, the second step may represent a slight rearrangement within the protein, which might take place after formation of metNGB and which thus may lead to only minor spectral changes.
In analogy to the reaction with Hb 2 , the peroxynitrite-medi- Interestingly, the rate constant for the first step of the reaction between Fe(II)NO and peroxynitrite is almost two orders of magnitude higher for NGB than for Hb, whereas NO ⅐ dissociates from NGBFe(III)NO at a much lower rate than from HbFe(III)NO (Table I). Therefore, it was possible to transiently observe the spectrum of NGBFe(III)NO (Fig. 6), which could not be detected during reductive nitrosylation (Equation 1; Fig.  2). Only one exponential equation was required to fit the first reaction step of Equation 3. Thus, in contrast to the metNgb or the CO form (8), the NGBFe(II)NO derivative does not seem to assume multiple protein conformations under our experimental conditions, as judged from the kinetic of its reaction with peroxynitrite and possibly with O 2 .
Physiological Significance-Under ischemic conditions, NO ⅐ concentration in the brain increases dramatically, from the basal level of 1-10 nM (45) to the low M range (27). In the mitochondria-rich segments of the retina, and possibly in the brain, the local NGB concentration may reach values as high as 0.1 mM (4). Thus, under ischemic conditions, NGB may act as a NO ⅐ scavenger. The kinetic data presented here indicate that binding of NO ⅐ to metNGB may be too slow to have a significant role in the acute phase of hypoxia. In contrast, rapid NG-BFe(II)NO formation may occur by the binding of NO ⅐ to the Fe(II) form of the protein. This rapid reaction has been observed for recombinant Ngb in intact E. coli cells (32) and thus is likely to occur also under hypoxic conditions in neurons exposed to high levels of NO ⅐ . A number of studies, reviewed by Lipton (27), have shown that the major fate of NO ⅐ during ischemia is indeed the diffusion-limited reaction with superoxide to generate peroxynitrite. A major target of peroxynitrite in vivo is CO 2 , which is present in mM concentrations in most tissues. The products of the rapid reaction between peroxynitrite and CO 2 (3 ϫ 10 4 M Ϫ1 s Ϫ1 at 24°C) (46) are even stronger nitrating agents than peroxynitrite. Because of the high value of the second-order rate constant, NGBFe(II)NO may efficiently compete with CO 2 and protect against peroxynitritemediated damage, in particular in the mitochondria-rich segments of the retina and in the brain.
In conclusion, as a distinctive feature of NGB, the Fe(II)NO form is a much better scavenger of peroxynitrite in NGB than in Mb. In addition, metNGB, in contrast to metMb and metHb, does not readily form ferryl species upon reaction with H 2 O 2 . This lack of reactivity seems to be related to heme hexacoordination and is essential to prevent generation of ferryl heme and to avoid the onset of oxidative reactions mediated by this highly reactive species. This feature may be physiologically important because, as observed in the retina (4), Ngb is apparently localized in the vicinity of mitochondria, i.e. close to the potential sites of superoxide and, consequently, of H 2 O 2 and peroxynitrite production.